Hospital Medicine

A 75-year-old man with type 2 diabetes and hypothyroidism underwent bilateral total knee replacement at our hospital.

His functional capacity had been moderately limited by knee pain, but he could easily climb one flight of stairs without symptoms. His medications at that time included levothyroxine (Synthroid) and metformin (Glucophage). He had no known cardiac or pulmonary disease. The preoperative evaluation, including laboratory tests and electrocardiography, was within normal limits.

Spinal anesthesia was used for surgery, and he was given 2 mg of midazolam (Versed) intravenously for sedation. No additional sedation was given. He was given oxygen via nasal cannula at 2 L/min.

All vital signs were stable at the start of the procedure. However, about halfway through, when the thigh tourniquet was released, his oxygen saturation dropped abruptly from 100% to 92%. All other vital signs remained stable, and he was asymptomatic, was oriented to person, time, and place, was conversing freely, and was in no distress. The oxygen flow was increased to 6 L/min, his oxygen saturation improved, and the procedure was then completed as planned.

At the conclusion of the surgery, before the patient was transported to the postanesthesia care unit (PACU) and while his oxygen flow rate was still 6 L/min, his oxygen saturation again dropped to 92%. A simple face mask was placed, and the oxygen flow rate was increased to 10 L/min. His oxygen saturation stayed low, near 90%.

Bleeding during surgery had been nominal. He had received 2 L of lactated Ringer’s solution and 500 mL of hetastarch (Hextend) during surgery. He continued to be asymptomatic in the PACU.

1. What is the most likely cause of oxygen desaturation during bilateral total knee arthroplasty?

• Fat embolism
• Intraoperative pneumonia
• Venous thromboembolism with pulmonary embolism
• Acute myocardial infarction
• Acute pulmonary edema
• Excessive sedation

The differential diagnosis of oxygen desaturation during orthopedic procedures is listed in Table 1.

Fat embolism is the most likely cause, particularly given the greater fatty embolic load that occurs with bilateral total knee arthroplasty than with unilateral total knee arthroplasty.

At what point the maximal showering of fat emboli occurs is not known. Fat may be released into the circulation with pressurization of the medullary canal during surgery or with manipulation of a fracture. The emboli may collect in the leg veins and then be released in a shower when the thigh tourniquet is released. Vasoactive mediators and methylmethacrylate cement released into the circulatory system after tourniquet deflation may also cause vasodilation, hypotension, and increased dead-space ventilation, resulting in hypoxia and a drop in end-tidal CO₂.

Pneumonia during surgery is rare without an apparent aspiration event.

Venous thromboembolism is possible but is more likely later in the postoperative period after major orthopedic surgery.

Acute myocardial infarction could present with hypoxia, particularly in a diabetic patient, who may not experience chest pain. However, intraoperative electrocardiographic changes would likely be seen. If myocardial infarction is suspected, postoperative serial electrocardiograms and measuring troponin and cardiac enzyme levels aid in the diagnosis.

Acute pulmonary edema is possible but not as highly suspected, as the patient had no history of congestive heart failure and received an appropriate amount of fluid for this type of surgery.

Excessive sedation could cause hypoventilation and, thus, oxygen desaturatation. However, this patient’s oxygen desaturatation began more than an hour after the midazolam was given. Midazolam is a short-acting benzodiazepine. It is unlikely that the patient would show signs of hypoventilation and oversedation an hour after the drug was given. Our patient also did not show any signs of excessive sedation, as he was awake and conversing during the surgery.

Fat emboli vs fat embolism syndrome

Fat embolism is the presence of fat drops within the systemic and pulmonary microcirculation, with or without clinical sequelae.¹ Fat embolism syndrome, on the other hand, is defined as injury to and dysfunction of one or more organs as a result of the embolization of fat, usually within 24 hours of injury or orthopedic surgery.²

Fat embolism syndrome is an unpredictable condition with a varied presentation. Fat droplets are thought to embolize via the venous circulation into the pulmonary arteries, occluding small blood vessels in the lung. However, they also get into the arterial circulation and occlude arteries in the brain, kidney, heart, and liver (more on this phenomenon below).

Fat embolism is reported to originate primarily from fractures of the femur, tibia, and pelvis.^2,3 As many as 90% of trauma patients have been shown to have evidence of fat embolism on autopsy.⁴ However, only a small number of patients develop the classic fat embolism syndrome,^2,3,5 Why some develop the syndrome and others do not is still unknown.

Orthopedic procedures associated with fat embolization include knee arthroplasty and hip arthroplasty, particularly if it involves intramedullary manipulation or medullary fixation.⁶ It has also been reported during spinal procedures in which pedicular screws are used.⁷ The syndrome occurs in 0.25% to 30% of patients following multiple fractures and in 0.1% to 12% of patients during or following knee or hip arthroplasty.

One study⁸ showed evidence of fat on transesophageal echocardiography in 88% of patients undergoing medullary reaming of lower-extremity fractures and hip hemiarthroplasty. Blood sampling from the right atrium confirmed that fat was responsible for the echocardiographic abnormalities. The study also showed that the severity of the embolic showering correlated with the severity of hypoxia and the decrease in end-tidal CO₂.⁸

CASE CONTINUED

On arrival at the PACU, our patient’s oxygen saturation was 94% while he was breathing oxygen via a simple face mask at a flow rate of 10 L/min. His heart rate was 60 bpm, blood pressure 110/60 mm Hg, and temperature 37.5°C (96.3°F). Chest sounds were normal on auscultation.

However, 3 hours later, his mental status rapidly deteriorated. He was oriented only to person, and he was drowsy. He had escalating respiratory distress with a rapid respiratory rate and decreasing oxygen saturation. At this point, auscultation of his chest wall revealed bilateral crackles and rales.

He was promptly intubated. Profuse fluid and secretions were noted to be coming from his lungs, filling the endotracheal tube. Arterial blood gas measurement showed a pH of 7.22, Pao₂ 64 mm Hg, and Paco₂ 56 mm Hg on 100% fraction of inspired oxygen, with no increased anion gap.

2. Which consequence of fat embolism is most likely at this time in this patient?

• Coexisting sepsis
• Fat embolism syndrome
• Acute cardioembolic stroke
• Anaphylaxis

Fat embolism syndrome should be highly suspected in this patient. As mentioned, it can affect many different organs. It is the most serious condition resulting from fat embolization after surgery or trauma.

Sepsis was unlikely in our patient, since he presented for his surgery in good health and with no preexisting signs or symptoms of infection. Acute cardioembolic stroke could have caused the neurologic signs, but this would not necessarily explain the coexisting hypoxia. An anaphylactic reaction to drugs or surgical cement would most likely present intraoperatively, shortly after exposure occurred, rather than several hours after surgery.

How common is fat embolism syndrome?

The occurrence rate of fat embolism syndrome has been reported to be 0.25% to 30% after multiple fractures and 0.1% to 12% after knee and hip joint surgery, with a mortality rate of 13% to 36%.^2,9–14 The rate of occurrence after unilateral total knee joint replacement has been reported to be 1.8% to 5%, and 4% to 12% after bilateral total knee replacement.^15–19

The syndrome is relatively more common with traumatic fractures of the lower extremities. However, it has also been reported with liposuction, total parenteral nutrition, bone marrow harvest and transplantation, burns, and acute pancreatitis, to mention a few.¹⁰

The broad range of reported incidence rates can be attributed to the fact that many studies were in patients with multiple trauma, whose concomitant injuries may have made it difficult to clearly define the contribution of fat embolism syndrome to the overall rates of morbidity and mortality. Also, different studies used different criteria to define the syndrome.

How does fat embolism syndrome occur?

Two hypotheses for how this syndrome occurs were proposed nearly a century ago.^20,21

The “mechanical” theory is that fat emboli are formed as a result of trauma and disruption of adipose tissue and other cells in the bone marrow. Increases in intramedullary pressure force the fat emboli through damaged medullary venous channels in the bone and into the circulation of the lower extremities. This embolization of fat causes an initial mechanical pulmonary obstruction. Mechanical obstruction by fat emboli in the pulmonary system leads to increased pulmonary pressures and an increase in right heart outflow pressure. The right heart becomes strained, leading to a decreased right-sided cardiac output. As a result, the left heart filling pressures diminish and hypotension ensues.²⁰

The “biochemical” theory, on the other hand, is that chylomicrons within the vascular system are modified and their stability is compromised as a result of stress. These traumatized chylomicrons then coalesce to form droplets of fat that accumulate in the pulmonary circulation and produce a mechanical obstruction. This would explain why nontraumatic, nonorthopedic insults can produce this syndrome.

Autopsy studies show that there is little correlation between the presence and quantity of intravascular fat and the severity of clinical symptoms, thus implying that the syndrome is caused by more than just mechanical obstruction. The biochemical theory postulates that fat globules within the circulatory system then cause the release of lipase from the pulmonary alveolar cells, which then hydrolyses the fat into free fatty acids. These free fatty acids cause an inflammatory reaction, complementmediated leukocyte aggregation, chemotoxin release, and subsequently endothelial damage. These vasoactive substances damage type 2 pneumocytes and lead to an increased permeability of the pulmonary capillary beds. Acute respiratory distress syndrome (ARDS) may ensue. Disseminated intravascular coagulation may occur as a result of the formation of microthrombi involving lipids, platelets, and fibrin.^21,22

Embolization of fat to the central nervous system can occur as fat globules cross into the systemic circulation via a patent foramen ovale, an atrioventricular shunt, or the pulmonary capillaries. This can then result in cerebral ischemia.²³

Although patent foramen ovale may seem the most direct route for cerebral embolization, the neurologic impairment and signs of cerebral emboli in fat embolism syndrome may occur in the absence of patent foramen ovale.^24,25 The fat globules may actually go through the lung capillaries, being flexible and forced through by increased pulmonary pressure.

But whether the cause of fat embolism syndrome is occlusion by globules, the release of biochemical mediators, or a combination of both is unknown. Both mechanisms are likely responsible. We can only suspect that the degree of fat load and intrinsic metabolic differences between individuals account for the variation in susceptibility.

FAT EMBOLI AFFECT THE LUNGS, SKIN, AND BRAIN

3. Where on the body is the rash associated with fat embolism syndrome usually seen?

• Face
• Near a site of fracture or surgery
• Chest, axilla, conjunctiva
• Distal extremities

Petechiae are part of the classic presenting triad of fat embolism syndrome, which also includes pulmonary and cerebral dysfunction.

Petechiae usually appear on the 2nd to 4th day after injury.²⁶ They are usually found across the chest, the anterior axillary folds, and the neck, as well as on the oral mucosa and the conjunctiva. The rash is caused by occlusion of dermal capillaries by fat, which increases their fragility.¹⁰

Pulmonary changes usually begin with tachypnea, dyspnea, and a drop in oxygen saturation, leading to generalized hypoxia. Respiratory symptoms are present in 100% of cases.² Respiratory symptoms can acutely develop with the sudden manipulation of a fracture, reaming of bone, or release of a limb tourniquet.²⁷

Body systems affected by fat embolism syndrome are summarized in Table 2.

4. How many hours after injury does fat embolism syndrome typically manifest?

• 1 to 2 hours
• 6 to 12 hours
• 12 to 20 hours
• 24 to 48 hours
• 72 to 84 hours

Most patients develop signs and symptoms 24 to 48 hours after injury. Patients presenting earlier than 12 hours usually have a more fulminant course.²⁹

The time between fat embolization and the development of fat embolism syndrome is thought to be related to the time required for the metabolic conversion of fat to free fatty acids.³⁰ We suspect that the early desaturation seen in our patient was the result of a heavy showering of fat intraoperatively. However, this could only be concluded after we had ruled out other causes of acute hypoxia and hypotension.

Fat embolism syndrome is a diagnosis of exclusion and is based on clinical criteria. No specific sign, symptom, or test is pathognomonic. It may often be confused with other conditions such as systemic inflammatory response syndrome or sepsis. However, the triad of respiratory and neurologic symptoms and petechiae coupled with the clinical picture of recent trauma or orthopedic surgery almost assures the diagnosis.

Fat embolism syndrome can range from subclinical to fulminating, with the more fulminating course attributable to a huge load of fat emboli, which leads to acute cor pulmonale.

Regardless of the criteria used, one must have a high index of suspicion for fat embolization syndrome in patients undergoing orthopedic procedures, particularly hip and knee surgery, and in patients with fractures, especially fractures of the femur, tibia, or pelvis and multiple, concomitant fractures.

CASE CONTINUED

Our patient was given furosemide (Lasix) empirically for diuresis and to improve oxygenation. However, his oxygen saturation remained low.

Chest radiography 4 hours after surgery showed bilateral pulmonary infiltrates. Serial electrocardiography showed no acute changes. Levels of cardiac enzymes and troponins were normal. Transthoracic echocardiography showed no left ventricular dysfunction, a normal right ventricle, and no evidence of valvular lesions. Urine and blood fat stains were negative, but the sputum stain was positive for copious extracellular fat. The patient became comatose 5 hours postoperatively. Computed tomography of the brain was normal. He was transferred to the surgical intensive care unit.

The clinical course was marked by hemodynamic instability requiring norepinephrine (Levophed) and vasopressin (Pitressin) for hypotension. Right ventricular filling pressures via central venous pressure monitoring showed no evidence of hypovolemia. The hemoglobin concentration and the hematocrit were stable, with no evidence of acute or ongoing bleeding. Blood, urine, and sputum cultures remained negative. Acute myocardial infarction was ruled out by serial electrocardiography, cardiac enzyme testing, and troponin testing.

Magnetic resonance imaging (MRI) of the brain on postoperative day 2 showed foci of acute ischemia suggestive of embolic phenomena consistent with fat embolism syndrome (Figure  1). Transthoracic echocardiography was repeated but again showed no evidence of a patent foramen ovale. Electroencephalography on postoperative day 4 showed severe, diffuse encephalopathy. There was no petechial skin rash. Other laboratory studies showed progressive thrombocytopenia with a platelet count of 53 × 199/L on postoperative day 3.

TESTS THAT AID THE CLINICAL DIAGNOSIS

Although no single laboratory test is pathognomonic for fat embolism syndrome, several tests may help raise suspicion of it, especially in the setting of fracture or an orthopedic surgical procedure.

Arterial blood gases must be measured. A Pao₂ of less than 60 mm Hg with no other obvious lung pathology in an orthopedic surgery patient is highly suspicious.¹² An alveolar-arterial gradient of greater than 100 mm Hg may further increase suspicion.

Tests for fat. The blood and urine may be examined for fat, although positive findings are not specific for fat embolism syndrome.³³ Fat in the urine indicates the occurrence of massive fat embolism, but this is not always accompanied by the syndrome.³⁴ Gurd and Wilson¹³ found fat globules larger than 8 μm circulating in the serum in all documented cases. They stated that, even though the relationship of large fat globules to the pathogenesis of the clinical picture remains obscure, the demonstration of their presence can be helpful in the diagnosis.¹³

Also, samples obtained with bronchoalveolar lavage may be examined for fat. The macrophages may be stained for fat using the oil red O stain. Again, this is a nonspecific marker, as fat-stained macrophages are seen in trauma patients,³⁵ but the finding has a very high negative predictive value.³⁶ Anemia, thrombocytopenia, hypofibrinogenemia, an elevated lipase level, and a high erythrocyte sedimentation rate may be found in fat embolism syndrome.¹³

Chest radiography may show bilateral infiltrates, as in ARDS, but this is not diagnostic for fat embolism syndrome.

Electrocardiography may show changes in ST and T waves and signs of right heart strain.

Transesophageal echocardiography may show increased right heart and pulmonary artery pressures.

Computed tomography is often negative,^37,38 but T2-weighted MRI is useful in the diagnosis of cerebral fat embolism syndrome, as it can show intracerebral microinfarcts as early as 4 hours after the onset of neurologic symptoms, and these findings correlate well with the clinical severity of brain injury.

Diffusion-weighted MRI may enhance the sensitivity and specificity of the neuroradiologic diagnosis. Diffusion-weighted MRI typically shows multiple nonconfluent areas of high-intensity signals or bright spots on a dark background, known as a “starfield pattern.” This pattern has been suggested to be pathognomonic of acute cerebral microinfarction. The abnormalities presumably reflect foci of cytotoxic edema that develops immediately, unlike vasogenic edema, seen in T2-weighted images, which may take up to several days to develop. Although these images are not necessarily specific for fat emboli, they are useful in helping make the diagnosis. Thus, diffusionweighted MRI should be done if fat embolism syndrome is suspected.^38,39

CASE CONCLUDED

The patient’s course in the intensive care unit was further complicated by gastrointestinal bleeding and renal failure. His neurologic status did not improve. Repeated MRI of the brain showed evolving bilateral watershed infarction throughout the cortices. The neurologic consult service diagnosed the patient as having severe encephalopathy with a very poor prognosis. The decision was made to withdraw care. He was placed under palliative care and died on postoperative day 22.

DRUG TREATMENT OF FAT EMBOLISM SYNDROME

5. Which of the following drugs has been proven to be effective in treating fat embolism syndrome?

• Intravenous ethanol
• Steroids
• Heparin
• Dextran
• Aspirin
• None of the above

None of the above has been proven to be effective in treating this disorder. The management is largely supportive. Thus, prevention, early diagnosis, and symptom management are vital.

Pulmonary and hemodynamic support are the cornerstones of successful treatment. Aggressive respiratory support is often needed. Management of acute lung injury and ARDS focuses on achieving acceptable gas exchange while preventing ventilator-associated lung injury. Intravascular volume must be supported. Inotropes and pulmonary vasodilators may be required to maintain hemodynamics. Exacerbation of central nervous system ischemia from hypotension or hypoxia should be avoided.

If the thrombocytopenia leads to clinical bleeding, platelet transfusions may be warranted.

Supportive care should include prophylaxis of deep venous thrombosis and of gastrointestinal bleeding, and maintenance of nutrition.⁴⁰ Patients who receive supportive care generally have a favorable outcome, with a mortality rate of less than 10%.²⁸

Drug studies have been inconclusive

Drugs suggested in the treatment of fat embolism syndrome include heparin, aspirin, dextran, hypertonic glucose, and alcohol, but the results have been inconclusive.^{3,11,23,40–43}

Heparin stimulates lipase activity, consequently decreasing the concentration of circulating fat globules. However, the increase in levels of free fatty acids may actually worsen the clinical picture. For this reason, and because of anticoagulation concerns and evidence of increased mortality rates, heparin is now contraindicated in the treatment of fat embolism syndrome.^2,41,43

Alcohol. Patients with a higher blood alcohol level at the time of injury have been reported to have a lower incidence of fat embolism syndrome. Alcohol inhibits lipase, suppressing the rise of free fatty acids. In experimental studies, the incidence of fat embolism syndrome was lower when the blood alcohol level was maintained at 20 mg/dL. However, no prospective randomized trial has been done to determine the clinical efficacy of ethanol as a treatment for this condition.^5,42

Dextran has been advocated, owing to its ability to improve small-vessel perfusion, but bleeding risk and acute renal failure associated with this drug have limited its use.⁵

N-acetylcysteine has been shown to attenuate fat-induced lung injury in a study of rats with induced fat embolism syndrome.⁴⁴

Corticosteroid treatment for this condition is controversial. Studies in patients with femoral and tibial fractures show that steroids reduce the incidence of fat embolism syndrome when given prophylactically, and those treated with steroids had a higher Pao₂ than controls. Doses of methylprednisolone in these studies ranged between 9 mg/kg to 90 mg/kg. A drawback of these studies is their small number of patients.^12,32,45,46

A meta-analysis⁴⁷ of randomized trials of corticosteroids to prevent fat embolism syndrome in patients with long-bone fractures identified 104 such studies. Only 7 of the 104 were considered adequate. In 389 patients with long-bone fractures, prophylactic corticosteroids reduced the risk of fat embolism syndrome by 78% (95% confidence interval 43%–92%) and corticosteroids also significantly reduced the risk of hypoxia with no difference in rates of infection or death. However, the overall quality of the trials was poor, and the authors of the meta-analysis concluded that more study is needed before corticosteroids could be formally recommended.⁴⁷

There is no evidence that steroids improve the overall clinical course of already established fat embolism syndrome.^12,32,45 The dosing and optimal timing of administration have also not been established. High doses pose a risk of septic complications, which may be devastating for the posttrauma or postoperative patient.

A 75-year-old man with type 2 diabetes and hypothyroidism underwent bilateral total knee replacement at our hospital.

His functional capacity had been moderately limited by knee pain, but he could easily climb one flight of stairs without symptoms. His medications at that time included levothyroxine (Synthroid) and metformin (Glucophage). He had no known cardiac or pulmonary disease. The preoperative evaluation, including laboratory tests and electrocardiography, was within normal limits.

Spinal anesthesia was used for surgery, and he was given 2 mg of midazolam (Versed) intravenously for sedation. No additional sedation was given. He was given oxygen via nasal cannula at 2 L/min.

All vital signs were stable at the start of the procedure. However, about halfway through, when the thigh tourniquet was released, his oxygen saturation dropped abruptly from 100% to 92%. All other vital signs remained stable, and he was asymptomatic, was oriented to person, time, and place, was conversing freely, and was in no distress. The oxygen flow was increased to 6 L/min, his oxygen saturation improved, and the procedure was then completed as planned.

At the conclusion of the surgery, before the patient was transported to the postanesthesia care unit (PACU) and while his oxygen flow rate was still 6 L/min, his oxygen saturation again dropped to 92%. A simple face mask was placed, and the oxygen flow rate was increased to 10 L/min. His oxygen saturation stayed low, near 90%.

Bleeding during surgery had been nominal. He had received 2 L of lactated Ringer’s solution and 500 mL of hetastarch (Hextend) during surgery. He continued to be asymptomatic in the PACU.

1. What is the most likely cause of oxygen desaturation during bilateral total knee arthroplasty?

• Fat embolism
• Intraoperative pneumonia
• Venous thromboembolism with pulmonary embolism
• Acute myocardial infarction
• Acute pulmonary edema
• Excessive sedation

The differential diagnosis of oxygen desaturation during orthopedic procedures is listed in Table 1.

Fat embolism is the most likely cause, particularly given the greater fatty embolic load that occurs with bilateral total knee arthroplasty than with unilateral total knee arthroplasty.

At what point the maximal showering of fat emboli occurs is not known. Fat may be released into the circulation with pressurization of the medullary canal during surgery or with manipulation of a fracture. The emboli may collect in the leg veins and then be released in a shower when the thigh tourniquet is released. Vasoactive mediators and methylmethacrylate cement released into the circulatory system after tourniquet deflation may also cause vasodilation, hypotension, and increased dead-space ventilation, resulting in hypoxia and a drop in end-tidal CO₂.

Pneumonia during surgery is rare without an apparent aspiration event.

Venous thromboembolism is possible but is more likely later in the postoperative period after major orthopedic surgery.

Acute myocardial infarction could present with hypoxia, particularly in a diabetic patient, who may not experience chest pain. However, intraoperative electrocardiographic changes would likely be seen. If myocardial infarction is suspected, postoperative serial electrocardiograms and measuring troponin and cardiac enzyme levels aid in the diagnosis.

Acute pulmonary edema is possible but not as highly suspected, as the patient had no history of congestive heart failure and received an appropriate amount of fluid for this type of surgery.

Excessive sedation could cause hypoventilation and, thus, oxygen desaturatation. However, this patient’s oxygen desaturatation began more than an hour after the midazolam was given. Midazolam is a short-acting benzodiazepine. It is unlikely that the patient would show signs of hypoventilation and oversedation an hour after the drug was given. Our patient also did not show any signs of excessive sedation, as he was awake and conversing during the surgery.

Fat emboli vs fat embolism syndrome

Fat embolism is the presence of fat drops within the systemic and pulmonary microcirculation, with or without clinical sequelae.¹ Fat embolism syndrome, on the other hand, is defined as injury to and dysfunction of one or more organs as a result of the embolization of fat, usually within 24 hours of injury or orthopedic surgery.²

Fat embolism syndrome is an unpredictable condition with a varied presentation. Fat droplets are thought to embolize via the venous circulation into the pulmonary arteries, occluding small blood vessels in the lung. However, they also get into the arterial circulation and occlude arteries in the brain, kidney, heart, and liver (more on this phenomenon below).

Fat embolism is reported to originate primarily from fractures of the femur, tibia, and pelvis.^2,3 As many as 90% of trauma patients have been shown to have evidence of fat embolism on autopsy.⁴ However, only a small number of patients develop the classic fat embolism syndrome,^2,3,5 Why some develop the syndrome and others do not is still unknown.

Orthopedic procedures associated with fat embolization include knee arthroplasty and hip arthroplasty, particularly if it involves intramedullary manipulation or medullary fixation.⁶ It has also been reported during spinal procedures in which pedicular screws are used.⁷ The syndrome occurs in 0.25% to 30% of patients following multiple fractures and in 0.1% to 12% of patients during or following knee or hip arthroplasty.

One study⁸ showed evidence of fat on transesophageal echocardiography in 88% of patients undergoing medullary reaming of lower-extremity fractures and hip hemiarthroplasty. Blood sampling from the right atrium confirmed that fat was responsible for the echocardiographic abnormalities. The study also showed that the severity of the embolic showering correlated with the severity of hypoxia and the decrease in end-tidal CO₂.⁸

CASE CONTINUED

On arrival at the PACU, our patient’s oxygen saturation was 94% while he was breathing oxygen via a simple face mask at a flow rate of 10 L/min. His heart rate was 60 bpm, blood pressure 110/60 mm Hg, and temperature 37.5°C (96.3°F). Chest sounds were normal on auscultation.

However, 3 hours later, his mental status rapidly deteriorated. He was oriented only to person, and he was drowsy. He had escalating respiratory distress with a rapid respiratory rate and decreasing oxygen saturation. At this point, auscultation of his chest wall revealed bilateral crackles and rales.

He was promptly intubated. Profuse fluid and secretions were noted to be coming from his lungs, filling the endotracheal tube. Arterial blood gas measurement showed a pH of 7.22, Pao₂ 64 mm Hg, and Paco₂ 56 mm Hg on 100% fraction of inspired oxygen, with no increased anion gap.

2. Which consequence of fat embolism is most likely at this time in this patient?

• Coexisting sepsis
• Fat embolism syndrome
• Acute cardioembolic stroke
• Anaphylaxis

Fat embolism syndrome should be highly suspected in this patient. As mentioned, it can affect many different organs. It is the most serious condition resulting from fat embolization after surgery or trauma.

Sepsis was unlikely in our patient, since he presented for his surgery in good health and with no preexisting signs or symptoms of infection. Acute cardioembolic stroke could have caused the neurologic signs, but this would not necessarily explain the coexisting hypoxia. An anaphylactic reaction to drugs or surgical cement would most likely present intraoperatively, shortly after exposure occurred, rather than several hours after surgery.

How common is fat embolism syndrome?

The occurrence rate of fat embolism syndrome has been reported to be 0.25% to 30% after multiple fractures and 0.1% to 12% after knee and hip joint surgery, with a mortality rate of 13% to 36%.^2,9–14 The rate of occurrence after unilateral total knee joint replacement has been reported to be 1.8% to 5%, and 4% to 12% after bilateral total knee replacement.^15–19

The syndrome is relatively more common with traumatic fractures of the lower extremities. However, it has also been reported with liposuction, total parenteral nutrition, bone marrow harvest and transplantation, burns, and acute pancreatitis, to mention a few.¹⁰

The broad range of reported incidence rates can be attributed to the fact that many studies were in patients with multiple trauma, whose concomitant injuries may have made it difficult to clearly define the contribution of fat embolism syndrome to the overall rates of morbidity and mortality. Also, different studies used different criteria to define the syndrome.

How does fat embolism syndrome occur?

Two hypotheses for how this syndrome occurs were proposed nearly a century ago.^20,21

The “mechanical” theory is that fat emboli are formed as a result of trauma and disruption of adipose tissue and other cells in the bone marrow. Increases in intramedullary pressure force the fat emboli through damaged medullary venous channels in the bone and into the circulation of the lower extremities. This embolization of fat causes an initial mechanical pulmonary obstruction. Mechanical obstruction by fat emboli in the pulmonary system leads to increased pulmonary pressures and an increase in right heart outflow pressure. The right heart becomes strained, leading to a decreased right-sided cardiac output. As a result, the left heart filling pressures diminish and hypotension ensues.²⁰

The “biochemical” theory, on the other hand, is that chylomicrons within the vascular system are modified and their stability is compromised as a result of stress. These traumatized chylomicrons then coalesce to form droplets of fat that accumulate in the pulmonary circulation and produce a mechanical obstruction. This would explain why nontraumatic, nonorthopedic insults can produce this syndrome.

Autopsy studies show that there is little correlation between the presence and quantity of intravascular fat and the severity of clinical symptoms, thus implying that the syndrome is caused by more than just mechanical obstruction. The biochemical theory postulates that fat globules within the circulatory system then cause the release of lipase from the pulmonary alveolar cells, which then hydrolyses the fat into free fatty acids. These free fatty acids cause an inflammatory reaction, complementmediated leukocyte aggregation, chemotoxin release, and subsequently endothelial damage. These vasoactive substances damage type 2 pneumocytes and lead to an increased permeability of the pulmonary capillary beds. Acute respiratory distress syndrome (ARDS) may ensue. Disseminated intravascular coagulation may occur as a result of the formation of microthrombi involving lipids, platelets, and fibrin.^21,22

Embolization of fat to the central nervous system can occur as fat globules cross into the systemic circulation via a patent foramen ovale, an atrioventricular shunt, or the pulmonary capillaries. This can then result in cerebral ischemia.²³

Although patent foramen ovale may seem the most direct route for cerebral embolization, the neurologic impairment and signs of cerebral emboli in fat embolism syndrome may occur in the absence of patent foramen ovale.^24,25 The fat globules may actually go through the lung capillaries, being flexible and forced through by increased pulmonary pressure.

But whether the cause of fat embolism syndrome is occlusion by globules, the release of biochemical mediators, or a combination of both is unknown. Both mechanisms are likely responsible. We can only suspect that the degree of fat load and intrinsic metabolic differences between individuals account for the variation in susceptibility.

FAT EMBOLI AFFECT THE LUNGS, SKIN, AND BRAIN

3. Where on the body is the rash associated with fat embolism syndrome usually seen?

• Face
• Near a site of fracture or surgery
• Chest, axilla, conjunctiva
• Distal extremities

Petechiae are part of the classic presenting triad of fat embolism syndrome, which also includes pulmonary and cerebral dysfunction.

Petechiae usually appear on the 2nd to 4th day after injury.²⁶ They are usually found across the chest, the anterior axillary folds, and the neck, as well as on the oral mucosa and the conjunctiva. The rash is caused by occlusion of dermal capillaries by fat, which increases their fragility.¹⁰

Pulmonary changes usually begin with tachypnea, dyspnea, and a drop in oxygen saturation, leading to generalized hypoxia. Respiratory symptoms are present in 100% of cases.² Respiratory symptoms can acutely develop with the sudden manipulation of a fracture, reaming of bone, or release of a limb tourniquet.²⁷

Body systems affected by fat embolism syndrome are summarized in Table 2.

4. How many hours after injury does fat embolism syndrome typically manifest?

• 1 to 2 hours
• 6 to 12 hours
• 12 to 20 hours
• 24 to 48 hours
• 72 to 84 hours

Most patients develop signs and symptoms 24 to 48 hours after injury. Patients presenting earlier than 12 hours usually have a more fulminant course.²⁹

The time between fat embolization and the development of fat embolism syndrome is thought to be related to the time required for the metabolic conversion of fat to free fatty acids.³⁰ We suspect that the early desaturation seen in our patient was the result of a heavy showering of fat intraoperatively. However, this could only be concluded after we had ruled out other causes of acute hypoxia and hypotension.

Fat embolism syndrome is a diagnosis of exclusion and is based on clinical criteria. No specific sign, symptom, or test is pathognomonic. It may often be confused with other conditions such as systemic inflammatory response syndrome or sepsis. However, the triad of respiratory and neurologic symptoms and petechiae coupled with the clinical picture of recent trauma or orthopedic surgery almost assures the diagnosis.

Fat embolism syndrome can range from subclinical to fulminating, with the more fulminating course attributable to a huge load of fat emboli, which leads to acute cor pulmonale.

Regardless of the criteria used, one must have a high index of suspicion for fat embolization syndrome in patients undergoing orthopedic procedures, particularly hip and knee surgery, and in patients with fractures, especially fractures of the femur, tibia, or pelvis and multiple, concomitant fractures.

CASE CONTINUED

Our patient was given furosemide (Lasix) empirically for diuresis and to improve oxygenation. However, his oxygen saturation remained low.

Chest radiography 4 hours after surgery showed bilateral pulmonary infiltrates. Serial electrocardiography showed no acute changes. Levels of cardiac enzymes and troponins were normal. Transthoracic echocardiography showed no left ventricular dysfunction, a normal right ventricle, and no evidence of valvular lesions. Urine and blood fat stains were negative, but the sputum stain was positive for copious extracellular fat. The patient became comatose 5 hours postoperatively. Computed tomography of the brain was normal. He was transferred to the surgical intensive care unit.

The clinical course was marked by hemodynamic instability requiring norepinephrine (Levophed) and vasopressin (Pitressin) for hypotension. Right ventricular filling pressures via central venous pressure monitoring showed no evidence of hypovolemia. The hemoglobin concentration and the hematocrit were stable, with no evidence of acute or ongoing bleeding. Blood, urine, and sputum cultures remained negative. Acute myocardial infarction was ruled out by serial electrocardiography, cardiac enzyme testing, and troponin testing.

Magnetic resonance imaging (MRI) of the brain on postoperative day 2 showed foci of acute ischemia suggestive of embolic phenomena consistent with fat embolism syndrome (Figure  1). Transthoracic echocardiography was repeated but again showed no evidence of a patent foramen ovale. Electroencephalography on postoperative day 4 showed severe, diffuse encephalopathy. There was no petechial skin rash. Other laboratory studies showed progressive thrombocytopenia with a platelet count of 53 × 199/L on postoperative day 3.

TESTS THAT AID THE CLINICAL DIAGNOSIS

Although no single laboratory test is pathognomonic for fat embolism syndrome, several tests may help raise suspicion of it, especially in the setting of fracture or an orthopedic surgical procedure.

Arterial blood gases must be measured. A Pao₂ of less than 60 mm Hg with no other obvious lung pathology in an orthopedic surgery patient is highly suspicious.¹² An alveolar-arterial gradient of greater than 100 mm Hg may further increase suspicion.

Tests for fat. The blood and urine may be examined for fat, although positive findings are not specific for fat embolism syndrome.³³ Fat in the urine indicates the occurrence of massive fat embolism, but this is not always accompanied by the syndrome.³⁴ Gurd and Wilson¹³ found fat globules larger than 8 μm circulating in the serum in all documented cases. They stated that, even though the relationship of large fat globules to the pathogenesis of the clinical picture remains obscure, the demonstration of their presence can be helpful in the diagnosis.¹³

Also, samples obtained with bronchoalveolar lavage may be examined for fat. The macrophages may be stained for fat using the oil red O stain. Again, this is a nonspecific marker, as fat-stained macrophages are seen in trauma patients,³⁵ but the finding has a very high negative predictive value.³⁶ Anemia, thrombocytopenia, hypofibrinogenemia, an elevated lipase level, and a high erythrocyte sedimentation rate may be found in fat embolism syndrome.¹³

Chest radiography may show bilateral infiltrates, as in ARDS, but this is not diagnostic for fat embolism syndrome.

Electrocardiography may show changes in ST and T waves and signs of right heart strain.

Transesophageal echocardiography may show increased right heart and pulmonary artery pressures.

Computed tomography is often negative,^37,38 but T2-weighted MRI is useful in the diagnosis of cerebral fat embolism syndrome, as it can show intracerebral microinfarcts as early as 4 hours after the onset of neurologic symptoms, and these findings correlate well with the clinical severity of brain injury.

Diffusion-weighted MRI may enhance the sensitivity and specificity of the neuroradiologic diagnosis. Diffusion-weighted MRI typically shows multiple nonconfluent areas of high-intensity signals or bright spots on a dark background, known as a “starfield pattern.” This pattern has been suggested to be pathognomonic of acute cerebral microinfarction. The abnormalities presumably reflect foci of cytotoxic edema that develops immediately, unlike vasogenic edema, seen in T2-weighted images, which may take up to several days to develop. Although these images are not necessarily specific for fat emboli, they are useful in helping make the diagnosis. Thus, diffusionweighted MRI should be done if fat embolism syndrome is suspected.^38,39

CASE CONCLUDED

The patient’s course in the intensive care unit was further complicated by gastrointestinal bleeding and renal failure. His neurologic status did not improve. Repeated MRI of the brain showed evolving bilateral watershed infarction throughout the cortices. The neurologic consult service diagnosed the patient as having severe encephalopathy with a very poor prognosis. The decision was made to withdraw care. He was placed under palliative care and died on postoperative day 22.

DRUG TREATMENT OF FAT EMBOLISM SYNDROME

5. Which of the following drugs has been proven to be effective in treating fat embolism syndrome?

• Intravenous ethanol
• Steroids
• Heparin
• Dextran
• Aspirin
• None of the above

None of the above has been proven to be effective in treating this disorder. The management is largely supportive. Thus, prevention, early diagnosis, and symptom management are vital.

Pulmonary and hemodynamic support are the cornerstones of successful treatment. Aggressive respiratory support is often needed. Management of acute lung injury and ARDS focuses on achieving acceptable gas exchange while preventing ventilator-associated lung injury. Intravascular volume must be supported. Inotropes and pulmonary vasodilators may be required to maintain hemodynamics. Exacerbation of central nervous system ischemia from hypotension or hypoxia should be avoided.

If the thrombocytopenia leads to clinical bleeding, platelet transfusions may be warranted.

Supportive care should include prophylaxis of deep venous thrombosis and of gastrointestinal bleeding, and maintenance of nutrition.⁴⁰ Patients who receive supportive care generally have a favorable outcome, with a mortality rate of less than 10%.²⁸

Drug studies have been inconclusive

Drugs suggested in the treatment of fat embolism syndrome include heparin, aspirin, dextran, hypertonic glucose, and alcohol, but the results have been inconclusive.^{3,11,23,40–43}

Heparin stimulates lipase activity, consequently decreasing the concentration of circulating fat globules. However, the increase in levels of free fatty acids may actually worsen the clinical picture. For this reason, and because of anticoagulation concerns and evidence of increased mortality rates, heparin is now contraindicated in the treatment of fat embolism syndrome.^2,41,43

Alcohol. Patients with a higher blood alcohol level at the time of injury have been reported to have a lower incidence of fat embolism syndrome. Alcohol inhibits lipase, suppressing the rise of free fatty acids. In experimental studies, the incidence of fat embolism syndrome was lower when the blood alcohol level was maintained at 20 mg/dL. However, no prospective randomized trial has been done to determine the clinical efficacy of ethanol as a treatment for this condition.^5,42

Dextran has been advocated, owing to its ability to improve small-vessel perfusion, but bleeding risk and acute renal failure associated with this drug have limited its use.⁵

N-acetylcysteine has been shown to attenuate fat-induced lung injury in a study of rats with induced fat embolism syndrome.⁴⁴

Corticosteroid treatment for this condition is controversial. Studies in patients with femoral and tibial fractures show that steroids reduce the incidence of fat embolism syndrome when given prophylactically, and those treated with steroids had a higher Pao₂ than controls. Doses of methylprednisolone in these studies ranged between 9 mg/kg to 90 mg/kg. A drawback of these studies is their small number of patients.^12,32,45,46

A meta-analysis⁴⁷ of randomized trials of corticosteroids to prevent fat embolism syndrome in patients with long-bone fractures identified 104 such studies. Only 7 of the 104 were considered adequate. In 389 patients with long-bone fractures, prophylactic corticosteroids reduced the risk of fat embolism syndrome by 78% (95% confidence interval 43%–92%) and corticosteroids also significantly reduced the risk of hypoxia with no difference in rates of infection or death. However, the overall quality of the trials was poor, and the authors of the meta-analysis concluded that more study is needed before corticosteroids could be formally recommended.⁴⁷

There is no evidence that steroids improve the overall clinical course of already established fat embolism syndrome.^12,32,45 The dosing and optimal timing of administration have also not been established. High doses pose a risk of septic complications, which may be devastating for the posttrauma or postoperative patient.

A 75-year-old man with type 2 diabetes and hypothyroidism underwent bilateral total knee replacement at our hospital.

His functional capacity had been moderately limited by knee pain, but he could easily climb one flight of stairs without symptoms. His medications at that time included levothyroxine (Synthroid) and metformin (Glucophage). He had no known cardiac or pulmonary disease. The preoperative evaluation, including laboratory tests and electrocardiography, was within normal limits.

Spinal anesthesia was used for surgery, and he was given 2 mg of midazolam (Versed) intravenously for sedation. No additional sedation was given. He was given oxygen via nasal cannula at 2 L/min.

All vital signs were stable at the start of the procedure. However, about halfway through, when the thigh tourniquet was released, his oxygen saturation dropped abruptly from 100% to 92%. All other vital signs remained stable, and he was asymptomatic, was oriented to person, time, and place, was conversing freely, and was in no distress. The oxygen flow was increased to 6 L/min, his oxygen saturation improved, and the procedure was then completed as planned.

At the conclusion of the surgery, before the patient was transported to the postanesthesia care unit (PACU) and while his oxygen flow rate was still 6 L/min, his oxygen saturation again dropped to 92%. A simple face mask was placed, and the oxygen flow rate was increased to 10 L/min. His oxygen saturation stayed low, near 90%.

Bleeding during surgery had been nominal. He had received 2 L of lactated Ringer’s solution and 500 mL of hetastarch (Hextend) during surgery. He continued to be asymptomatic in the PACU.

1. What is the most likely cause of oxygen desaturation during bilateral total knee arthroplasty?

• Fat embolism
• Intraoperative pneumonia
• Venous thromboembolism with pulmonary embolism
• Acute myocardial infarction
• Acute pulmonary edema
• Excessive sedation

The differential diagnosis of oxygen desaturation during orthopedic procedures is listed in Table 1.

Fat embolism is the most likely cause, particularly given the greater fatty embolic load that occurs with bilateral total knee arthroplasty than with unilateral total knee arthroplasty.

At what point the maximal showering of fat emboli occurs is not known. Fat may be released into the circulation with pressurization of the medullary canal during surgery or with manipulation of a fracture. The emboli may collect in the leg veins and then be released in a shower when the thigh tourniquet is released. Vasoactive mediators and methylmethacrylate cement released into the circulatory system after tourniquet deflation may also cause vasodilation, hypotension, and increased dead-space ventilation, resulting in hypoxia and a drop in end-tidal CO₂.

Pneumonia during surgery is rare without an apparent aspiration event.

Venous thromboembolism is possible but is more likely later in the postoperative period after major orthopedic surgery.

Acute myocardial infarction could present with hypoxia, particularly in a diabetic patient, who may not experience chest pain. However, intraoperative electrocardiographic changes would likely be seen. If myocardial infarction is suspected, postoperative serial electrocardiograms and measuring troponin and cardiac enzyme levels aid in the diagnosis.

Acute pulmonary edema is possible but not as highly suspected, as the patient had no history of congestive heart failure and received an appropriate amount of fluid for this type of surgery.

Excessive sedation could cause hypoventilation and, thus, oxygen desaturatation. However, this patient’s oxygen desaturatation began more than an hour after the midazolam was given. Midazolam is a short-acting benzodiazepine. It is unlikely that the patient would show signs of hypoventilation and oversedation an hour after the drug was given. Our patient also did not show any signs of excessive sedation, as he was awake and conversing during the surgery.

Fat emboli vs fat embolism syndrome

Fat embolism is the presence of fat drops within the systemic and pulmonary microcirculation, with or without clinical sequelae.¹ Fat embolism syndrome, on the other hand, is defined as injury to and dysfunction of one or more organs as a result of the embolization of fat, usually within 24 hours of injury or orthopedic surgery.²

Fat embolism syndrome is an unpredictable condition with a varied presentation. Fat droplets are thought to embolize via the venous circulation into the pulmonary arteries, occluding small blood vessels in the lung. However, they also get into the arterial circulation and occlude arteries in the brain, kidney, heart, and liver (more on this phenomenon below).

Fat embolism is reported to originate primarily from fractures of the femur, tibia, and pelvis.^2,3 As many as 90% of trauma patients have been shown to have evidence of fat embolism on autopsy.⁴ However, only a small number of patients develop the classic fat embolism syndrome,^2,3,5 Why some develop the syndrome and others do not is still unknown.

Orthopedic procedures associated with fat embolization include knee arthroplasty and hip arthroplasty, particularly if it involves intramedullary manipulation or medullary fixation.⁶ It has also been reported during spinal procedures in which pedicular screws are used.⁷ The syndrome occurs in 0.25% to 30% of patients following multiple fractures and in 0.1% to 12% of patients during or following knee or hip arthroplasty.

One study⁸ showed evidence of fat on transesophageal echocardiography in 88% of patients undergoing medullary reaming of lower-extremity fractures and hip hemiarthroplasty. Blood sampling from the right atrium confirmed that fat was responsible for the echocardiographic abnormalities. The study also showed that the severity of the embolic showering correlated with the severity of hypoxia and the decrease in end-tidal CO₂.⁸

CASE CONTINUED

On arrival at the PACU, our patient’s oxygen saturation was 94% while he was breathing oxygen via a simple face mask at a flow rate of 10 L/min. His heart rate was 60 bpm, blood pressure 110/60 mm Hg, and temperature 37.5°C (96.3°F). Chest sounds were normal on auscultation.

However, 3 hours later, his mental status rapidly deteriorated. He was oriented only to person, and he was drowsy. He had escalating respiratory distress with a rapid respiratory rate and decreasing oxygen saturation. At this point, auscultation of his chest wall revealed bilateral crackles and rales.

He was promptly intubated. Profuse fluid and secretions were noted to be coming from his lungs, filling the endotracheal tube. Arterial blood gas measurement showed a pH of 7.22, Pao₂ 64 mm Hg, and Paco₂ 56 mm Hg on 100% fraction of inspired oxygen, with no increased anion gap.

2. Which consequence of fat embolism is most likely at this time in this patient?

• Coexisting sepsis
• Fat embolism syndrome
• Acute cardioembolic stroke
• Anaphylaxis

Fat embolism syndrome should be highly suspected in this patient. As mentioned, it can affect many different organs. It is the most serious condition resulting from fat embolization after surgery or trauma.

Sepsis was unlikely in our patient, since he presented for his surgery in good health and with no preexisting signs or symptoms of infection. Acute cardioembolic stroke could have caused the neurologic signs, but this would not necessarily explain the coexisting hypoxia. An anaphylactic reaction to drugs or surgical cement would most likely present intraoperatively, shortly after exposure occurred, rather than several hours after surgery.

How common is fat embolism syndrome?

The occurrence rate of fat embolism syndrome has been reported to be 0.25% to 30% after multiple fractures and 0.1% to 12% after knee and hip joint surgery, with a mortality rate of 13% to 36%.^2,9–14 The rate of occurrence after unilateral total knee joint replacement has been reported to be 1.8% to 5%, and 4% to 12% after bilateral total knee replacement.^15–19

The syndrome is relatively more common with traumatic fractures of the lower extremities. However, it has also been reported with liposuction, total parenteral nutrition, bone marrow harvest and transplantation, burns, and acute pancreatitis, to mention a few.¹⁰

The broad range of reported incidence rates can be attributed to the fact that many studies were in patients with multiple trauma, whose concomitant injuries may have made it difficult to clearly define the contribution of fat embolism syndrome to the overall rates of morbidity and mortality. Also, different studies used different criteria to define the syndrome.

How does fat embolism syndrome occur?

Two hypotheses for how this syndrome occurs were proposed nearly a century ago.^20,21

The “mechanical” theory is that fat emboli are formed as a result of trauma and disruption of adipose tissue and other cells in the bone marrow. Increases in intramedullary pressure force the fat emboli through damaged medullary venous channels in the bone and into the circulation of the lower extremities. This embolization of fat causes an initial mechanical pulmonary obstruction. Mechanical obstruction by fat emboli in the pulmonary system leads to increased pulmonary pressures and an increase in right heart outflow pressure. The right heart becomes strained, leading to a decreased right-sided cardiac output. As a result, the left heart filling pressures diminish and hypotension ensues.²⁰

The “biochemical” theory, on the other hand, is that chylomicrons within the vascular system are modified and their stability is compromised as a result of stress. These traumatized chylomicrons then coalesce to form droplets of fat that accumulate in the pulmonary circulation and produce a mechanical obstruction. This would explain why nontraumatic, nonorthopedic insults can produce this syndrome.

Autopsy studies show that there is little correlation between the presence and quantity of intravascular fat and the severity of clinical symptoms, thus implying that the syndrome is caused by more than just mechanical obstruction. The biochemical theory postulates that fat globules within the circulatory system then cause the release of lipase from the pulmonary alveolar cells, which then hydrolyses the fat into free fatty acids. These free fatty acids cause an inflammatory reaction, complementmediated leukocyte aggregation, chemotoxin release, and subsequently endothelial damage. These vasoactive substances damage type 2 pneumocytes and lead to an increased permeability of the pulmonary capillary beds. Acute respiratory distress syndrome (ARDS) may ensue. Disseminated intravascular coagulation may occur as a result of the formation of microthrombi involving lipids, platelets, and fibrin.^21,22

Embolization of fat to the central nervous system can occur as fat globules cross into the systemic circulation via a patent foramen ovale, an atrioventricular shunt, or the pulmonary capillaries. This can then result in cerebral ischemia.²³

Although patent foramen ovale may seem the most direct route for cerebral embolization, the neurologic impairment and signs of cerebral emboli in fat embolism syndrome may occur in the absence of patent foramen ovale.^24,25 The fat globules may actually go through the lung capillaries, being flexible and forced through by increased pulmonary pressure.

But whether the cause of fat embolism syndrome is occlusion by globules, the release of biochemical mediators, or a combination of both is unknown. Both mechanisms are likely responsible. We can only suspect that the degree of fat load and intrinsic metabolic differences between individuals account for the variation in susceptibility.

FAT EMBOLI AFFECT THE LUNGS, SKIN, AND BRAIN

3. Where on the body is the rash associated with fat embolism syndrome usually seen?

• Face
• Near a site of fracture or surgery
• Chest, axilla, conjunctiva
• Distal extremities

Petechiae are part of the classic presenting triad of fat embolism syndrome, which also includes pulmonary and cerebral dysfunction.

Petechiae usually appear on the 2nd to 4th day after injury.²⁶ They are usually found across the chest, the anterior axillary folds, and the neck, as well as on the oral mucosa and the conjunctiva. The rash is caused by occlusion of dermal capillaries by fat, which increases their fragility.¹⁰

Pulmonary changes usually begin with tachypnea, dyspnea, and a drop in oxygen saturation, leading to generalized hypoxia. Respiratory symptoms are present in 100% of cases.² Respiratory symptoms can acutely develop with the sudden manipulation of a fracture, reaming of bone, or release of a limb tourniquet.²⁷

Body systems affected by fat embolism syndrome are summarized in Table 2.

4. How many hours after injury does fat embolism syndrome typically manifest?

• 1 to 2 hours
• 6 to 12 hours
• 12 to 20 hours
• 24 to 48 hours
• 72 to 84 hours

Most patients develop signs and symptoms 24 to 48 hours after injury. Patients presenting earlier than 12 hours usually have a more fulminant course.²⁹

The time between fat embolization and the development of fat embolism syndrome is thought to be related to the time required for the metabolic conversion of fat to free fatty acids.³⁰ We suspect that the early desaturation seen in our patient was the result of a heavy showering of fat intraoperatively. However, this could only be concluded after we had ruled out other causes of acute hypoxia and hypotension.

Fat embolism syndrome is a diagnosis of exclusion and is based on clinical criteria. No specific sign, symptom, or test is pathognomonic. It may often be confused with other conditions such as systemic inflammatory response syndrome or sepsis. However, the triad of respiratory and neurologic symptoms and petechiae coupled with the clinical picture of recent trauma or orthopedic surgery almost assures the diagnosis.

Fat embolism syndrome can range from subclinical to fulminating, with the more fulminating course attributable to a huge load of fat emboli, which leads to acute cor pulmonale.

Regardless of the criteria used, one must have a high index of suspicion for fat embolization syndrome in patients undergoing orthopedic procedures, particularly hip and knee surgery, and in patients with fractures, especially fractures of the femur, tibia, or pelvis and multiple, concomitant fractures.

CASE CONTINUED

Our patient was given furosemide (Lasix) empirically for diuresis and to improve oxygenation. However, his oxygen saturation remained low.

Chest radiography 4 hours after surgery showed bilateral pulmonary infiltrates. Serial electrocardiography showed no acute changes. Levels of cardiac enzymes and troponins were normal. Transthoracic echocardiography showed no left ventricular dysfunction, a normal right ventricle, and no evidence of valvular lesions. Urine and blood fat stains were negative, but the sputum stain was positive for copious extracellular fat. The patient became comatose 5 hours postoperatively. Computed tomography of the brain was normal. He was transferred to the surgical intensive care unit.

The clinical course was marked by hemodynamic instability requiring norepinephrine (Levophed) and vasopressin (Pitressin) for hypotension. Right ventricular filling pressures via central venous pressure monitoring showed no evidence of hypovolemia. The hemoglobin concentration and the hematocrit were stable, with no evidence of acute or ongoing bleeding. Blood, urine, and sputum cultures remained negative. Acute myocardial infarction was ruled out by serial electrocardiography, cardiac enzyme testing, and troponin testing.

Magnetic resonance imaging (MRI) of the brain on postoperative day 2 showed foci of acute ischemia suggestive of embolic phenomena consistent with fat embolism syndrome (Figure  1). Transthoracic echocardiography was repeated but again showed no evidence of a patent foramen ovale. Electroencephalography on postoperative day 4 showed severe, diffuse encephalopathy. There was no petechial skin rash. Other laboratory studies showed progressive thrombocytopenia with a platelet count of 53 × 199/L on postoperative day 3.

TESTS THAT AID THE CLINICAL DIAGNOSIS

Although no single laboratory test is pathognomonic for fat embolism syndrome, several tests may help raise suspicion of it, especially in the setting of fracture or an orthopedic surgical procedure.

Arterial blood gases must be measured. A Pao₂ of less than 60 mm Hg with no other obvious lung pathology in an orthopedic surgery patient is highly suspicious.¹² An alveolar-arterial gradient of greater than 100 mm Hg may further increase suspicion.

Tests for fat. The blood and urine may be examined for fat, although positive findings are not specific for fat embolism syndrome.³³ Fat in the urine indicates the occurrence of massive fat embolism, but this is not always accompanied by the syndrome.³⁴ Gurd and Wilson¹³ found fat globules larger than 8 μm circulating in the serum in all documented cases. They stated that, even though the relationship of large fat globules to the pathogenesis of the clinical picture remains obscure, the demonstration of their presence can be helpful in the diagnosis.¹³

Also, samples obtained with bronchoalveolar lavage may be examined for fat. The macrophages may be stained for fat using the oil red O stain. Again, this is a nonspecific marker, as fat-stained macrophages are seen in trauma patients,³⁵ but the finding has a very high negative predictive value.³⁶ Anemia, thrombocytopenia, hypofibrinogenemia, an elevated lipase level, and a high erythrocyte sedimentation rate may be found in fat embolism syndrome.¹³

Chest radiography may show bilateral infiltrates, as in ARDS, but this is not diagnostic for fat embolism syndrome.

Electrocardiography may show changes in ST and T waves and signs of right heart strain.

Transesophageal echocardiography may show increased right heart and pulmonary artery pressures.

Computed tomography is often negative,^37,38 but T2-weighted MRI is useful in the diagnosis of cerebral fat embolism syndrome, as it can show intracerebral microinfarcts as early as 4 hours after the onset of neurologic symptoms, and these findings correlate well with the clinical severity of brain injury.

Diffusion-weighted MRI may enhance the sensitivity and specificity of the neuroradiologic diagnosis. Diffusion-weighted MRI typically shows multiple nonconfluent areas of high-intensity signals or bright spots on a dark background, known as a “starfield pattern.” This pattern has been suggested to be pathognomonic of acute cerebral microinfarction. The abnormalities presumably reflect foci of cytotoxic edema that develops immediately, unlike vasogenic edema, seen in T2-weighted images, which may take up to several days to develop. Although these images are not necessarily specific for fat emboli, they are useful in helping make the diagnosis. Thus, diffusionweighted MRI should be done if fat embolism syndrome is suspected.^38,39

CASE CONCLUDED

The patient’s course in the intensive care unit was further complicated by gastrointestinal bleeding and renal failure. His neurologic status did not improve. Repeated MRI of the brain showed evolving bilateral watershed infarction throughout the cortices. The neurologic consult service diagnosed the patient as having severe encephalopathy with a very poor prognosis. The decision was made to withdraw care. He was placed under palliative care and died on postoperative day 22.

DRUG TREATMENT OF FAT EMBOLISM SYNDROME

5. Which of the following drugs has been proven to be effective in treating fat embolism syndrome?

• Intravenous ethanol
• Steroids
• Heparin
• Dextran
• Aspirin
• None of the above

None of the above has been proven to be effective in treating this disorder. The management is largely supportive. Thus, prevention, early diagnosis, and symptom management are vital.

Pulmonary and hemodynamic support are the cornerstones of successful treatment. Aggressive respiratory support is often needed. Management of acute lung injury and ARDS focuses on achieving acceptable gas exchange while preventing ventilator-associated lung injury. Intravascular volume must be supported. Inotropes and pulmonary vasodilators may be required to maintain hemodynamics. Exacerbation of central nervous system ischemia from hypotension or hypoxia should be avoided.

If the thrombocytopenia leads to clinical bleeding, platelet transfusions may be warranted.

Supportive care should include prophylaxis of deep venous thrombosis and of gastrointestinal bleeding, and maintenance of nutrition.⁴⁰ Patients who receive supportive care generally have a favorable outcome, with a mortality rate of less than 10%.²⁸

Drug studies have been inconclusive

Drugs suggested in the treatment of fat embolism syndrome include heparin, aspirin, dextran, hypertonic glucose, and alcohol, but the results have been inconclusive.^{3,11,23,40–43}

Heparin stimulates lipase activity, consequently decreasing the concentration of circulating fat globules. However, the increase in levels of free fatty acids may actually worsen the clinical picture. For this reason, and because of anticoagulation concerns and evidence of increased mortality rates, heparin is now contraindicated in the treatment of fat embolism syndrome.^2,41,43

Alcohol. Patients with a higher blood alcohol level at the time of injury have been reported to have a lower incidence of fat embolism syndrome. Alcohol inhibits lipase, suppressing the rise of free fatty acids. In experimental studies, the incidence of fat embolism syndrome was lower when the blood alcohol level was maintained at 20 mg/dL. However, no prospective randomized trial has been done to determine the clinical efficacy of ethanol as a treatment for this condition.^5,42

Dextran has been advocated, owing to its ability to improve small-vessel perfusion, but bleeding risk and acute renal failure associated with this drug have limited its use.⁵

N-acetylcysteine has been shown to attenuate fat-induced lung injury in a study of rats with induced fat embolism syndrome.⁴⁴

Corticosteroid treatment for this condition is controversial. Studies in patients with femoral and tibial fractures show that steroids reduce the incidence of fat embolism syndrome when given prophylactically, and those treated with steroids had a higher Pao₂ than controls. Doses of methylprednisolone in these studies ranged between 9 mg/kg to 90 mg/kg. A drawback of these studies is their small number of patients.^12,32,45,46

A meta-analysis⁴⁷ of randomized trials of corticosteroids to prevent fat embolism syndrome in patients with long-bone fractures identified 104 such studies. Only 7 of the 104 were considered adequate. In 389 patients with long-bone fractures, prophylactic corticosteroids reduced the risk of fat embolism syndrome by 78% (95% confidence interval 43%–92%) and corticosteroids also significantly reduced the risk of hypoxia with no difference in rates of infection or death. However, the overall quality of the trials was poor, and the authors of the meta-analysis concluded that more study is needed before corticosteroids could be formally recommended.⁴⁷

There is no evidence that steroids improve the overall clinical course of already established fat embolism syndrome.^12,32,45 The dosing and optimal timing of administration have also not been established. High doses pose a risk of septic complications, which may be devastating for the posttrauma or postoperative patient.

After nearly a half-century, the subspecialty of critical care medicine—uniquely trained physicians caring for critically ill or injured patients in specialized, discrete nursing units—continues to suffer from an identity crisis.

Too often, the role of the intensivist in caring for the patient is unclear, to the patient, to the family, and to other physicians. Is the intensivist merely a consultant, or does he or she have a larger role?

The time has come to end the identity crisis with a fundamental paradigm shift, to identify intensivists as the principal caregivers of critically ill patients, ie, the “primary critical care physicians,” or PCCPs. We think this is necessary based not only on evidence from clinical studies, but also on our decades of experience as intensivist caregivers in a high-intensity, closed-staffing model.

REASONS FOR THE IDENTITY CRISIS

The reasons for the continued identity crisis of intensivists are many and complex.

To begin with, other physicians tend to be ambiguous about the duties of intensivists, and the general population is mostly unaware of the subspecialty. In contrast to mature subspecialties such as cardiology or gastroenterology, where responsibilities are generally known to physicians and the lay public alike, or in contrast even to recently evolved specialties such as emergency medicine, the enigmatic roles of an intensivist may differ depending on primary specialty (anesthesiology, internal medicine, surgery) and the patient population, or even among intensive care units (ICUs) within the same hospital.

Moreover, that an identity crisis exists is even more surprising given the disproportionately large consumption by critical care medicine of finite economic resources. One would expect that a sector of health care that expends 1% of the GNP¹ would have clearly explicit roles and responsibilities for its physicians.

Nearly three-quarters of the care by intensivists in the United States is delivered in what is considered an “open” or “low-intensity” ICU staffing model²: an intensivist makes treatment recommendations but otherwise has no overarching authority over patient care. In this model, the admitting physician is not trained in critical care and is not available throughout the day to make decisions concerning the management of the patient. In addition, various consulting physicians and single-organ specialists may not be aware of the overall management plan, resulting in potentially unnecessary or conflicting orders and increased expense.² What is more, in an open ICU model, critical care nurses are often left to detect and correct a significant change in a patient’s status without the necessary immediate physician availability, resulting not only in a stressful working environment for nursing staff, but also in potential harm associated with individuals providing care outside their scope of practice.³

In only a small percentage of ICUs—mostly medical ICUs and ICUs in teaching hospitals—is critical care provided in a “high-intensity” or “closed” staffing pattern, in which treatment decisions are cohesively managed under the guidance of an intensivist.²

EVIDENCE IN THE MEDICAL LITERATURE

Staffing patterns in the ICU

Several studies have attempted to identify the consequences of these different ICU staffing patterns on patient care.

Hanson et al⁴ examined two concurrent patient cohorts admitted to a surgical ICU. The study cohort was cared for by an on-site critical care team supervised by an intensivist, while the control cohort received care from a team with patient care responsibilities in multiple sites, supervised by a general surgeon. The results showed that patients cared for by the critical care team spent less time in the ICU, used fewer resources, had fewer complications, and had lower total hospital charges. The difference between the two cohorts was most evident in patients with the worst Acute Physiology and Chronic Health Evaluation (APACHE) II scores.

According to Hanson et al, the lack of an accepted prototype for the delivery of critical care is due to factors such as the relative youth of the discipline, contention over control of individual patient management, and the absence of a single academic advocate.⁴

Moreover, Pronovost et al⁵ concluded that high-intensity staffing (mandatory intensivist consultation or closed ICU) was associated with lower ICU mortality rates in 93% of studies and with a reduced ICU length of stay in the high-intensity staffing units when compared with ICUs with low-intensity staffing (no intensivist or elective intensivist consultation).

Critics of our PCCP paradigm may point to a study by Levy et al⁶ that, using a database of more than 100,000 patients, could not demonstrate any survival benefit with management by critical care physicians. Indeed the study found that patients managed by intensivists had a higher mortality rate than patients managed by physicians not trained in critical care. However, they also showed that more patients managed for the entire stay by intensivists received interventions such as intravenous drugs, mechanical ventilation, and continuous sedation and that they had a higher mean severity of illness as measured by the expanded Simplified Acute Physiology Score (SAPS II) and higher hospital mortality rates than patients who were not managed by a critical care team.

According to Levy et al, most ICUs in the United States are structured as completely open units in which the admitting physicians retain full clinical and decisional responsibility and thus have the option to care for their patients with or without input from intensivists.⁶

However, a recent study by Kim et al⁷ likely rebuts the findings of Levy et al. Kim et al analyzed more than 100,000 ICU admissions and found that the lowest odds of death within 30 days were in ICUs that had high-intensity physician staffing and multidisciplinary care teams, suggesting that the presence of an intensivist confers a survival benefit.

Other studies have also shown that high-intensity staffing improves patient outcomes in the ICU.^5,8,9

Issues of cost and use of resources

Issues concerning cost and human resources for staffing ICUs have acquired increasing importance. According to Angus et al,¹⁰ intensivists provided care to only 36.8% of all ICU patients. The demand for critical care services will continue to grow rapidly as the population ages. It is this shift in the care of the critically ill that requires intensivists to take on the role of the PCCP, so as to provide high-quality, evidence-based critical care and to promote a long-term sustainable model of physician and nursing care.

OUR EXPERIENCE

Our intensivist group has been providing a near-primary-care style of critical care practice for almost 40 years, from its inception in 1977 by one of the authors (A.B.), to our current group of 15 board-certified intensivists. We can easily cite the clinical value of our practice approach, with outcome data showing consistent and better-than-expected Standardized Mortality Ratio accounts from our APACHE IV data (personal communication, Cleveland Clinic Cerner/APACHE IV report), or with reports showing that the presence of a full-time, attending-level, in-house staff physician ensures that patients, surgeons, and consultants have confidence and respect for the care provided. However, we feel that the intangible components are what make our practice a prototype for the PCCP model.

A dedicated team with a low turnover rate

First, we have a team of anesthesiology- and surgery-based intensivists dedicated to ICU practice, with a very low turnover or burnout rate, in contrast to most ICUs in the United States, where intensivists tend to practice part-time (at other times either providing operating-room-based anesthesia or surgical care or working in a pulmonary- or sleep-lab-based practice). We believe this point should not go unstressed: we have a team of physicians who have dedicated their career to working in the ICU full-time, and some have done so in excess of 20 years, even as long as 30 years! It is our opinion that we are able to provide such a highly desirable working environment by a unique daily staffing model that does not utilize the conventional practice style of one intensivist on-call per week.

We also feel that our model dramatically reduces the risk of burnout by permitting our attending intensivists to break up on-call sequences so that there are days on which work in the ICU is not also associated with on-call responsibilities.

A successful fellowship program

Second, we have an extremely successful fellowship program, which began in 1974 when one of the authors (A.B.) advocated the training of anesthesiology residents as intensivists.¹¹ The American Board of Anesthesiology certifies on average 55 candidates per year in critical care medicine, and our program trains about 10% of the physicians applying for certification. In most years, there are actually more candidates for our program than there are available positions, which is atypical for anesthesiology-based critical care training programs. This wealth of young, talented candidates interested in critical care as a career is, again, in contrast to most anesthesiology-based programs, which find it difficult to enroll even one fellow per year.

Critical care programs grounded in anesthesiology typically struggle because of the realities of economics.¹² The payoff of operating-room-based anesthesiology practices generally outshines those in critical care, yet we already have three times as many candidates as there are positions to start our training program in the next 2 years. We feel that candidates are attracted to our program simply because our environment (dedicated staffing, equal clinical footing with surgeons, low burnout rates) is seen as an exciting, positively charged role-modeling atmosphere for young physicians who may have a career interest that involves more than just their original base specialty.

A collegial working relationship

Third, we have a thriving, collegial working relationship—including daily bedside and weekly bioethics rounds with our nursing staff—which has fueled a high degree of professional satisfaction among nurses. This is evidenced by the extremely low turnover rate of nurses (less than 5% per year in the last 5 years) and by national recognition for nursing excellence (Beacon Award for Critical Care Excellence, American Association of Critical Care Nurses) (personal communication, S. Wilson, Nurse Manager). In 2009, the four nurses out of 174 who left did so to further their careers.

While low turnover rates among nurses and award-winning practices are surely a testament to a highly motivated and skilled nursing team, there is no question that a constructive collegiality among the physicians and nurses has provided an environment to allow these positive aspects to flourish.

OVERCOMING ROADBLOCKS

Obviously, although in theory it is easy to proclaim a PCCP paradigm, in reality the roadblocks are many.

For example, standardization of education and credentialing would be an essential hurdle to overcome. The current educational arrangement of the various adult specialties (anesthesiology, internal medicine, surgery), each offering disparate subspecialty critical care training and certification, is deeply rooted in interdisciplinary politics, but without any demonstration of improved patient care.¹³ As described recently by Kaplan and Shaw,¹⁴ an all-encompassing training and credentialing standard for critical care is essential for 21st century medicine and would go a long way toward development of the PCCP paradigm.

Another major roadblock is the shortage of intensivists in the United States.¹³ There are many reasons why physicians opt not to select critical care as a career, such as a non-straight-forward training pathway (as described above), recognition that the 24-hours per day, 7-days-per-week nature of critical care affects lifestyle issues, and inconsistent physician compensation.¹³

However, technological and personnel advances, including the use of electronic (e-ICU)¹⁵ and mid-level practitioner models, have led to creative approaches to extend critical care coverage.¹³

Additionally, the multitude of physician specialty stakeholders and the overall flux of the future of medical care in the United States all would contribute to the difficulties of prioritizing the implementation of the PCCP concept. Also, our practice style—a large intensivist group working in an ostensibly closed surgical ICU in a tertiary-care hospital—is one possible model, as is the even more highly evolved Cleveland Clinic medical ICU, where medical intensivists are already essentially PCCPs. But these models of care may not be generalizable among the local care patterns and medical politics across hospitals or ICUs.

Based on the described successes of our practice model, coupled with evidence in the literature, we have proposed a paradigm shift toward the concept of a PCCP. To be sure, paradigm shifts nearly always require time, effort, and wherewithal. In the end, however, we feel that embracement of the PCCP paradigm would result in a concise, discrete understanding of the role of intensivist, eliminate the specialty’s identity crisis, and ultimately improve patient care.

After nearly a half-century, the subspecialty of critical care medicine—uniquely trained physicians caring for critically ill or injured patients in specialized, discrete nursing units—continues to suffer from an identity crisis.

Too often, the role of the intensivist in caring for the patient is unclear, to the patient, to the family, and to other physicians. Is the intensivist merely a consultant, or does he or she have a larger role?

The time has come to end the identity crisis with a fundamental paradigm shift, to identify intensivists as the principal caregivers of critically ill patients, ie, the “primary critical care physicians,” or PCCPs. We think this is necessary based not only on evidence from clinical studies, but also on our decades of experience as intensivist caregivers in a high-intensity, closed-staffing model.

REASONS FOR THE IDENTITY CRISIS

The reasons for the continued identity crisis of intensivists are many and complex.

To begin with, other physicians tend to be ambiguous about the duties of intensivists, and the general population is mostly unaware of the subspecialty. In contrast to mature subspecialties such as cardiology or gastroenterology, where responsibilities are generally known to physicians and the lay public alike, or in contrast even to recently evolved specialties such as emergency medicine, the enigmatic roles of an intensivist may differ depending on primary specialty (anesthesiology, internal medicine, surgery) and the patient population, or even among intensive care units (ICUs) within the same hospital.

Moreover, that an identity crisis exists is even more surprising given the disproportionately large consumption by critical care medicine of finite economic resources. One would expect that a sector of health care that expends 1% of the GNP¹ would have clearly explicit roles and responsibilities for its physicians.

Nearly three-quarters of the care by intensivists in the United States is delivered in what is considered an “open” or “low-intensity” ICU staffing model²: an intensivist makes treatment recommendations but otherwise has no overarching authority over patient care. In this model, the admitting physician is not trained in critical care and is not available throughout the day to make decisions concerning the management of the patient. In addition, various consulting physicians and single-organ specialists may not be aware of the overall management plan, resulting in potentially unnecessary or conflicting orders and increased expense.² What is more, in an open ICU model, critical care nurses are often left to detect and correct a significant change in a patient’s status without the necessary immediate physician availability, resulting not only in a stressful working environment for nursing staff, but also in potential harm associated with individuals providing care outside their scope of practice.³

In only a small percentage of ICUs—mostly medical ICUs and ICUs in teaching hospitals—is critical care provided in a “high-intensity” or “closed” staffing pattern, in which treatment decisions are cohesively managed under the guidance of an intensivist.²

EVIDENCE IN THE MEDICAL LITERATURE

Staffing patterns in the ICU

Several studies have attempted to identify the consequences of these different ICU staffing patterns on patient care.

Hanson et al⁴ examined two concurrent patient cohorts admitted to a surgical ICU. The study cohort was cared for by an on-site critical care team supervised by an intensivist, while the control cohort received care from a team with patient care responsibilities in multiple sites, supervised by a general surgeon. The results showed that patients cared for by the critical care team spent less time in the ICU, used fewer resources, had fewer complications, and had lower total hospital charges. The difference between the two cohorts was most evident in patients with the worst Acute Physiology and Chronic Health Evaluation (APACHE) II scores.

According to Hanson et al, the lack of an accepted prototype for the delivery of critical care is due to factors such as the relative youth of the discipline, contention over control of individual patient management, and the absence of a single academic advocate.⁴

Moreover, Pronovost et al⁵ concluded that high-intensity staffing (mandatory intensivist consultation or closed ICU) was associated with lower ICU mortality rates in 93% of studies and with a reduced ICU length of stay in the high-intensity staffing units when compared with ICUs with low-intensity staffing (no intensivist or elective intensivist consultation).

Critics of our PCCP paradigm may point to a study by Levy et al⁶ that, using a database of more than 100,000 patients, could not demonstrate any survival benefit with management by critical care physicians. Indeed the study found that patients managed by intensivists had a higher mortality rate than patients managed by physicians not trained in critical care. However, they also showed that more patients managed for the entire stay by intensivists received interventions such as intravenous drugs, mechanical ventilation, and continuous sedation and that they had a higher mean severity of illness as measured by the expanded Simplified Acute Physiology Score (SAPS II) and higher hospital mortality rates than patients who were not managed by a critical care team.

According to Levy et al, most ICUs in the United States are structured as completely open units in which the admitting physicians retain full clinical and decisional responsibility and thus have the option to care for their patients with or without input from intensivists.⁶

However, a recent study by Kim et al⁷ likely rebuts the findings of Levy et al. Kim et al analyzed more than 100,000 ICU admissions and found that the lowest odds of death within 30 days were in ICUs that had high-intensity physician staffing and multidisciplinary care teams, suggesting that the presence of an intensivist confers a survival benefit.

Other studies have also shown that high-intensity staffing improves patient outcomes in the ICU.^5,8,9

Issues of cost and use of resources

Issues concerning cost and human resources for staffing ICUs have acquired increasing importance. According to Angus et al,¹⁰ intensivists provided care to only 36.8% of all ICU patients. The demand for critical care services will continue to grow rapidly as the population ages. It is this shift in the care of the critically ill that requires intensivists to take on the role of the PCCP, so as to provide high-quality, evidence-based critical care and to promote a long-term sustainable model of physician and nursing care.

OUR EXPERIENCE

Our intensivist group has been providing a near-primary-care style of critical care practice for almost 40 years, from its inception in 1977 by one of the authors (A.B.), to our current group of 15 board-certified intensivists. We can easily cite the clinical value of our practice approach, with outcome data showing consistent and better-than-expected Standardized Mortality Ratio accounts from our APACHE IV data (personal communication, Cleveland Clinic Cerner/APACHE IV report), or with reports showing that the presence of a full-time, attending-level, in-house staff physician ensures that patients, surgeons, and consultants have confidence and respect for the care provided. However, we feel that the intangible components are what make our practice a prototype for the PCCP model.

A dedicated team with a low turnover rate

First, we have a team of anesthesiology- and surgery-based intensivists dedicated to ICU practice, with a very low turnover or burnout rate, in contrast to most ICUs in the United States, where intensivists tend to practice part-time (at other times either providing operating-room-based anesthesia or surgical care or working in a pulmonary- or sleep-lab-based practice). We believe this point should not go unstressed: we have a team of physicians who have dedicated their career to working in the ICU full-time, and some have done so in excess of 20 years, even as long as 30 years! It is our opinion that we are able to provide such a highly desirable working environment by a unique daily staffing model that does not utilize the conventional practice style of one intensivist on-call per week.

We also feel that our model dramatically reduces the risk of burnout by permitting our attending intensivists to break up on-call sequences so that there are days on which work in the ICU is not also associated with on-call responsibilities.

A successful fellowship program

Second, we have an extremely successful fellowship program, which began in 1974 when one of the authors (A.B.) advocated the training of anesthesiology residents as intensivists.¹¹ The American Board of Anesthesiology certifies on average 55 candidates per year in critical care medicine, and our program trains about 10% of the physicians applying for certification. In most years, there are actually more candidates for our program than there are available positions, which is atypical for anesthesiology-based critical care training programs. This wealth of young, talented candidates interested in critical care as a career is, again, in contrast to most anesthesiology-based programs, which find it difficult to enroll even one fellow per year.

Critical care programs grounded in anesthesiology typically struggle because of the realities of economics.¹² The payoff of operating-room-based anesthesiology practices generally outshines those in critical care, yet we already have three times as many candidates as there are positions to start our training program in the next 2 years. We feel that candidates are attracted to our program simply because our environment (dedicated staffing, equal clinical footing with surgeons, low burnout rates) is seen as an exciting, positively charged role-modeling atmosphere for young physicians who may have a career interest that involves more than just their original base specialty.

A collegial working relationship

Third, we have a thriving, collegial working relationship—including daily bedside and weekly bioethics rounds with our nursing staff—which has fueled a high degree of professional satisfaction among nurses. This is evidenced by the extremely low turnover rate of nurses (less than 5% per year in the last 5 years) and by national recognition for nursing excellence (Beacon Award for Critical Care Excellence, American Association of Critical Care Nurses) (personal communication, S. Wilson, Nurse Manager). In 2009, the four nurses out of 174 who left did so to further their careers.

While low turnover rates among nurses and award-winning practices are surely a testament to a highly motivated and skilled nursing team, there is no question that a constructive collegiality among the physicians and nurses has provided an environment to allow these positive aspects to flourish.

OVERCOMING ROADBLOCKS

Obviously, although in theory it is easy to proclaim a PCCP paradigm, in reality the roadblocks are many.

For example, standardization of education and credentialing would be an essential hurdle to overcome. The current educational arrangement of the various adult specialties (anesthesiology, internal medicine, surgery), each offering disparate subspecialty critical care training and certification, is deeply rooted in interdisciplinary politics, but without any demonstration of improved patient care.¹³ As described recently by Kaplan and Shaw,¹⁴ an all-encompassing training and credentialing standard for critical care is essential for 21st century medicine and would go a long way toward development of the PCCP paradigm.

Another major roadblock is the shortage of intensivists in the United States.¹³ There are many reasons why physicians opt not to select critical care as a career, such as a non-straight-forward training pathway (as described above), recognition that the 24-hours per day, 7-days-per-week nature of critical care affects lifestyle issues, and inconsistent physician compensation.¹³

However, technological and personnel advances, including the use of electronic (e-ICU)¹⁵ and mid-level practitioner models, have led to creative approaches to extend critical care coverage.¹³

Additionally, the multitude of physician specialty stakeholders and the overall flux of the future of medical care in the United States all would contribute to the difficulties of prioritizing the implementation of the PCCP concept. Also, our practice style—a large intensivist group working in an ostensibly closed surgical ICU in a tertiary-care hospital—is one possible model, as is the even more highly evolved Cleveland Clinic medical ICU, where medical intensivists are already essentially PCCPs. But these models of care may not be generalizable among the local care patterns and medical politics across hospitals or ICUs.

Based on the described successes of our practice model, coupled with evidence in the literature, we have proposed a paradigm shift toward the concept of a PCCP. To be sure, paradigm shifts nearly always require time, effort, and wherewithal. In the end, however, we feel that embracement of the PCCP paradigm would result in a concise, discrete understanding of the role of intensivist, eliminate the specialty’s identity crisis, and ultimately improve patient care.

After nearly a half-century, the subspecialty of critical care medicine—uniquely trained physicians caring for critically ill or injured patients in specialized, discrete nursing units—continues to suffer from an identity crisis.

Too often, the role of the intensivist in caring for the patient is unclear, to the patient, to the family, and to other physicians. Is the intensivist merely a consultant, or does he or she have a larger role?

The time has come to end the identity crisis with a fundamental paradigm shift, to identify intensivists as the principal caregivers of critically ill patients, ie, the “primary critical care physicians,” or PCCPs. We think this is necessary based not only on evidence from clinical studies, but also on our decades of experience as intensivist caregivers in a high-intensity, closed-staffing model.

REASONS FOR THE IDENTITY CRISIS

The reasons for the continued identity crisis of intensivists are many and complex.

To begin with, other physicians tend to be ambiguous about the duties of intensivists, and the general population is mostly unaware of the subspecialty. In contrast to mature subspecialties such as cardiology or gastroenterology, where responsibilities are generally known to physicians and the lay public alike, or in contrast even to recently evolved specialties such as emergency medicine, the enigmatic roles of an intensivist may differ depending on primary specialty (anesthesiology, internal medicine, surgery) and the patient population, or even among intensive care units (ICUs) within the same hospital.

Moreover, that an identity crisis exists is even more surprising given the disproportionately large consumption by critical care medicine of finite economic resources. One would expect that a sector of health care that expends 1% of the GNP¹ would have clearly explicit roles and responsibilities for its physicians.

Nearly three-quarters of the care by intensivists in the United States is delivered in what is considered an “open” or “low-intensity” ICU staffing model²: an intensivist makes treatment recommendations but otherwise has no overarching authority over patient care. In this model, the admitting physician is not trained in critical care and is not available throughout the day to make decisions concerning the management of the patient. In addition, various consulting physicians and single-organ specialists may not be aware of the overall management plan, resulting in potentially unnecessary or conflicting orders and increased expense.² What is more, in an open ICU model, critical care nurses are often left to detect and correct a significant change in a patient’s status without the necessary immediate physician availability, resulting not only in a stressful working environment for nursing staff, but also in potential harm associated with individuals providing care outside their scope of practice.³

In only a small percentage of ICUs—mostly medical ICUs and ICUs in teaching hospitals—is critical care provided in a “high-intensity” or “closed” staffing pattern, in which treatment decisions are cohesively managed under the guidance of an intensivist.²

EVIDENCE IN THE MEDICAL LITERATURE

Staffing patterns in the ICU

Several studies have attempted to identify the consequences of these different ICU staffing patterns on patient care.

Hanson et al⁴ examined two concurrent patient cohorts admitted to a surgical ICU. The study cohort was cared for by an on-site critical care team supervised by an intensivist, while the control cohort received care from a team with patient care responsibilities in multiple sites, supervised by a general surgeon. The results showed that patients cared for by the critical care team spent less time in the ICU, used fewer resources, had fewer complications, and had lower total hospital charges. The difference between the two cohorts was most evident in patients with the worst Acute Physiology and Chronic Health Evaluation (APACHE) II scores.

According to Hanson et al, the lack of an accepted prototype for the delivery of critical care is due to factors such as the relative youth of the discipline, contention over control of individual patient management, and the absence of a single academic advocate.⁴

Moreover, Pronovost et al⁵ concluded that high-intensity staffing (mandatory intensivist consultation or closed ICU) was associated with lower ICU mortality rates in 93% of studies and with a reduced ICU length of stay in the high-intensity staffing units when compared with ICUs with low-intensity staffing (no intensivist or elective intensivist consultation).

Critics of our PCCP paradigm may point to a study by Levy et al⁶ that, using a database of more than 100,000 patients, could not demonstrate any survival benefit with management by critical care physicians. Indeed the study found that patients managed by intensivists had a higher mortality rate than patients managed by physicians not trained in critical care. However, they also showed that more patients managed for the entire stay by intensivists received interventions such as intravenous drugs, mechanical ventilation, and continuous sedation and that they had a higher mean severity of illness as measured by the expanded Simplified Acute Physiology Score (SAPS II) and higher hospital mortality rates than patients who were not managed by a critical care team.

According to Levy et al, most ICUs in the United States are structured as completely open units in which the admitting physicians retain full clinical and decisional responsibility and thus have the option to care for their patients with or without input from intensivists.⁶

However, a recent study by Kim et al⁷ likely rebuts the findings of Levy et al. Kim et al analyzed more than 100,000 ICU admissions and found that the lowest odds of death within 30 days were in ICUs that had high-intensity physician staffing and multidisciplinary care teams, suggesting that the presence of an intensivist confers a survival benefit.

Other studies have also shown that high-intensity staffing improves patient outcomes in the ICU.^5,8,9

Issues of cost and use of resources

Issues concerning cost and human resources for staffing ICUs have acquired increasing importance. According to Angus et al,¹⁰ intensivists provided care to only 36.8% of all ICU patients. The demand for critical care services will continue to grow rapidly as the population ages. It is this shift in the care of the critically ill that requires intensivists to take on the role of the PCCP, so as to provide high-quality, evidence-based critical care and to promote a long-term sustainable model of physician and nursing care.

OUR EXPERIENCE

Our intensivist group has been providing a near-primary-care style of critical care practice for almost 40 years, from its inception in 1977 by one of the authors (A.B.), to our current group of 15 board-certified intensivists. We can easily cite the clinical value of our practice approach, with outcome data showing consistent and better-than-expected Standardized Mortality Ratio accounts from our APACHE IV data (personal communication, Cleveland Clinic Cerner/APACHE IV report), or with reports showing that the presence of a full-time, attending-level, in-house staff physician ensures that patients, surgeons, and consultants have confidence and respect for the care provided. However, we feel that the intangible components are what make our practice a prototype for the PCCP model.

A dedicated team with a low turnover rate

First, we have a team of anesthesiology- and surgery-based intensivists dedicated to ICU practice, with a very low turnover or burnout rate, in contrast to most ICUs in the United States, where intensivists tend to practice part-time (at other times either providing operating-room-based anesthesia or surgical care or working in a pulmonary- or sleep-lab-based practice). We believe this point should not go unstressed: we have a team of physicians who have dedicated their career to working in the ICU full-time, and some have done so in excess of 20 years, even as long as 30 years! It is our opinion that we are able to provide such a highly desirable working environment by a unique daily staffing model that does not utilize the conventional practice style of one intensivist on-call per week.

We also feel that our model dramatically reduces the risk of burnout by permitting our attending intensivists to break up on-call sequences so that there are days on which work in the ICU is not also associated with on-call responsibilities.

A successful fellowship program

Second, we have an extremely successful fellowship program, which began in 1974 when one of the authors (A.B.) advocated the training of anesthesiology residents as intensivists.¹¹ The American Board of Anesthesiology certifies on average 55 candidates per year in critical care medicine, and our program trains about 10% of the physicians applying for certification. In most years, there are actually more candidates for our program than there are available positions, which is atypical for anesthesiology-based critical care training programs. This wealth of young, talented candidates interested in critical care as a career is, again, in contrast to most anesthesiology-based programs, which find it difficult to enroll even one fellow per year.

Critical care programs grounded in anesthesiology typically struggle because of the realities of economics.¹² The payoff of operating-room-based anesthesiology practices generally outshines those in critical care, yet we already have three times as many candidates as there are positions to start our training program in the next 2 years. We feel that candidates are attracted to our program simply because our environment (dedicated staffing, equal clinical footing with surgeons, low burnout rates) is seen as an exciting, positively charged role-modeling atmosphere for young physicians who may have a career interest that involves more than just their original base specialty.

A collegial working relationship

Third, we have a thriving, collegial working relationship—including daily bedside and weekly bioethics rounds with our nursing staff—which has fueled a high degree of professional satisfaction among nurses. This is evidenced by the extremely low turnover rate of nurses (less than 5% per year in the last 5 years) and by national recognition for nursing excellence (Beacon Award for Critical Care Excellence, American Association of Critical Care Nurses) (personal communication, S. Wilson, Nurse Manager). In 2009, the four nurses out of 174 who left did so to further their careers.

While low turnover rates among nurses and award-winning practices are surely a testament to a highly motivated and skilled nursing team, there is no question that a constructive collegiality among the physicians and nurses has provided an environment to allow these positive aspects to flourish.

OVERCOMING ROADBLOCKS

Obviously, although in theory it is easy to proclaim a PCCP paradigm, in reality the roadblocks are many.

For example, standardization of education and credentialing would be an essential hurdle to overcome. The current educational arrangement of the various adult specialties (anesthesiology, internal medicine, surgery), each offering disparate subspecialty critical care training and certification, is deeply rooted in interdisciplinary politics, but without any demonstration of improved patient care.¹³ As described recently by Kaplan and Shaw,¹⁴ an all-encompassing training and credentialing standard for critical care is essential for 21st century medicine and would go a long way toward development of the PCCP paradigm.

Another major roadblock is the shortage of intensivists in the United States.¹³ There are many reasons why physicians opt not to select critical care as a career, such as a non-straight-forward training pathway (as described above), recognition that the 24-hours per day, 7-days-per-week nature of critical care affects lifestyle issues, and inconsistent physician compensation.¹³

However, technological and personnel advances, including the use of electronic (e-ICU)¹⁵ and mid-level practitioner models, have led to creative approaches to extend critical care coverage.¹³

Additionally, the multitude of physician specialty stakeholders and the overall flux of the future of medical care in the United States all would contribute to the difficulties of prioritizing the implementation of the PCCP concept. Also, our practice style—a large intensivist group working in an ostensibly closed surgical ICU in a tertiary-care hospital—is one possible model, as is the even more highly evolved Cleveland Clinic medical ICU, where medical intensivists are already essentially PCCPs. But these models of care may not be generalizable among the local care patterns and medical politics across hospitals or ICUs.

Based on the described successes of our practice model, coupled with evidence in the literature, we have proposed a paradigm shift toward the concept of a PCCP. To be sure, paradigm shifts nearly always require time, effort, and wherewithal. In the end, however, we feel that embracement of the PCCP paradigm would result in a concise, discrete understanding of the role of intensivist, eliminate the specialty’s identity crisis, and ultimately improve patient care.

We have seen significant growth in clinical research in critical care medicine in the last decade. Advances have been made in many important areas in this field; of these, advances in treating septic shock and acute respiratory distress syndrome (ARDS), and also in supportive therapies for critically ill patients (eg, sedatives, insulin), have perhaps received the most attention.

Of note, several once-established therapies in these areas have failed the test of time, as the result of evidence from more-recent clinical trials. For example, recent studies have shown that a pulmonary arterial catheter does not improve outcomes in patients with ARDS. Similarly, what used to be “optimal” fluid management in patients with ARDS is no longer considered appropriate.

In this review, we summarize eight major studies in critical care medicine published in the last 5 years, studies that have contributed to changes in our practice in the intensive care unit (ICU).

FLUID MANAGEMENT IN ARDS

Key points

• In patients with acute lung injury (ALI) and ARDS, fluid restriction is associated with better outcomes than a liberal fluid policy.
• A pulmonary arterial catheter is not necessary and, compared with a central venous catheter, may result in more complications in patients with ALI and ARDS.

Background

Fluid management practices in patients with ARDS have been extremely variable. Two different approaches are commonly used: the liberal or “wet” approach to optimize tissue perfusion and the “dry” approach, which focuses on reducing lung edema. Given that most deaths attributed to ARDS result from extrapulmonary organ failure, aggressive fluid restriction has been the less popular approach.

Additionally, although earlier studies and meta-analyses suggested that the use of a pulmonary arterial catheter was not associated with better outcomes in critically ill patients,¹ controversy remained regarding the value of a pulmonary arterial catheter compared with a central venous catheter in guiding fluid management in patients with ARDS, and data were insufficient to prove one strategy better than the other.

The Fluids and Catheter Treatment Trial (FACTT)

NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK; WIEDEMANN HP, WHEELER AP, BERNARD GR, ET AL. COMPARISON OF TWO FLUID-MANAGEMENT STRATEGIES IN ACUTE LUNG INJURY. N ENGL J MED 2006; 354:2564–2575.

NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK; WHEELER AP, BERNARD GR, THOMPSON BT, ET AL. PULMONARY-ARTERY VERSUS CENTRAL VENOUS CATHETER TO GUIDE TREATMENT OF ACUTE LUNG INJURY. N ENGL J MED 2006; 354:2213–2224.

The Fluids and Catheter Treatment Trial (FACTT) compared two fluid strategies² and also the utility of a pulmonary arterial catheter vs a central venous catheter³ in patients with ALI or ARDS.

This two-by-two factorial trial randomized 1,000 patients to be treated according to either a conservative (fluid-restrictive or “dry”) or a liberal (“wet”) fluid management strategy for 7 days. Additionally, they were randomly assigned to receive either a central venous catheter or a pulmonary arterial catheter. The trial thus had four treatment groups:

• Fluid-restricted and a central venous catheter, with a goal of keeping the central venous pressure below 4 mm Hg
• Fluid-restricted and a pulmonary arterial catheter: fluids were restricted and diuretics were given to keep the pulmonary artery occlusion pressure below 8 mm Hg
• Fluid-liberal and a central venous catheter: fluids were given to keep the central venous pressure between 10 and 14 mm Hg
• Fluid-liberal and a pulmonary arterial catheter: fluids were given to keep the pulmonary artery occlusion pressure between 14 and 18 mm Hg.

The primary end point was the mortality rate at 60 days. Secondary end points included the number of ventilator-free days and organ-failure-free days and parameters of lung physiology. All patients were managed with a low-tidal-volume strategy.

The ‘dry’ strategy was better

The cumulative fluid balance was −136 mL ± 491 mL in the “dry” group and 6,992 mL ± 502 mL in the “wet” group, a difference of more than 7 L (P < .0001). Of note, before randomization, the patients were already fluid-positive, with a mean total fluid balance of +2,700 mL).²

At 60 days, no statistically significant difference in mortality rate was seen between the fluid-management groups (25.5% in the dry group vs 28.4% in the wet group (P = .30). Nevertheless, patients in the dry group had better oxygenation indices and lung injury scores (including lower plateau airway pressure), resulting in more ventilator-free days (14.6 ± 0.5 vs 12.1 ± 0.5; P = .0002) and ICU-free days (13.4 ± 0.4 vs 11.2 ± 0.4; P = .0003).²

Although those in the dry-strategy group had a slightly lower cardiac index and mean arterial pressure, they did not have a higher incidence of shock.

More importantly, the dry group did not have a higher rate of nonpulmonary organ failure. Serum creatinine and blood urea nitrogen concentrations were slightly higher in this group, but this was not associated with a higher incidence of renal failure or the use of dialysis: 10% in the dry-strategy group vs 14% in the wet-strategy group; P = .0642).²

No advantage with a pulmonary arterial catheter

The mortality rate did not differ between the catheter groups. However, the patients who received a pulmonary arterial catheter stayed in the ICU 0.2 days longer and had twice as many nonfatal cardiac arrhythmias as those who received a central venous catheter.³

Comments

The liberal fluid-strategy group had fluid balances similar to those seen in previous National Institutes of Health ARDS Network trials in which fluid management was not controlled. This suggests that the liberal fluid strategy reflects usual clinical practice.

Although the goals used in this study (central venous pressure < 4 mm Hg or pulmonary artery occlusion pressure < 8 mm Hg) could be difficult to achieve in clinical practice, a conservative strategy of fluid management is preferred in patients with ALI or ARDS, given the benefits observed in this trial.

A pulmonary arterial catheter is not indicated to guide hemodynamic management of patients with ARDS.

CORTICOSTEROID USE IN ARDS

Key points

• In selected patients with ARDS, the prolonged use of corticosteroids may result in better oxygenation and a shorter duration of mechanical ventilation.
• Late use of corticosteroids in patients with ARDS (> 14 days after diagnosis) is not indicated and may increase the risk of death.
• The role of corticosteroids in early ARDS (< 7 days after diagnosis) remains controversial.

Background

Systemic corticosteroid therapy was commonly used in ARDS patients in the 1970s and 1980s. However, a single-center study published in the late 1980s showed that a corticosteroid in high doses (methylprednisolone 30 mg/kg) resulted in more complications and was not associated with a lower mortality rate.⁴ On the other hand, a small study that included only patients with persistent ARDS (defined as ARDS lasting for more than 7 days) subsequently showed that oxygenation was significantly better and that fewer patients died while in the hospital with the use of methylprednisolone 2 mg/kg for 32 days.⁵

In view of these divergent findings, the ARDS Network decided to perform a study to help understand the role of corticosteroids in ARDS.

The Late Steroid Rescue Study (LaSRS)

STEINBERG KP, HUDSON LD, GOODMAN RB, ET AL; NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK. EFFICACY AND SAFETY OF CORTICOSTEROIDS FOR PERSISTENT ACUTE RESPIRATORY DISTRESS SYNDROME. N ENGL J MED 2006; 354:1671–1684.

The Late Steroid Rescue Study (LaSRS),⁶ a double-blind, multicenter trial, randomly assigned 180 patients with persistent ARDS (defined as ongoing disease 7–28 days after its onset) to receive methylprednisolone or placebo for 21 days.

Methylprednisolone was given in an initial dose of 2 mg/kg of predicted body weight followed by a dose of 0.5 mg/kg every 6 hours for 14 days and then a dose of 0.5 mg/kg every 12 hours for 7 days, and then it was tapered over 2 to 4 days and discontinued. It could be discontinued if 21 days of treatment were completed or if the patient was able to breathe without assistance.

The primary end point was the mortality rate at 60 days. Secondary end points included the number of ventilator-free days, organ-failure-free days, and complications and the levels of biomarkers of inflammation.

No reduction in mortality rates with steroids

The mortality rates did not differ significantly in the corticosteroid group vs the placebo group at 60 days:

• 29.2% with methylprednisolone (95% confidence interval [CI] 20.8–39.4)
• 28.6% with placebo (95% CI 20.3–38.6, P = 1.0).

Mortality rates at 180 days were also similar between the groups:

• 31.5% with methylprednisolone (95% CI 22.8–41.7)
• 31.9% with placebo (95% CI 23.2–42.0, P = 1.0).

In patients randomized between 7 and 13 days after the onset of ARDS, the mortality rates were lower in the methylprednisolone group than in the placebo group but the differences were not statistically significant. The mortality rate in this subgroup was 27% vs 36% (P = .26) at 60 days and was 27% vs 39% (P = .14) at 180 days.

However, in patients randomized more than 14 days after the onset of ARDS, the mortality rate was significantly higher in the methylprednisolone group than in the placebo group at 60 days (35% vs 8%, P = .02) and at 180 days (44% vs 12%, P = .01).

Some benefit in secondary outcomes

At day 28, methylprednisolone was associated with:

• More ventilator-free days (11.2 ± 9.4 vs 6.8 ± 8.5, P < .001)
• More shock-free days (20.7 ± 8.9 vs 17.9 ± 10.2, P = .04)
• More ICU-free days (8.9 ± 8.2 vs 6.7 ± 7.8, P = .02).

Similarly, pulmonary physiologic indices were better with methylprednisolone, specifically:

• The ratio of Pao₂ to the fraction of inspired oxygen at days 3, 4, and 14 (P < .05)
• Plateau pressure at days 4, 5, and 7 (P < .05)
• Static compliance at days 7 and 14 (P < .05).

In terms of side effects, methylprednisolone was associated with more events associated with myopathy or neuropathy (9 vs 0, P = .001), but there were no differences in the number of serious infections or in glycemic control.

Comments

Although other recent studies suggested that corticosteroid use may be associated with a reduction in mortality rates,^7–9 LaSRS did not confirm this effect. Although the doses and length of therapy were similar in these studies, LaSRS was much larger and included patients from the ARDS Network.

Nevertheless, LaSRS was criticized because of strict exclusion criteria and poor enrollment (only 5% of eligible patients were included). Additionally, it was conducted over a period of time when some ICU practices varied significantly (eg, low vs high tidal volume ventilation, tight vs loose glucose control).

INTERRUPTING SEDATION DURING MECHANICAL VENTILATION

Key points

• Daily awakening of mechanically ventilated patients is safe.
• Daily interruption of sedation in mechanically ventilated patients is associated with a shorter length of mechanical ventilation.

Background

Sedatives are a central component of critical care. Continuous infusions of narcotics, benzodiazepines, and anesthetic agents are frequently used to promote comfort in patients receiving mechanical ventilation.

Despite its widespread use in the ICU, there is little evidence that such sedation improves outcomes. Observational and randomized trials^10–12 have shown that patients who receive continuous infusions of sedatives need to be on mechanical ventilation longer than those who receive intermittent dosing. Additionally, an earlier randomized controlled trial¹³ showed that daily interruption of sedative drug infusions decreased the duration of mechanical ventilation by almost 50% and resulted in a reduction in the length of stay in the ICU.

Despite these findings, many ICU physicians remain skeptical of the value of daily interruption of sedative medications and question the safety of this practice.

The Awakening and Breathing Controlled (ABC) trial

GIRARD TD, KRESS JP, FUCHS BD, ET AL. EFFICACY AND SAFETY OF A PAIRED SEDATION AND VENTILATOR WEANING PROTOCOL FOR MECHANICALLY VENTILATED PATIENTS IN INTENSIVE CARE (AWAKENING AND BREATHING CONTROLLED TRIAL): A RANDOMISED CONTROLLED TRIAL. LANCET 2008; 371:126–134.

The Awakening and Breathing Controlled (ABC) trial¹⁴ was a multicenter, randomized controlled trial that included 336 patients who required at least 12 consecutive hours of mechanical ventilation. All patients had to be receiving patient-targeted sedation.

Those in the intervention group (n = 168) had their sedation interrupted every day, followed by a clinical assessment to determine whether they could be allowed to try breathing spontaneously. The control group (n = 168) also received a clinical assessment for a trial of spontaneous breathing, while their sedation was continued as usual.

In patients in the intervention group who failed the screening for a spontaneous breathing trial, the sedatives were resumed at half the previous dose. Criteria for failure on the spontaneous breathing trial included any of the following: anxiety, agitation, respiratory rate more than 35 breaths per minute for 5 minutes or longer, cardiac arrhythmia, oxygen saturation less than 88% for 5 minutes or longer, or two or more signs of respiratory distress, tachycardia, bradycardia, paradoxical breathing, accessory muscle use, diaphoresis, or marked dyspnea.

The combination of sedation interruption and a spontaneous breathing trial was superior to a spontaneous breathing trial alone. The mean number of ventilator-free days:

• 14.7 ± 0.9 with sedation interruption
• 11.6 ± 0.9 days with usual care (P = .02).

The median time to ICU discharge:

• 9.1 days with sedation interruption (interquartile range 5.1 to 17.8)
• 12.9 days with usual care (interquartile range 6.0 to 24.2, P = .01).

The mortality rate at 28 days:

• 28% with sedation interruption
• 35% with usual care (P = .21).

The mortality rate at 1 year:

• 44% with sedation interruption
• 58% with usual care (hazard ratio [HR] in the intervention group 0.68, 95% CI 0.50–0.92, P = .01).

Of note, patients in the intervention group had a higher rate of self-extubation (9.6% vs 3.6%, P = .03), but the rate of reintubation was similar between the groups (14% vs 13%, P = .47).

Comments

The addition of daily awakenings to spontaneous breathing trials results in a further reduction in the number of ICU days and increases the number of ventilator-free days.

Of note, the protocol allowed patients in the control group to undergo a spontaneous breathing trial while on sedatives (69% of the patients were receiving sedation at the time). Therefore, a bias effect in favor of the intervention group cannot be excluded. However, both groups had to meet criteria for readiness for spontaneous breathing.

The study demonstrates the safety of daily awakenings and confirms previous findings suggesting that a daily trial of spontaneous breathing results in better ICU outcomes.

GLUCOSE CONTROL IN THE ICU

Key points

• Although earlier studies suggested that intensive insulin therapy might be beneficial in critically ill patients, new findings show that strict glucose control can lead to complications without improving outcomes.

Background

A previous study¹⁵ found that intensive insulin therapy to maintain a blood glucose level between 80 and 110 mg/dL (compared with 180–200 mg/dL) reduced the mortality rate in surgical critical care patients. The mortality rate in the ICU was 4.6% with intensive insulin therapy vs 8.0% with conventional therapy (P < .04), and the effect was more robust for patients who remained longer than 5 days in the ICU (10.6% vs 20.2%).

Importantly, however, hypoglycemia (defined as blood glucose ≤ 40 mg/dL) occurred in 39 patients in the intensive-treatment group vs 6 patients in the conventional-treatment group.

The NICE-SUGAR trial

NICE-SUGAR STUDY INVESTIGATORS; FINFER S, CHITTOCK DR, SU SY, ET AL. INTENSIVE VERSUS CONVENTIONAL GLUCOSE CONTROL IN CRITICALLY ILL PATIENTS. N ENGL J MED 2009; 360:1283–1297.

The Normoglycemia in Intensive Care Evaluation-Survival Using Glucose Algorithm Regulation (NICE-SUGAR) trial¹⁶ randomized 6,104 patients in medical and surgical ICUs to receive either intensive glucose control (blood glucose 81–108 mg/dL) with insulin therapy or conventional glucose control (blood glucose < 180 mg/dL). In the conventional-control group, insulin was discontinued if the blood glucose level dropped below 144 mg/dL.

A higher mortality rate with intensive glucose control

As expected, the intensive-control group achieved lower blood glucose levels: 115 vs 144 mg/dL.

Nevertheless, intensive glucose control was associated with a higher incidence of severe hypoglycemia, defined as a blood glucose level lower than 40 mg/dL: 6.8% vs 0.5%.

More importantly, compared with conventional insulin therapy, intensive glucose control was associated with a higher 90-day mortality rate: 27.5% vs 24.9% (odds ratio 1.14, 95% CI 1.02–1.28). These findings were similar in the subgroup of surgical patients (24.4% vs 19.8%, odds ratio 1.31, 95% CI 1.07–1.61).

Comments

Of note, the conventional-control group had more patients who discontinued the treatment protocol prematurely. Additionally, more patients in this group received corticosteroids.

These results widely differ from those of a previous study by van den Berghe et al,¹⁵ which showed that tight glycemic control is associated with a survival benefit. The differences in outcomes are probably largely related to different patient populations, as van den Berghe et al included patients who had undergone cardiac surgery, who were more likely to benefit from strict blood glucose control.

The VISEP trial

BRUNKHORST FM, ENGEL C, BLOOS F, ET AL; GERMAN COMPETENCE NETWORK SEPSIS (SEPNET). INTENSIVE INSULIN THERAPY AND PENTASTARCH RESUSCITATION IN SEVERE SEPSIS. N ENGL J MED 2008; 358:125–139.

The Volume Substitution and Insulin Therapy in Severe Sepsis (VISEP) trial was a multicenter study designed to compare intensive insulin therapy (target blood glucose level 80–110 mg/dL) and conventional glucose control (target blood glucose level 180–200 mg/dL) in patients with severe sepsis.¹⁷ It also compared two fluids for volume resuscitation: 10% pentastarch vs modified Ringer's lactate. It included both medical and surgical patients.

Trial halted early for safety reasons

The mean morning blood glucose level was significantly lower in the intensive insulin group (112 vs 151 mg/dL).

Severe hypoglycemia (blood glucose ≤ 40 mg/dL) was more common in the group that received intensive insulin therapy (17% vs 4.1%, P < .001).

Mortality rates at 28 days did not differ significantly: 24.7% with intensive control vs 26.0% with conventional glucose control. The mortality rate at 90 days was 39.7% in the intensive therapy group and 35.4% in the conventional therapy group, but the difference was not statistically significant.

The intensive insulin arm of the trial was stopped after 488 patients were enrolled because of a higher rate of hypoglycemia (12.1% vs 2.1%) and of serious adverse events (10.9% vs 5.2%).

Additionally, the fluid resuscitation arm of the study was suspended at the first planned interim analysis because of a higher risk of organ failure in the 10% pentastarch group.

CORTICOSTEROID THERAPY IN SEPTIC SHOCK

Key points

• Corticosteroid therapy improves hemodynamic outcomes in patients with severe septic shock.
• Although meta-analyses suggest the mortality rate is lower with corticosteroid therapy, there is not enough evidence from randomized controlled trials to prove that the use of low-dose corticosteroids lowers the mortality rate in patients with septic shock.
• The corticotropin (ACTH) stimulation test should not be used to determine the need for corticosteroids in patients with septic shock.

Background

A previous multicenter study,¹⁸ performed in France, found that the use of corticosteroids in patients with septic shock resulted in lower rates of death at 28 days, in the ICU, and in the hospital and a shorter time to vasopressor withdrawal. Nevertheless, the beneficial effects were not observed in patients with adequate adrenal reserve (based on an ACTH stimulation test).

This study was criticized because of a high mortality rate in the placebo group.

The CORTICUS study

SPRUNG CL, ANNANE D, KEH D, ET AL; CORTICUS STUDY GROUP. HYDROCORTISONE THERAPY FOR PATIENTS WITH SEPTIC SHOCK. N ENGL J MED 2008; 358:111–124.

The Corticosteroid Therapy of Septic Shock (CORTICUS) study was a multicenter trial that randomly assigned 499 patients with septic shock to receive hydrocortisone (50 mg intravenously every 6 hours for 5 days, followed by a 6-day taper period) or placebo.¹⁹

Patients were eligible to be enrolled within 72 hours of onset of shock. Similar to previous studies, the CORTICUS trial classified patients on the basis of an ACTH stimulation test as having inadequate adrenal reserve (a cortisol increase of ≤ 9 μg/dL) or adequate adrenal reserve (a cortisol increase of > 9 μg/dL).

Faster reversal of shock with steroids

At baseline, the mean Simplified Acute Physiologic Score II (SAPS II) was 49 (the range of possible scores is 0 to 163; the higher the score the worse the organ dysfunction).

Hydrocortisone use resulted in a shorter duration of vasopressor use and a faster reversal of shock (3.3 days vs 5.8 days, P < .001).

This association was the same when patients were divided according to response to ACTH stimulation test. Time to reversal of shock in responders:

• 2.8 days with hydrocortisone
• 5.8 days with placebo (P < .001).

Time to reversal of shock in nonresponders:

• 3.9 days with hydrocortisone
• 6.0 days with placebo (P = .06).

Nevertheless, the treatment did not reduce the mortality rate at 28 days overall (34.3% vs 31.5% P = .51), or in the subgroups based on response to ACTH, or at any other time point. A post hoc analysis suggested that patients who had a systolic blood pressure of less than 90 mm Hg within 30 minutes of enrollment had a greater benefit in terms of mortality rate, but the effect was not statistically significant: the absolute difference was −11.2% (P = 0.28). Similarly, post hoc analyses also revealed a higher rate of death at 28 days in patients who received etomidate (Amidate) before randomization in both groups (P = .03).

Importantly, patients who received corticosteroids had a higher incidence of superinfections, including new episodes of sepsis or septic shock, with a combined odds ratio of 1.37 (95% CI 1.05–1.79).

Length of stay in the hospital or in the ICU was similar in patients who received corticosteroids and in those who received placebo. The ICU length of stay was 19 ± 31 days with hydrocortisone vs 18 ± 17 days with placebo (P = .51).

Comments

The CORTICUS trial showed that low-dose corticosteroid therapy results in faster reversal of shock in patients with severe septic shock. The hemodynamic benefits are present in all patients regardless of response to the ACTH stimulation test.

Nevertheless, contrary to previous findings,¹⁸ corticosteroid use was not associated with an improvement in mortality rates. Important differences exist between these two studies:

• The mortality rates in the placebo groups were significantly different (> 50% in the French study vs 30% in CORTICUS).
• The SAPS II scores were different in these two trials (55 vs 49), suggesting a greater severity of illness in the French study.
• The criteria for enrollment were different: the French study included patients who had a systolic blood pressure lower than 90 mm Hg for more than 1 hour despite fluid administration and vasopressor use, whereas the CORTICUS trial included patients who had a systolic blood pressure lower than 90 mm Hg for more than 1 hour despite fluid administration or vasopressor use.
• The time of enrollment was different: patients were enrolled much faster in the French study (within 8 hours) than in the CORTICUS trial (within 72 hours).

A recent meta-analysis of 17 randomized trials (including the CORTICUS study), found that, compared with those who received placebo, patients who received corticosteroids had a small reduction in the 28-day mortality rate (HR 0.84, 95% CI 0.71–1.00, P < .05).²⁰ Of note, this meta-analysis has been criticized for possible publication bias and also for a large degree of heterogeneity in its results.²¹

VASOPRESSOR THERAPY IN SHOCK

Key points

• Vasopressin use in patients with severe septic shock is not associated with an improvement in mortality rates.
• Vasopressin should not be used as a first-line agent in patients with septic shock.
• Norepinephrine should be considered a first-line agent in patients with shock.
• Compared with norepinephrine, the use of dopamine in patients with shock is associated with similar mortality rates, although its use may result in a greater number of cardiac adverse events.

Background

Vasopressin gained popularity in critical care in the last 10 years because several small studies showed that adding it improves hemodynamics and results in a reduction in the doses of catecholamines in patients with refractory septic shock.²² Furthermore, the Surviving Sepsis Campaign guidelines recommended the use of vasopressin in patients who have refractory shock despite fluid resuscitation and the use of other “conventional” vasopressors.²³

Despite these positive findings, it remained unknown if the use of vasopressin increases the survival rate in patients with septic shock.

The Vasopressin and Septic Shock Trial (VASST)

RUSSELL JA, WALLEY KR, SINGER J, ET AL; VASST INVESTIGATORS. VASOPRESSIN VERSUS NOREPINEPHRINE INFUSION IN PATIENTS WITH SEPTIC SHOCK. N ENGL J MED 2008; 358:877–887.

The Vasopressin and Septic Shock Trial (VASST)²⁴ was a multicenter randomized, double-blind, controlled trial that included 778 patients with refractory septic shock. Refractory shock was defined as the lack of a response to a normal saline fluid bolus of 500 mL or the need for vasopressors (norepinephrine in doses of at least 5 μg/minute or its equivalent for 6 hours or more in the 24 hours before randomization).

Two subgroups were identified: those with severe septic shock (requiring norepinephrine in doses of 15 μg/minute or higher) and those with less-severe septic shock (needing norepinephrine in doses of 5 to 14 μg/minute). Patients with unstable coronary artery disease (acute myocardial infarction, angina) and severe congestive heart failure were excluded.

Patients were randomized to receive an intravenous infusion of vasopressin (0.01–0.03 U/minute) or norepinephrine (5–15 mg/minute) in addition to open-labeled vasopressors (excluding vasopressin). The primary outcome was the all-cause mortality rate at 28 days.

Results

At 28 days, fewer patients had died in the vasopressin group than in the norepinephrine group (35.4% vs 39.3%), but the difference was not statistically significant (P = .26). The trend was the same at 90 days (mortality rate 43.9% vs 49.6%, P = .11).

Subgroup analysis showed that in patients with less-severe septic shock, those who received vasopressin had a lower mortality rate at 28 days (26.5% vs 35.7%, P = .05; relative risk 0.74; 95% CI 0.55–1.01) and at 90 days (35.8% vs 46.1%, P = .04; relative risk 0.78, 95% CI 0.61–0.99).

There were no statistically significant differences in any of the other secondary outcomes or in serious adverse events.

Comments

The study has been criticized for several reasons:

• The mean arterial blood pressure at baseline before initiation of vasopressin was 72 mm Hg (and some argue that vasopressin was therefore not needed by the time it was started).
• The time from screening to infusion of the study drug was very long (12 hours).
• The observed mortality rate was lower than expected (37%).

Despite these considerations, the VASST trial showed that vasopressin is not associated with an increased number of adverse events in patients without active cardiovascular disease. The possible benefit in terms of the mortality rate in the subgroup of patients with less-severe septic shock requires further investigation.

Is dopamine equivalent to norepinephrine?

Previously, the Sepsis Occurrence in Acutely Ill Patients (SOAP) study, a multicenter, observational cohort study, found that dopamine use was associated with a higher all-cause mortality rate in the ICU compared with no dopamine.²⁵ This finding had not been reproduced, as few well-designed studies had compared the effects of dopamine and norepinephrine.

The SOAP II study

DE BACKER D, BISTON P, DEVRIENDT J, ET AL; SOAP II INVESTIGATORS.. COMPARISON OF DOPAMINE AND NOREPINEPHRINE IN THE TREATMENT OF SHOCK. N ENGL J MED 2010; 362:779–789.

The SOAP II study,²⁶ a multicenter, randomized trial, compared dopamine vs norepinephrine as first-line vasopressor therapy. In patients with refractory shock despite use of dopamine 20 μg/kg/minute or norepinephrine 0.19 μg/kg/minute, open-label norepinephrine, epinephrine, or vasopressin was added.

The primary outcome was the mortality rate at 28 days after randomization; secondary end points included the number of days without need for organ support and the occurrence of adverse events.

Results

A total of 1,679 patients were included; 858 were assigned to dopamine and 821 to norepinephrine. Most (1,044, 62%) of the patients had a diagnosis of septic shock.

No significant difference in mortality rates was noted at 28 days: 52.5% with dopamine vs 48.5% with norepinephrine (P = .10).

However, there were more arrhythmias in the patients treated with dopamine: 207 events (24.1%) vs 102 events (12.4%) (P < .001). The number of other adverse events such as renal failure, myocardial infarction, arterial occlusion, or skin necrosis was not different between the groups.

A subgroup analysis showed that dopamine was associated with more deaths at 28 days in patients with cardiogenic shock (P = .03) but not in patients with septic shock (P = .19) or with hypovolemic shock (P = .84).

Comments

The study was criticized because the patients may not have received adequate fluid resuscitation (the study considered adequate resuscitation to be equivalent to 1 L of crystalloids or 500 mL of colloids), as different degrees of volume depletion among patients make direct comparisons of vasopressor effects difficult.

Additionally, the study defined dopamine 20 μg/kg/minute as being equipotent with norepinephrine 0.19 μg/kg/minute. Comparisons of potency between drugs are difficult to establish, as there are no available data.

Nevertheless, this study further confirms previous findings suggesting that norepinephrine is not associated with more end-organ damage (such as renal failure or skin ischemia), and shows that dopamine may increase the number of adverse events, particularly in patients with cardiac disease.

We have seen significant growth in clinical research in critical care medicine in the last decade. Advances have been made in many important areas in this field; of these, advances in treating septic shock and acute respiratory distress syndrome (ARDS), and also in supportive therapies for critically ill patients (eg, sedatives, insulin), have perhaps received the most attention.

Of note, several once-established therapies in these areas have failed the test of time, as the result of evidence from more-recent clinical trials. For example, recent studies have shown that a pulmonary arterial catheter does not improve outcomes in patients with ARDS. Similarly, what used to be “optimal” fluid management in patients with ARDS is no longer considered appropriate.

In this review, we summarize eight major studies in critical care medicine published in the last 5 years, studies that have contributed to changes in our practice in the intensive care unit (ICU).

FLUID MANAGEMENT IN ARDS

Key points

• In patients with acute lung injury (ALI) and ARDS, fluid restriction is associated with better outcomes than a liberal fluid policy.
• A pulmonary arterial catheter is not necessary and, compared with a central venous catheter, may result in more complications in patients with ALI and ARDS.

Background

Fluid management practices in patients with ARDS have been extremely variable. Two different approaches are commonly used: the liberal or “wet” approach to optimize tissue perfusion and the “dry” approach, which focuses on reducing lung edema. Given that most deaths attributed to ARDS result from extrapulmonary organ failure, aggressive fluid restriction has been the less popular approach.

Additionally, although earlier studies and meta-analyses suggested that the use of a pulmonary arterial catheter was not associated with better outcomes in critically ill patients,¹ controversy remained regarding the value of a pulmonary arterial catheter compared with a central venous catheter in guiding fluid management in patients with ARDS, and data were insufficient to prove one strategy better than the other.

The Fluids and Catheter Treatment Trial (FACTT)

NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK; WIEDEMANN HP, WHEELER AP, BERNARD GR, ET AL. COMPARISON OF TWO FLUID-MANAGEMENT STRATEGIES IN ACUTE LUNG INJURY. N ENGL J MED 2006; 354:2564–2575.

NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK; WHEELER AP, BERNARD GR, THOMPSON BT, ET AL. PULMONARY-ARTERY VERSUS CENTRAL VENOUS CATHETER TO GUIDE TREATMENT OF ACUTE LUNG INJURY. N ENGL J MED 2006; 354:2213–2224.

The Fluids and Catheter Treatment Trial (FACTT) compared two fluid strategies² and also the utility of a pulmonary arterial catheter vs a central venous catheter³ in patients with ALI or ARDS.

This two-by-two factorial trial randomized 1,000 patients to be treated according to either a conservative (fluid-restrictive or “dry”) or a liberal (“wet”) fluid management strategy for 7 days. Additionally, they were randomly assigned to receive either a central venous catheter or a pulmonary arterial catheter. The trial thus had four treatment groups:

• Fluid-restricted and a central venous catheter, with a goal of keeping the central venous pressure below 4 mm Hg
• Fluid-restricted and a pulmonary arterial catheter: fluids were restricted and diuretics were given to keep the pulmonary artery occlusion pressure below 8 mm Hg
• Fluid-liberal and a central venous catheter: fluids were given to keep the central venous pressure between 10 and 14 mm Hg
• Fluid-liberal and a pulmonary arterial catheter: fluids were given to keep the pulmonary artery occlusion pressure between 14 and 18 mm Hg.

The primary end point was the mortality rate at 60 days. Secondary end points included the number of ventilator-free days and organ-failure-free days and parameters of lung physiology. All patients were managed with a low-tidal-volume strategy.

The ‘dry’ strategy was better

The cumulative fluid balance was −136 mL ± 491 mL in the “dry” group and 6,992 mL ± 502 mL in the “wet” group, a difference of more than 7 L (P < .0001). Of note, before randomization, the patients were already fluid-positive, with a mean total fluid balance of +2,700 mL).²

At 60 days, no statistically significant difference in mortality rate was seen between the fluid-management groups (25.5% in the dry group vs 28.4% in the wet group (P = .30). Nevertheless, patients in the dry group had better oxygenation indices and lung injury scores (including lower plateau airway pressure), resulting in more ventilator-free days (14.6 ± 0.5 vs 12.1 ± 0.5; P = .0002) and ICU-free days (13.4 ± 0.4 vs 11.2 ± 0.4; P = .0003).²

Although those in the dry-strategy group had a slightly lower cardiac index and mean arterial pressure, they did not have a higher incidence of shock.

More importantly, the dry group did not have a higher rate of nonpulmonary organ failure. Serum creatinine and blood urea nitrogen concentrations were slightly higher in this group, but this was not associated with a higher incidence of renal failure or the use of dialysis: 10% in the dry-strategy group vs 14% in the wet-strategy group; P = .0642).²

No advantage with a pulmonary arterial catheter

The mortality rate did not differ between the catheter groups. However, the patients who received a pulmonary arterial catheter stayed in the ICU 0.2 days longer and had twice as many nonfatal cardiac arrhythmias as those who received a central venous catheter.³

Comments

The liberal fluid-strategy group had fluid balances similar to those seen in previous National Institutes of Health ARDS Network trials in which fluid management was not controlled. This suggests that the liberal fluid strategy reflects usual clinical practice.

Although the goals used in this study (central venous pressure < 4 mm Hg or pulmonary artery occlusion pressure < 8 mm Hg) could be difficult to achieve in clinical practice, a conservative strategy of fluid management is preferred in patients with ALI or ARDS, given the benefits observed in this trial.

A pulmonary arterial catheter is not indicated to guide hemodynamic management of patients with ARDS.

CORTICOSTEROID USE IN ARDS

Key points

• In selected patients with ARDS, the prolonged use of corticosteroids may result in better oxygenation and a shorter duration of mechanical ventilation.
• Late use of corticosteroids in patients with ARDS (> 14 days after diagnosis) is not indicated and may increase the risk of death.
• The role of corticosteroids in early ARDS (< 7 days after diagnosis) remains controversial.

Background

Systemic corticosteroid therapy was commonly used in ARDS patients in the 1970s and 1980s. However, a single-center study published in the late 1980s showed that a corticosteroid in high doses (methylprednisolone 30 mg/kg) resulted in more complications and was not associated with a lower mortality rate.⁴ On the other hand, a small study that included only patients with persistent ARDS (defined as ARDS lasting for more than 7 days) subsequently showed that oxygenation was significantly better and that fewer patients died while in the hospital with the use of methylprednisolone 2 mg/kg for 32 days.⁵

In view of these divergent findings, the ARDS Network decided to perform a study to help understand the role of corticosteroids in ARDS.

The Late Steroid Rescue Study (LaSRS)

STEINBERG KP, HUDSON LD, GOODMAN RB, ET AL; NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK. EFFICACY AND SAFETY OF CORTICOSTEROIDS FOR PERSISTENT ACUTE RESPIRATORY DISTRESS SYNDROME. N ENGL J MED 2006; 354:1671–1684.

The Late Steroid Rescue Study (LaSRS),⁶ a double-blind, multicenter trial, randomly assigned 180 patients with persistent ARDS (defined as ongoing disease 7–28 days after its onset) to receive methylprednisolone or placebo for 21 days.

Methylprednisolone was given in an initial dose of 2 mg/kg of predicted body weight followed by a dose of 0.5 mg/kg every 6 hours for 14 days and then a dose of 0.5 mg/kg every 12 hours for 7 days, and then it was tapered over 2 to 4 days and discontinued. It could be discontinued if 21 days of treatment were completed or if the patient was able to breathe without assistance.

The primary end point was the mortality rate at 60 days. Secondary end points included the number of ventilator-free days, organ-failure-free days, and complications and the levels of biomarkers of inflammation.

No reduction in mortality rates with steroids

The mortality rates did not differ significantly in the corticosteroid group vs the placebo group at 60 days:

• 29.2% with methylprednisolone (95% confidence interval [CI] 20.8–39.4)
• 28.6% with placebo (95% CI 20.3–38.6, P = 1.0).

Mortality rates at 180 days were also similar between the groups:

• 31.5% with methylprednisolone (95% CI 22.8–41.7)
• 31.9% with placebo (95% CI 23.2–42.0, P = 1.0).

In patients randomized between 7 and 13 days after the onset of ARDS, the mortality rates were lower in the methylprednisolone group than in the placebo group but the differences were not statistically significant. The mortality rate in this subgroup was 27% vs 36% (P = .26) at 60 days and was 27% vs 39% (P = .14) at 180 days.

However, in patients randomized more than 14 days after the onset of ARDS, the mortality rate was significantly higher in the methylprednisolone group than in the placebo group at 60 days (35% vs 8%, P = .02) and at 180 days (44% vs 12%, P = .01).

Some benefit in secondary outcomes

At day 28, methylprednisolone was associated with:

• More ventilator-free days (11.2 ± 9.4 vs 6.8 ± 8.5, P < .001)
• More shock-free days (20.7 ± 8.9 vs 17.9 ± 10.2, P = .04)
• More ICU-free days (8.9 ± 8.2 vs 6.7 ± 7.8, P = .02).

Similarly, pulmonary physiologic indices were better with methylprednisolone, specifically:

• The ratio of Pao₂ to the fraction of inspired oxygen at days 3, 4, and 14 (P < .05)
• Plateau pressure at days 4, 5, and 7 (P < .05)
• Static compliance at days 7 and 14 (P < .05).

In terms of side effects, methylprednisolone was associated with more events associated with myopathy or neuropathy (9 vs 0, P = .001), but there were no differences in the number of serious infections or in glycemic control.

Comments

Although other recent studies suggested that corticosteroid use may be associated with a reduction in mortality rates,^7–9 LaSRS did not confirm this effect. Although the doses and length of therapy were similar in these studies, LaSRS was much larger and included patients from the ARDS Network.

Nevertheless, LaSRS was criticized because of strict exclusion criteria and poor enrollment (only 5% of eligible patients were included). Additionally, it was conducted over a period of time when some ICU practices varied significantly (eg, low vs high tidal volume ventilation, tight vs loose glucose control).

INTERRUPTING SEDATION DURING MECHANICAL VENTILATION

Key points

• Daily awakening of mechanically ventilated patients is safe.
• Daily interruption of sedation in mechanically ventilated patients is associated with a shorter length of mechanical ventilation.

Background

Sedatives are a central component of critical care. Continuous infusions of narcotics, benzodiazepines, and anesthetic agents are frequently used to promote comfort in patients receiving mechanical ventilation.

Despite its widespread use in the ICU, there is little evidence that such sedation improves outcomes. Observational and randomized trials^10–12 have shown that patients who receive continuous infusions of sedatives need to be on mechanical ventilation longer than those who receive intermittent dosing. Additionally, an earlier randomized controlled trial¹³ showed that daily interruption of sedative drug infusions decreased the duration of mechanical ventilation by almost 50% and resulted in a reduction in the length of stay in the ICU.

Despite these findings, many ICU physicians remain skeptical of the value of daily interruption of sedative medications and question the safety of this practice.

The Awakening and Breathing Controlled (ABC) trial

GIRARD TD, KRESS JP, FUCHS BD, ET AL. EFFICACY AND SAFETY OF A PAIRED SEDATION AND VENTILATOR WEANING PROTOCOL FOR MECHANICALLY VENTILATED PATIENTS IN INTENSIVE CARE (AWAKENING AND BREATHING CONTROLLED TRIAL): A RANDOMISED CONTROLLED TRIAL. LANCET 2008; 371:126–134.

The Awakening and Breathing Controlled (ABC) trial¹⁴ was a multicenter, randomized controlled trial that included 336 patients who required at least 12 consecutive hours of mechanical ventilation. All patients had to be receiving patient-targeted sedation.

Those in the intervention group (n = 168) had their sedation interrupted every day, followed by a clinical assessment to determine whether they could be allowed to try breathing spontaneously. The control group (n = 168) also received a clinical assessment for a trial of spontaneous breathing, while their sedation was continued as usual.

In patients in the intervention group who failed the screening for a spontaneous breathing trial, the sedatives were resumed at half the previous dose. Criteria for failure on the spontaneous breathing trial included any of the following: anxiety, agitation, respiratory rate more than 35 breaths per minute for 5 minutes or longer, cardiac arrhythmia, oxygen saturation less than 88% for 5 minutes or longer, or two or more signs of respiratory distress, tachycardia, bradycardia, paradoxical breathing, accessory muscle use, diaphoresis, or marked dyspnea.

The combination of sedation interruption and a spontaneous breathing trial was superior to a spontaneous breathing trial alone. The mean number of ventilator-free days:

• 14.7 ± 0.9 with sedation interruption
• 11.6 ± 0.9 days with usual care (P = .02).

The median time to ICU discharge:

• 9.1 days with sedation interruption (interquartile range 5.1 to 17.8)
• 12.9 days with usual care (interquartile range 6.0 to 24.2, P = .01).

The mortality rate at 28 days:

• 28% with sedation interruption
• 35% with usual care (P = .21).

The mortality rate at 1 year:

• 44% with sedation interruption
• 58% with usual care (hazard ratio [HR] in the intervention group 0.68, 95% CI 0.50–0.92, P = .01).

Of note, patients in the intervention group had a higher rate of self-extubation (9.6% vs 3.6%, P = .03), but the rate of reintubation was similar between the groups (14% vs 13%, P = .47).

Comments

The addition of daily awakenings to spontaneous breathing trials results in a further reduction in the number of ICU days and increases the number of ventilator-free days.

Of note, the protocol allowed patients in the control group to undergo a spontaneous breathing trial while on sedatives (69% of the patients were receiving sedation at the time). Therefore, a bias effect in favor of the intervention group cannot be excluded. However, both groups had to meet criteria for readiness for spontaneous breathing.

The study demonstrates the safety of daily awakenings and confirms previous findings suggesting that a daily trial of spontaneous breathing results in better ICU outcomes.

GLUCOSE CONTROL IN THE ICU

Key points

• Although earlier studies suggested that intensive insulin therapy might be beneficial in critically ill patients, new findings show that strict glucose control can lead to complications without improving outcomes.

Background

A previous study¹⁵ found that intensive insulin therapy to maintain a blood glucose level between 80 and 110 mg/dL (compared with 180–200 mg/dL) reduced the mortality rate in surgical critical care patients. The mortality rate in the ICU was 4.6% with intensive insulin therapy vs 8.0% with conventional therapy (P < .04), and the effect was more robust for patients who remained longer than 5 days in the ICU (10.6% vs 20.2%).

Importantly, however, hypoglycemia (defined as blood glucose ≤ 40 mg/dL) occurred in 39 patients in the intensive-treatment group vs 6 patients in the conventional-treatment group.

The NICE-SUGAR trial

NICE-SUGAR STUDY INVESTIGATORS; FINFER S, CHITTOCK DR, SU SY, ET AL. INTENSIVE VERSUS CONVENTIONAL GLUCOSE CONTROL IN CRITICALLY ILL PATIENTS. N ENGL J MED 2009; 360:1283–1297.

The Normoglycemia in Intensive Care Evaluation-Survival Using Glucose Algorithm Regulation (NICE-SUGAR) trial¹⁶ randomized 6,104 patients in medical and surgical ICUs to receive either intensive glucose control (blood glucose 81–108 mg/dL) with insulin therapy or conventional glucose control (blood glucose < 180 mg/dL). In the conventional-control group, insulin was discontinued if the blood glucose level dropped below 144 mg/dL.

A higher mortality rate with intensive glucose control

As expected, the intensive-control group achieved lower blood glucose levels: 115 vs 144 mg/dL.

Nevertheless, intensive glucose control was associated with a higher incidence of severe hypoglycemia, defined as a blood glucose level lower than 40 mg/dL: 6.8% vs 0.5%.

More importantly, compared with conventional insulin therapy, intensive glucose control was associated with a higher 90-day mortality rate: 27.5% vs 24.9% (odds ratio 1.14, 95% CI 1.02–1.28). These findings were similar in the subgroup of surgical patients (24.4% vs 19.8%, odds ratio 1.31, 95% CI 1.07–1.61).

Comments

Of note, the conventional-control group had more patients who discontinued the treatment protocol prematurely. Additionally, more patients in this group received corticosteroids.

These results widely differ from those of a previous study by van den Berghe et al,¹⁵ which showed that tight glycemic control is associated with a survival benefit. The differences in outcomes are probably largely related to different patient populations, as van den Berghe et al included patients who had undergone cardiac surgery, who were more likely to benefit from strict blood glucose control.

The VISEP trial

BRUNKHORST FM, ENGEL C, BLOOS F, ET AL; GERMAN COMPETENCE NETWORK SEPSIS (SEPNET). INTENSIVE INSULIN THERAPY AND PENTASTARCH RESUSCITATION IN SEVERE SEPSIS. N ENGL J MED 2008; 358:125–139.

The Volume Substitution and Insulin Therapy in Severe Sepsis (VISEP) trial was a multicenter study designed to compare intensive insulin therapy (target blood glucose level 80–110 mg/dL) and conventional glucose control (target blood glucose level 180–200 mg/dL) in patients with severe sepsis.¹⁷ It also compared two fluids for volume resuscitation: 10% pentastarch vs modified Ringer's lactate. It included both medical and surgical patients.

Trial halted early for safety reasons

The mean morning blood glucose level was significantly lower in the intensive insulin group (112 vs 151 mg/dL).

Severe hypoglycemia (blood glucose ≤ 40 mg/dL) was more common in the group that received intensive insulin therapy (17% vs 4.1%, P < .001).

Mortality rates at 28 days did not differ significantly: 24.7% with intensive control vs 26.0% with conventional glucose control. The mortality rate at 90 days was 39.7% in the intensive therapy group and 35.4% in the conventional therapy group, but the difference was not statistically significant.

The intensive insulin arm of the trial was stopped after 488 patients were enrolled because of a higher rate of hypoglycemia (12.1% vs 2.1%) and of serious adverse events (10.9% vs 5.2%).

Additionally, the fluid resuscitation arm of the study was suspended at the first planned interim analysis because of a higher risk of organ failure in the 10% pentastarch group.

CORTICOSTEROID THERAPY IN SEPTIC SHOCK

Key points

• Corticosteroid therapy improves hemodynamic outcomes in patients with severe septic shock.
• Although meta-analyses suggest the mortality rate is lower with corticosteroid therapy, there is not enough evidence from randomized controlled trials to prove that the use of low-dose corticosteroids lowers the mortality rate in patients with septic shock.
• The corticotropin (ACTH) stimulation test should not be used to determine the need for corticosteroids in patients with septic shock.

Background

A previous multicenter study,¹⁸ performed in France, found that the use of corticosteroids in patients with septic shock resulted in lower rates of death at 28 days, in the ICU, and in the hospital and a shorter time to vasopressor withdrawal. Nevertheless, the beneficial effects were not observed in patients with adequate adrenal reserve (based on an ACTH stimulation test).

This study was criticized because of a high mortality rate in the placebo group.

The CORTICUS study

SPRUNG CL, ANNANE D, KEH D, ET AL; CORTICUS STUDY GROUP. HYDROCORTISONE THERAPY FOR PATIENTS WITH SEPTIC SHOCK. N ENGL J MED 2008; 358:111–124.

The Corticosteroid Therapy of Septic Shock (CORTICUS) study was a multicenter trial that randomly assigned 499 patients with septic shock to receive hydrocortisone (50 mg intravenously every 6 hours for 5 days, followed by a 6-day taper period) or placebo.¹⁹

Patients were eligible to be enrolled within 72 hours of onset of shock. Similar to previous studies, the CORTICUS trial classified patients on the basis of an ACTH stimulation test as having inadequate adrenal reserve (a cortisol increase of ≤ 9 μg/dL) or adequate adrenal reserve (a cortisol increase of > 9 μg/dL).

Faster reversal of shock with steroids

At baseline, the mean Simplified Acute Physiologic Score II (SAPS II) was 49 (the range of possible scores is 0 to 163; the higher the score the worse the organ dysfunction).

Hydrocortisone use resulted in a shorter duration of vasopressor use and a faster reversal of shock (3.3 days vs 5.8 days, P < .001).

This association was the same when patients were divided according to response to ACTH stimulation test. Time to reversal of shock in responders:

• 2.8 days with hydrocortisone
• 5.8 days with placebo (P < .001).

Time to reversal of shock in nonresponders:

• 3.9 days with hydrocortisone
• 6.0 days with placebo (P = .06).

Nevertheless, the treatment did not reduce the mortality rate at 28 days overall (34.3% vs 31.5% P = .51), or in the subgroups based on response to ACTH, or at any other time point. A post hoc analysis suggested that patients who had a systolic blood pressure of less than 90 mm Hg within 30 minutes of enrollment had a greater benefit in terms of mortality rate, but the effect was not statistically significant: the absolute difference was −11.2% (P = 0.28). Similarly, post hoc analyses also revealed a higher rate of death at 28 days in patients who received etomidate (Amidate) before randomization in both groups (P = .03).

Importantly, patients who received corticosteroids had a higher incidence of superinfections, including new episodes of sepsis or septic shock, with a combined odds ratio of 1.37 (95% CI 1.05–1.79).

Length of stay in the hospital or in the ICU was similar in patients who received corticosteroids and in those who received placebo. The ICU length of stay was 19 ± 31 days with hydrocortisone vs 18 ± 17 days with placebo (P = .51).

Comments

The CORTICUS trial showed that low-dose corticosteroid therapy results in faster reversal of shock in patients with severe septic shock. The hemodynamic benefits are present in all patients regardless of response to the ACTH stimulation test.

Nevertheless, contrary to previous findings,¹⁸ corticosteroid use was not associated with an improvement in mortality rates. Important differences exist between these two studies:

• The mortality rates in the placebo groups were significantly different (> 50% in the French study vs 30% in CORTICUS).
• The SAPS II scores were different in these two trials (55 vs 49), suggesting a greater severity of illness in the French study.
• The criteria for enrollment were different: the French study included patients who had a systolic blood pressure lower than 90 mm Hg for more than 1 hour despite fluid administration and vasopressor use, whereas the CORTICUS trial included patients who had a systolic blood pressure lower than 90 mm Hg for more than 1 hour despite fluid administration or vasopressor use.
• The time of enrollment was different: patients were enrolled much faster in the French study (within 8 hours) than in the CORTICUS trial (within 72 hours).

A recent meta-analysis of 17 randomized trials (including the CORTICUS study), found that, compared with those who received placebo, patients who received corticosteroids had a small reduction in the 28-day mortality rate (HR 0.84, 95% CI 0.71–1.00, P < .05).²⁰ Of note, this meta-analysis has been criticized for possible publication bias and also for a large degree of heterogeneity in its results.²¹

VASOPRESSOR THERAPY IN SHOCK

Key points

• Vasopressin use in patients with severe septic shock is not associated with an improvement in mortality rates.
• Vasopressin should not be used as a first-line agent in patients with septic shock.
• Norepinephrine should be considered a first-line agent in patients with shock.
• Compared with norepinephrine, the use of dopamine in patients with shock is associated with similar mortality rates, although its use may result in a greater number of cardiac adverse events.

Background

Vasopressin gained popularity in critical care in the last 10 years because several small studies showed that adding it improves hemodynamics and results in a reduction in the doses of catecholamines in patients with refractory septic shock.²² Furthermore, the Surviving Sepsis Campaign guidelines recommended the use of vasopressin in patients who have refractory shock despite fluid resuscitation and the use of other “conventional” vasopressors.²³

Despite these positive findings, it remained unknown if the use of vasopressin increases the survival rate in patients with septic shock.

The Vasopressin and Septic Shock Trial (VASST)

RUSSELL JA, WALLEY KR, SINGER J, ET AL; VASST INVESTIGATORS. VASOPRESSIN VERSUS NOREPINEPHRINE INFUSION IN PATIENTS WITH SEPTIC SHOCK. N ENGL J MED 2008; 358:877–887.

The Vasopressin and Septic Shock Trial (VASST)²⁴ was a multicenter randomized, double-blind, controlled trial that included 778 patients with refractory septic shock. Refractory shock was defined as the lack of a response to a normal saline fluid bolus of 500 mL or the need for vasopressors (norepinephrine in doses of at least 5 μg/minute or its equivalent for 6 hours or more in the 24 hours before randomization).

Two subgroups were identified: those with severe septic shock (requiring norepinephrine in doses of 15 μg/minute or higher) and those with less-severe septic shock (needing norepinephrine in doses of 5 to 14 μg/minute). Patients with unstable coronary artery disease (acute myocardial infarction, angina) and severe congestive heart failure were excluded.

Patients were randomized to receive an intravenous infusion of vasopressin (0.01–0.03 U/minute) or norepinephrine (5–15 mg/minute) in addition to open-labeled vasopressors (excluding vasopressin). The primary outcome was the all-cause mortality rate at 28 days.

Results

At 28 days, fewer patients had died in the vasopressin group than in the norepinephrine group (35.4% vs 39.3%), but the difference was not statistically significant (P = .26). The trend was the same at 90 days (mortality rate 43.9% vs 49.6%, P = .11).

Subgroup analysis showed that in patients with less-severe septic shock, those who received vasopressin had a lower mortality rate at 28 days (26.5% vs 35.7%, P = .05; relative risk 0.74; 95% CI 0.55–1.01) and at 90 days (35.8% vs 46.1%, P = .04; relative risk 0.78, 95% CI 0.61–0.99).

There were no statistically significant differences in any of the other secondary outcomes or in serious adverse events.

Comments

The study has been criticized for several reasons:

• The mean arterial blood pressure at baseline before initiation of vasopressin was 72 mm Hg (and some argue that vasopressin was therefore not needed by the time it was started).
• The time from screening to infusion of the study drug was very long (12 hours).
• The observed mortality rate was lower than expected (37%).

Despite these considerations, the VASST trial showed that vasopressin is not associated with an increased number of adverse events in patients without active cardiovascular disease. The possible benefit in terms of the mortality rate in the subgroup of patients with less-severe septic shock requires further investigation.

Is dopamine equivalent to norepinephrine?

Previously, the Sepsis Occurrence in Acutely Ill Patients (SOAP) study, a multicenter, observational cohort study, found that dopamine use was associated with a higher all-cause mortality rate in the ICU compared with no dopamine.²⁵ This finding had not been reproduced, as few well-designed studies had compared the effects of dopamine and norepinephrine.

The SOAP II study

DE BACKER D, BISTON P, DEVRIENDT J, ET AL; SOAP II INVESTIGATORS.. COMPARISON OF DOPAMINE AND NOREPINEPHRINE IN THE TREATMENT OF SHOCK. N ENGL J MED 2010; 362:779–789.

The SOAP II study,²⁶ a multicenter, randomized trial, compared dopamine vs norepinephrine as first-line vasopressor therapy. In patients with refractory shock despite use of dopamine 20 μg/kg/minute or norepinephrine 0.19 μg/kg/minute, open-label norepinephrine, epinephrine, or vasopressin was added.

The primary outcome was the mortality rate at 28 days after randomization; secondary end points included the number of days without need for organ support and the occurrence of adverse events.

Results

A total of 1,679 patients were included; 858 were assigned to dopamine and 821 to norepinephrine. Most (1,044, 62%) of the patients had a diagnosis of septic shock.

No significant difference in mortality rates was noted at 28 days: 52.5% with dopamine vs 48.5% with norepinephrine (P = .10).

However, there were more arrhythmias in the patients treated with dopamine: 207 events (24.1%) vs 102 events (12.4%) (P < .001). The number of other adverse events such as renal failure, myocardial infarction, arterial occlusion, or skin necrosis was not different between the groups.

A subgroup analysis showed that dopamine was associated with more deaths at 28 days in patients with cardiogenic shock (P = .03) but not in patients with septic shock (P = .19) or with hypovolemic shock (P = .84).

Comments

The study was criticized because the patients may not have received adequate fluid resuscitation (the study considered adequate resuscitation to be equivalent to 1 L of crystalloids or 500 mL of colloids), as different degrees of volume depletion among patients make direct comparisons of vasopressor effects difficult.

Additionally, the study defined dopamine 20 μg/kg/minute as being equipotent with norepinephrine 0.19 μg/kg/minute. Comparisons of potency between drugs are difficult to establish, as there are no available data.

Nevertheless, this study further confirms previous findings suggesting that norepinephrine is not associated with more end-organ damage (such as renal failure or skin ischemia), and shows that dopamine may increase the number of adverse events, particularly in patients with cardiac disease.

We have seen significant growth in clinical research in critical care medicine in the last decade. Advances have been made in many important areas in this field; of these, advances in treating septic shock and acute respiratory distress syndrome (ARDS), and also in supportive therapies for critically ill patients (eg, sedatives, insulin), have perhaps received the most attention.

Of note, several once-established therapies in these areas have failed the test of time, as the result of evidence from more-recent clinical trials. For example, recent studies have shown that a pulmonary arterial catheter does not improve outcomes in patients with ARDS. Similarly, what used to be “optimal” fluid management in patients with ARDS is no longer considered appropriate.

In this review, we summarize eight major studies in critical care medicine published in the last 5 years, studies that have contributed to changes in our practice in the intensive care unit (ICU).

FLUID MANAGEMENT IN ARDS

Key points

• In patients with acute lung injury (ALI) and ARDS, fluid restriction is associated with better outcomes than a liberal fluid policy.
• A pulmonary arterial catheter is not necessary and, compared with a central venous catheter, may result in more complications in patients with ALI and ARDS.

Background

Fluid management practices in patients with ARDS have been extremely variable. Two different approaches are commonly used: the liberal or “wet” approach to optimize tissue perfusion and the “dry” approach, which focuses on reducing lung edema. Given that most deaths attributed to ARDS result from extrapulmonary organ failure, aggressive fluid restriction has been the less popular approach.

Additionally, although earlier studies and meta-analyses suggested that the use of a pulmonary arterial catheter was not associated with better outcomes in critically ill patients,¹ controversy remained regarding the value of a pulmonary arterial catheter compared with a central venous catheter in guiding fluid management in patients with ARDS, and data were insufficient to prove one strategy better than the other.

The Fluids and Catheter Treatment Trial (FACTT)

NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK; WIEDEMANN HP, WHEELER AP, BERNARD GR, ET AL. COMPARISON OF TWO FLUID-MANAGEMENT STRATEGIES IN ACUTE LUNG INJURY. N ENGL J MED 2006; 354:2564–2575.

NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK; WHEELER AP, BERNARD GR, THOMPSON BT, ET AL. PULMONARY-ARTERY VERSUS CENTRAL VENOUS CATHETER TO GUIDE TREATMENT OF ACUTE LUNG INJURY. N ENGL J MED 2006; 354:2213–2224.

The Fluids and Catheter Treatment Trial (FACTT) compared two fluid strategies² and also the utility of a pulmonary arterial catheter vs a central venous catheter³ in patients with ALI or ARDS.

This two-by-two factorial trial randomized 1,000 patients to be treated according to either a conservative (fluid-restrictive or “dry”) or a liberal (“wet”) fluid management strategy for 7 days. Additionally, they were randomly assigned to receive either a central venous catheter or a pulmonary arterial catheter. The trial thus had four treatment groups:

• Fluid-restricted and a central venous catheter, with a goal of keeping the central venous pressure below 4 mm Hg
• Fluid-restricted and a pulmonary arterial catheter: fluids were restricted and diuretics were given to keep the pulmonary artery occlusion pressure below 8 mm Hg
• Fluid-liberal and a central venous catheter: fluids were given to keep the central venous pressure between 10 and 14 mm Hg
• Fluid-liberal and a pulmonary arterial catheter: fluids were given to keep the pulmonary artery occlusion pressure between 14 and 18 mm Hg.

The primary end point was the mortality rate at 60 days. Secondary end points included the number of ventilator-free days and organ-failure-free days and parameters of lung physiology. All patients were managed with a low-tidal-volume strategy.

The ‘dry’ strategy was better

The cumulative fluid balance was −136 mL ± 491 mL in the “dry” group and 6,992 mL ± 502 mL in the “wet” group, a difference of more than 7 L (P < .0001). Of note, before randomization, the patients were already fluid-positive, with a mean total fluid balance of +2,700 mL).²

At 60 days, no statistically significant difference in mortality rate was seen between the fluid-management groups (25.5% in the dry group vs 28.4% in the wet group (P = .30). Nevertheless, patients in the dry group had better oxygenation indices and lung injury scores (including lower plateau airway pressure), resulting in more ventilator-free days (14.6 ± 0.5 vs 12.1 ± 0.5; P = .0002) and ICU-free days (13.4 ± 0.4 vs 11.2 ± 0.4; P = .0003).²

Although those in the dry-strategy group had a slightly lower cardiac index and mean arterial pressure, they did not have a higher incidence of shock.

More importantly, the dry group did not have a higher rate of nonpulmonary organ failure. Serum creatinine and blood urea nitrogen concentrations were slightly higher in this group, but this was not associated with a higher incidence of renal failure or the use of dialysis: 10% in the dry-strategy group vs 14% in the wet-strategy group; P = .0642).²

No advantage with a pulmonary arterial catheter

The mortality rate did not differ between the catheter groups. However, the patients who received a pulmonary arterial catheter stayed in the ICU 0.2 days longer and had twice as many nonfatal cardiac arrhythmias as those who received a central venous catheter.³

Comments

The liberal fluid-strategy group had fluid balances similar to those seen in previous National Institutes of Health ARDS Network trials in which fluid management was not controlled. This suggests that the liberal fluid strategy reflects usual clinical practice.

Although the goals used in this study (central venous pressure < 4 mm Hg or pulmonary artery occlusion pressure < 8 mm Hg) could be difficult to achieve in clinical practice, a conservative strategy of fluid management is preferred in patients with ALI or ARDS, given the benefits observed in this trial.

A pulmonary arterial catheter is not indicated to guide hemodynamic management of patients with ARDS.

CORTICOSTEROID USE IN ARDS

Key points

• In selected patients with ARDS, the prolonged use of corticosteroids may result in better oxygenation and a shorter duration of mechanical ventilation.
• Late use of corticosteroids in patients with ARDS (> 14 days after diagnosis) is not indicated and may increase the risk of death.
• The role of corticosteroids in early ARDS (< 7 days after diagnosis) remains controversial.

Background

Systemic corticosteroid therapy was commonly used in ARDS patients in the 1970s and 1980s. However, a single-center study published in the late 1980s showed that a corticosteroid in high doses (methylprednisolone 30 mg/kg) resulted in more complications and was not associated with a lower mortality rate.⁴ On the other hand, a small study that included only patients with persistent ARDS (defined as ARDS lasting for more than 7 days) subsequently showed that oxygenation was significantly better and that fewer patients died while in the hospital with the use of methylprednisolone 2 mg/kg for 32 days.⁵

In view of these divergent findings, the ARDS Network decided to perform a study to help understand the role of corticosteroids in ARDS.

The Late Steroid Rescue Study (LaSRS)

STEINBERG KP, HUDSON LD, GOODMAN RB, ET AL; NATIONAL HEART, LUNG, AND BLOOD INSTITUTE ACUTE RESPIRATORY DISTRESS SYNDROME (ARDS) CLINICAL TRIALS NETWORK. EFFICACY AND SAFETY OF CORTICOSTEROIDS FOR PERSISTENT ACUTE RESPIRATORY DISTRESS SYNDROME. N ENGL J MED 2006; 354:1671–1684.

The Late Steroid Rescue Study (LaSRS),⁶ a double-blind, multicenter trial, randomly assigned 180 patients with persistent ARDS (defined as ongoing disease 7–28 days after its onset) to receive methylprednisolone or placebo for 21 days.

Methylprednisolone was given in an initial dose of 2 mg/kg of predicted body weight followed by a dose of 0.5 mg/kg every 6 hours for 14 days and then a dose of 0.5 mg/kg every 12 hours for 7 days, and then it was tapered over 2 to 4 days and discontinued. It could be discontinued if 21 days of treatment were completed or if the patient was able to breathe without assistance.

The primary end point was the mortality rate at 60 days. Secondary end points included the number of ventilator-free days, organ-failure-free days, and complications and the levels of biomarkers of inflammation.

No reduction in mortality rates with steroids

The mortality rates did not differ significantly in the corticosteroid group vs the placebo group at 60 days:

• 29.2% with methylprednisolone (95% confidence interval [CI] 20.8–39.4)
• 28.6% with placebo (95% CI 20.3–38.6, P = 1.0).

Mortality rates at 180 days were also similar between the groups:

• 31.5% with methylprednisolone (95% CI 22.8–41.7)
• 31.9% with placebo (95% CI 23.2–42.0, P = 1.0).

In patients randomized between 7 and 13 days after the onset of ARDS, the mortality rates were lower in the methylprednisolone group than in the placebo group but the differences were not statistically significant. The mortality rate in this subgroup was 27% vs 36% (P = .26) at 60 days and was 27% vs 39% (P = .14) at 180 days.

However, in patients randomized more than 14 days after the onset of ARDS, the mortality rate was significantly higher in the methylprednisolone group than in the placebo group at 60 days (35% vs 8%, P = .02) and at 180 days (44% vs 12%, P = .01).

Some benefit in secondary outcomes

At day 28, methylprednisolone was associated with:

• More ventilator-free days (11.2 ± 9.4 vs 6.8 ± 8.5, P < .001)
• More shock-free days (20.7 ± 8.9 vs 17.9 ± 10.2, P = .04)
• More ICU-free days (8.9 ± 8.2 vs 6.7 ± 7.8, P = .02).

Similarly, pulmonary physiologic indices were better with methylprednisolone, specifically:

• The ratio of Pao₂ to the fraction of inspired oxygen at days 3, 4, and 14 (P < .05)
• Plateau pressure at days 4, 5, and 7 (P < .05)
• Static compliance at days 7 and 14 (P < .05).

In terms of side effects, methylprednisolone was associated with more events associated with myopathy or neuropathy (9 vs 0, P = .001), but there were no differences in the number of serious infections or in glycemic control.

Comments

Although other recent studies suggested that corticosteroid use may be associated with a reduction in mortality rates,^7–9 LaSRS did not confirm this effect. Although the doses and length of therapy were similar in these studies, LaSRS was much larger and included patients from the ARDS Network.

Nevertheless, LaSRS was criticized because of strict exclusion criteria and poor enrollment (only 5% of eligible patients were included). Additionally, it was conducted over a period of time when some ICU practices varied significantly (eg, low vs high tidal volume ventilation, tight vs loose glucose control).

INTERRUPTING SEDATION DURING MECHANICAL VENTILATION

Key points

• Daily awakening of mechanically ventilated patients is safe.
• Daily interruption of sedation in mechanically ventilated patients is associated with a shorter length of mechanical ventilation.

Background

Sedatives are a central component of critical care. Continuous infusions of narcotics, benzodiazepines, and anesthetic agents are frequently used to promote comfort in patients receiving mechanical ventilation.

Despite its widespread use in the ICU, there is little evidence that such sedation improves outcomes. Observational and randomized trials^10–12 have shown that patients who receive continuous infusions of sedatives need to be on mechanical ventilation longer than those who receive intermittent dosing. Additionally, an earlier randomized controlled trial¹³ showed that daily interruption of sedative drug infusions decreased the duration of mechanical ventilation by almost 50% and resulted in a reduction in the length of stay in the ICU.

Despite these findings, many ICU physicians remain skeptical of the value of daily interruption of sedative medications and question the safety of this practice.

The Awakening and Breathing Controlled (ABC) trial

GIRARD TD, KRESS JP, FUCHS BD, ET AL. EFFICACY AND SAFETY OF A PAIRED SEDATION AND VENTILATOR WEANING PROTOCOL FOR MECHANICALLY VENTILATED PATIENTS IN INTENSIVE CARE (AWAKENING AND BREATHING CONTROLLED TRIAL): A RANDOMISED CONTROLLED TRIAL. LANCET 2008; 371:126–134.

The Awakening and Breathing Controlled (ABC) trial¹⁴ was a multicenter, randomized controlled trial that included 336 patients who required at least 12 consecutive hours of mechanical ventilation. All patients had to be receiving patient-targeted sedation.

Those in the intervention group (n = 168) had their sedation interrupted every day, followed by a clinical assessment to determine whether they could be allowed to try breathing spontaneously. The control group (n = 168) also received a clinical assessment for a trial of spontaneous breathing, while their sedation was continued as usual.

In patients in the intervention group who failed the screening for a spontaneous breathing trial, the sedatives were resumed at half the previous dose. Criteria for failure on the spontaneous breathing trial included any of the following: anxiety, agitation, respiratory rate more than 35 breaths per minute for 5 minutes or longer, cardiac arrhythmia, oxygen saturation less than 88% for 5 minutes or longer, or two or more signs of respiratory distress, tachycardia, bradycardia, paradoxical breathing, accessory muscle use, diaphoresis, or marked dyspnea.

The combination of sedation interruption and a spontaneous breathing trial was superior to a spontaneous breathing trial alone. The mean number of ventilator-free days:

• 14.7 ± 0.9 with sedation interruption
• 11.6 ± 0.9 days with usual care (P = .02).

The median time to ICU discharge:

• 9.1 days with sedation interruption (interquartile range 5.1 to 17.8)
• 12.9 days with usual care (interquartile range 6.0 to 24.2, P = .01).

The mortality rate at 28 days:

• 28% with sedation interruption
• 35% with usual care (P = .21).

The mortality rate at 1 year:

• 44% with sedation interruption
• 58% with usual care (hazard ratio [HR] in the intervention group 0.68, 95% CI 0.50–0.92, P = .01).

Of note, patients in the intervention group had a higher rate of self-extubation (9.6% vs 3.6%, P = .03), but the rate of reintubation was similar between the groups (14% vs 13%, P = .47).

Comments

The addition of daily awakenings to spontaneous breathing trials results in a further reduction in the number of ICU days and increases the number of ventilator-free days.

Of note, the protocol allowed patients in the control group to undergo a spontaneous breathing trial while on sedatives (69% of the patients were receiving sedation at the time). Therefore, a bias effect in favor of the intervention group cannot be excluded. However, both groups had to meet criteria for readiness for spontaneous breathing.

The study demonstrates the safety of daily awakenings and confirms previous findings suggesting that a daily trial of spontaneous breathing results in better ICU outcomes.

GLUCOSE CONTROL IN THE ICU

Key points

• Although earlier studies suggested that intensive insulin therapy might be beneficial in critically ill patients, new findings show that strict glucose control can lead to complications without improving outcomes.

Background

A previous study¹⁵ found that intensive insulin therapy to maintain a blood glucose level between 80 and 110 mg/dL (compared with 180–200 mg/dL) reduced the mortality rate in surgical critical care patients. The mortality rate in the ICU was 4.6% with intensive insulin therapy vs 8.0% with conventional therapy (P < .04), and the effect was more robust for patients who remained longer than 5 days in the ICU (10.6% vs 20.2%).

Importantly, however, hypoglycemia (defined as blood glucose ≤ 40 mg/dL) occurred in 39 patients in the intensive-treatment group vs 6 patients in the conventional-treatment group.

The NICE-SUGAR trial

NICE-SUGAR STUDY INVESTIGATORS; FINFER S, CHITTOCK DR, SU SY, ET AL. INTENSIVE VERSUS CONVENTIONAL GLUCOSE CONTROL IN CRITICALLY ILL PATIENTS. N ENGL J MED 2009; 360:1283–1297.

The Normoglycemia in Intensive Care Evaluation-Survival Using Glucose Algorithm Regulation (NICE-SUGAR) trial¹⁶ randomized 6,104 patients in medical and surgical ICUs to receive either intensive glucose control (blood glucose 81–108 mg/dL) with insulin therapy or conventional glucose control (blood glucose < 180 mg/dL). In the conventional-control group, insulin was discontinued if the blood glucose level dropped below 144 mg/dL.

A higher mortality rate with intensive glucose control

As expected, the intensive-control group achieved lower blood glucose levels: 115 vs 144 mg/dL.

Nevertheless, intensive glucose control was associated with a higher incidence of severe hypoglycemia, defined as a blood glucose level lower than 40 mg/dL: 6.8% vs 0.5%.

More importantly, compared with conventional insulin therapy, intensive glucose control was associated with a higher 90-day mortality rate: 27.5% vs 24.9% (odds ratio 1.14, 95% CI 1.02–1.28). These findings were similar in the subgroup of surgical patients (24.4% vs 19.8%, odds ratio 1.31, 95% CI 1.07–1.61).

Comments

Of note, the conventional-control group had more patients who discontinued the treatment protocol prematurely. Additionally, more patients in this group received corticosteroids.

These results widely differ from those of a previous study by van den Berghe et al,¹⁵ which showed that tight glycemic control is associated with a survival benefit. The differences in outcomes are probably largely related to different patient populations, as van den Berghe et al included patients who had undergone cardiac surgery, who were more likely to benefit from strict blood glucose control.

The VISEP trial

BRUNKHORST FM, ENGEL C, BLOOS F, ET AL; GERMAN COMPETENCE NETWORK SEPSIS (SEPNET). INTENSIVE INSULIN THERAPY AND PENTASTARCH RESUSCITATION IN SEVERE SEPSIS. N ENGL J MED 2008; 358:125–139.

The Volume Substitution and Insulin Therapy in Severe Sepsis (VISEP) trial was a multicenter study designed to compare intensive insulin therapy (target blood glucose level 80–110 mg/dL) and conventional glucose control (target blood glucose level 180–200 mg/dL) in patients with severe sepsis.¹⁷ It also compared two fluids for volume resuscitation: 10% pentastarch vs modified Ringer's lactate. It included both medical and surgical patients.

Trial halted early for safety reasons

The mean morning blood glucose level was significantly lower in the intensive insulin group (112 vs 151 mg/dL).

Severe hypoglycemia (blood glucose ≤ 40 mg/dL) was more common in the group that received intensive insulin therapy (17% vs 4.1%, P < .001).

Mortality rates at 28 days did not differ significantly: 24.7% with intensive control vs 26.0% with conventional glucose control. The mortality rate at 90 days was 39.7% in the intensive therapy group and 35.4% in the conventional therapy group, but the difference was not statistically significant.

The intensive insulin arm of the trial was stopped after 488 patients were enrolled because of a higher rate of hypoglycemia (12.1% vs 2.1%) and of serious adverse events (10.9% vs 5.2%).

Additionally, the fluid resuscitation arm of the study was suspended at the first planned interim analysis because of a higher risk of organ failure in the 10% pentastarch group.

CORTICOSTEROID THERAPY IN SEPTIC SHOCK

Key points

• Corticosteroid therapy improves hemodynamic outcomes in patients with severe septic shock.
• Although meta-analyses suggest the mortality rate is lower with corticosteroid therapy, there is not enough evidence from randomized controlled trials to prove that the use of low-dose corticosteroids lowers the mortality rate in patients with septic shock.
• The corticotropin (ACTH) stimulation test should not be used to determine the need for corticosteroids in patients with septic shock.

Background

A previous multicenter study,¹⁸ performed in France, found that the use of corticosteroids in patients with septic shock resulted in lower rates of death at 28 days, in the ICU, and in the hospital and a shorter time to vasopressor withdrawal. Nevertheless, the beneficial effects were not observed in patients with adequate adrenal reserve (based on an ACTH stimulation test).

This study was criticized because of a high mortality rate in the placebo group.

The CORTICUS study

SPRUNG CL, ANNANE D, KEH D, ET AL; CORTICUS STUDY GROUP. HYDROCORTISONE THERAPY FOR PATIENTS WITH SEPTIC SHOCK. N ENGL J MED 2008; 358:111–124.

The Corticosteroid Therapy of Septic Shock (CORTICUS) study was a multicenter trial that randomly assigned 499 patients with septic shock to receive hydrocortisone (50 mg intravenously every 6 hours for 5 days, followed by a 6-day taper period) or placebo.¹⁹

Patients were eligible to be enrolled within 72 hours of onset of shock. Similar to previous studies, the CORTICUS trial classified patients on the basis of an ACTH stimulation test as having inadequate adrenal reserve (a cortisol increase of ≤ 9 μg/dL) or adequate adrenal reserve (a cortisol increase of > 9 μg/dL).

Faster reversal of shock with steroids

At baseline, the mean Simplified Acute Physiologic Score II (SAPS II) was 49 (the range of possible scores is 0 to 163; the higher the score the worse the organ dysfunction).

Hydrocortisone use resulted in a shorter duration of vasopressor use and a faster reversal of shock (3.3 days vs 5.8 days, P < .001).

This association was the same when patients were divided according to response to ACTH stimulation test. Time to reversal of shock in responders:

• 2.8 days with hydrocortisone
• 5.8 days with placebo (P < .001).

Time to reversal of shock in nonresponders:

• 3.9 days with hydrocortisone
• 6.0 days with placebo (P = .06).

Nevertheless, the treatment did not reduce the mortality rate at 28 days overall (34.3% vs 31.5% P = .51), or in the subgroups based on response to ACTH, or at any other time point. A post hoc analysis suggested that patients who had a systolic blood pressure of less than 90 mm Hg within 30 minutes of enrollment had a greater benefit in terms of mortality rate, but the effect was not statistically significant: the absolute difference was −11.2% (P = 0.28). Similarly, post hoc analyses also revealed a higher rate of death at 28 days in patients who received etomidate (Amidate) before randomization in both groups (P = .03).

Importantly, patients who received corticosteroids had a higher incidence of superinfections, including new episodes of sepsis or septic shock, with a combined odds ratio of 1.37 (95% CI 1.05–1.79).

Length of stay in the hospital or in the ICU was similar in patients who received corticosteroids and in those who received placebo. The ICU length of stay was 19 ± 31 days with hydrocortisone vs 18 ± 17 days with placebo (P = .51).

Comments

The CORTICUS trial showed that low-dose corticosteroid therapy results in faster reversal of shock in patients with severe septic shock. The hemodynamic benefits are present in all patients regardless of response to the ACTH stimulation test.

Nevertheless, contrary to previous findings,¹⁸ corticosteroid use was not associated with an improvement in mortality rates. Important differences exist between these two studies:

• The mortality rates in the placebo groups were significantly different (> 50% in the French study vs 30% in CORTICUS).
• The SAPS II scores were different in these two trials (55 vs 49), suggesting a greater severity of illness in the French study.
• The criteria for enrollment were different: the French study included patients who had a systolic blood pressure lower than 90 mm Hg for more than 1 hour despite fluid administration and vasopressor use, whereas the CORTICUS trial included patients who had a systolic blood pressure lower than 90 mm Hg for more than 1 hour despite fluid administration or vasopressor use.
• The time of enrollment was different: patients were enrolled much faster in the French study (within 8 hours) than in the CORTICUS trial (within 72 hours).

A recent meta-analysis of 17 randomized trials (including the CORTICUS study), found that, compared with those who received placebo, patients who received corticosteroids had a small reduction in the 28-day mortality rate (HR 0.84, 95% CI 0.71–1.00, P < .05).²⁰ Of note, this meta-analysis has been criticized for possible publication bias and also for a large degree of heterogeneity in its results.²¹

VASOPRESSOR THERAPY IN SHOCK

Key points

• Vasopressin use in patients with severe septic shock is not associated with an improvement in mortality rates.
• Vasopressin should not be used as a first-line agent in patients with septic shock.
• Norepinephrine should be considered a first-line agent in patients with shock.
• Compared with norepinephrine, the use of dopamine in patients with shock is associated with similar mortality rates, although its use may result in a greater number of cardiac adverse events.

Background

Vasopressin gained popularity in critical care in the last 10 years because several small studies showed that adding it improves hemodynamics and results in a reduction in the doses of catecholamines in patients with refractory septic shock.²² Furthermore, the Surviving Sepsis Campaign guidelines recommended the use of vasopressin in patients who have refractory shock despite fluid resuscitation and the use of other “conventional” vasopressors.²³

Despite these positive findings, it remained unknown if the use of vasopressin increases the survival rate in patients with septic shock.

The Vasopressin and Septic Shock Trial (VASST)

RUSSELL JA, WALLEY KR, SINGER J, ET AL; VASST INVESTIGATORS. VASOPRESSIN VERSUS NOREPINEPHRINE INFUSION IN PATIENTS WITH SEPTIC SHOCK. N ENGL J MED 2008; 358:877–887.

The Vasopressin and Septic Shock Trial (VASST)²⁴ was a multicenter randomized, double-blind, controlled trial that included 778 patients with refractory septic shock. Refractory shock was defined as the lack of a response to a normal saline fluid bolus of 500 mL or the need for vasopressors (norepinephrine in doses of at least 5 μg/minute or its equivalent for 6 hours or more in the 24 hours before randomization).

Two subgroups were identified: those with severe septic shock (requiring norepinephrine in doses of 15 μg/minute or higher) and those with less-severe septic shock (needing norepinephrine in doses of 5 to 14 μg/minute). Patients with unstable coronary artery disease (acute myocardial infarction, angina) and severe congestive heart failure were excluded.

Patients were randomized to receive an intravenous infusion of vasopressin (0.01–0.03 U/minute) or norepinephrine (5–15 mg/minute) in addition to open-labeled vasopressors (excluding vasopressin). The primary outcome was the all-cause mortality rate at 28 days.

Results

At 28 days, fewer patients had died in the vasopressin group than in the norepinephrine group (35.4% vs 39.3%), but the difference was not statistically significant (P = .26). The trend was the same at 90 days (mortality rate 43.9% vs 49.6%, P = .11).

Subgroup analysis showed that in patients with less-severe septic shock, those who received vasopressin had a lower mortality rate at 28 days (26.5% vs 35.7%, P = .05; relative risk 0.74; 95% CI 0.55–1.01) and at 90 days (35.8% vs 46.1%, P = .04; relative risk 0.78, 95% CI 0.61–0.99).

There were no statistically significant differences in any of the other secondary outcomes or in serious adverse events.

Comments

The study has been criticized for several reasons:

• The mean arterial blood pressure at baseline before initiation of vasopressin was 72 mm Hg (and some argue that vasopressin was therefore not needed by the time it was started).
• The time from screening to infusion of the study drug was very long (12 hours).
• The observed mortality rate was lower than expected (37%).

Despite these considerations, the VASST trial showed that vasopressin is not associated with an increased number of adverse events in patients without active cardiovascular disease. The possible benefit in terms of the mortality rate in the subgroup of patients with less-severe septic shock requires further investigation.

Is dopamine equivalent to norepinephrine?

Previously, the Sepsis Occurrence in Acutely Ill Patients (SOAP) study, a multicenter, observational cohort study, found that dopamine use was associated with a higher all-cause mortality rate in the ICU compared with no dopamine.²⁵ This finding had not been reproduced, as few well-designed studies had compared the effects of dopamine and norepinephrine.

The SOAP II study

DE BACKER D, BISTON P, DEVRIENDT J, ET AL; SOAP II INVESTIGATORS.. COMPARISON OF DOPAMINE AND NOREPINEPHRINE IN THE TREATMENT OF SHOCK. N ENGL J MED 2010; 362:779–789.

The SOAP II study,²⁶ a multicenter, randomized trial, compared dopamine vs norepinephrine as first-line vasopressor therapy. In patients with refractory shock despite use of dopamine 20 μg/kg/minute or norepinephrine 0.19 μg/kg/minute, open-label norepinephrine, epinephrine, or vasopressin was added.

The primary outcome was the mortality rate at 28 days after randomization; secondary end points included the number of days without need for organ support and the occurrence of adverse events.

Results

A total of 1,679 patients were included; 858 were assigned to dopamine and 821 to norepinephrine. Most (1,044, 62%) of the patients had a diagnosis of septic shock.

No significant difference in mortality rates was noted at 28 days: 52.5% with dopamine vs 48.5% with norepinephrine (P = .10).

However, there were more arrhythmias in the patients treated with dopamine: 207 events (24.1%) vs 102 events (12.4%) (P < .001). The number of other adverse events such as renal failure, myocardial infarction, arterial occlusion, or skin necrosis was not different between the groups.

A subgroup analysis showed that dopamine was associated with more deaths at 28 days in patients with cardiogenic shock (P = .03) but not in patients with septic shock (P = .19) or with hypovolemic shock (P = .84).

Comments

The study was criticized because the patients may not have received adequate fluid resuscitation (the study considered adequate resuscitation to be equivalent to 1 L of crystalloids or 500 mL of colloids), as different degrees of volume depletion among patients make direct comparisons of vasopressor effects difficult.

Additionally, the study defined dopamine 20 μg/kg/minute as being equipotent with norepinephrine 0.19 μg/kg/minute. Comparisons of potency between drugs are difficult to establish, as there are no available data.

Nevertheless, this study further confirms previous findings suggesting that norepinephrine is not associated with more end-organ damage (such as renal failure or skin ischemia), and shows that dopamine may increase the number of adverse events, particularly in patients with cardiac disease.

A number of studies published in the last few years will likely affect the way we practice medicine in the hospital. Here, we will use a hypothetical case scenario to focus on the issues of anticoagulants, patient safety, quality improvement, critical care, transitions of care, and perioperative medicine.

AN ELDERLY MAN WITH NEW-ONSET ATRIAL FIBRILLATION

P.G. is an 80-year-old man with a history of hypertension and type 2 diabetes mellitus who is admitted with new-onset atrial fibrillation. In the hospital, his heart rate is brought under control with intravenous metoprolol (Lopressor). On discharge, he will be followed by his primary care physician (PCP). He does not have access to an anticoagulation clinic.

1. What are this patient’s options for stroke prevention?

• Aspirin 81 mg daily and clopidogrel (Plavix) 75 mg daily
• Warfarin (Coumadin) with a target international normalized ratio (INR) of 2.0 to 3.0
• Aspirin mg daily by itself
• Dabigatran (Pradaxa) 150 mg daily

A new oral anticoagulant agent

In deciding what type of anticoagulation to give to a patient with atrial fibrillation, it is useful to look at the CHADS₂ score (1 point each for congestive heart failure, hypertension, age 75 or older, and diabetes mellitus; 2 points for prior stroke or transient ischemic attack. This patient has a CHADS₂ score of 3, indicating that he should receive warfarin. An alternative is dabigatran, the first new anticoagulant agent in more than 50 years.

In a multicenter, international trial, Connolly et al¹ randomized 18,113 patients (mean age 71, 64% men) to receive dabigatran 110 mg twice daily, dabigatran 150 mg twice daily, or warfarin with a target INR of 2.0 to 3.0. In this noninferiority trial, dabigatran was given in a blinded manner, but the use of warfarin was open-label. Patients were eligible if they had atrial fibrillation at screening or within the previous 6 months and were at risk of stroke—ie, if they had at least one of the following: a history of stroke or transient ischemic attack, a left ventricular ejection fraction of less than 40%, symptoms of congestive heart failure (New York Heart Association class II or higher), and an age of 75 or older or an age of 65 to 74 with diabetes mellitus, hypertension, or coronary artery disease.

At a mean follow-up of 2 years, the rate of stroke or systolic embolism was 1.69% per year in the warfarin group compared with 1.1% in the higher-dose dabigatran group (relative risk 0.66, 95% confidence interval [CI] 0.53–0.82, P < .001). The rates of major hemorrhage were similar between these two groups. Comparing lower-dose dabigatran and warfarin, the rates of stroke or systolic embolism were not significantly different, but the rate of major bleeding was significantly lower with lower-dose dabigatran.

In a trial in patients with acute venous thromboembolism, Schulman et al² found that dabigatran was not inferior to warfarin in preventing venous thromboembolism.

Guidelines from the American College of Cardiology Foundation and the American Heart Association now endorse dabigatran as an alternative to warfarin for patients with atrial fibrillation.³ However, the guidelines state that it should be reserved for those patients who:

• Do not have a prosthetic heart valve or hemodynamically significant valve disease
• Have good kidney function (dabigatran is cleared by the kidney; the creatinine clearance rate should be greater than 30 mL/min for patients to receive dabigatran 150 mg twice a day, and at least 15 mL/min to receive 75 mg twice a day)
• Do not have severe hepatic dysfunction (which would impair baseline clotting function).

They note that other factors to consider are whether the patient:

• Can comply with the twice-daily dosing required
• Can afford the drug
• Has access to an anticoagulation management program (which would argue in favor of using warfarin).

Dabigatran is not yet approved to prevent venous thromboembolism.

CASE CONTINUED: HE GETS AN INFECTION

P.G. is started on dabigatran 150 mg by mouth twice a day.

While in the hospital he develops shortness of breath and needs intravenous furosemide (Lasix). Because he has bad veins, a percutaneous intravenous central catheter (PICC) line is placed. However, 2 days later, his temperature is 101.5°F, and his systolic blood pressure is 70 mm Hg. He is transferred to the medical intensive care unit (ICU) for treatment of sepsis. The anticoagulant is held, the PICC line is removed, and a new central catheter is inserted.

2. Which of the following directions is incorrect?

• Wash your hands before inserting the catheter. The accompanying nurse is required to directly observe this procedure or, if this step is not observed, to confirm that the physician did it.
• Before inserting the catheter, clean the patient’s skin with chlorhexidine antiseptic.
• Place sterile drapes over the entire patient.
• Wear any mask, hat, gown, and gloves available.
• Put a sterile dressing over the catheter.

A checklist can prevent infections when inserting central catheters

A checklist developed at Johns Hopkins Hospital consists of the five statements above, except for the second to last one—you should wear a sterile mask, hat, gown and gloves. This is important to ensure that sterility is not broken at any point during the procedure.

Pronovost et al⁴ launched a multicenter initiative at 90 ICUs, predominantly in the state of Michigan, to implement interventions to improve staff culture and teamwork and to translate research into practice by increasing the extent to which these five evidence-based recommendations were applied. The mean rate of catheter-related blood stream infections at baseline was 7.7%; this dropped to 2.8% during the implementation period, 2.3% in the first 3 months after implementation, 1.3% in months 16 through 18, and 1.1% in months 34 through 36, demonstrating that the gains from this quality-improvement project were sustainable.

If this intervention and collaborative model were implemented in all ICUs across the United States and if similar success rates were achieved, substantial and sustained reductions could be made in the 82,000 infections, 28,000 deaths, and $2.3 billion in costs attributed to these infections annually.

CASE CONTINUED: HE IS RESUSCITATED

P.G. is started on a 1-L fluid bolus but he remains hypotensive, necessitating a norepinephrine drip. He does well for about 6 hours, but in the middle of the night he develops ventricular tachycardia and ventricular fibrillation, and a code is called. He is successfully resuscitated, but the family is looking for prognostic information.

3. What are P.G.’s chances of surviving and leaving the hospital?

• 5%
• 8%
• 15%
• 23%

A registry of cardiopulmonary resuscitation

Tian et al⁵ evaluated outcomes in the largest registry of cardiopulmonary resuscitation to date. In this analysis, 49,656 adult patients with a first cardiopulmonary arrest occurring in an ICU between January 1, 2000, and August 26, 2008, were evaluated for their outcomes on pressors vs those not on pressors.

Other independent predictors of a lower survival rate were nonwhite race, mechanical ventilation, having three or more immediate causes of cardiopulmonary arrest, age 65 years or older, and cardiopulmonary arrest occurring at night or over the weekend.

Fortunately, for our patient, survival rates were higher for patients with ventricular tachycardia or fibrillation than with other causes of cardiopulmonary arrest: 22.6% for those on pressors (like our patient) and 40.7% for those on no pressors.

CASE CONTINUED: HE RECOVERS AND GOES HOME

P.G. makes a remarkable recovery and is now ready to go home. It is the weekend, and you are unable to schedule a follow-up appointment before his discharge, so you ask him to make an appointment with his PCP.

4. What is the likelihood that P.G. will be readmitted within 1 month?

• 5%
• 12%
• 20%
• 25%
• 30%

The importance of follow-up with a primary care physician

Misky et al,⁶ in a small study, attempted to identify the characteristics and outcomes of discharged patients who lack timely follow-up with a PCP. They prospectively enrolled 65 patients admitted to University of Colorado Hospital, an urban 425-bed tertiary care center, collecting information about patient demographics, diagnosis, payer source, and PCPs. After discharge, they called the patients to determine their PCP follow-up and readmission status. Thirty-day readmission rates and hospital length of stay were compared in patients with and without timely PCP follow-up (ie, within 4 weeks).

Patients lacking timely PCP follow-up were 10 times more likely to be readmitted (odds ratio [OR] = 9.9, P = .04): the rate was 21% in patients lacking timely PCP follow-up vs 3% in patients with timely PCP follow-up, P = .03. Lack of insurance was associated with lower rates of timely PCP follow-up: 29% vs 56% (P = .06), but did not independently increase the readmission rate or length of stay (OR = 1.0, P = .96). Index hospital length of stay was longer in patients lacking timely PCP follow-up: 4.4 days vs 6.3 days, P = 0.11.

Comment. Nearly half of the patients in this study, who were discharged from a large urban academic center, lacked timely follow-up with a PCP, resulting in higher rates of readmission and a nonsignificant trend toward longer length of stay. Timely follow-up is necessary for vulnerable patients.

Since the lack of timely PCP follow-up results in higher readmission rates and possibly a longer length of stay, a PCP appointment at discharge should perhaps be considered a core quality measure. This would be problematic in our American health care system, in which many patients lack health insurance and do not have a PCP.

A MAN UNDERGOING GASTRIC BYPASS SURGERY

A 55-year-old morbidly obese man (body mass index 45 kg/m²) with a history of type 2 diabetes mellitus, chronic renal insufficiency (serum creatinine level 2.1 mg/dL), hypercholesterolemia, and previous stroke is scheduled for gastric bypass surgery. His functional capacity is low, but he is able to do his activities of daily living. He reports having dyspnea on exertion and intermittently at rest, but no chest pain. His medications include insulin, atorvastatin (Lipitor), aspirin, and atenolol (Tenormin). He is afebrile; his blood pressure is 130/80 mm Hg, pulse 75, and oxygen saturation 97% on room air. His baseline electrocardiogram shows no Q waves.

5. Which of the following is an appropriate next step before proceeding to surgery?

• Echocardiography
• Cardiac catheterization
• Dobutamine stress echocardiography or adenosine thallium scanning
• No cardiac testing is necessary before surgery

Wijeysundera et al⁷ performed a retrospective cohort study of patients who underwent elective surgery at acute care hospitals in Ontario, Canada, in the years 1994 through 2004. The aim was to determine the association of noninvasive cardiac stress testing before surgery with survival rates and length of hospital stay. Included were 271,082 patients, of whom 23,991 (8.9%) underwent stress testing less than 6 months before surgery. These patients were matched with 46,120 who did not undergo testing.

One year after surgery, fewer patients who underwent stress testing had died: 1,622 (7.0%) vs 1,738 (7.5%); hazard ratio 0.92, 95% CI 0.86–0.99, P = .03. The number needed to treat (ie, to be tested) to prevent one death was 221. The tested patients also had a shorter mean hospital stay: 8.72 vs 8.96 days, a difference of 0.24 days (95% CI −0.07 to −0.43; P < .001).

However, the elderly patients (ie, older than 66 years) who underwent testing were more likely to be on beta-blockers and statins than those who did not undergo testing, which may be a confounding factor.

Furthermore, the benefit was all in the patients at intermediate or high risk. The authors performed a subgroup analysis, dividing the patients on the basis of their Revised Cardiac Risk Index (RCRI; 1 point each for ischemic heart disease, congestive heart failure, cerebrovascular disease, diabetes, renal insufficiency, and high-risk surgery).⁸ Patients with an RCRI of 0 points (indicating low risk) actually had a higher risk of death with testing than without testing: hazard ratio 1.35 (95% CI 1.03–1.74), number needed to harm 179—ie, for every 179 low-risk patients tested, one excess death occurred. Those with an RCRI of 1 or 2 points (indicating intermediate risk) had a hazard ratio of 0.92 with testing (95% CI 085–0.99), and those with an RCRI of 3 to 6 points (indicating high risk) had a hazard ratio of 0.80 with testing (95% CI 0.67- 0.97; number needed to treat = 38).

Comment. These findings indicate that cardiac stress testing should be done selectively before noncardiac surgery, and primarily for patients at high risk (with an RCRI of 3 or higher) and in some patients at intermediate risk, but not in patients at low risk, in whom it may be harmful. Stress testing may change patient management because a positive stress test allows one to start a beta-blocker or a statin, use more aggressive intraoperative and postoperative care, and identify patients who have indications for revascularization.

A number of studies published in the last few years will likely affect the way we practice medicine in the hospital. Here, we will use a hypothetical case scenario to focus on the issues of anticoagulants, patient safety, quality improvement, critical care, transitions of care, and perioperative medicine.

AN ELDERLY MAN WITH NEW-ONSET ATRIAL FIBRILLATION

P.G. is an 80-year-old man with a history of hypertension and type 2 diabetes mellitus who is admitted with new-onset atrial fibrillation. In the hospital, his heart rate is brought under control with intravenous metoprolol (Lopressor). On discharge, he will be followed by his primary care physician (PCP). He does not have access to an anticoagulation clinic.

1. What are this patient’s options for stroke prevention?

• Aspirin 81 mg daily and clopidogrel (Plavix) 75 mg daily
• Warfarin (Coumadin) with a target international normalized ratio (INR) of 2.0 to 3.0
• Aspirin mg daily by itself
• Dabigatran (Pradaxa) 150 mg daily

A new oral anticoagulant agent

In deciding what type of anticoagulation to give to a patient with atrial fibrillation, it is useful to look at the CHADS₂ score (1 point each for congestive heart failure, hypertension, age 75 or older, and diabetes mellitus; 2 points for prior stroke or transient ischemic attack. This patient has a CHADS₂ score of 3, indicating that he should receive warfarin. An alternative is dabigatran, the first new anticoagulant agent in more than 50 years.

In a multicenter, international trial, Connolly et al¹ randomized 18,113 patients (mean age 71, 64% men) to receive dabigatran 110 mg twice daily, dabigatran 150 mg twice daily, or warfarin with a target INR of 2.0 to 3.0. In this noninferiority trial, dabigatran was given in a blinded manner, but the use of warfarin was open-label. Patients were eligible if they had atrial fibrillation at screening or within the previous 6 months and were at risk of stroke—ie, if they had at least one of the following: a history of stroke or transient ischemic attack, a left ventricular ejection fraction of less than 40%, symptoms of congestive heart failure (New York Heart Association class II or higher), and an age of 75 or older or an age of 65 to 74 with diabetes mellitus, hypertension, or coronary artery disease.

At a mean follow-up of 2 years, the rate of stroke or systolic embolism was 1.69% per year in the warfarin group compared with 1.1% in the higher-dose dabigatran group (relative risk 0.66, 95% confidence interval [CI] 0.53–0.82, P < .001). The rates of major hemorrhage were similar between these two groups. Comparing lower-dose dabigatran and warfarin, the rates of stroke or systolic embolism were not significantly different, but the rate of major bleeding was significantly lower with lower-dose dabigatran.

In a trial in patients with acute venous thromboembolism, Schulman et al² found that dabigatran was not inferior to warfarin in preventing venous thromboembolism.

Guidelines from the American College of Cardiology Foundation and the American Heart Association now endorse dabigatran as an alternative to warfarin for patients with atrial fibrillation.³ However, the guidelines state that it should be reserved for those patients who:

• Do not have a prosthetic heart valve or hemodynamically significant valve disease
• Have good kidney function (dabigatran is cleared by the kidney; the creatinine clearance rate should be greater than 30 mL/min for patients to receive dabigatran 150 mg twice a day, and at least 15 mL/min to receive 75 mg twice a day)
• Do not have severe hepatic dysfunction (which would impair baseline clotting function).

They note that other factors to consider are whether the patient:

• Can comply with the twice-daily dosing required
• Can afford the drug
• Has access to an anticoagulation management program (which would argue in favor of using warfarin).

Dabigatran is not yet approved to prevent venous thromboembolism.

CASE CONTINUED: HE GETS AN INFECTION

P.G. is started on dabigatran 150 mg by mouth twice a day.

While in the hospital he develops shortness of breath and needs intravenous furosemide (Lasix). Because he has bad veins, a percutaneous intravenous central catheter (PICC) line is placed. However, 2 days later, his temperature is 101.5°F, and his systolic blood pressure is 70 mm Hg. He is transferred to the medical intensive care unit (ICU) for treatment of sepsis. The anticoagulant is held, the PICC line is removed, and a new central catheter is inserted.

2. Which of the following directions is incorrect?

• Wash your hands before inserting the catheter. The accompanying nurse is required to directly observe this procedure or, if this step is not observed, to confirm that the physician did it.
• Before inserting the catheter, clean the patient’s skin with chlorhexidine antiseptic.
• Place sterile drapes over the entire patient.
• Wear any mask, hat, gown, and gloves available.
• Put a sterile dressing over the catheter.

A checklist can prevent infections when inserting central catheters

A checklist developed at Johns Hopkins Hospital consists of the five statements above, except for the second to last one—you should wear a sterile mask, hat, gown and gloves. This is important to ensure that sterility is not broken at any point during the procedure.

Pronovost et al⁴ launched a multicenter initiative at 90 ICUs, predominantly in the state of Michigan, to implement interventions to improve staff culture and teamwork and to translate research into practice by increasing the extent to which these five evidence-based recommendations were applied. The mean rate of catheter-related blood stream infections at baseline was 7.7%; this dropped to 2.8% during the implementation period, 2.3% in the first 3 months after implementation, 1.3% in months 16 through 18, and 1.1% in months 34 through 36, demonstrating that the gains from this quality-improvement project were sustainable.

If this intervention and collaborative model were implemented in all ICUs across the United States and if similar success rates were achieved, substantial and sustained reductions could be made in the 82,000 infections, 28,000 deaths, and $2.3 billion in costs attributed to these infections annually.

CASE CONTINUED: HE IS RESUSCITATED

P.G. is started on a 1-L fluid bolus but he remains hypotensive, necessitating a norepinephrine drip. He does well for about 6 hours, but in the middle of the night he develops ventricular tachycardia and ventricular fibrillation, and a code is called. He is successfully resuscitated, but the family is looking for prognostic information.

3. What are P.G.’s chances of surviving and leaving the hospital?

• 5%
• 8%
• 15%
• 23%

A registry of cardiopulmonary resuscitation

Tian et al⁵ evaluated outcomes in the largest registry of cardiopulmonary resuscitation to date. In this analysis, 49,656 adult patients with a first cardiopulmonary arrest occurring in an ICU between January 1, 2000, and August 26, 2008, were evaluated for their outcomes on pressors vs those not on pressors.

Other independent predictors of a lower survival rate were nonwhite race, mechanical ventilation, having three or more immediate causes of cardiopulmonary arrest, age 65 years or older, and cardiopulmonary arrest occurring at night or over the weekend.

Fortunately, for our patient, survival rates were higher for patients with ventricular tachycardia or fibrillation than with other causes of cardiopulmonary arrest: 22.6% for those on pressors (like our patient) and 40.7% for those on no pressors.

CASE CONTINUED: HE RECOVERS AND GOES HOME

P.G. makes a remarkable recovery and is now ready to go home. It is the weekend, and you are unable to schedule a follow-up appointment before his discharge, so you ask him to make an appointment with his PCP.

4. What is the likelihood that P.G. will be readmitted within 1 month?

• 5%
• 12%
• 20%
• 25%
• 30%

The importance of follow-up with a primary care physician

Misky et al,⁶ in a small study, attempted to identify the characteristics and outcomes of discharged patients who lack timely follow-up with a PCP. They prospectively enrolled 65 patients admitted to University of Colorado Hospital, an urban 425-bed tertiary care center, collecting information about patient demographics, diagnosis, payer source, and PCPs. After discharge, they called the patients to determine their PCP follow-up and readmission status. Thirty-day readmission rates and hospital length of stay were compared in patients with and without timely PCP follow-up (ie, within 4 weeks).

Patients lacking timely PCP follow-up were 10 times more likely to be readmitted (odds ratio [OR] = 9.9, P = .04): the rate was 21% in patients lacking timely PCP follow-up vs 3% in patients with timely PCP follow-up, P = .03. Lack of insurance was associated with lower rates of timely PCP follow-up: 29% vs 56% (P = .06), but did not independently increase the readmission rate or length of stay (OR = 1.0, P = .96). Index hospital length of stay was longer in patients lacking timely PCP follow-up: 4.4 days vs 6.3 days, P = 0.11.

Comment. Nearly half of the patients in this study, who were discharged from a large urban academic center, lacked timely follow-up with a PCP, resulting in higher rates of readmission and a nonsignificant trend toward longer length of stay. Timely follow-up is necessary for vulnerable patients.

Since the lack of timely PCP follow-up results in higher readmission rates and possibly a longer length of stay, a PCP appointment at discharge should perhaps be considered a core quality measure. This would be problematic in our American health care system, in which many patients lack health insurance and do not have a PCP.

A MAN UNDERGOING GASTRIC BYPASS SURGERY

A 55-year-old morbidly obese man (body mass index 45 kg/m²) with a history of type 2 diabetes mellitus, chronic renal insufficiency (serum creatinine level 2.1 mg/dL), hypercholesterolemia, and previous stroke is scheduled for gastric bypass surgery. His functional capacity is low, but he is able to do his activities of daily living. He reports having dyspnea on exertion and intermittently at rest, but no chest pain. His medications include insulin, atorvastatin (Lipitor), aspirin, and atenolol (Tenormin). He is afebrile; his blood pressure is 130/80 mm Hg, pulse 75, and oxygen saturation 97% on room air. His baseline electrocardiogram shows no Q waves.

5. Which of the following is an appropriate next step before proceeding to surgery?

• Echocardiography
• Cardiac catheterization
• Dobutamine stress echocardiography or adenosine thallium scanning
• No cardiac testing is necessary before surgery

Wijeysundera et al⁷ performed a retrospective cohort study of patients who underwent elective surgery at acute care hospitals in Ontario, Canada, in the years 1994 through 2004. The aim was to determine the association of noninvasive cardiac stress testing before surgery with survival rates and length of hospital stay. Included were 271,082 patients, of whom 23,991 (8.9%) underwent stress testing less than 6 months before surgery. These patients were matched with 46,120 who did not undergo testing.

One year after surgery, fewer patients who underwent stress testing had died: 1,622 (7.0%) vs 1,738 (7.5%); hazard ratio 0.92, 95% CI 0.86–0.99, P = .03. The number needed to treat (ie, to be tested) to prevent one death was 221. The tested patients also had a shorter mean hospital stay: 8.72 vs 8.96 days, a difference of 0.24 days (95% CI −0.07 to −0.43; P < .001).

However, the elderly patients (ie, older than 66 years) who underwent testing were more likely to be on beta-blockers and statins than those who did not undergo testing, which may be a confounding factor.

Furthermore, the benefit was all in the patients at intermediate or high risk. The authors performed a subgroup analysis, dividing the patients on the basis of their Revised Cardiac Risk Index (RCRI; 1 point each for ischemic heart disease, congestive heart failure, cerebrovascular disease, diabetes, renal insufficiency, and high-risk surgery).⁸ Patients with an RCRI of 0 points (indicating low risk) actually had a higher risk of death with testing than without testing: hazard ratio 1.35 (95% CI 1.03–1.74), number needed to harm 179—ie, for every 179 low-risk patients tested, one excess death occurred. Those with an RCRI of 1 or 2 points (indicating intermediate risk) had a hazard ratio of 0.92 with testing (95% CI 085–0.99), and those with an RCRI of 3 to 6 points (indicating high risk) had a hazard ratio of 0.80 with testing (95% CI 0.67- 0.97; number needed to treat = 38).

Comment. These findings indicate that cardiac stress testing should be done selectively before noncardiac surgery, and primarily for patients at high risk (with an RCRI of 3 or higher) and in some patients at intermediate risk, but not in patients at low risk, in whom it may be harmful. Stress testing may change patient management because a positive stress test allows one to start a beta-blocker or a statin, use more aggressive intraoperative and postoperative care, and identify patients who have indications for revascularization.

A number of studies published in the last few years will likely affect the way we practice medicine in the hospital. Here, we will use a hypothetical case scenario to focus on the issues of anticoagulants, patient safety, quality improvement, critical care, transitions of care, and perioperative medicine.

AN ELDERLY MAN WITH NEW-ONSET ATRIAL FIBRILLATION

P.G. is an 80-year-old man with a history of hypertension and type 2 diabetes mellitus who is admitted with new-onset atrial fibrillation. In the hospital, his heart rate is brought under control with intravenous metoprolol (Lopressor). On discharge, he will be followed by his primary care physician (PCP). He does not have access to an anticoagulation clinic.

1. What are this patient’s options for stroke prevention?

• Aspirin 81 mg daily and clopidogrel (Plavix) 75 mg daily
• Warfarin (Coumadin) with a target international normalized ratio (INR) of 2.0 to 3.0
• Aspirin mg daily by itself
• Dabigatran (Pradaxa) 150 mg daily

A new oral anticoagulant agent

In deciding what type of anticoagulation to give to a patient with atrial fibrillation, it is useful to look at the CHADS₂ score (1 point each for congestive heart failure, hypertension, age 75 or older, and diabetes mellitus; 2 points for prior stroke or transient ischemic attack. This patient has a CHADS₂ score of 3, indicating that he should receive warfarin. An alternative is dabigatran, the first new anticoagulant agent in more than 50 years.

In a multicenter, international trial, Connolly et al¹ randomized 18,113 patients (mean age 71, 64% men) to receive dabigatran 110 mg twice daily, dabigatran 150 mg twice daily, or warfarin with a target INR of 2.0 to 3.0. In this noninferiority trial, dabigatran was given in a blinded manner, but the use of warfarin was open-label. Patients were eligible if they had atrial fibrillation at screening or within the previous 6 months and were at risk of stroke—ie, if they had at least one of the following: a history of stroke or transient ischemic attack, a left ventricular ejection fraction of less than 40%, symptoms of congestive heart failure (New York Heart Association class II or higher), and an age of 75 or older or an age of 65 to 74 with diabetes mellitus, hypertension, or coronary artery disease.

At a mean follow-up of 2 years, the rate of stroke or systolic embolism was 1.69% per year in the warfarin group compared with 1.1% in the higher-dose dabigatran group (relative risk 0.66, 95% confidence interval [CI] 0.53–0.82, P < .001). The rates of major hemorrhage were similar between these two groups. Comparing lower-dose dabigatran and warfarin, the rates of stroke or systolic embolism were not significantly different, but the rate of major bleeding was significantly lower with lower-dose dabigatran.

In a trial in patients with acute venous thromboembolism, Schulman et al² found that dabigatran was not inferior to warfarin in preventing venous thromboembolism.

Guidelines from the American College of Cardiology Foundation and the American Heart Association now endorse dabigatran as an alternative to warfarin for patients with atrial fibrillation.³ However, the guidelines state that it should be reserved for those patients who:

• Do not have a prosthetic heart valve or hemodynamically significant valve disease
• Have good kidney function (dabigatran is cleared by the kidney; the creatinine clearance rate should be greater than 30 mL/min for patients to receive dabigatran 150 mg twice a day, and at least 15 mL/min to receive 75 mg twice a day)
• Do not have severe hepatic dysfunction (which would impair baseline clotting function).

They note that other factors to consider are whether the patient:

• Can comply with the twice-daily dosing required
• Can afford the drug
• Has access to an anticoagulation management program (which would argue in favor of using warfarin).

Dabigatran is not yet approved to prevent venous thromboembolism.

CASE CONTINUED: HE GETS AN INFECTION

P.G. is started on dabigatran 150 mg by mouth twice a day.

While in the hospital he develops shortness of breath and needs intravenous furosemide (Lasix). Because he has bad veins, a percutaneous intravenous central catheter (PICC) line is placed. However, 2 days later, his temperature is 101.5°F, and his systolic blood pressure is 70 mm Hg. He is transferred to the medical intensive care unit (ICU) for treatment of sepsis. The anticoagulant is held, the PICC line is removed, and a new central catheter is inserted.

2. Which of the following directions is incorrect?

• Wash your hands before inserting the catheter. The accompanying nurse is required to directly observe this procedure or, if this step is not observed, to confirm that the physician did it.
• Before inserting the catheter, clean the patient’s skin with chlorhexidine antiseptic.
• Place sterile drapes over the entire patient.
• Wear any mask, hat, gown, and gloves available.
• Put a sterile dressing over the catheter.

A checklist can prevent infections when inserting central catheters

A checklist developed at Johns Hopkins Hospital consists of the five statements above, except for the second to last one—you should wear a sterile mask, hat, gown and gloves. This is important to ensure that sterility is not broken at any point during the procedure.

Pronovost et al⁴ launched a multicenter initiative at 90 ICUs, predominantly in the state of Michigan, to implement interventions to improve staff culture and teamwork and to translate research into practice by increasing the extent to which these five evidence-based recommendations were applied. The mean rate of catheter-related blood stream infections at baseline was 7.7%; this dropped to 2.8% during the implementation period, 2.3% in the first 3 months after implementation, 1.3% in months 16 through 18, and 1.1% in months 34 through 36, demonstrating that the gains from this quality-improvement project were sustainable.

If this intervention and collaborative model were implemented in all ICUs across the United States and if similar success rates were achieved, substantial and sustained reductions could be made in the 82,000 infections, 28,000 deaths, and $2.3 billion in costs attributed to these infections annually.

CASE CONTINUED: HE IS RESUSCITATED

P.G. is started on a 1-L fluid bolus but he remains hypotensive, necessitating a norepinephrine drip. He does well for about 6 hours, but in the middle of the night he develops ventricular tachycardia and ventricular fibrillation, and a code is called. He is successfully resuscitated, but the family is looking for prognostic information.

3. What are P.G.’s chances of surviving and leaving the hospital?

• 5%
• 8%
• 15%
• 23%

A registry of cardiopulmonary resuscitation

Tian et al⁵ evaluated outcomes in the largest registry of cardiopulmonary resuscitation to date. In this analysis, 49,656 adult patients with a first cardiopulmonary arrest occurring in an ICU between January 1, 2000, and August 26, 2008, were evaluated for their outcomes on pressors vs those not on pressors.

Other independent predictors of a lower survival rate were nonwhite race, mechanical ventilation, having three or more immediate causes of cardiopulmonary arrest, age 65 years or older, and cardiopulmonary arrest occurring at night or over the weekend.

Fortunately, for our patient, survival rates were higher for patients with ventricular tachycardia or fibrillation than with other causes of cardiopulmonary arrest: 22.6% for those on pressors (like our patient) and 40.7% for those on no pressors.

CASE CONTINUED: HE RECOVERS AND GOES HOME

P.G. makes a remarkable recovery and is now ready to go home. It is the weekend, and you are unable to schedule a follow-up appointment before his discharge, so you ask him to make an appointment with his PCP.

4. What is the likelihood that P.G. will be readmitted within 1 month?

• 5%
• 12%
• 20%
• 25%
• 30%

The importance of follow-up with a primary care physician

Misky et al,⁶ in a small study, attempted to identify the characteristics and outcomes of discharged patients who lack timely follow-up with a PCP. They prospectively enrolled 65 patients admitted to University of Colorado Hospital, an urban 425-bed tertiary care center, collecting information about patient demographics, diagnosis, payer source, and PCPs. After discharge, they called the patients to determine their PCP follow-up and readmission status. Thirty-day readmission rates and hospital length of stay were compared in patients with and without timely PCP follow-up (ie, within 4 weeks).

Patients lacking timely PCP follow-up were 10 times more likely to be readmitted (odds ratio [OR] = 9.9, P = .04): the rate was 21% in patients lacking timely PCP follow-up vs 3% in patients with timely PCP follow-up, P = .03. Lack of insurance was associated with lower rates of timely PCP follow-up: 29% vs 56% (P = .06), but did not independently increase the readmission rate or length of stay (OR = 1.0, P = .96). Index hospital length of stay was longer in patients lacking timely PCP follow-up: 4.4 days vs 6.3 days, P = 0.11.

Comment. Nearly half of the patients in this study, who were discharged from a large urban academic center, lacked timely follow-up with a PCP, resulting in higher rates of readmission and a nonsignificant trend toward longer length of stay. Timely follow-up is necessary for vulnerable patients.

Since the lack of timely PCP follow-up results in higher readmission rates and possibly a longer length of stay, a PCP appointment at discharge should perhaps be considered a core quality measure. This would be problematic in our American health care system, in which many patients lack health insurance and do not have a PCP.

A MAN UNDERGOING GASTRIC BYPASS SURGERY

A 55-year-old morbidly obese man (body mass index 45 kg/m²) with a history of type 2 diabetes mellitus, chronic renal insufficiency (serum creatinine level 2.1 mg/dL), hypercholesterolemia, and previous stroke is scheduled for gastric bypass surgery. His functional capacity is low, but he is able to do his activities of daily living. He reports having dyspnea on exertion and intermittently at rest, but no chest pain. His medications include insulin, atorvastatin (Lipitor), aspirin, and atenolol (Tenormin). He is afebrile; his blood pressure is 130/80 mm Hg, pulse 75, and oxygen saturation 97% on room air. His baseline electrocardiogram shows no Q waves.

5. Which of the following is an appropriate next step before proceeding to surgery?

• Echocardiography
• Cardiac catheterization
• Dobutamine stress echocardiography or adenosine thallium scanning
• No cardiac testing is necessary before surgery

Wijeysundera et al⁷ performed a retrospective cohort study of patients who underwent elective surgery at acute care hospitals in Ontario, Canada, in the years 1994 through 2004. The aim was to determine the association of noninvasive cardiac stress testing before surgery with survival rates and length of hospital stay. Included were 271,082 patients, of whom 23,991 (8.9%) underwent stress testing less than 6 months before surgery. These patients were matched with 46,120 who did not undergo testing.

One year after surgery, fewer patients who underwent stress testing had died: 1,622 (7.0%) vs 1,738 (7.5%); hazard ratio 0.92, 95% CI 0.86–0.99, P = .03. The number needed to treat (ie, to be tested) to prevent one death was 221. The tested patients also had a shorter mean hospital stay: 8.72 vs 8.96 days, a difference of 0.24 days (95% CI −0.07 to −0.43; P < .001).

However, the elderly patients (ie, older than 66 years) who underwent testing were more likely to be on beta-blockers and statins than those who did not undergo testing, which may be a confounding factor.

Furthermore, the benefit was all in the patients at intermediate or high risk. The authors performed a subgroup analysis, dividing the patients on the basis of their Revised Cardiac Risk Index (RCRI; 1 point each for ischemic heart disease, congestive heart failure, cerebrovascular disease, diabetes, renal insufficiency, and high-risk surgery).⁸ Patients with an RCRI of 0 points (indicating low risk) actually had a higher risk of death with testing than without testing: hazard ratio 1.35 (95% CI 1.03–1.74), number needed to harm 179—ie, for every 179 low-risk patients tested, one excess death occurred. Those with an RCRI of 1 or 2 points (indicating intermediate risk) had a hazard ratio of 0.92 with testing (95% CI 085–0.99), and those with an RCRI of 3 to 6 points (indicating high risk) had a hazard ratio of 0.80 with testing (95% CI 0.67- 0.97; number needed to treat = 38).

Comment. These findings indicate that cardiac stress testing should be done selectively before noncardiac surgery, and primarily for patients at high risk (with an RCRI of 3 or higher) and in some patients at intermediate risk, but not in patients at low risk, in whom it may be harmful. Stress testing may change patient management because a positive stress test allows one to start a beta-blocker or a statin, use more aggressive intraoperative and postoperative care, and identify patients who have indications for revascularization.

Hyperglycemia and diabetes mellitus are very common in hospitalized patients. Although more data are available on the prevalence of this problem and on how to manage it in the intensive care unit (ICU) than on regular hospital floors, the situation is changing. Information is emerging on the prevalence and impact of hyperglycemia and diabetes in the non-ICU setting, which is the focus of this paper.

HYPERGLYCEMIA IS COMMON AND PREDICTS POOR OUTCOMES

Cook et al,¹ in a survey of 126 US hospitals, found that the prevalence of hyperglycemia (blood glucose > 180 mg/dL) was 46% in the ICU and 32% in regular wards.

Kosiborod et al² reported that hyperglycemia (blood glucose > 140 mg/dL) was present in 78% of diabetic patients hospitalized with acute coronary syndrome and 26% of similar hospitalized nondiabetic patients.

Hyperglycemia is a common comorbidity in medical-surgical patients in community hospitals. Our group³ found that, in our hospital, 62% of patients were normoglycemic (ie, had a fasting blood glucose < 126 mg/dL or a random blood glucose < 200 mg/dL on two occasions), 26% had known diabetes, and 12% had new hyperglycemia. Further, new hyperglycemia was associated with a higher in-hospital death rate than the other two conditions.

Failure to identify diabetes is a predictor of rehospitalization. Robbins and Webb⁴ reported that 30.6% of those who had diabetes that was missed during hospitalization were readmitted within 30 days, compared with 9.4% of patients with diabetes first diagnosed during hospitalization.

WHAT DIAGNOSTIC CRITERIA SHOULD WE USE?

Blood glucose greater than 140 mg/dL

A consensus statement from the American Association of Clinical Endocrinologists (ACE) and the American Diabetes Association (ADA)⁵ defines in-hospital hyperglycemia as a blood glucose level greater than 140 mg/dL on admission or in the hospital. If the blood glucose is higher than this, the question arises as to whether the patient has preexisting diabetes or has stress hyperglycemia.

Hemoglobin A_1c of 6.5% or higher

In view of the uncertainty as to whether a patient with an elevated blood glucose level has preexisting diabetes or stress hyperglycemia, upcoming guidelines will recommend measuring the hemoglobin A_1c level if the blood glucose level is higher than 140 mg/dL.

A patient with an elevated blood glucose level (>140 mg/dL) whose hemoglobin A_1c level is 6.5% or higher can be identified as having diabetes that preceded the hospitalization. Hemoglobin A_1c testing can also be useful to assess glycemic control before admission and in designing an optional regimen at the time of discharge. In patients with newly recognized hyperglycemia, a hemoglobin A_1c measurement can help differentiate patients with previously undiagnosed diabetes from those with stress-induced hyperglycemia.

Clinicians should keep in mind that a hemoglobin A_1c cutoff of 6.5% identifies fewer cases of undiagnosed diabetes than does a high fasting glucose concentration, and that a level less than 6.5% does not rule out the diagnosis of diabetes. Several epidemiologic studies⁶ have reported a low sensitivity (44% to 66%) but a high specificity (76% to 99%) for hemoglobin A_1c values higher than 6.5% in an outpatient population. The high specificity therefore supports the use of hemoglobin A_1c to confirm the diagnosis of diabetes in patients with hyperglycemia, but the low sensitivity indicates that this test should not be used for universal screening in the hospital.

Many factors can influence the hemoglobin A_1c level, such as anemia, iron deficiency, blood transfusions, hemolytic anemia, and renal failure.

Until now, if patients had hyperglycemia but no prior diagnosis of diabetes, the recommendation was for an oral 2-hour glucose tolerance test shortly after discharge to confirm the diagnosis of diabetes. Norhammar et al⁷ performed oral glucose tolerance tests in patients admitted with acute myocardial infarction, and Matz et al⁸ performed glucose tolerance tests in patients with acute stroke. They found that impaired glucose tolerance and undiagnosed type 2 diabetes were very common in these two groups. However, physicians rarely order oral glucose tolerance tests. We believe that hemoglobin A_1c will be a better tool than an oral glucose tolerance test to confirm diabetes in hyperglycemic patients in the hospital setting.

WHAT IS THE ASSOCIATION BETWEEN HYPERGLYCEMIA AND OUTCOMES?

In 2,471 patients admitted to the hospital with community-acquired pneumonia, McAlister et al¹⁰ found that the rates of hospital complications and of death rose with blood glucose levels.

Falguera et al¹¹ found that, in 660 episodes of community-acquired pneumonia, the rates of hospitalization, death, pleural effusion, and concomitant illnesses were all significantly higher in diabetic patients than in nondiabetic patients.

Noordzij et al¹² performed a case-control study of 108,593 patients who underwent noncardiac surgery. The odds ratio for perioperative death was 1.19 (95% confidence interval [CI] 1.1–1.3) for every 1-mmol/L increase in the glucose level.

Frisch et al,¹³ in patients undergoing noncardiac surgery, found that the 30-day rates of death and of in-hospital complications were all higher in patients with diabetes than without diabetes.

Our group³ identified hyperglycemia as an independent marker of in-hospital death in patients with undiagnosed diabetes. The rates of death were 1.7% in those with normoglycemia, 3.0% in those with known diabetes, and 16.0% (P < .01) in those with new hyperglycemia.

The ACE/ADA consensus panel¹⁴ set the following glucose targets for patients in the non-ICU setting:

• Pre-meal blood glucose < 140 mg/dL
• Random blood glucose < 180 mg/dL.

On the other hand, hypoglycemia is also associated with adverse outcomes. Therefore, to avoid hypoglycemia, the insulin regimen should be reassessed if blood glucose levels fall below 100 mg/dL. New guidelines will suggest keeping the blood glucose between 100 and 140 mg/dL.

HOW SHOULD WE MANAGE HYPERGLYCEMIA IN THE NON-ICU SETTING?

The ACE/ADA guidelines recommend subcutaneous insulin therapy for most medical-surgical patients with diabetes, reserving intravenous insulin therapy for hyperglycemic crises and uncontrolled hyperglycemia.¹⁴

Oral antidiabetic agents are not generally recommended, as we have no data to support their use in the hospital. Another argument against using noninsulin therapies in the hospital is that sulfonylureas, especially glyburide (Diabeta, Micronase) are a major cause of hypoglycemia. Metformin (Glucophage) is contraindicated in decreased renal function, in hemodynamic instability, in surgical patients, and with the use of iodinated contrast dye. Thiazolidinediones are associated with edema and congestive heart failure, and they take up to 12 weeks to lower blood glucose levels. Alpha-glucosidase inhibitors are weak glucose-lowering agents. Also, therapies directed at glucagon-like-protein 1 can cause nausea and have a greater effect on postprandial glucose.¹⁴

The two main options for managing hyperglycemia and diabetes in the non-ICU setting are short-acting insulin on a sliding scale and basal-bolus therapy, the latter with either NPH plus regular insulin or long-acting plus rapid-acting insulin analogues.

Basal-bolus vs sliding scale insulin: The RABBIT-2 trial

In the RABBIT 2 trial (Randomized Basal Bolus Versus Sliding Scale Regular Insulin in Patients With Type 2 Diabetes Mellitus),¹⁵ our group compared the efficacy and safety of a basal-bolus regimen and a sliding-scale regimen in 130 hospitalized patients with type 2 diabetes treated with diet, with oral hypoglycemic agents, or with both. Oral antidiabetic drugs were discontinued on admission, and patients were randomized to one of the treatment groups.

In the basal-bolus group, the starting total daily dose was 0.4 U/kg/day if the blood glucose level on admission was between 140 and 200 mg/dL, or 0.5 U/kg/day if the glucose level was between 201 and 400 mg/dL. Half of the total daily dose was given as insulin glargine (Lantus) once daily, and the other half was given as insulin glulisine (Apidra) before meals. These doses were adjusted if the patient’s fasting or pre-meal blood glucose levels rose above 140 mg/dL or fell below 70 mg/dL.

The sliding-scale group received regular insulin four times daily (before meals and at bedtime) for glucose levels higher than 140 mg/dL; the higher the level, the more they got.

The basal-bolus regimen was better than sliding-scale regular insulin. At admission, the mean glucose values and hemoglobin A_1c values were similar in both groups, but the mean glucose level on therapy was significantly lower in the basal-bolus group than in the sliding-scale group, 166 ± 32 mg/dL vs 193 ± 54 mg/dL, P < .001). About two-thirds of the basal-bolus group achieved a blood glucose target of less than 140 mg/dL, compared with only about one-third of the sliding-scale group. The basal-bolus group received more insulin, a mean of 42 units per day vs 12.5 units per day in the sliding-scale group. Yet the incidence of hypoglycemia was 3% in both groups.

NPH plus regular vs detemir plus aspart: The DEAN trial

Several long-acting insulin analogues are available and have a longer duration of action than NPH. Similarly, several newer rapid-acting analogues act more rapidly than regular insulin. Do these pharmacokinetic advantages matter? And do they justify the higher costs of the newer agents?

In the randomized Insulin Detemir Versus NPH Insulin in Hospitalized Patients With Diabetes (DEAN) trial,¹⁶ we compared two regimens: detemir plus aspart in a basal-bolus regimen, and NPH plus regular insulin in two divided doses, two-thirds of the total daily dose in the morning before breakfast and one-third before dinner, both doses in a ratio of two-thirds NPH and one-third regular, mixed in the same syringe. We recruited 130 patients with type 2 diabetes mellitus who were on oral hypoglycemic agents or insulin therapy.

NPH plus regular was just as good as detemir plus aspart in improving glycemic control. Blood glucose levels fell during the first day of therapy and were similar in both groups throughout the trial, as measured before breakfast, lunch, and dinner and at bedtime. The mean total daily insulin dose was not significantly different between treatment groups: 56 ± 45 units in the basal-bolus detemir-aspart group and 45 ± 32 units in the NPH-regular group. However, the basal-bolus group received significantly more short-acting insulin: 27 ± 20 units a day of aspart vs 18 ± 14 units of regular.

Somewhat fewer patients in the NPH-regular group had episodes of hypoglycemia, although the difference between groups was not statistically significant.

In a univariate analysis of the RABBIT-2 and DEAN trials,¹⁷ factors that predicted a blood glucose level less than 60 mg/dL were older age, lower body weight, higher serum creatinine level, and previous insulin therapy. Factors that were not predictive were the hemoglobin A_1c level and the enrollment blood glucose level. Based on these data, we believe that to reduce the rate of hypoglycemia, lower insulin doses are needed in elderly patients and patients with renal impairment, and that if patients have been taking insulin before they come to the hospital, the dose should be cut back by about 25% while they are hospitalized.

Does better glucose control in surgical patients affect outcomes in patients undergoing general surgery? To find out, we performed a prospective, multicenter, randomized, open-label trial in general surgery patients not in the ICU.¹⁸ We recruited and randomized 211 patients with type 2 diabetes who were on diet therapy or oral hypoglycemic agents or insulin in low doses (< 0.4 U/kg/day).

Oral drugs were discontinued on admission, and patients were randomized to receive either a basal-bolus regimen of glargine plus glulisine or regular insulin on a sliding scale. The basal-bolus group got 0.5 U/kg/day, half of it as glargine once daily and half as glulisine before meals. The total daily dose was reduced to 0.3 U/kg/day in patients age 70 and older or who had a serum creatinine level of 2.0 mg/dL or higher.

The goal was to maintain fasting and pre-meal glucose concentrations between 100 and 140 mg/dL. The total daily dose was raised by 10% (mostly in the glargine dose) if the blood glucose level was in the range of 141 to 180 mg/dL, and by 20% if the glucose level was higher than 181 mg/dL. The dose was decreased by 10% for glucose levels between 70 and 99 mg/dL, was decreased by 20% if the glucose level was between 40 and 69, and was held if the glucose level was lower than 40 mg/dL. If a patient was not able to eat, insulin glulisine was held until meals were resumed.

The sliding-scale group received regular insulin four times a day for blood glucose levels higher than 140 mg/dL.

The primary outcomes measured were the difference between groups in mean daily blood glucose concentration and a composite of hospital complications including postoperative wound infection, pneumonia, respiratory failure, acute renal failure, and bacteremia. Secondary outcomes were differences between groups in mean fasting and pre-meal blood glucose, number of hypoglycemic episodes (blood glucose < 70 mg/dL), hyperglycemic episodes (blood glucose > 200 mg/dL), length of hospital stay, need for intensive care, and rate of complications including wound infection, pneumonia, acute renal failure, and death.

Blood glucose levels were significantly lower in the basal-bolus group through the first 7 days after randomization, as measured before breakfast, lunch, and dinner, and at bedtime, and then they converged.

More patients in the sliding-scale group had hospital complications, 26 vs 9, P = .003. On the other hand, more patients in the basal-bolus group had episodes of hypoglycemia: 24 (23%) vs 5 (4.7%) had episodes of less than 70 mg/dL (P < .001), 12 (12%) vs 2 (1.9%) had episodes of less than 60 mg/dL (P = .005), and 4 (3.8%) vs 0 had episodes of less than 40 mg/dL (P = .057). The mean total daily dose of insulin was 33.4 units in the basal-bolus group and 12.3 units in the sliding-scale group.

WHAT HAVE WE LEARNED?

Don’t use a sliding-scale regimen as a single agent in patients with diabetes. Glycemic control is better with a basal-bolus regimen than with a sliding-scale regimen, and a basal-bolus insulin regimen is preferred for most patients with hyperglycemia.

The old human insulins (ie, regular and NPH) are still good and improve glycemic control as well as the new basal insulin analogues (detemir and aspart) do.

Improved control may reduce the rate of hospital complications, according to preliminary evidence. More studies are under way.

One size does not fit all. Those who are elderly or who have impaired renal function should receive lower doses of insulin, eg, 0.3 U/kg/day instead of 0.5 U/kg/day. Those who are on insulin should have their dose decreased when they are admitted to the hospital. Perhaps lean patients with type 2 diabetes should also have a lower dose.

Most hospitalized patients with diabetes and elevated blood glucose values (or hyperglycemia) should receive subcutaneous insulin treatment with a basal-bolus regimen or a multidose combination of NPH plus regular insulin. Selected patients with severe insulin resistance and persistent hyperglycemia despite subcutaneous insulin may benefit from continuous intravenous insulin infusion.

Patients treated with insulin at home should continue to receive insulin therapy in the hospital. However, the insulin dosage should be reduced by about 25% to allow for lower food intake.

QUESTIONS FOR FURTHER STUDY

Should we modify the standard basal-bolus regimen?

In a typical basal-bolus regimen, patients get 50% of their total daily insulin dose in the form of a basal injection and 50% in the form of rapid-acting boluses before meals. However, for a variety of reasons, hospitalized patients do not eat very much. Thus, a 50-50 basal-bolus regimen may not be ideal for patients with poor oral intake.

In the Basal-PLUS trial, currently under way, we are comparing the safety and efficacy of a daily dose of basal insulin (glargine) plus correction doses of a rapid-acting insulin analogue (glulisine) on a sliding scale and a standard basal-bolus regimen in medical and surgical patients.

Does one glycemic target fit all patients?

Falciglia et al¹⁹ found an association between hyperglycemia and death in patients with unstable angina, arrhythmias, stroke, pneumonia, gastrointestinal bleeding, respiratory failure, sepsis, acute renal failure, and congestive heart failure. However, they found no such association in patients with chronic obstructive pulmonary disease, liver failure, diabetic ketoacidosis, gastrointestinal neoplasm, musculoskeletal disease, peripheral vascular disease with bypass, hip fracture, amputation due to peripheral vascular disease, or prostate surgery. Should patients in this second group be treated with a less-intensive insulin regimen?

What is the best regimen after hospital discharge?

We are conducting a prospective clinical trial to assess the impact of insulin after hospital discharge. Our current practice when a patient is discharged from the hospital is as follows:

• If the admission hemoglobin A_1c level is less than 7%, we restart the previous outpatient treatment regimen of oral antidiabetic agents, or insulin, or both.
• If the admission hemoglobin A_1c is between 7% and 9%, we restart the outpatient oral agents and continue glargine once daily at 50% to 80% of the hospital dose.
• If the hemoglobin A_1c level is higher than 9%, we discharge the patient on a basal-bolus regimen at the same dosage as in the hospital. As an alternative, we could restart the oral agents and add glargine once daily at 80% of the hospital dose.

Hyperglycemia and diabetes mellitus are very common in hospitalized patients. Although more data are available on the prevalence of this problem and on how to manage it in the intensive care unit (ICU) than on regular hospital floors, the situation is changing. Information is emerging on the prevalence and impact of hyperglycemia and diabetes in the non-ICU setting, which is the focus of this paper.

HYPERGLYCEMIA IS COMMON AND PREDICTS POOR OUTCOMES

Cook et al,¹ in a survey of 126 US hospitals, found that the prevalence of hyperglycemia (blood glucose > 180 mg/dL) was 46% in the ICU and 32% in regular wards.

Kosiborod et al² reported that hyperglycemia (blood glucose > 140 mg/dL) was present in 78% of diabetic patients hospitalized with acute coronary syndrome and 26% of similar hospitalized nondiabetic patients.

Hyperglycemia is a common comorbidity in medical-surgical patients in community hospitals. Our group³ found that, in our hospital, 62% of patients were normoglycemic (ie, had a fasting blood glucose < 126 mg/dL or a random blood glucose < 200 mg/dL on two occasions), 26% had known diabetes, and 12% had new hyperglycemia. Further, new hyperglycemia was associated with a higher in-hospital death rate than the other two conditions.

Failure to identify diabetes is a predictor of rehospitalization. Robbins and Webb⁴ reported that 30.6% of those who had diabetes that was missed during hospitalization were readmitted within 30 days, compared with 9.4% of patients with diabetes first diagnosed during hospitalization.

WHAT DIAGNOSTIC CRITERIA SHOULD WE USE?

Blood glucose greater than 140 mg/dL

A consensus statement from the American Association of Clinical Endocrinologists (ACE) and the American Diabetes Association (ADA)⁵ defines in-hospital hyperglycemia as a blood glucose level greater than 140 mg/dL on admission or in the hospital. If the blood glucose is higher than this, the question arises as to whether the patient has preexisting diabetes or has stress hyperglycemia.

Hemoglobin A_1c of 6.5% or higher

In view of the uncertainty as to whether a patient with an elevated blood glucose level has preexisting diabetes or stress hyperglycemia, upcoming guidelines will recommend measuring the hemoglobin A_1c level if the blood glucose level is higher than 140 mg/dL.

A patient with an elevated blood glucose level (>140 mg/dL) whose hemoglobin A_1c level is 6.5% or higher can be identified as having diabetes that preceded the hospitalization. Hemoglobin A_1c testing can also be useful to assess glycemic control before admission and in designing an optional regimen at the time of discharge. In patients with newly recognized hyperglycemia, a hemoglobin A_1c measurement can help differentiate patients with previously undiagnosed diabetes from those with stress-induced hyperglycemia.

Clinicians should keep in mind that a hemoglobin A_1c cutoff of 6.5% identifies fewer cases of undiagnosed diabetes than does a high fasting glucose concentration, and that a level less than 6.5% does not rule out the diagnosis of diabetes. Several epidemiologic studies⁶ have reported a low sensitivity (44% to 66%) but a high specificity (76% to 99%) for hemoglobin A_1c values higher than 6.5% in an outpatient population. The high specificity therefore supports the use of hemoglobin A_1c to confirm the diagnosis of diabetes in patients with hyperglycemia, but the low sensitivity indicates that this test should not be used for universal screening in the hospital.

Many factors can influence the hemoglobin A_1c level, such as anemia, iron deficiency, blood transfusions, hemolytic anemia, and renal failure.

Until now, if patients had hyperglycemia but no prior diagnosis of diabetes, the recommendation was for an oral 2-hour glucose tolerance test shortly after discharge to confirm the diagnosis of diabetes. Norhammar et al⁷ performed oral glucose tolerance tests in patients admitted with acute myocardial infarction, and Matz et al⁸ performed glucose tolerance tests in patients with acute stroke. They found that impaired glucose tolerance and undiagnosed type 2 diabetes were very common in these two groups. However, physicians rarely order oral glucose tolerance tests. We believe that hemoglobin A_1c will be a better tool than an oral glucose tolerance test to confirm diabetes in hyperglycemic patients in the hospital setting.

WHAT IS THE ASSOCIATION BETWEEN HYPERGLYCEMIA AND OUTCOMES?

In 2,471 patients admitted to the hospital with community-acquired pneumonia, McAlister et al¹⁰ found that the rates of hospital complications and of death rose with blood glucose levels.

Falguera et al¹¹ found that, in 660 episodes of community-acquired pneumonia, the rates of hospitalization, death, pleural effusion, and concomitant illnesses were all significantly higher in diabetic patients than in nondiabetic patients.

Noordzij et al¹² performed a case-control study of 108,593 patients who underwent noncardiac surgery. The odds ratio for perioperative death was 1.19 (95% confidence interval [CI] 1.1–1.3) for every 1-mmol/L increase in the glucose level.

Frisch et al,¹³ in patients undergoing noncardiac surgery, found that the 30-day rates of death and of in-hospital complications were all higher in patients with diabetes than without diabetes.

Our group³ identified hyperglycemia as an independent marker of in-hospital death in patients with undiagnosed diabetes. The rates of death were 1.7% in those with normoglycemia, 3.0% in those with known diabetes, and 16.0% (P < .01) in those with new hyperglycemia.

The ACE/ADA consensus panel¹⁴ set the following glucose targets for patients in the non-ICU setting:

• Pre-meal blood glucose < 140 mg/dL
• Random blood glucose < 180 mg/dL.

On the other hand, hypoglycemia is also associated with adverse outcomes. Therefore, to avoid hypoglycemia, the insulin regimen should be reassessed if blood glucose levels fall below 100 mg/dL. New guidelines will suggest keeping the blood glucose between 100 and 140 mg/dL.

HOW SHOULD WE MANAGE HYPERGLYCEMIA IN THE NON-ICU SETTING?

The ACE/ADA guidelines recommend subcutaneous insulin therapy for most medical-surgical patients with diabetes, reserving intravenous insulin therapy for hyperglycemic crises and uncontrolled hyperglycemia.¹⁴

Oral antidiabetic agents are not generally recommended, as we have no data to support their use in the hospital. Another argument against using noninsulin therapies in the hospital is that sulfonylureas, especially glyburide (Diabeta, Micronase) are a major cause of hypoglycemia. Metformin (Glucophage) is contraindicated in decreased renal function, in hemodynamic instability, in surgical patients, and with the use of iodinated contrast dye. Thiazolidinediones are associated with edema and congestive heart failure, and they take up to 12 weeks to lower blood glucose levels. Alpha-glucosidase inhibitors are weak glucose-lowering agents. Also, therapies directed at glucagon-like-protein 1 can cause nausea and have a greater effect on postprandial glucose.¹⁴

The two main options for managing hyperglycemia and diabetes in the non-ICU setting are short-acting insulin on a sliding scale and basal-bolus therapy, the latter with either NPH plus regular insulin or long-acting plus rapid-acting insulin analogues.

Basal-bolus vs sliding scale insulin: The RABBIT-2 trial

In the RABBIT 2 trial (Randomized Basal Bolus Versus Sliding Scale Regular Insulin in Patients With Type 2 Diabetes Mellitus),¹⁵ our group compared the efficacy and safety of a basal-bolus regimen and a sliding-scale regimen in 130 hospitalized patients with type 2 diabetes treated with diet, with oral hypoglycemic agents, or with both. Oral antidiabetic drugs were discontinued on admission, and patients were randomized to one of the treatment groups.

In the basal-bolus group, the starting total daily dose was 0.4 U/kg/day if the blood glucose level on admission was between 140 and 200 mg/dL, or 0.5 U/kg/day if the glucose level was between 201 and 400 mg/dL. Half of the total daily dose was given as insulin glargine (Lantus) once daily, and the other half was given as insulin glulisine (Apidra) before meals. These doses were adjusted if the patient’s fasting or pre-meal blood glucose levels rose above 140 mg/dL or fell below 70 mg/dL.

The sliding-scale group received regular insulin four times daily (before meals and at bedtime) for glucose levels higher than 140 mg/dL; the higher the level, the more they got.

The basal-bolus regimen was better than sliding-scale regular insulin. At admission, the mean glucose values and hemoglobin A_1c values were similar in both groups, but the mean glucose level on therapy was significantly lower in the basal-bolus group than in the sliding-scale group, 166 ± 32 mg/dL vs 193 ± 54 mg/dL, P < .001). About two-thirds of the basal-bolus group achieved a blood glucose target of less than 140 mg/dL, compared with only about one-third of the sliding-scale group. The basal-bolus group received more insulin, a mean of 42 units per day vs 12.5 units per day in the sliding-scale group. Yet the incidence of hypoglycemia was 3% in both groups.

NPH plus regular vs detemir plus aspart: The DEAN trial

Several long-acting insulin analogues are available and have a longer duration of action than NPH. Similarly, several newer rapid-acting analogues act more rapidly than regular insulin. Do these pharmacokinetic advantages matter? And do they justify the higher costs of the newer agents?

In the randomized Insulin Detemir Versus NPH Insulin in Hospitalized Patients With Diabetes (DEAN) trial,¹⁶ we compared two regimens: detemir plus aspart in a basal-bolus regimen, and NPH plus regular insulin in two divided doses, two-thirds of the total daily dose in the morning before breakfast and one-third before dinner, both doses in a ratio of two-thirds NPH and one-third regular, mixed in the same syringe. We recruited 130 patients with type 2 diabetes mellitus who were on oral hypoglycemic agents or insulin therapy.

NPH plus regular was just as good as detemir plus aspart in improving glycemic control. Blood glucose levels fell during the first day of therapy and were similar in both groups throughout the trial, as measured before breakfast, lunch, and dinner and at bedtime. The mean total daily insulin dose was not significantly different between treatment groups: 56 ± 45 units in the basal-bolus detemir-aspart group and 45 ± 32 units in the NPH-regular group. However, the basal-bolus group received significantly more short-acting insulin: 27 ± 20 units a day of aspart vs 18 ± 14 units of regular.

Somewhat fewer patients in the NPH-regular group had episodes of hypoglycemia, although the difference between groups was not statistically significant.

In a univariate analysis of the RABBIT-2 and DEAN trials,¹⁷ factors that predicted a blood glucose level less than 60 mg/dL were older age, lower body weight, higher serum creatinine level, and previous insulin therapy. Factors that were not predictive were the hemoglobin A_1c level and the enrollment blood glucose level. Based on these data, we believe that to reduce the rate of hypoglycemia, lower insulin doses are needed in elderly patients and patients with renal impairment, and that if patients have been taking insulin before they come to the hospital, the dose should be cut back by about 25% while they are hospitalized.

Does better glucose control in surgical patients affect outcomes in patients undergoing general surgery? To find out, we performed a prospective, multicenter, randomized, open-label trial in general surgery patients not in the ICU.¹⁸ We recruited and randomized 211 patients with type 2 diabetes who were on diet therapy or oral hypoglycemic agents or insulin in low doses (< 0.4 U/kg/day).

Oral drugs were discontinued on admission, and patients were randomized to receive either a basal-bolus regimen of glargine plus glulisine or regular insulin on a sliding scale. The basal-bolus group got 0.5 U/kg/day, half of it as glargine once daily and half as glulisine before meals. The total daily dose was reduced to 0.3 U/kg/day in patients age 70 and older or who had a serum creatinine level of 2.0 mg/dL or higher.

The goal was to maintain fasting and pre-meal glucose concentrations between 100 and 140 mg/dL. The total daily dose was raised by 10% (mostly in the glargine dose) if the blood glucose level was in the range of 141 to 180 mg/dL, and by 20% if the glucose level was higher than 181 mg/dL. The dose was decreased by 10% for glucose levels between 70 and 99 mg/dL, was decreased by 20% if the glucose level was between 40 and 69, and was held if the glucose level was lower than 40 mg/dL. If a patient was not able to eat, insulin glulisine was held until meals were resumed.

The sliding-scale group received regular insulin four times a day for blood glucose levels higher than 140 mg/dL.

The primary outcomes measured were the difference between groups in mean daily blood glucose concentration and a composite of hospital complications including postoperative wound infection, pneumonia, respiratory failure, acute renal failure, and bacteremia. Secondary outcomes were differences between groups in mean fasting and pre-meal blood glucose, number of hypoglycemic episodes (blood glucose < 70 mg/dL), hyperglycemic episodes (blood glucose > 200 mg/dL), length of hospital stay, need for intensive care, and rate of complications including wound infection, pneumonia, acute renal failure, and death.

Blood glucose levels were significantly lower in the basal-bolus group through the first 7 days after randomization, as measured before breakfast, lunch, and dinner, and at bedtime, and then they converged.

More patients in the sliding-scale group had hospital complications, 26 vs 9, P = .003. On the other hand, more patients in the basal-bolus group had episodes of hypoglycemia: 24 (23%) vs 5 (4.7%) had episodes of less than 70 mg/dL (P < .001), 12 (12%) vs 2 (1.9%) had episodes of less than 60 mg/dL (P = .005), and 4 (3.8%) vs 0 had episodes of less than 40 mg/dL (P = .057). The mean total daily dose of insulin was 33.4 units in the basal-bolus group and 12.3 units in the sliding-scale group.

WHAT HAVE WE LEARNED?

Don’t use a sliding-scale regimen as a single agent in patients with diabetes. Glycemic control is better with a basal-bolus regimen than with a sliding-scale regimen, and a basal-bolus insulin regimen is preferred for most patients with hyperglycemia.

The old human insulins (ie, regular and NPH) are still good and improve glycemic control as well as the new basal insulin analogues (detemir and aspart) do.

Improved control may reduce the rate of hospital complications, according to preliminary evidence. More studies are under way.

One size does not fit all. Those who are elderly or who have impaired renal function should receive lower doses of insulin, eg, 0.3 U/kg/day instead of 0.5 U/kg/day. Those who are on insulin should have their dose decreased when they are admitted to the hospital. Perhaps lean patients with type 2 diabetes should also have a lower dose.

Most hospitalized patients with diabetes and elevated blood glucose values (or hyperglycemia) should receive subcutaneous insulin treatment with a basal-bolus regimen or a multidose combination of NPH plus regular insulin. Selected patients with severe insulin resistance and persistent hyperglycemia despite subcutaneous insulin may benefit from continuous intravenous insulin infusion.

Patients treated with insulin at home should continue to receive insulin therapy in the hospital. However, the insulin dosage should be reduced by about 25% to allow for lower food intake.

QUESTIONS FOR FURTHER STUDY

Should we modify the standard basal-bolus regimen?

In a typical basal-bolus regimen, patients get 50% of their total daily insulin dose in the form of a basal injection and 50% in the form of rapid-acting boluses before meals. However, for a variety of reasons, hospitalized patients do not eat very much. Thus, a 50-50 basal-bolus regimen may not be ideal for patients with poor oral intake.

In the Basal-PLUS trial, currently under way, we are comparing the safety and efficacy of a daily dose of basal insulin (glargine) plus correction doses of a rapid-acting insulin analogue (glulisine) on a sliding scale and a standard basal-bolus regimen in medical and surgical patients.

Does one glycemic target fit all patients?

Falciglia et al¹⁹ found an association between hyperglycemia and death in patients with unstable angina, arrhythmias, stroke, pneumonia, gastrointestinal bleeding, respiratory failure, sepsis, acute renal failure, and congestive heart failure. However, they found no such association in patients with chronic obstructive pulmonary disease, liver failure, diabetic ketoacidosis, gastrointestinal neoplasm, musculoskeletal disease, peripheral vascular disease with bypass, hip fracture, amputation due to peripheral vascular disease, or prostate surgery. Should patients in this second group be treated with a less-intensive insulin regimen?

What is the best regimen after hospital discharge?

We are conducting a prospective clinical trial to assess the impact of insulin after hospital discharge. Our current practice when a patient is discharged from the hospital is as follows:

• If the admission hemoglobin A_1c level is less than 7%, we restart the previous outpatient treatment regimen of oral antidiabetic agents, or insulin, or both.
• If the admission hemoglobin A_1c is between 7% and 9%, we restart the outpatient oral agents and continue glargine once daily at 50% to 80% of the hospital dose.
• If the hemoglobin A_1c level is higher than 9%, we discharge the patient on a basal-bolus regimen at the same dosage as in the hospital. As an alternative, we could restart the oral agents and add glargine once daily at 80% of the hospital dose.

Hyperglycemia and diabetes mellitus are very common in hospitalized patients. Although more data are available on the prevalence of this problem and on how to manage it in the intensive care unit (ICU) than on regular hospital floors, the situation is changing. Information is emerging on the prevalence and impact of hyperglycemia and diabetes in the non-ICU setting, which is the focus of this paper.

HYPERGLYCEMIA IS COMMON AND PREDICTS POOR OUTCOMES

Cook et al,¹ in a survey of 126 US hospitals, found that the prevalence of hyperglycemia (blood glucose > 180 mg/dL) was 46% in the ICU and 32% in regular wards.

Kosiborod et al² reported that hyperglycemia (blood glucose > 140 mg/dL) was present in 78% of diabetic patients hospitalized with acute coronary syndrome and 26% of similar hospitalized nondiabetic patients.

Hyperglycemia is a common comorbidity in medical-surgical patients in community hospitals. Our group³ found that, in our hospital, 62% of patients were normoglycemic (ie, had a fasting blood glucose < 126 mg/dL or a random blood glucose < 200 mg/dL on two occasions), 26% had known diabetes, and 12% had new hyperglycemia. Further, new hyperglycemia was associated with a higher in-hospital death rate than the other two conditions.

Failure to identify diabetes is a predictor of rehospitalization. Robbins and Webb⁴ reported that 30.6% of those who had diabetes that was missed during hospitalization were readmitted within 30 days, compared with 9.4% of patients with diabetes first diagnosed during hospitalization.

WHAT DIAGNOSTIC CRITERIA SHOULD WE USE?

Blood glucose greater than 140 mg/dL

A consensus statement from the American Association of Clinical Endocrinologists (ACE) and the American Diabetes Association (ADA)⁵ defines in-hospital hyperglycemia as a blood glucose level greater than 140 mg/dL on admission or in the hospital. If the blood glucose is higher than this, the question arises as to whether the patient has preexisting diabetes or has stress hyperglycemia.

Hemoglobin A_1c of 6.5% or higher

In view of the uncertainty as to whether a patient with an elevated blood glucose level has preexisting diabetes or stress hyperglycemia, upcoming guidelines will recommend measuring the hemoglobin A_1c level if the blood glucose level is higher than 140 mg/dL.

A patient with an elevated blood glucose level (>140 mg/dL) whose hemoglobin A_1c level is 6.5% or higher can be identified as having diabetes that preceded the hospitalization. Hemoglobin A_1c testing can also be useful to assess glycemic control before admission and in designing an optional regimen at the time of discharge. In patients with newly recognized hyperglycemia, a hemoglobin A_1c measurement can help differentiate patients with previously undiagnosed diabetes from those with stress-induced hyperglycemia.

Clinicians should keep in mind that a hemoglobin A_1c cutoff of 6.5% identifies fewer cases of undiagnosed diabetes than does a high fasting glucose concentration, and that a level less than 6.5% does not rule out the diagnosis of diabetes. Several epidemiologic studies⁶ have reported a low sensitivity (44% to 66%) but a high specificity (76% to 99%) for hemoglobin A_1c values higher than 6.5% in an outpatient population. The high specificity therefore supports the use of hemoglobin A_1c to confirm the diagnosis of diabetes in patients with hyperglycemia, but the low sensitivity indicates that this test should not be used for universal screening in the hospital.

Many factors can influence the hemoglobin A_1c level, such as anemia, iron deficiency, blood transfusions, hemolytic anemia, and renal failure.

Until now, if patients had hyperglycemia but no prior diagnosis of diabetes, the recommendation was for an oral 2-hour glucose tolerance test shortly after discharge to confirm the diagnosis of diabetes. Norhammar et al⁷ performed oral glucose tolerance tests in patients admitted with acute myocardial infarction, and Matz et al⁸ performed glucose tolerance tests in patients with acute stroke. They found that impaired glucose tolerance and undiagnosed type 2 diabetes were very common in these two groups. However, physicians rarely order oral glucose tolerance tests. We believe that hemoglobin A_1c will be a better tool than an oral glucose tolerance test to confirm diabetes in hyperglycemic patients in the hospital setting.

WHAT IS THE ASSOCIATION BETWEEN HYPERGLYCEMIA AND OUTCOMES?

In 2,471 patients admitted to the hospital with community-acquired pneumonia, McAlister et al¹⁰ found that the rates of hospital complications and of death rose with blood glucose levels.

Falguera et al¹¹ found that, in 660 episodes of community-acquired pneumonia, the rates of hospitalization, death, pleural effusion, and concomitant illnesses were all significantly higher in diabetic patients than in nondiabetic patients.

Noordzij et al¹² performed a case-control study of 108,593 patients who underwent noncardiac surgery. The odds ratio for perioperative death was 1.19 (95% confidence interval [CI] 1.1–1.3) for every 1-mmol/L increase in the glucose level.

Frisch et al,¹³ in patients undergoing noncardiac surgery, found that the 30-day rates of death and of in-hospital complications were all higher in patients with diabetes than without diabetes.

Our group³ identified hyperglycemia as an independent marker of in-hospital death in patients with undiagnosed diabetes. The rates of death were 1.7% in those with normoglycemia, 3.0% in those with known diabetes, and 16.0% (P < .01) in those with new hyperglycemia.

The ACE/ADA consensus panel¹⁴ set the following glucose targets for patients in the non-ICU setting:

• Pre-meal blood glucose < 140 mg/dL
• Random blood glucose < 180 mg/dL.

On the other hand, hypoglycemia is also associated with adverse outcomes. Therefore, to avoid hypoglycemia, the insulin regimen should be reassessed if blood glucose levels fall below 100 mg/dL. New guidelines will suggest keeping the blood glucose between 100 and 140 mg/dL.

HOW SHOULD WE MANAGE HYPERGLYCEMIA IN THE NON-ICU SETTING?

The ACE/ADA guidelines recommend subcutaneous insulin therapy for most medical-surgical patients with diabetes, reserving intravenous insulin therapy for hyperglycemic crises and uncontrolled hyperglycemia.¹⁴

Oral antidiabetic agents are not generally recommended, as we have no data to support their use in the hospital. Another argument against using noninsulin therapies in the hospital is that sulfonylureas, especially glyburide (Diabeta, Micronase) are a major cause of hypoglycemia. Metformin (Glucophage) is contraindicated in decreased renal function, in hemodynamic instability, in surgical patients, and with the use of iodinated contrast dye. Thiazolidinediones are associated with edema and congestive heart failure, and they take up to 12 weeks to lower blood glucose levels. Alpha-glucosidase inhibitors are weak glucose-lowering agents. Also, therapies directed at glucagon-like-protein 1 can cause nausea and have a greater effect on postprandial glucose.¹⁴

The two main options for managing hyperglycemia and diabetes in the non-ICU setting are short-acting insulin on a sliding scale and basal-bolus therapy, the latter with either NPH plus regular insulin or long-acting plus rapid-acting insulin analogues.

Basal-bolus vs sliding scale insulin: The RABBIT-2 trial

In the RABBIT 2 trial (Randomized Basal Bolus Versus Sliding Scale Regular Insulin in Patients With Type 2 Diabetes Mellitus),¹⁵ our group compared the efficacy and safety of a basal-bolus regimen and a sliding-scale regimen in 130 hospitalized patients with type 2 diabetes treated with diet, with oral hypoglycemic agents, or with both. Oral antidiabetic drugs were discontinued on admission, and patients were randomized to one of the treatment groups.

In the basal-bolus group, the starting total daily dose was 0.4 U/kg/day if the blood glucose level on admission was between 140 and 200 mg/dL, or 0.5 U/kg/day if the glucose level was between 201 and 400 mg/dL. Half of the total daily dose was given as insulin glargine (Lantus) once daily, and the other half was given as insulin glulisine (Apidra) before meals. These doses were adjusted if the patient’s fasting or pre-meal blood glucose levels rose above 140 mg/dL or fell below 70 mg/dL.

The sliding-scale group received regular insulin four times daily (before meals and at bedtime) for glucose levels higher than 140 mg/dL; the higher the level, the more they got.

The basal-bolus regimen was better than sliding-scale regular insulin. At admission, the mean glucose values and hemoglobin A_1c values were similar in both groups, but the mean glucose level on therapy was significantly lower in the basal-bolus group than in the sliding-scale group, 166 ± 32 mg/dL vs 193 ± 54 mg/dL, P < .001). About two-thirds of the basal-bolus group achieved a blood glucose target of less than 140 mg/dL, compared with only about one-third of the sliding-scale group. The basal-bolus group received more insulin, a mean of 42 units per day vs 12.5 units per day in the sliding-scale group. Yet the incidence of hypoglycemia was 3% in both groups.

NPH plus regular vs detemir plus aspart: The DEAN trial

Several long-acting insulin analogues are available and have a longer duration of action than NPH. Similarly, several newer rapid-acting analogues act more rapidly than regular insulin. Do these pharmacokinetic advantages matter? And do they justify the higher costs of the newer agents?

In the randomized Insulin Detemir Versus NPH Insulin in Hospitalized Patients With Diabetes (DEAN) trial,¹⁶ we compared two regimens: detemir plus aspart in a basal-bolus regimen, and NPH plus regular insulin in two divided doses, two-thirds of the total daily dose in the morning before breakfast and one-third before dinner, both doses in a ratio of two-thirds NPH and one-third regular, mixed in the same syringe. We recruited 130 patients with type 2 diabetes mellitus who were on oral hypoglycemic agents or insulin therapy.

NPH plus regular was just as good as detemir plus aspart in improving glycemic control. Blood glucose levels fell during the first day of therapy and were similar in both groups throughout the trial, as measured before breakfast, lunch, and dinner and at bedtime. The mean total daily insulin dose was not significantly different between treatment groups: 56 ± 45 units in the basal-bolus detemir-aspart group and 45 ± 32 units in the NPH-regular group. However, the basal-bolus group received significantly more short-acting insulin: 27 ± 20 units a day of aspart vs 18 ± 14 units of regular.

Somewhat fewer patients in the NPH-regular group had episodes of hypoglycemia, although the difference between groups was not statistically significant.

In a univariate analysis of the RABBIT-2 and DEAN trials,¹⁷ factors that predicted a blood glucose level less than 60 mg/dL were older age, lower body weight, higher serum creatinine level, and previous insulin therapy. Factors that were not predictive were the hemoglobin A_1c level and the enrollment blood glucose level. Based on these data, we believe that to reduce the rate of hypoglycemia, lower insulin doses are needed in elderly patients and patients with renal impairment, and that if patients have been taking insulin before they come to the hospital, the dose should be cut back by about 25% while they are hospitalized.

Does better glucose control in surgical patients affect outcomes in patients undergoing general surgery? To find out, we performed a prospective, multicenter, randomized, open-label trial in general surgery patients not in the ICU.¹⁸ We recruited and randomized 211 patients with type 2 diabetes who were on diet therapy or oral hypoglycemic agents or insulin in low doses (< 0.4 U/kg/day).

Oral drugs were discontinued on admission, and patients were randomized to receive either a basal-bolus regimen of glargine plus glulisine or regular insulin on a sliding scale. The basal-bolus group got 0.5 U/kg/day, half of it as glargine once daily and half as glulisine before meals. The total daily dose was reduced to 0.3 U/kg/day in patients age 70 and older or who had a serum creatinine level of 2.0 mg/dL or higher.

The goal was to maintain fasting and pre-meal glucose concentrations between 100 and 140 mg/dL. The total daily dose was raised by 10% (mostly in the glargine dose) if the blood glucose level was in the range of 141 to 180 mg/dL, and by 20% if the glucose level was higher than 181 mg/dL. The dose was decreased by 10% for glucose levels between 70 and 99 mg/dL, was decreased by 20% if the glucose level was between 40 and 69, and was held if the glucose level was lower than 40 mg/dL. If a patient was not able to eat, insulin glulisine was held until meals were resumed.

The sliding-scale group received regular insulin four times a day for blood glucose levels higher than 140 mg/dL.

The primary outcomes measured were the difference between groups in mean daily blood glucose concentration and a composite of hospital complications including postoperative wound infection, pneumonia, respiratory failure, acute renal failure, and bacteremia. Secondary outcomes were differences between groups in mean fasting and pre-meal blood glucose, number of hypoglycemic episodes (blood glucose < 70 mg/dL), hyperglycemic episodes (blood glucose > 200 mg/dL), length of hospital stay, need for intensive care, and rate of complications including wound infection, pneumonia, acute renal failure, and death.

Blood glucose levels were significantly lower in the basal-bolus group through the first 7 days after randomization, as measured before breakfast, lunch, and dinner, and at bedtime, and then they converged.

More patients in the sliding-scale group had hospital complications, 26 vs 9, P = .003. On the other hand, more patients in the basal-bolus group had episodes of hypoglycemia: 24 (23%) vs 5 (4.7%) had episodes of less than 70 mg/dL (P < .001), 12 (12%) vs 2 (1.9%) had episodes of less than 60 mg/dL (P = .005), and 4 (3.8%) vs 0 had episodes of less than 40 mg/dL (P = .057). The mean total daily dose of insulin was 33.4 units in the basal-bolus group and 12.3 units in the sliding-scale group.

WHAT HAVE WE LEARNED?

Don’t use a sliding-scale regimen as a single agent in patients with diabetes. Glycemic control is better with a basal-bolus regimen than with a sliding-scale regimen, and a basal-bolus insulin regimen is preferred for most patients with hyperglycemia.

The old human insulins (ie, regular and NPH) are still good and improve glycemic control as well as the new basal insulin analogues (detemir and aspart) do.

Improved control may reduce the rate of hospital complications, according to preliminary evidence. More studies are under way.

One size does not fit all. Those who are elderly or who have impaired renal function should receive lower doses of insulin, eg, 0.3 U/kg/day instead of 0.5 U/kg/day. Those who are on insulin should have their dose decreased when they are admitted to the hospital. Perhaps lean patients with type 2 diabetes should also have a lower dose.

Most hospitalized patients with diabetes and elevated blood glucose values (or hyperglycemia) should receive subcutaneous insulin treatment with a basal-bolus regimen or a multidose combination of NPH plus regular insulin. Selected patients with severe insulin resistance and persistent hyperglycemia despite subcutaneous insulin may benefit from continuous intravenous insulin infusion.

Patients treated with insulin at home should continue to receive insulin therapy in the hospital. However, the insulin dosage should be reduced by about 25% to allow for lower food intake.

QUESTIONS FOR FURTHER STUDY

Should we modify the standard basal-bolus regimen?

In a typical basal-bolus regimen, patients get 50% of their total daily insulin dose in the form of a basal injection and 50% in the form of rapid-acting boluses before meals. However, for a variety of reasons, hospitalized patients do not eat very much. Thus, a 50-50 basal-bolus regimen may not be ideal for patients with poor oral intake.

In the Basal-PLUS trial, currently under way, we are comparing the safety and efficacy of a daily dose of basal insulin (glargine) plus correction doses of a rapid-acting insulin analogue (glulisine) on a sliding scale and a standard basal-bolus regimen in medical and surgical patients.

Does one glycemic target fit all patients?

Falciglia et al¹⁹ found an association between hyperglycemia and death in patients with unstable angina, arrhythmias, stroke, pneumonia, gastrointestinal bleeding, respiratory failure, sepsis, acute renal failure, and congestive heart failure. However, they found no such association in patients with chronic obstructive pulmonary disease, liver failure, diabetic ketoacidosis, gastrointestinal neoplasm, musculoskeletal disease, peripheral vascular disease with bypass, hip fracture, amputation due to peripheral vascular disease, or prostate surgery. Should patients in this second group be treated with a less-intensive insulin regimen?

What is the best regimen after hospital discharge?

We are conducting a prospective clinical trial to assess the impact of insulin after hospital discharge. Our current practice when a patient is discharged from the hospital is as follows:

• If the admission hemoglobin A_1c level is less than 7%, we restart the previous outpatient treatment regimen of oral antidiabetic agents, or insulin, or both.
• If the admission hemoglobin A_1c is between 7% and 9%, we restart the outpatient oral agents and continue glargine once daily at 50% to 80% of the hospital dose.
• If the hemoglobin A_1c level is higher than 9%, we discharge the patient on a basal-bolus regimen at the same dosage as in the hospital. As an alternative, we could restart the oral agents and add glargine once daily at 80% of the hospital dose.

Summit Director:
Amir K. Jaffer, MD

Contents

Summit Faculty

Summit Program

Abstract 1: Application of 2007 ACC/AHA guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery using decision support tools
BobbieJean Sweitzer, Michael Vigoda, Vicente Behrens, Nikola Miljkovic, and Kris Arheart

Abstract 2: Prevalence of obstructive dleep spnea in patients presenting for hip or knee replacement surgery
Micah Beachy, DO; Jason Shiffermiller, MD; and Chad Vokoun, MD

Abstract 3: A protocol to triage preoperative assessments to either nurses or nurse practitioners/physician assistants
Anthony Basil, RN; Pamela Pennigar, FNP; David R. Wright, MD; and Ronald P. Olson, MD

Abstract 4: Application of 2007 ACC/AHA guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery using decision support tools
BobbieJean Sweitzer, Michael Vigoda, Vicente Behrens, Nikola Miljkovic, and Kris Arheart

Abstract 5: Most anesthesiologists don’t correctly apply 2007 ACC/AHA guidelines on perioperative cardiac evaluation
BobbieJean Sweitzer, Michael Vigoda, Vicente Behrens, Nikola Miljkovic, Kris Arheart, and Richard Dutton

Abstract 6: Anesthesiology residents do not agree with their training programs on the degree to which the 2007 ACC/AHA guidelines are emphasized
BobbieJean Sweitzer, Michael Vigoda, Vicente Behrens, Nikola Miljkovic, and Kris Arheart

Abstract 7: Prevalence of obstructive sleep apnea in patients presenting for hip or knee replacement surgery
Micah Beachy, DO; Jason Shiffermiller, MD; and Chad Vokoun, MD

Abstract 8: A protocol to triage preoperative assessments to either nurses or nurse practitioners/physician assistants
Anthony Basil, RN; Pamela Pennigar, FNP; David R. Wright, MD; and Ronald P. Olson, MD

Abstract 9: Do ACEIs on the morning of surgery increase risk of intraoperative hypotension?
Steven L. Cohn, MD, and Kalia Skeete, MD

Abstract 10: One-year incidence of postoperative troponin revations in patients undergoing major orthopedic surgery
Michael Urban, MD, PhD; Stephen Wolfe, BS; Niel Sanghevi, BS; and Steven Magid, MD

Abstract 11: A review of preoperative clinic cardiology referrals for adults undergoing intermediate- and low-risk surgery
Susan Calderwood, MD; Jennifer Lee Morse, MS; and Damon R. Michaels, CCRP

Abstract 12: Patterns of preoperative consultation by risk and surgical specialty in a large health care system
Stephan Thilen, MD, MS; Christopher Bryson, MD, MS; Robert Reid, MD, PhD; and Miriam Treggiari, MD, MPH, PhD

Abstract 13: One-year incidence for admission to a critical care unit after major orthopedic surgery
Michael Urban, MD, PhD; Steven Magid, MD; and Michele Mangini, DNP

Abstract 14: Determination of the causes of long patient wait times in a preoperative evaluation clinic
Jean Kwo, MD; Devon Price, BS; Mary Elizabeth Ellbeg, RN; and Retsef Levi, PhD

Abstract 15: Does perioperative statin treatment affect hospital and ICU length of stay rollowing cardiac surgery: A systematic review
Vineet Chopra, MD, FACP, FHM; David Wesorick, MD; and Kim A. Eagle, MD

Abstract 16: Assessment of patient satisfaction of nurse screening vs complete preoperative assessment
Ronald Olson, MD, and Kathy Bock, RN

Abstract 17: Traumatic subdural hematoma: An update on morbidity
Rachel Thompson, MD; Christina Ryan, MD; Nancy Temkin, PhD; Richard Ellenbogen, MD; and Joann G. Elmore, MD, MPH

Abstract 18: Lipid emulsion as a lifesaving treatment for local anesthetic systemic toxicity (LAST)
Deepti Sachdev and Guy Weinberg, MD

Abstract 19: Perioperative ACLS recommendations should be modified for the treatment of local anesthetic toxicity
Adam Haas, MD, and Alexia Beccue, MD

Abstract 20: Preoperative EMR containing smart-set reminders improve accuracy of documentation by nonanesthesia clinicians during preoperative assessments
Angela Edwards, MD; Jill Grant, PA; and Ruth Hyde, MD

Abstract 21: POET: Procedure outcomes evaluation tool
Ahmad AbuSalah, MSc, and Terrence Adam, MD, PhD

Abstract 22: Results of a multidisciplinary preoperative assessment process for high-risk orthopedic patients
Terrence Adam, MD, PhD; Connie Parenti, MD; Terence Gioe, MD; and Karen Ringsred, MD

Abstract 23: Practical algorithm for preoperative evaluation of patients with liver disease
Madalina A. Vlase, PA-C, and Deborah C. Richman, MBChB, FFA(SA)

Abstract 24: Evaluation and management of isolated elevated aPTT
Sheila Hassan, MSN, NP; Patricia Kidik, MSN, NP; Catherine McGowan, MSN, NP; and Angela M. Bader, MD

Abstract 25: A perioperative triage plan for obstructive sleep apnea patients
Christian Altman, MD; R. Michael Boyer, DO; and Peter G. Kallas, MD

Abstract 26: Quantitative evaluation of handoff checklists
Jay Joshi, MD, and David Mayer, MD

Abstract 27: To deflate or not to deflate: Lap-Band® management in subsequent surgeries
Arjun Reddy, MD, and Deborah C. Richman, MBChB, FFA(SA)

Abstract 28: Takotsubo cardiomyopathy and resultant cardiogenic shock after mitral valve repair
Adam Evans, MD, MBA; Daniel B. Sims, MD; Nir Uriel, MD; Ulrich P. Jorde, MD; and Craig R. Smith, MD

Abstract 29: Intravenous vitamin K: Rapid reversal of warfarin and lack of subsequent warfarin resistance
Feras Abdul Khalek, MD; Interdeep Dhaliwal, MD; and Twylla Tassava, MD

Abstract 30: Cervical spine surgery: When not to extubate postoperatively
Carlos Mateo Mijares, MD; Doris Debs, ARNP, MSN-BC; Nicole Martin, MD; and Ronald Lee Samson, MD

Abstract 31: Total occlusion of oral cavity by mandibular sarcoma for resection: To intubate nasally or proceed to an awake tracheostomy?
Carlos Mateo Mijares, MD, and Maria DeLapena, MD

Abstract 32: Perioperative fatal embolic stroke associated with iron deficiency anemia and thrombocytosis
Carlos Mateo Mijares, MD; Nicole Martin, MD; and Ricardo Martinez-Ruiz, MD

Abstract 33: Conservative approach saves the day anesthesia-wise and surgical-wise
Carlos Mateo Mijares, MD; Bradley Shore, MD; Edward Zalkind, CRNA; and Nicole Martin, MD

Abstract 34: Predictors of acute kidney injury in patients undergoing total knee replacement surgery
Vishal Sehgal, MD; Pardeep Bansal, MD; Praveen Reddy, MD; Vishal Sharma, MD; Samuel Lesko, MD; John H. Doherty, MD; Theodore Tomaszewski, MD; Jack Prior, MD; Roger Getts, MD; and Jeremiah Eagan, MD

Abstract 35: Perioperative medical management of the Marfan patient undergoing repeat cardiothoracic surgery
Aashish Shah, MD, and Adam Skrzynski, MD

Index of abstract authors

Summit Director:
Amir K. Jaffer, MD

Contents

Summit Faculty

Summit Program

Abstract 1: Application of 2007 ACC/AHA guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery using decision support tools
BobbieJean Sweitzer, Michael Vigoda, Vicente Behrens, Nikola Miljkovic, and Kris Arheart

Abstract 2: Prevalence of obstructive dleep spnea in patients presenting for hip or knee replacement surgery
Micah Beachy, DO; Jason Shiffermiller, MD; and Chad Vokoun, MD

Abstract 3: A protocol to triage preoperative assessments to either nurses or nurse practitioners/physician assistants
Anthony Basil, RN; Pamela Pennigar, FNP; David R. Wright, MD; and Ronald P. Olson, MD