Emergency Medicine

ANSWER

The radiograph demonstrates a compression fracture of T12 with moderate loss of height approaching 50% anteriorly. In addition, there is a vertically oriented fracture within the middle of the vertebral body, causing the back portion to be posteriorly displaced.

This type of burst fracture is potentially unstable and should be treated as such. The patient was placed on strict spine precautions and transferred to a facility where trauma and neurosurgical services were available.

ANSWER

The radiograph demonstrates a compression fracture of T12 with moderate loss of height approaching 50% anteriorly. In addition, there is a vertically oriented fracture within the middle of the vertebral body, causing the back portion to be posteriorly displaced.

This type of burst fracture is potentially unstable and should be treated as such. The patient was placed on strict spine precautions and transferred to a facility where trauma and neurosurgical services were available.

ANSWER

The radiograph demonstrates a compression fracture of T12 with moderate loss of height approaching 50% anteriorly. In addition, there is a vertically oriented fracture within the middle of the vertebral body, causing the back portion to be posteriorly displaced.

This type of burst fracture is potentially unstable and should be treated as such. The patient was placed on strict spine precautions and transferred to a facility where trauma and neurosurgical services were available.

WASHINGTON – The Medicare Hospital Readmissions Reduction Program is working, according to an original analysis of Medicare claims data presented at a meeting of the Medicare Payment Advisory Commission.

“First, readmissions declined,” MedPAC staff member Jeff Stensland, PhD, said during a congressionally mandated staff report to the commissioners. “Second, while observation stays increased, they did not fully offset the decrease in readmissions. Third, while [emergency department] visits also increased, those increases appear to largely be due to factors other than the readmission program. And fourth, in addition, all the evidence we examined suggests that the readmissions program did not result in increased mortality.”

Copyright Kimberly Pack/Thinkstock

gtwachtman@frontlinemedcom.com

WASHINGTON – The Medicare Hospital Readmissions Reduction Program is working, according to an original analysis of Medicare claims data presented at a meeting of the Medicare Payment Advisory Commission.

“First, readmissions declined,” MedPAC staff member Jeff Stensland, PhD, said during a congressionally mandated staff report to the commissioners. “Second, while observation stays increased, they did not fully offset the decrease in readmissions. Third, while [emergency department] visits also increased, those increases appear to largely be due to factors other than the readmission program. And fourth, in addition, all the evidence we examined suggests that the readmissions program did not result in increased mortality.”

Copyright Kimberly Pack/Thinkstock

gtwachtman@frontlinemedcom.com

WASHINGTON – The Medicare Hospital Readmissions Reduction Program is working, according to an original analysis of Medicare claims data presented at a meeting of the Medicare Payment Advisory Commission.

“First, readmissions declined,” MedPAC staff member Jeff Stensland, PhD, said during a congressionally mandated staff report to the commissioners. “Second, while observation stays increased, they did not fully offset the decrease in readmissions. Third, while [emergency department] visits also increased, those increases appear to largely be due to factors other than the readmission program. And fourth, in addition, all the evidence we examined suggests that the readmissions program did not result in increased mortality.”

Copyright Kimberly Pack/Thinkstock

gtwachtman@frontlinemedcom.com

IN THIS ARTICLE

• Diagnosis
• Management
• Outcome for the case patient

An otherwise healthy 9-year-old boy is brought to the emergency department (ED) by his father for evaluation of severe pain, blurry vision, and four hours of tearing in his right eye. The patient was in school when he experienced sudden-onset irritation and scratching pain that caused him to rub his eye. He says it “feels like there is something in my eye,” but he denies any known substance or foreign body. He has no medical or surgical history, does not wear contact lenses or eyeglasses, and denies loss of vision. There is no history of recent illness or travel.

On evaluation, the patient is in no acute distress but is holding his right eye closed due to foreign-body sensation and increased photosensitivity and tearing. There is no obvious erythema or swelling in the upper or lower eyelids bilaterally. A visual acuity test with a Snellen eye chart shows 20/20 vision in the left eye and 20/50 in the right, secondary to pain, photophobia, and excessive tearing. The patient’s right sclera is significantly injected. Intraocular pressure, measured with a tonometer, is 12 to 14 mm Hg. A fluorescein stain of the eye yields no significant findings. The globe is intact.

At first glance, a slit-lamp exam shows no obvious signs of a foreign body. But much higher magnification reveals substantial conjunctival injection and numerous intracorneal linear foreign bodies in the right eye (see Figure 1 for example [not the case patient]). The anterior chamber shows no inflammatory reaction, and findings in the posterior segment are unremarkable.

The initial diagnosis is simple conjunctivitis—but closer examination reveals multiple fine, barbed hairs embedded in the patient’s right cornea. Upon further questioning, the patient reports that prior to symptom onset, he had been holding the classroom pet, a Chilean Rose tarantula, in the palm of his hands.

DISCUSSION

Foreign body injury is a common cause of ocular pain and corneal damage, which can lead to challenging complications. Ophthalmic emergencies account for 2% of ED visits in the US annually and are a major cause of visual impairment.¹ But when a painful eye is the chief complaint, contact with insects, plants, or spiders is rarely included in the differential. Tarantulas are popular classroom and household pets, however, and ocular injury should be suspected in anyone who has been holding a tarantula prior to onset of pain.

Ophthalmia nodosa

Tarantulas are one of the most common arachnids known to cause ophthalmia nodosa—a granulomatous reaction of the conjunctiva or cornea to an implanted plant, insect, or spider hair that typically manifests with photophobia, irritation, and chemosis.^2,3 Tarantulas, when scared or defending their eggs, shoot urticating setae at the threat—a defensive mechanism largely unknown to parents, tarantula owners, and medical professionals.

Urticating setae are found in roughly 90% of tarantula species throughout tropical and subtropical regions.⁴ Depending on the species, setae can be located on the distal prolateral surface of the palpal femur or the dorsum of the abdomen. They can be released when the tarantula scratches its legs against the abdominal urticating setae patch or scratches the palps against the chelicerae (appendages in front of the mouth), or when direct exterior contact is made with the abdominal setae.⁴

There are six types of urticating hairs. Each is attached to the spider’s cuticle by either a stalk (which represents the break-off region) or a socket.⁴ Tarantula hairs range in size from 0.1 mm to 0.3 mm and have a sharp, pointed head and numerous barbs, which help embed them in the target.⁵ They are long and thin, to facilitate deep tissue penetration, and can enter the eyes, lungs, or other body parts (see Figure 2).

Diagnosis and management

Ophthalmic emergencies can affect the visual system and, if left untreated, can lead to permanent vision loss. Affected patients require immediate medical attention and should be referred to an ophthalmologist for follow-up care.

Diagnosis. A thorough history and physical exam are of utmost importance; tiny setae can be easily overlooked if the examiner is not diligent, and the similar symptomatology can lead to misdiagnosis as simple conjunctivitis.³ A visual acuity test and slit-lamp exam are useful for confirmation.

Treatment. Once the diagnosis is confirmed, treatment should consist of mild topical antibiotics and steroids to effectively control infection and inflammation. While topical steroids may be appropriate, local adverse events associated with their use (eg, glaucoma, cataracts) can be problematic. Gentle eye irrigation has been noted by some researchers as contraindicated, while others find it useful to flush out some of the hairs.^5,8,9

Most of the visible protruding tarantula hairs can and should be removed under microscopy during slit-lamp exam. Hairs that are buried in the cornea, however, are nearly impossible to remove and pose a threat of further complications, as described. Conservative management with careful observation is therefore recommended. If the patient develops a granuloma, excision—along with a course of systemic steroids and setae removal via vitrectomy—may be needed.⁹

The good news is that, in many cases, deeper hairs are absorbed without complication, making their removal unnecessary.⁵ Factors that encourage leaving the setae untouched include a large number of hairs, deep corneal penetration, lack of patient tolerance for the procedure, and risk for perforation.³

More invasive treatments (eg, laser photocoagulation, intraocular surgery) to remove offending hairs are possible, but literature on the outcome of these interventions is limited. One report to date used argon laser photocoagulation to treat endophthalmitis from vitreous hairs.¹⁰ The laser can fragment the hairs so that they lose their barbed characteristic and cannot penetrate deeper.⁶

Follow-up. Close follow-up is advised, and patients should be educated on the importance of medication compliance and return visits for reevaluation. Given the potential dangers of handling these spiders, tarantula owners should be advised to use protective gloving and goggles.^2,5,8,9

OUTCOME FOR THE CASE PATIENT

The case patient was sent to an ophthalmologist on day 1. Proparacaine was placed in his right eye, and all of the superficial tarantula hairs were removed using 25- and 30-gauge needles with jeweler forceps under slit-lamp microscopy. Most of the hairs were removed from the superior cornea; fewer were found in the paracentral and inferior regions of the cornea. Approximately five hairs in the paracentral area of the cornea were embedded in the midstromal depth and could not be removed. One drop of ciprofloxacin was administered.

The patient was sent home with an eye shield and instructions to use tobramycin/dexamethasone eye drops (qid in his right eye) and avoid rubbing the eye. (The eye shield, though not technically necessary, was deemed beneficial to help the patient avoid touching the eye.) He was scheduled to return to the clinic one week later.

On follow-up, a careful exam performed under microscopy showed that the five tarantula hairs were still embedded, and an additional six hairs were found in the deep stroma. Superficial punctate keratitis—an eye disorder caused by epithelial cell death on the surface of the cornea—was noted, but no anterior chamber cells were seen. The patient was instructed to continue using the eye drops as prescribed until finished, then start using loteprednol (tid) and artificial lubricating tears (every 2 h).

He returned to the clinic every two weeks for a total of 10 visits. At the end of the treatment course, the remaining tarantula hairs were unable to be removed. The patient used tapering doses of topical eye steroids and antibiotic drops secondary to flare-up.

CONCLUSION

Determining the etiology of ophthalmic emergencies is essential to timely and appropriate management. In this case, a recognized but often overlooked cause, tarantula hairs, made the diagnosis more complicated than simple conjunctivitis. When ocular injury is suspected, the provider must obtain an accurate and detailed history along with a thorough physical exam. Since patients must comply with medication regimens to prevent acute and chronic infection, a clear treatment and follow-up plan should be established. With these in place, ophthalmia nodosa caused by urticating setae can be effectively managed.

IN THIS ARTICLE

• Diagnosis
• Management
• Outcome for the case patient

An otherwise healthy 9-year-old boy is brought to the emergency department (ED) by his father for evaluation of severe pain, blurry vision, and four hours of tearing in his right eye. The patient was in school when he experienced sudden-onset irritation and scratching pain that caused him to rub his eye. He says it “feels like there is something in my eye,” but he denies any known substance or foreign body. He has no medical or surgical history, does not wear contact lenses or eyeglasses, and denies loss of vision. There is no history of recent illness or travel.

On evaluation, the patient is in no acute distress but is holding his right eye closed due to foreign-body sensation and increased photosensitivity and tearing. There is no obvious erythema or swelling in the upper or lower eyelids bilaterally. A visual acuity test with a Snellen eye chart shows 20/20 vision in the left eye and 20/50 in the right, secondary to pain, photophobia, and excessive tearing. The patient’s right sclera is significantly injected. Intraocular pressure, measured with a tonometer, is 12 to 14 mm Hg. A fluorescein stain of the eye yields no significant findings. The globe is intact.

At first glance, a slit-lamp exam shows no obvious signs of a foreign body. But much higher magnification reveals substantial conjunctival injection and numerous intracorneal linear foreign bodies in the right eye (see Figure 1 for example [not the case patient]). The anterior chamber shows no inflammatory reaction, and findings in the posterior segment are unremarkable.

The initial diagnosis is simple conjunctivitis—but closer examination reveals multiple fine, barbed hairs embedded in the patient’s right cornea. Upon further questioning, the patient reports that prior to symptom onset, he had been holding the classroom pet, a Chilean Rose tarantula, in the palm of his hands.

DISCUSSION

Foreign body injury is a common cause of ocular pain and corneal damage, which can lead to challenging complications. Ophthalmic emergencies account for 2% of ED visits in the US annually and are a major cause of visual impairment.¹ But when a painful eye is the chief complaint, contact with insects, plants, or spiders is rarely included in the differential. Tarantulas are popular classroom and household pets, however, and ocular injury should be suspected in anyone who has been holding a tarantula prior to onset of pain.

Ophthalmia nodosa

Tarantulas are one of the most common arachnids known to cause ophthalmia nodosa—a granulomatous reaction of the conjunctiva or cornea to an implanted plant, insect, or spider hair that typically manifests with photophobia, irritation, and chemosis.^2,3 Tarantulas, when scared or defending their eggs, shoot urticating setae at the threat—a defensive mechanism largely unknown to parents, tarantula owners, and medical professionals.

Urticating setae are found in roughly 90% of tarantula species throughout tropical and subtropical regions.⁴ Depending on the species, setae can be located on the distal prolateral surface of the palpal femur or the dorsum of the abdomen. They can be released when the tarantula scratches its legs against the abdominal urticating setae patch or scratches the palps against the chelicerae (appendages in front of the mouth), or when direct exterior contact is made with the abdominal setae.⁴

There are six types of urticating hairs. Each is attached to the spider’s cuticle by either a stalk (which represents the break-off region) or a socket.⁴ Tarantula hairs range in size from 0.1 mm to 0.3 mm and have a sharp, pointed head and numerous barbs, which help embed them in the target.⁵ They are long and thin, to facilitate deep tissue penetration, and can enter the eyes, lungs, or other body parts (see Figure 2).

Diagnosis and management

Ophthalmic emergencies can affect the visual system and, if left untreated, can lead to permanent vision loss. Affected patients require immediate medical attention and should be referred to an ophthalmologist for follow-up care.

Diagnosis. A thorough history and physical exam are of utmost importance; tiny setae can be easily overlooked if the examiner is not diligent, and the similar symptomatology can lead to misdiagnosis as simple conjunctivitis.³ A visual acuity test and slit-lamp exam are useful for confirmation.

Treatment. Once the diagnosis is confirmed, treatment should consist of mild topical antibiotics and steroids to effectively control infection and inflammation. While topical steroids may be appropriate, local adverse events associated with their use (eg, glaucoma, cataracts) can be problematic. Gentle eye irrigation has been noted by some researchers as contraindicated, while others find it useful to flush out some of the hairs.^5,8,9

Most of the visible protruding tarantula hairs can and should be removed under microscopy during slit-lamp exam. Hairs that are buried in the cornea, however, are nearly impossible to remove and pose a threat of further complications, as described. Conservative management with careful observation is therefore recommended. If the patient develops a granuloma, excision—along with a course of systemic steroids and setae removal via vitrectomy—may be needed.⁹

The good news is that, in many cases, deeper hairs are absorbed without complication, making their removal unnecessary.⁵ Factors that encourage leaving the setae untouched include a large number of hairs, deep corneal penetration, lack of patient tolerance for the procedure, and risk for perforation.³

More invasive treatments (eg, laser photocoagulation, intraocular surgery) to remove offending hairs are possible, but literature on the outcome of these interventions is limited. One report to date used argon laser photocoagulation to treat endophthalmitis from vitreous hairs.¹⁰ The laser can fragment the hairs so that they lose their barbed characteristic and cannot penetrate deeper.⁶

Follow-up. Close follow-up is advised, and patients should be educated on the importance of medication compliance and return visits for reevaluation. Given the potential dangers of handling these spiders, tarantula owners should be advised to use protective gloving and goggles.^2,5,8,9

OUTCOME FOR THE CASE PATIENT

The case patient was sent to an ophthalmologist on day 1. Proparacaine was placed in his right eye, and all of the superficial tarantula hairs were removed using 25- and 30-gauge needles with jeweler forceps under slit-lamp microscopy. Most of the hairs were removed from the superior cornea; fewer were found in the paracentral and inferior regions of the cornea. Approximately five hairs in the paracentral area of the cornea were embedded in the midstromal depth and could not be removed. One drop of ciprofloxacin was administered.

The patient was sent home with an eye shield and instructions to use tobramycin/dexamethasone eye drops (qid in his right eye) and avoid rubbing the eye. (The eye shield, though not technically necessary, was deemed beneficial to help the patient avoid touching the eye.) He was scheduled to return to the clinic one week later.

On follow-up, a careful exam performed under microscopy showed that the five tarantula hairs were still embedded, and an additional six hairs were found in the deep stroma. Superficial punctate keratitis—an eye disorder caused by epithelial cell death on the surface of the cornea—was noted, but no anterior chamber cells were seen. The patient was instructed to continue using the eye drops as prescribed until finished, then start using loteprednol (tid) and artificial lubricating tears (every 2 h).

He returned to the clinic every two weeks for a total of 10 visits. At the end of the treatment course, the remaining tarantula hairs were unable to be removed. The patient used tapering doses of topical eye steroids and antibiotic drops secondary to flare-up.

CONCLUSION

Determining the etiology of ophthalmic emergencies is essential to timely and appropriate management. In this case, a recognized but often overlooked cause, tarantula hairs, made the diagnosis more complicated than simple conjunctivitis. When ocular injury is suspected, the provider must obtain an accurate and detailed history along with a thorough physical exam. Since patients must comply with medication regimens to prevent acute and chronic infection, a clear treatment and follow-up plan should be established. With these in place, ophthalmia nodosa caused by urticating setae can be effectively managed.

IN THIS ARTICLE

• Diagnosis
• Management
• Outcome for the case patient

An otherwise healthy 9-year-old boy is brought to the emergency department (ED) by his father for evaluation of severe pain, blurry vision, and four hours of tearing in his right eye. The patient was in school when he experienced sudden-onset irritation and scratching pain that caused him to rub his eye. He says it “feels like there is something in my eye,” but he denies any known substance or foreign body. He has no medical or surgical history, does not wear contact lenses or eyeglasses, and denies loss of vision. There is no history of recent illness or travel.

On evaluation, the patient is in no acute distress but is holding his right eye closed due to foreign-body sensation and increased photosensitivity and tearing. There is no obvious erythema or swelling in the upper or lower eyelids bilaterally. A visual acuity test with a Snellen eye chart shows 20/20 vision in the left eye and 20/50 in the right, secondary to pain, photophobia, and excessive tearing. The patient’s right sclera is significantly injected. Intraocular pressure, measured with a tonometer, is 12 to 14 mm Hg. A fluorescein stain of the eye yields no significant findings. The globe is intact.

At first glance, a slit-lamp exam shows no obvious signs of a foreign body. But much higher magnification reveals substantial conjunctival injection and numerous intracorneal linear foreign bodies in the right eye (see Figure 1 for example [not the case patient]). The anterior chamber shows no inflammatory reaction, and findings in the posterior segment are unremarkable.

The initial diagnosis is simple conjunctivitis—but closer examination reveals multiple fine, barbed hairs embedded in the patient’s right cornea. Upon further questioning, the patient reports that prior to symptom onset, he had been holding the classroom pet, a Chilean Rose tarantula, in the palm of his hands.

DISCUSSION

Foreign body injury is a common cause of ocular pain and corneal damage, which can lead to challenging complications. Ophthalmic emergencies account for 2% of ED visits in the US annually and are a major cause of visual impairment.¹ But when a painful eye is the chief complaint, contact with insects, plants, or spiders is rarely included in the differential. Tarantulas are popular classroom and household pets, however, and ocular injury should be suspected in anyone who has been holding a tarantula prior to onset of pain.

Ophthalmia nodosa

Tarantulas are one of the most common arachnids known to cause ophthalmia nodosa—a granulomatous reaction of the conjunctiva or cornea to an implanted plant, insect, or spider hair that typically manifests with photophobia, irritation, and chemosis.^2,3 Tarantulas, when scared or defending their eggs, shoot urticating setae at the threat—a defensive mechanism largely unknown to parents, tarantula owners, and medical professionals.

Urticating setae are found in roughly 90% of tarantula species throughout tropical and subtropical regions.⁴ Depending on the species, setae can be located on the distal prolateral surface of the palpal femur or the dorsum of the abdomen. They can be released when the tarantula scratches its legs against the abdominal urticating setae patch or scratches the palps against the chelicerae (appendages in front of the mouth), or when direct exterior contact is made with the abdominal setae.⁴

There are six types of urticating hairs. Each is attached to the spider’s cuticle by either a stalk (which represents the break-off region) or a socket.⁴ Tarantula hairs range in size from 0.1 mm to 0.3 mm and have a sharp, pointed head and numerous barbs, which help embed them in the target.⁵ They are long and thin, to facilitate deep tissue penetration, and can enter the eyes, lungs, or other body parts (see Figure 2).

Diagnosis and management

Ophthalmic emergencies can affect the visual system and, if left untreated, can lead to permanent vision loss. Affected patients require immediate medical attention and should be referred to an ophthalmologist for follow-up care.

Diagnosis. A thorough history and physical exam are of utmost importance; tiny setae can be easily overlooked if the examiner is not diligent, and the similar symptomatology can lead to misdiagnosis as simple conjunctivitis.³ A visual acuity test and slit-lamp exam are useful for confirmation.

Treatment. Once the diagnosis is confirmed, treatment should consist of mild topical antibiotics and steroids to effectively control infection and inflammation. While topical steroids may be appropriate, local adverse events associated with their use (eg, glaucoma, cataracts) can be problematic. Gentle eye irrigation has been noted by some researchers as contraindicated, while others find it useful to flush out some of the hairs.^5,8,9

Most of the visible protruding tarantula hairs can and should be removed under microscopy during slit-lamp exam. Hairs that are buried in the cornea, however, are nearly impossible to remove and pose a threat of further complications, as described. Conservative management with careful observation is therefore recommended. If the patient develops a granuloma, excision—along with a course of systemic steroids and setae removal via vitrectomy—may be needed.⁹

The good news is that, in many cases, deeper hairs are absorbed without complication, making their removal unnecessary.⁵ Factors that encourage leaving the setae untouched include a large number of hairs, deep corneal penetration, lack of patient tolerance for the procedure, and risk for perforation.³

More invasive treatments (eg, laser photocoagulation, intraocular surgery) to remove offending hairs are possible, but literature on the outcome of these interventions is limited. One report to date used argon laser photocoagulation to treat endophthalmitis from vitreous hairs.¹⁰ The laser can fragment the hairs so that they lose their barbed characteristic and cannot penetrate deeper.⁶

Follow-up. Close follow-up is advised, and patients should be educated on the importance of medication compliance and return visits for reevaluation. Given the potential dangers of handling these spiders, tarantula owners should be advised to use protective gloving and goggles.^2,5,8,9

OUTCOME FOR THE CASE PATIENT

The case patient was sent to an ophthalmologist on day 1. Proparacaine was placed in his right eye, and all of the superficial tarantula hairs were removed using 25- and 30-gauge needles with jeweler forceps under slit-lamp microscopy. Most of the hairs were removed from the superior cornea; fewer were found in the paracentral and inferior regions of the cornea. Approximately five hairs in the paracentral area of the cornea were embedded in the midstromal depth and could not be removed. One drop of ciprofloxacin was administered.

The patient was sent home with an eye shield and instructions to use tobramycin/dexamethasone eye drops (qid in his right eye) and avoid rubbing the eye. (The eye shield, though not technically necessary, was deemed beneficial to help the patient avoid touching the eye.) He was scheduled to return to the clinic one week later.

On follow-up, a careful exam performed under microscopy showed that the five tarantula hairs were still embedded, and an additional six hairs were found in the deep stroma. Superficial punctate keratitis—an eye disorder caused by epithelial cell death on the surface of the cornea—was noted, but no anterior chamber cells were seen. The patient was instructed to continue using the eye drops as prescribed until finished, then start using loteprednol (tid) and artificial lubricating tears (every 2 h).

He returned to the clinic every two weeks for a total of 10 visits. At the end of the treatment course, the remaining tarantula hairs were unable to be removed. The patient used tapering doses of topical eye steroids and antibiotic drops secondary to flare-up.

CONCLUSION

Determining the etiology of ophthalmic emergencies is essential to timely and appropriate management. In this case, a recognized but often overlooked cause, tarantula hairs, made the diagnosis more complicated than simple conjunctivitis. When ocular injury is suspected, the provider must obtain an accurate and detailed history along with a thorough physical exam. Since patients must comply with medication regimens to prevent acute and chronic infection, a clear treatment and follow-up plan should be established. With these in place, ophthalmia nodosa caused by urticating setae can be effectively managed.

A 66-year-old man presented to the hospital with 3 days of nausea, vomiting, and abdominal pain. He had come to the emergency department several times during this period, but the cause of his symptoms had not been determined.

The patient was successfully treated with urgent right hemicolectomy.

THE CHILAIDITI SIGN AND SYNDROME

The Chilaiditi sign is an infrequent anomaly found incidentally on chest or abdominal radiography as a colonic interposition between the liver and right hemidiaphragm.¹ It is often asymptomatic but is sometimes accompanied by nausea, vomiting, abdominal pain, and constipation, ie, Chilaiditi syndrome.

Generally, after conservative treatment with fasting and pain control, symptoms may subside and follow-up should be sufficient. However, nasogastric decompression and laxatives are occasionally needed and are often effective in patients with Chilaiditi syndrome. Urgent abdominal surgery is indicated for patients with symptoms of volvulus of the colon, stomach, or small intestine.²

DISTINGUISHING CHARACTERISTICS

The Chilaiditi sign is often confused with pneumoperitoneum, which usually requires urgent abdominal surgery. But the presence of haustration or valvulae conniventes (folds in the small bowel mucosa) in the hepatodiaphragmatic space helps distinguish between intraluminal gas and free air. If the patient presents with abdominal pain without signs of peritonitis, and if imaging indicates the Chilaiditi sign, then supplementary imaging (eg, decubitus radiography, chest CT, abdominal CT) is recommended to make the definitive diagnosis and to avoid unnecessary surgery.

Gas under the diaphragm on standing chest radiography without signs of peritonitis may also be seen after laparotomy and after scuba diving as well as in cases of biliary enteric fistula, incompetent sphincter of Oddi, gallstone ileus, and pneumatosis cystoides intestinalis. The incidence rate of the Chilaiditi sign detected by radiography is between 0.025% and 0.28%.³

PREDISPOSING FACTORS

The cause of the Chilaiditi sign remains unknown. Predisposing factors can be categorized as diaphragmatic (diaphragmatic thinning, phrenic nerve injury, expanded thoracic cavity), intestinal (megacolon, increased intra-abdominal pressure), and hepatic (hepatic atrophy, cirrhosis, ascites).

In healthy people, Chilaiditi syndrome is usually attributed to a congenital abnormal lengthening of the colon or to undue looseness of ligaments of the colon and liver.

Recognizing the Chilaiditi sign is particularly important in patients scheduled to undergo a percutaneous transhepatic procedure or colonoscopic examination, as these procedures increase the risk of perforation.

A 66-year-old man presented to the hospital with 3 days of nausea, vomiting, and abdominal pain. He had come to the emergency department several times during this period, but the cause of his symptoms had not been determined.

The patient was successfully treated with urgent right hemicolectomy.

THE CHILAIDITI SIGN AND SYNDROME

The Chilaiditi sign is an infrequent anomaly found incidentally on chest or abdominal radiography as a colonic interposition between the liver and right hemidiaphragm.¹ It is often asymptomatic but is sometimes accompanied by nausea, vomiting, abdominal pain, and constipation, ie, Chilaiditi syndrome.

Generally, after conservative treatment with fasting and pain control, symptoms may subside and follow-up should be sufficient. However, nasogastric decompression and laxatives are occasionally needed and are often effective in patients with Chilaiditi syndrome. Urgent abdominal surgery is indicated for patients with symptoms of volvulus of the colon, stomach, or small intestine.²

DISTINGUISHING CHARACTERISTICS

The Chilaiditi sign is often confused with pneumoperitoneum, which usually requires urgent abdominal surgery. But the presence of haustration or valvulae conniventes (folds in the small bowel mucosa) in the hepatodiaphragmatic space helps distinguish between intraluminal gas and free air. If the patient presents with abdominal pain without signs of peritonitis, and if imaging indicates the Chilaiditi sign, then supplementary imaging (eg, decubitus radiography, chest CT, abdominal CT) is recommended to make the definitive diagnosis and to avoid unnecessary surgery.

Gas under the diaphragm on standing chest radiography without signs of peritonitis may also be seen after laparotomy and after scuba diving as well as in cases of biliary enteric fistula, incompetent sphincter of Oddi, gallstone ileus, and pneumatosis cystoides intestinalis. The incidence rate of the Chilaiditi sign detected by radiography is between 0.025% and 0.28%.³

PREDISPOSING FACTORS

The cause of the Chilaiditi sign remains unknown. Predisposing factors can be categorized as diaphragmatic (diaphragmatic thinning, phrenic nerve injury, expanded thoracic cavity), intestinal (megacolon, increased intra-abdominal pressure), and hepatic (hepatic atrophy, cirrhosis, ascites).

In healthy people, Chilaiditi syndrome is usually attributed to a congenital abnormal lengthening of the colon or to undue looseness of ligaments of the colon and liver.

Recognizing the Chilaiditi sign is particularly important in patients scheduled to undergo a percutaneous transhepatic procedure or colonoscopic examination, as these procedures increase the risk of perforation.

A 66-year-old man presented to the hospital with 3 days of nausea, vomiting, and abdominal pain. He had come to the emergency department several times during this period, but the cause of his symptoms had not been determined.

The patient was successfully treated with urgent right hemicolectomy.

THE CHILAIDITI SIGN AND SYNDROME

The Chilaiditi sign is an infrequent anomaly found incidentally on chest or abdominal radiography as a colonic interposition between the liver and right hemidiaphragm.¹ It is often asymptomatic but is sometimes accompanied by nausea, vomiting, abdominal pain, and constipation, ie, Chilaiditi syndrome.

Generally, after conservative treatment with fasting and pain control, symptoms may subside and follow-up should be sufficient. However, nasogastric decompression and laxatives are occasionally needed and are often effective in patients with Chilaiditi syndrome. Urgent abdominal surgery is indicated for patients with symptoms of volvulus of the colon, stomach, or small intestine.²

DISTINGUISHING CHARACTERISTICS

The Chilaiditi sign is often confused with pneumoperitoneum, which usually requires urgent abdominal surgery. But the presence of haustration or valvulae conniventes (folds in the small bowel mucosa) in the hepatodiaphragmatic space helps distinguish between intraluminal gas and free air. If the patient presents with abdominal pain without signs of peritonitis, and if imaging indicates the Chilaiditi sign, then supplementary imaging (eg, decubitus radiography, chest CT, abdominal CT) is recommended to make the definitive diagnosis and to avoid unnecessary surgery.

Gas under the diaphragm on standing chest radiography without signs of peritonitis may also be seen after laparotomy and after scuba diving as well as in cases of biliary enteric fistula, incompetent sphincter of Oddi, gallstone ileus, and pneumatosis cystoides intestinalis. The incidence rate of the Chilaiditi sign detected by radiography is between 0.025% and 0.28%.³

PREDISPOSING FACTORS

The cause of the Chilaiditi sign remains unknown. Predisposing factors can be categorized as diaphragmatic (diaphragmatic thinning, phrenic nerve injury, expanded thoracic cavity), intestinal (megacolon, increased intra-abdominal pressure), and hepatic (hepatic atrophy, cirrhosis, ascites).

In healthy people, Chilaiditi syndrome is usually attributed to a congenital abnormal lengthening of the colon or to undue looseness of ligaments of the colon and liver.

Recognizing the Chilaiditi sign is particularly important in patients scheduled to undergo a percutaneous transhepatic procedure or colonoscopic examination, as these procedures increase the risk of perforation.

A 75-year-old man presented to the emergency department for evaluation of abdominal pain. He had stage 3 chronic obstructive pulmonary disease (COPD), with a forced expiratory volume in 1 second of 33%.

PREVIOUS HOSPITALIZATION

Aside from his COPD, he had been healthy until 1 month earlier, when he had been hospitalized because of shortness of breath and chest pressure with exertion. His troponin T level had been elevated, peaking at 0.117 ng/mL (reference range 0–0.029).

Left heart catheterization had shown no significant coronary artery disease. A myocardial bridge of the distal left anterior descending coronary artery had been seen, so that the artery appeared to be narrowed by 50% to 60% with ventricular contraction. But this was not thought to have been the cause of his presentation.

On discharge, he required oxygen 4 L/min by nasal cannula. Previously, he had not needed supplemental oxygen.

CURRENT PRESENTATION

The patient described persistent and severe periumbilical abdominal pain during the previous day. It was not associated with eating, and he denied diarrhea, constipation, hematemesis, hematochezia, bright red blood per rectum, or melena. He continued to describe persistent shortness of breath and pleuritic chest pain. His vital signs were as follows:

• Heart rate 104 beats per minute
• Respiratory rate 16 to 20 breaths per minute
• Blood pressure 101–142/62–84 mm Hg
• Oxygen saturation 78% on room air.

His laboratory findings on presentation are shown in Table 1, and his electrocardiogram is shown in Figure 1.

WHAT DOES HIS ELECTROCARDIOGRAM SHOW?

1. Which of the following is the most accurate description of this patient’s electrocardiogram?

• Sinus tachycardia, peaked P waves (P pulmonale) in lead II, and T-wave inversions in the right precordial leads
• Sinus tachycardia and left bundle branch block
• Sinus tachycardia and poor R-wave progression
• Sinus tachycardia and ST elevation in the precordial leads

Our patient’s electrocardiogram shows sinus tachycardia, P pulmonale, T-wave inversion in the right precordial leads (V₁–V₃), and biphasic T waves in lead V_4,, which suggest right ventricular strain.

The rhythm most commonly seen in patients with pulmonary embolism is sinus tachycardia, followed by nonspecific ST-segment or T-wave abnormalities. In one series of patients with acute pulmonary embolism, the classic findings of P pulmonale, right ventricular hypertrophy, right axis deviation, and right bundle branch block were rare (< 6%).¹ Thus, these classic findings are not sensitive for the diagnosis of pulmonary embolism, and their absence does not rule it out.

Further studies for our patient

Transthoracic echocardiography was performed to look for evidence of right ventricular strain secondary to the pulmonary embolism.

2. Which of the following findings on transthoracic echocardiography would not suggest acute pulmonary embolism?

• Midright ventricular wall hypokinesis with apical sparing
• Severe tricuspid regurgitation
• Left ventricular dilation
• Lack of respiratory variation of the inferior vena cava
• Septal wall motion toward the left ventricle

Left ventricular dilation does not suggest acute pulmonary embolism. Echocardiograms of patients with acute submassive pulmonary embolism typically show evidence of right ventricular strain, such as the other entities listed above (midright ventricular hypokinesis with apical sparing, severe tricuspid regurgitation, lack of respiratory variation of the inferior vena cava, and septal wall motion toward the left ventricle).

The degree of right ventricular dysfunction is related to the extent of acute pulmonary vascular occlusion and aids in risk-stratification of patients with acute pulmonary embolism. Midright ventricular wall hypokinesis with apical sparing has been termed the McConnell sign.²

In our patient, transthoracic echocardiography showed:

• Normal left ventricular ejection fraction
• Mild diastolic dysfunction
• Right ventricular dilation with moderately decreased right ventricular systolic function and apical sparing
• Right ventricular systolic pressure 54 mm Hg, consistent with moderate pulmonary hypertension
• Right atrial pressure 10 mm Hg
• No inspiratory collapse of a dilated inferior vena cava
• Mild tricuspid valve regurgitation.

CLASSIFICATION OF ACUTE PULMONARY EMBOLISM

3. Given the above information, how would you classify the patient’s pulmonary embolism?

• Massive
• Submassive
• Low-risk
• Clinically stable

The patient’s pulmonary embolism is submassive.

Historically, the classification of pulmonary embolism was determined by the angiographic thrombus burden. However, this has limited utility because clinical factors (eg, hypotension on initial presentation) have been shown to be better predictors of short-term mortality risk.³

Our patient is characterized as having a submassive pulmonary embolism based on elevated biomarkers (troponin T, N-terminal pro-B-type natriuretic peptide) and right ventricular dysfunction in the absence of hypotension.

ULTRASONOGRAPHY FOR DIAGNOSIS OF DEEP VEIN THROMBOSIS

Venous duplex ultrasonography has become the standard for diagnosis of lower extremity deep vein thrombosis. However, its quality and diagnostic accuracy depend on the skill of the person performing the examination. It is further limited by certain patient characteristics, including severe obesity, edema, and wounds and dressings at the site being examined.⁵

Our patient underwent duplex ultrasonography of the lower extremities, which demonstrated acute proximal and calf deep vein thrombosis in the right femoral, popliteal, and peroneal veins and no deep vein thrombosis in the left leg.

Multiple models exist to estimate the risk of complications in patients with acute pulmonary embolism.

The Bova score⁶ is based on the following factors:

• Systolic blood pressure 90–100 mm Hg (2 points) (patients with systolic blood pressure lower than 90 mm Hg were excluded from the study from which this score was derived)
• Cardiac troponin elevation (2 points)
• Right ventricular dysfunction on echocardiography or computed tomography (2 points)
• Heart rate 100 beats/min or greater (1 point).

A total score of 0, 1, or 2 (stage I) denotes low risk, 3 or 4 points (stage II) intermediate risk, and more than 4 points (stage III) high risk.

The PESI score (Pulmonary Embolism Severity Index)⁷ is based on:

• Age (1 point per year)
• Sex (10 points for being male)
• Heart rate 110 per minute or greater (20 points)
• Cancer (30 points)
• Heart failure (10 points)
• Chronic lung disease (10 points)
• Systolic blood pressure less than 100 mm Hg (30 points)
• Respiratory rate at least 30 per minute (20 points)
• Temperature less than 36ºC (20 points)
• Altered mental status (60 points)
• Arterial oxygen saturation less than 90% (20 points).

The total score is broken down into 5 classes: I (< 65 points), II (65–85), III (86–105), IV (106–125), and V (> 126). Classes I and II are low risk, and the higher ones are high risk.

The simplified PESI score⁸ was developed to more rapidly risk-stratify patients and has been found to be similar to the PESI score in prognostic accuracy. Patients get 1 point for each of the following:

• Age over 80
• Cancer
• Chronic cardiopulmonary disease (heart failure or chronic lung disease)
• Heart rate 110 per minute or greater
• Systolic blood pressure less than 100 mm Hg
• Arterial oxygen saturation less than 90%.

A total score of 0 is low risk; anything higher is high risk.

Back to our patient

Our patient had proximal and calf deep vein thrombosis of the right leg, bilateral submassive pulmonary emboli with associated biomarker elevation and right ventricular dysfunction, and left renal artery thrombosis with infarction. Using the PESI score, his risk of death in the next 30 days was 13.7% and his 30-day risk of a complicated course was 27%. Using the Bova score, his 30-day risk of death was 15.5% and his 30-day risk of a complicated course was 29.2%.^6,7

Notably, the patient’s right ventricular function had also been impaired on the echocardiogram performed during his admission 1 month previously. On transthoracic echocardiography during the current admission, the patient was found to have a similar degree of right ventricular dysfunction. This finding, along with the oxygen requirement that developed during the earlier admission, suggested that his pulmonary embolism may have been subacute and that the diagnosis may have been missed during the earlier hospital stay.

The patient was treated with unfractionated heparin. After the hospital’s multidisciplinary pulmonary embolism response team discussed and weighed the above factors, they recommended to not pursue thrombolytic therapy or inferior vena cava filter placement.

Of note, the patient’s pulses in the left lower extremity continued to be weak but palpable, and the left leg was cooler to touch than the right leg.

ASSESSING PERIPHERAL ARTERY DISEASE

4. How should the finding of weak pulses in this patient’s left leg be initially investigated?

• Computed tomographic angiography with runoff
• Ankle-brachial indices with pulse-volume recordings
• Arterial duplex ultrasonography
• Magnetic resonance angiography of the lower extremities

The ankle-brachial index is the initial diagnostic test for assessment of pulse abnormalities and for diagnosis of lower-extremity peripheral artery disease. It is calculated by dividing the higher of the ankle systolic pressures (posterior tibial or dorsalis pedis) by the higher of the 2 brachial pressures (left or right).⁹ Normal values are between 1.00 and 1.40.

Ankle-brachial indices in our patient

Our patient underwent measurement of his brachial, dorsalis pedis, and posterior tibial artery systolic pressures using blood pressure cuffs and continuous-wave Doppler. Ankle pulse-volume recordings were also obtained.

Given the patient’s poor renal function and concern for acute renal infarction, we thought it best to avoid iodinated or gadolinium contrast, such as with magnetic resonance or computed tomographic angiography.

Segmental leg pressures and pulse-volume recordings can be performed to help localize the level of arterial disease in the extremities, but were not done in this case because of the extensive deep vein thrombosis in the right leg.^10,11

Arterial ultrasonography in our patient

Arterial duplex ultrasonography was performed to help determine the location of arterial disease. It showed patent arteries in the right leg. In the left lower extremity there was slow, monophasic blood flow in the distal superficial femoral artery. The popliteal artery was occluded. The posterior tibial artery was occluded at the origin, with reconstitution distally. The peroneal artery was occluded throughout. The anterior tibial artery was patent throughout. The ultrasonographic findings were thought to be suspicious for arterial thromboembolism.

5. Given that the patient had both arterial thrombosis (renal artery, lower-extremity arteries) and venous thromboembolism (deep vein thrombosis and pulmonary embolism), which of the following would be included in the differential diagnosis?

• Antiphospholipid antibody syndrome
• Protein C or protein S deficiency
• Malignancy
• Paradoxical embolization
• Factor V Leiden mutation

Correct answers include antiphospholipid antibody syndrome, malignancy, and paradoxical embolization.

The differential diagnosis for concomitant venous and arterial thrombosis is broad,¹² and includes the following:

• Structural factors: patent foramen ovale, popliteal artery aneurysm
• Malignancy
• Inflammatory diseases: Behçet disease, Buerger disease, inflammatory bowel disease, antiphospholipid antibody syndrome, elevated lipoprotein(a), elevated homocysteine
• Hematologic diseases: myelodysplastic syndrome, disseminated intravascular coagulation, paroxysmal nocturnal hemoglobinuria, heparin-induced thrombocytopenia.

Traditional risk factors for venous thromboembolism include protein C deficiency, protein S deficiency, factor V Leiden mutation, the prothrombin G20210A gene mutation, and others. These are relatively minor risk factors for venous thrombosis and do not pose a risk for arterial thrombosis.¹² In contrast, antiphospholipid antibody syndrome and malignancy pose a risk for both venous and arterial thrombosis. Paradoxical embolism is a mechanism by which arterial thrombosis (emboli) can develop in the setting of existing venous thrombosis.¹²

Our patient underwent testing for antiphospholipid antibodies and lupus anticoagulant, and he was encouraged to undergo age-appropriate cancer screening as an outpatient.¹²

ANTIPHOSPHOLIPID ANTIBODY SYNDROME

Antiphospholipid antibody syndrome is defined by both clinical and laboratory criteria. Clinical symptoms include vascular thrombosis (arterial, venous, or both) and pregnancy-related complications.¹³

Laboratory criteria require the presence of antiphospholipid antibodies or lupus anticoagulant. These must be confirmed with repeat testing in 12 weeks. Antiphospholipid antibodies are detected by an enzyme-linked immunosorbent assay; laboratory assessment for the presence of lupus anticoagulant is a stepwise process and relies on 4 criteria:

• There should be prolongation of a phospholipid-dependent clotting test (eg, activated partial thromboplastin time, dilute Russell viper venom time test).
• There must be evidence of an inhibitory activity with mixing study.
• The inhibitor must exhibit phospholipid dependence; that is, with more phospholipid there is shortening of clotting time.
• Specific inhibitors must be excluded, including factor VIII and anticoagulant drugs such as heparin.^14–17

From Houghton DE, Moll S. Antiphospholipid antibodies. Vasc Med 2017; 22:545–550.

Clinically, one should consider antiphospholipid antibody syndrome in patients who have arterial thrombosis, a history of pregnancy morbidity, or unexplained prolongation of activated partial thromboplastin time.¹³

Antiphospholipid antibodies may be present in up to a quarter of patients with venous thromboembolism, but it is persistent positivity of antibody assays that is associated with increased future risk of venous thromboembolism.¹⁹ Of note, the risk of venous thromboembolism in patients with confirmed antiphospholipid antibody syndrome is 10 times higher than in the general population.²⁰

ANTIPHOSPHOLIPID ANTIBODIES ARE NOT ALL THE SAME

6. Which of the following antiphospholipid antibodies have not been associated with an increased thrombotic risk?

• Anti-beta-2-glycoprotein I IgG
• Lupus anticoagulant
• Antiphosphatidylserine
• Anticardiolipin IgM
• Anticardiolipin IgG

The correct answer is antiphosphatidylserine.¹⁵

Antiphospholipid antibodies are directed against a portion of select plasma proteins that are uncovered upon phospholipid binding. While lupus anticoagulant, anti-beta-2-glycoprotein I, and anticardiolipin antibodies are associated with thrombosis, antiprothrombin antibodies (including antiprothrombin and antiphosphatidylserine antibodies) are not.^15,21

Patent foramen ovale, a communication between the right and left atrium in the interatrial septum, is associated with an increased risk of paradoxical embolization. The prevalence of patent foramen ovale is estimated to be 27% to 29% in the general population.²² Noncerebral systemic paradoxical embolism occurs less frequently than cerebral embolism, accounting for approximately 5% to 10% of paradoxical emboli.²²

To evaluate for patent foramen ovale, transthoracic echocardiography is performed with a bubble (agitated saline contrast) study to assess for interatrial shunting. Transesophageal echocardiography or transcranial Doppler bubble studies may also be performed.

Although patent foramen ovale is most commonly associated with cerebral embolism, peripheral emboli can occur. Some research suggests that this may be a more common cause of arterial thromboembolism in younger patients. There have also been reports of other sites of systemic embolization, including the renal artery.¹²

Back to our patient

Initial antiphospholipid antibody testing was positive for lupus anticoagulant. Anticardiolipin and anti-beta-2-glycoprotein I antibodies were not detected.

Transesophageal echocardiography revealed a patent foramen ovale with a highly mobile atrial septum (atrial septal aneurysm).

The patient was treated with intravenous unfractionated heparin with bridging to warfarin with a target international normalized ratio (INR) of 2 to 3. His renal artery infarction and his lower-extremity arterial thromboembolic event were conservatively managed. His respiratory status improved, and he no longer required supplemental oxygen. His creatinine peaked at 1.7 mg/dL during his admission and improved to 1.2 mg/dL before he was discharged.

At follow-up, repeat echocardiography showed that his right ventricular systolic pressure had improved (decreased) to 37 mm Hg from 54 mm Hg. Repeat confirmatory testing was positive for lupus anticoagulant 12 weeks later. He has been maintained on warfarin with an INR goal of 2 to 3 as well as low-dose aspirin with plans for long-term anticoagulation. We decided to keep the patient on anticoagulation indefinitely with warfarin; he was not a candidate for a direct oral anticoagulant, given limited data on the use of these agents in the setting of lupus anticoagulant and antiphospholipid antibody syndrome.

SUMMARY OF CASE

In summary, this patient was a 75-year-old man with COPD who presented with abdominal pain. He was noted to have a left renal infarction, extensive unprovoked lower-extremity deep vein thrombosis with pulmonary emboli, and lower limb arterial thromboembolism.

He also had an underlying hypercoagulable state—antiphospholipid antibody syndrome—that predisposed him to both arterial and venous thrombosis. He was ultimately found to have a patent foramen ovale, which further increased the risk of arterial thrombosis by facilitating paradoxical embolization of venous thrombi. It is not certain whether the renal infarction and leg artery thrombi were due to paradoxical embolism or to in situ thrombosis, but we believe that it was most likely paradoxical embolization.        

A 75-year-old man presented to the emergency department for evaluation of abdominal pain. He had stage 3 chronic obstructive pulmonary disease (COPD), with a forced expiratory volume in 1 second of 33%.

PREVIOUS HOSPITALIZATION

Aside from his COPD, he had been healthy until 1 month earlier, when he had been hospitalized because of shortness of breath and chest pressure with exertion. His troponin T level had been elevated, peaking at 0.117 ng/mL (reference range 0–0.029).

Left heart catheterization had shown no significant coronary artery disease. A myocardial bridge of the distal left anterior descending coronary artery had been seen, so that the artery appeared to be narrowed by 50% to 60% with ventricular contraction. But this was not thought to have been the cause of his presentation.

On discharge, he required oxygen 4 L/min by nasal cannula. Previously, he had not needed supplemental oxygen.

CURRENT PRESENTATION

The patient described persistent and severe periumbilical abdominal pain during the previous day. It was not associated with eating, and he denied diarrhea, constipation, hematemesis, hematochezia, bright red blood per rectum, or melena. He continued to describe persistent shortness of breath and pleuritic chest pain. His vital signs were as follows:

• Heart rate 104 beats per minute
• Respiratory rate 16 to 20 breaths per minute
• Blood pressure 101–142/62–84 mm Hg
• Oxygen saturation 78% on room air.

His laboratory findings on presentation are shown in Table 1, and his electrocardiogram is shown in Figure 1.

WHAT DOES HIS ELECTROCARDIOGRAM SHOW?

1. Which of the following is the most accurate description of this patient’s electrocardiogram?

• Sinus tachycardia, peaked P waves (P pulmonale) in lead II, and T-wave inversions in the right precordial leads
• Sinus tachycardia and left bundle branch block
• Sinus tachycardia and poor R-wave progression
• Sinus tachycardia and ST elevation in the precordial leads

Our patient’s electrocardiogram shows sinus tachycardia, P pulmonale, T-wave inversion in the right precordial leads (V₁–V₃), and biphasic T waves in lead V_4,, which suggest right ventricular strain.

The rhythm most commonly seen in patients with pulmonary embolism is sinus tachycardia, followed by nonspecific ST-segment or T-wave abnormalities. In one series of patients with acute pulmonary embolism, the classic findings of P pulmonale, right ventricular hypertrophy, right axis deviation, and right bundle branch block were rare (< 6%).¹ Thus, these classic findings are not sensitive for the diagnosis of pulmonary embolism, and their absence does not rule it out.

Further studies for our patient

Transthoracic echocardiography was performed to look for evidence of right ventricular strain secondary to the pulmonary embolism.

2. Which of the following findings on transthoracic echocardiography would not suggest acute pulmonary embolism?

• Midright ventricular wall hypokinesis with apical sparing
• Severe tricuspid regurgitation
• Left ventricular dilation
• Lack of respiratory variation of the inferior vena cava
• Septal wall motion toward the left ventricle

Left ventricular dilation does not suggest acute pulmonary embolism. Echocardiograms of patients with acute submassive pulmonary embolism typically show evidence of right ventricular strain, such as the other entities listed above (midright ventricular hypokinesis with apical sparing, severe tricuspid regurgitation, lack of respiratory variation of the inferior vena cava, and septal wall motion toward the left ventricle).

The degree of right ventricular dysfunction is related to the extent of acute pulmonary vascular occlusion and aids in risk-stratification of patients with acute pulmonary embolism. Midright ventricular wall hypokinesis with apical sparing has been termed the McConnell sign.²

In our patient, transthoracic echocardiography showed:

• Normal left ventricular ejection fraction
• Mild diastolic dysfunction
• Right ventricular dilation with moderately decreased right ventricular systolic function and apical sparing
• Right ventricular systolic pressure 54 mm Hg, consistent with moderate pulmonary hypertension
• Right atrial pressure 10 mm Hg
• No inspiratory collapse of a dilated inferior vena cava
• Mild tricuspid valve regurgitation.

CLASSIFICATION OF ACUTE PULMONARY EMBOLISM

3. Given the above information, how would you classify the patient’s pulmonary embolism?

• Massive
• Submassive
• Low-risk
• Clinically stable

The patient’s pulmonary embolism is submassive.

Historically, the classification of pulmonary embolism was determined by the angiographic thrombus burden. However, this has limited utility because clinical factors (eg, hypotension on initial presentation) have been shown to be better predictors of short-term mortality risk.³

Our patient is characterized as having a submassive pulmonary embolism based on elevated biomarkers (troponin T, N-terminal pro-B-type natriuretic peptide) and right ventricular dysfunction in the absence of hypotension.

ULTRASONOGRAPHY FOR DIAGNOSIS OF DEEP VEIN THROMBOSIS

Venous duplex ultrasonography has become the standard for diagnosis of lower extremity deep vein thrombosis. However, its quality and diagnostic accuracy depend on the skill of the person performing the examination. It is further limited by certain patient characteristics, including severe obesity, edema, and wounds and dressings at the site being examined.⁵

Our patient underwent duplex ultrasonography of the lower extremities, which demonstrated acute proximal and calf deep vein thrombosis in the right femoral, popliteal, and peroneal veins and no deep vein thrombosis in the left leg.

Multiple models exist to estimate the risk of complications in patients with acute pulmonary embolism.

The Bova score⁶ is based on the following factors:

• Systolic blood pressure 90–100 mm Hg (2 points) (patients with systolic blood pressure lower than 90 mm Hg were excluded from the study from which this score was derived)
• Cardiac troponin elevation (2 points)
• Right ventricular dysfunction on echocardiography or computed tomography (2 points)
• Heart rate 100 beats/min or greater (1 point).

A total score of 0, 1, or 2 (stage I) denotes low risk, 3 or 4 points (stage II) intermediate risk, and more than 4 points (stage III) high risk.

The PESI score (Pulmonary Embolism Severity Index)⁷ is based on:

• Age (1 point per year)
• Sex (10 points for being male)
• Heart rate 110 per minute or greater (20 points)
• Cancer (30 points)
• Heart failure (10 points)
• Chronic lung disease (10 points)
• Systolic blood pressure less than 100 mm Hg (30 points)
• Respiratory rate at least 30 per minute (20 points)
• Temperature less than 36ºC (20 points)
• Altered mental status (60 points)
• Arterial oxygen saturation less than 90% (20 points).

The total score is broken down into 5 classes: I (< 65 points), II (65–85), III (86–105), IV (106–125), and V (> 126). Classes I and II are low risk, and the higher ones are high risk.

The simplified PESI score⁸ was developed to more rapidly risk-stratify patients and has been found to be similar to the PESI score in prognostic accuracy. Patients get 1 point for each of the following:

• Age over 80
• Cancer
• Chronic cardiopulmonary disease (heart failure or chronic lung disease)
• Heart rate 110 per minute or greater
• Systolic blood pressure less than 100 mm Hg
• Arterial oxygen saturation less than 90%.

A total score of 0 is low risk; anything higher is high risk.

Back to our patient

Our patient had proximal and calf deep vein thrombosis of the right leg, bilateral submassive pulmonary emboli with associated biomarker elevation and right ventricular dysfunction, and left renal artery thrombosis with infarction. Using the PESI score, his risk of death in the next 30 days was 13.7% and his 30-day risk of a complicated course was 27%. Using the Bova score, his 30-day risk of death was 15.5% and his 30-day risk of a complicated course was 29.2%.^6,7

Notably, the patient’s right ventricular function had also been impaired on the echocardiogram performed during his admission 1 month previously. On transthoracic echocardiography during the current admission, the patient was found to have a similar degree of right ventricular dysfunction. This finding, along with the oxygen requirement that developed during the earlier admission, suggested that his pulmonary embolism may have been subacute and that the diagnosis may have been missed during the earlier hospital stay.

The patient was treated with unfractionated heparin. After the hospital’s multidisciplinary pulmonary embolism response team discussed and weighed the above factors, they recommended to not pursue thrombolytic therapy or inferior vena cava filter placement.

Of note, the patient’s pulses in the left lower extremity continued to be weak but palpable, and the left leg was cooler to touch than the right leg.

ASSESSING PERIPHERAL ARTERY DISEASE

4. How should the finding of weak pulses in this patient’s left leg be initially investigated?

• Computed tomographic angiography with runoff
• Ankle-brachial indices with pulse-volume recordings
• Arterial duplex ultrasonography
• Magnetic resonance angiography of the lower extremities

The ankle-brachial index is the initial diagnostic test for assessment of pulse abnormalities and for diagnosis of lower-extremity peripheral artery disease. It is calculated by dividing the higher of the ankle systolic pressures (posterior tibial or dorsalis pedis) by the higher of the 2 brachial pressures (left or right).⁹ Normal values are between 1.00 and 1.40.

Ankle-brachial indices in our patient

Our patient underwent measurement of his brachial, dorsalis pedis, and posterior tibial artery systolic pressures using blood pressure cuffs and continuous-wave Doppler. Ankle pulse-volume recordings were also obtained.

Given the patient’s poor renal function and concern for acute renal infarction, we thought it best to avoid iodinated or gadolinium contrast, such as with magnetic resonance or computed tomographic angiography.

Segmental leg pressures and pulse-volume recordings can be performed to help localize the level of arterial disease in the extremities, but were not done in this case because of the extensive deep vein thrombosis in the right leg.^10,11

Arterial ultrasonography in our patient

Arterial duplex ultrasonography was performed to help determine the location of arterial disease. It showed patent arteries in the right leg. In the left lower extremity there was slow, monophasic blood flow in the distal superficial femoral artery. The popliteal artery was occluded. The posterior tibial artery was occluded at the origin, with reconstitution distally. The peroneal artery was occluded throughout. The anterior tibial artery was patent throughout. The ultrasonographic findings were thought to be suspicious for arterial thromboembolism.

5. Given that the patient had both arterial thrombosis (renal artery, lower-extremity arteries) and venous thromboembolism (deep vein thrombosis and pulmonary embolism), which of the following would be included in the differential diagnosis?

• Antiphospholipid antibody syndrome
• Protein C or protein S deficiency
• Malignancy
• Paradoxical embolization
• Factor V Leiden mutation

Correct answers include antiphospholipid antibody syndrome, malignancy, and paradoxical embolization.

The differential diagnosis for concomitant venous and arterial thrombosis is broad,¹² and includes the following:

• Structural factors: patent foramen ovale, popliteal artery aneurysm
• Malignancy
• Inflammatory diseases: Behçet disease, Buerger disease, inflammatory bowel disease, antiphospholipid antibody syndrome, elevated lipoprotein(a), elevated homocysteine
• Hematologic diseases: myelodysplastic syndrome, disseminated intravascular coagulation, paroxysmal nocturnal hemoglobinuria, heparin-induced thrombocytopenia.

Traditional risk factors for venous thromboembolism include protein C deficiency, protein S deficiency, factor V Leiden mutation, the prothrombin G20210A gene mutation, and others. These are relatively minor risk factors for venous thrombosis and do not pose a risk for arterial thrombosis.¹² In contrast, antiphospholipid antibody syndrome and malignancy pose a risk for both venous and arterial thrombosis. Paradoxical embolism is a mechanism by which arterial thrombosis (emboli) can develop in the setting of existing venous thrombosis.¹²

Our patient underwent testing for antiphospholipid antibodies and lupus anticoagulant, and he was encouraged to undergo age-appropriate cancer screening as an outpatient.¹²

ANTIPHOSPHOLIPID ANTIBODY SYNDROME

Antiphospholipid antibody syndrome is defined by both clinical and laboratory criteria. Clinical symptoms include vascular thrombosis (arterial, venous, or both) and pregnancy-related complications.¹³

Laboratory criteria require the presence of antiphospholipid antibodies or lupus anticoagulant. These must be confirmed with repeat testing in 12 weeks. Antiphospholipid antibodies are detected by an enzyme-linked immunosorbent assay; laboratory assessment for the presence of lupus anticoagulant is a stepwise process and relies on 4 criteria:

• There should be prolongation of a phospholipid-dependent clotting test (eg, activated partial thromboplastin time, dilute Russell viper venom time test).
• There must be evidence of an inhibitory activity with mixing study.
• The inhibitor must exhibit phospholipid dependence; that is, with more phospholipid there is shortening of clotting time.
• Specific inhibitors must be excluded, including factor VIII and anticoagulant drugs such as heparin.^14–17

From Houghton DE, Moll S. Antiphospholipid antibodies. Vasc Med 2017; 22:545–550.

Clinically, one should consider antiphospholipid antibody syndrome in patients who have arterial thrombosis, a history of pregnancy morbidity, or unexplained prolongation of activated partial thromboplastin time.¹³

Antiphospholipid antibodies may be present in up to a quarter of patients with venous thromboembolism, but it is persistent positivity of antibody assays that is associated with increased future risk of venous thromboembolism.¹⁹ Of note, the risk of venous thromboembolism in patients with confirmed antiphospholipid antibody syndrome is 10 times higher than in the general population.²⁰

ANTIPHOSPHOLIPID ANTIBODIES ARE NOT ALL THE SAME

6. Which of the following antiphospholipid antibodies have not been associated with an increased thrombotic risk?

• Anti-beta-2-glycoprotein I IgG
• Lupus anticoagulant
• Antiphosphatidylserine
• Anticardiolipin IgM
• Anticardiolipin IgG

The correct answer is antiphosphatidylserine.¹⁵

Antiphospholipid antibodies are directed against a portion of select plasma proteins that are uncovered upon phospholipid binding. While lupus anticoagulant, anti-beta-2-glycoprotein I, and anticardiolipin antibodies are associated with thrombosis, antiprothrombin antibodies (including antiprothrombin and antiphosphatidylserine antibodies) are not.^15,21

Patent foramen ovale, a communication between the right and left atrium in the interatrial septum, is associated with an increased risk of paradoxical embolization. The prevalence of patent foramen ovale is estimated to be 27% to 29% in the general population.²² Noncerebral systemic paradoxical embolism occurs less frequently than cerebral embolism, accounting for approximately 5% to 10% of paradoxical emboli.²²

To evaluate for patent foramen ovale, transthoracic echocardiography is performed with a bubble (agitated saline contrast) study to assess for interatrial shunting. Transesophageal echocardiography or transcranial Doppler bubble studies may also be performed.

Although patent foramen ovale is most commonly associated with cerebral embolism, peripheral emboli can occur. Some research suggests that this may be a more common cause of arterial thromboembolism in younger patients. There have also been reports of other sites of systemic embolization, including the renal artery.¹²

Back to our patient

Initial antiphospholipid antibody testing was positive for lupus anticoagulant. Anticardiolipin and anti-beta-2-glycoprotein I antibodies were not detected.

Transesophageal echocardiography revealed a patent foramen ovale with a highly mobile atrial septum (atrial septal aneurysm).

The patient was treated with intravenous unfractionated heparin with bridging to warfarin with a target international normalized ratio (INR) of 2 to 3. His renal artery infarction and his lower-extremity arterial thromboembolic event were conservatively managed. His respiratory status improved, and he no longer required supplemental oxygen. His creatinine peaked at 1.7 mg/dL during his admission and improved to 1.2 mg/dL before he was discharged.

At follow-up, repeat echocardiography showed that his right ventricular systolic pressure had improved (decreased) to 37 mm Hg from 54 mm Hg. Repeat confirmatory testing was positive for lupus anticoagulant 12 weeks later. He has been maintained on warfarin with an INR goal of 2 to 3 as well as low-dose aspirin with plans for long-term anticoagulation. We decided to keep the patient on anticoagulation indefinitely with warfarin; he was not a candidate for a direct oral anticoagulant, given limited data on the use of these agents in the setting of lupus anticoagulant and antiphospholipid antibody syndrome.

SUMMARY OF CASE

In summary, this patient was a 75-year-old man with COPD who presented with abdominal pain. He was noted to have a left renal infarction, extensive unprovoked lower-extremity deep vein thrombosis with pulmonary emboli, and lower limb arterial thromboembolism.

He also had an underlying hypercoagulable state—antiphospholipid antibody syndrome—that predisposed him to both arterial and venous thrombosis. He was ultimately found to have a patent foramen ovale, which further increased the risk of arterial thrombosis by facilitating paradoxical embolization of venous thrombi. It is not certain whether the renal infarction and leg artery thrombi were due to paradoxical embolism or to in situ thrombosis, but we believe that it was most likely paradoxical embolization.        

A 75-year-old man presented to the emergency department for evaluation of abdominal pain. He had stage 3 chronic obstructive pulmonary disease (COPD), with a forced expiratory volume in 1 second of 33%.

PREVIOUS HOSPITALIZATION

Aside from his COPD, he had been healthy until 1 month earlier, when he had been hospitalized because of shortness of breath and chest pressure with exertion. His troponin T level had been elevated, peaking at 0.117 ng/mL (reference range 0–0.029).

Left heart catheterization had shown no significant coronary artery disease. A myocardial bridge of the distal left anterior descending coronary artery had been seen, so that the artery appeared to be narrowed by 50% to 60% with ventricular contraction. But this was not thought to have been the cause of his presentation.

On discharge, he required oxygen 4 L/min by nasal cannula. Previously, he had not needed supplemental oxygen.

CURRENT PRESENTATION

The patient described persistent and severe periumbilical abdominal pain during the previous day. It was not associated with eating, and he denied diarrhea, constipation, hematemesis, hematochezia, bright red blood per rectum, or melena. He continued to describe persistent shortness of breath and pleuritic chest pain. His vital signs were as follows:

• Heart rate 104 beats per minute
• Respiratory rate 16 to 20 breaths per minute
• Blood pressure 101–142/62–84 mm Hg
• Oxygen saturation 78% on room air.

His laboratory findings on presentation are shown in Table 1, and his electrocardiogram is shown in Figure 1.

WHAT DOES HIS ELECTROCARDIOGRAM SHOW?

1. Which of the following is the most accurate description of this patient’s electrocardiogram?

• Sinus tachycardia, peaked P waves (P pulmonale) in lead II, and T-wave inversions in the right precordial leads
• Sinus tachycardia and left bundle branch block
• Sinus tachycardia and poor R-wave progression
• Sinus tachycardia and ST elevation in the precordial leads

Our patient’s electrocardiogram shows sinus tachycardia, P pulmonale, T-wave inversion in the right precordial leads (V₁–V₃), and biphasic T waves in lead V_4,, which suggest right ventricular strain.

The rhythm most commonly seen in patients with pulmonary embolism is sinus tachycardia, followed by nonspecific ST-segment or T-wave abnormalities. In one series of patients with acute pulmonary embolism, the classic findings of P pulmonale, right ventricular hypertrophy, right axis deviation, and right bundle branch block were rare (< 6%).¹ Thus, these classic findings are not sensitive for the diagnosis of pulmonary embolism, and their absence does not rule it out.

Further studies for our patient

Transthoracic echocardiography was performed to look for evidence of right ventricular strain secondary to the pulmonary embolism.

2. Which of the following findings on transthoracic echocardiography would not suggest acute pulmonary embolism?

• Midright ventricular wall hypokinesis with apical sparing
• Severe tricuspid regurgitation
• Left ventricular dilation
• Lack of respiratory variation of the inferior vena cava
• Septal wall motion toward the left ventricle

Left ventricular dilation does not suggest acute pulmonary embolism. Echocardiograms of patients with acute submassive pulmonary embolism typically show evidence of right ventricular strain, such as the other entities listed above (midright ventricular hypokinesis with apical sparing, severe tricuspid regurgitation, lack of respiratory variation of the inferior vena cava, and septal wall motion toward the left ventricle).

The degree of right ventricular dysfunction is related to the extent of acute pulmonary vascular occlusion and aids in risk-stratification of patients with acute pulmonary embolism. Midright ventricular wall hypokinesis with apical sparing has been termed the McConnell sign.²

In our patient, transthoracic echocardiography showed:

• Normal left ventricular ejection fraction
• Mild diastolic dysfunction
• Right ventricular dilation with moderately decreased right ventricular systolic function and apical sparing
• Right ventricular systolic pressure 54 mm Hg, consistent with moderate pulmonary hypertension
• Right atrial pressure 10 mm Hg
• No inspiratory collapse of a dilated inferior vena cava
• Mild tricuspid valve regurgitation.

CLASSIFICATION OF ACUTE PULMONARY EMBOLISM

3. Given the above information, how would you classify the patient’s pulmonary embolism?

• Massive
• Submassive
• Low-risk
• Clinically stable

The patient’s pulmonary embolism is submassive.

Historically, the classification of pulmonary embolism was determined by the angiographic thrombus burden. However, this has limited utility because clinical factors (eg, hypotension on initial presentation) have been shown to be better predictors of short-term mortality risk.³

Our patient is characterized as having a submassive pulmonary embolism based on elevated biomarkers (troponin T, N-terminal pro-B-type natriuretic peptide) and right ventricular dysfunction in the absence of hypotension.

ULTRASONOGRAPHY FOR DIAGNOSIS OF DEEP VEIN THROMBOSIS

Venous duplex ultrasonography has become the standard for diagnosis of lower extremity deep vein thrombosis. However, its quality and diagnostic accuracy depend on the skill of the person performing the examination. It is further limited by certain patient characteristics, including severe obesity, edema, and wounds and dressings at the site being examined.⁵

Our patient underwent duplex ultrasonography of the lower extremities, which demonstrated acute proximal and calf deep vein thrombosis in the right femoral, popliteal, and peroneal veins and no deep vein thrombosis in the left leg.

Multiple models exist to estimate the risk of complications in patients with acute pulmonary embolism.

The Bova score⁶ is based on the following factors:

• Systolic blood pressure 90–100 mm Hg (2 points) (patients with systolic blood pressure lower than 90 mm Hg were excluded from the study from which this score was derived)
• Cardiac troponin elevation (2 points)
• Right ventricular dysfunction on echocardiography or computed tomography (2 points)
• Heart rate 100 beats/min or greater (1 point).

A total score of 0, 1, or 2 (stage I) denotes low risk, 3 or 4 points (stage II) intermediate risk, and more than 4 points (stage III) high risk.

The PESI score (Pulmonary Embolism Severity Index)⁷ is based on:

• Age (1 point per year)
• Sex (10 points for being male)
• Heart rate 110 per minute or greater (20 points)
• Cancer (30 points)
• Heart failure (10 points)
• Chronic lung disease (10 points)
• Systolic blood pressure less than 100 mm Hg (30 points)
• Respiratory rate at least 30 per minute (20 points)
• Temperature less than 36ºC (20 points)
• Altered mental status (60 points)
• Arterial oxygen saturation less than 90% (20 points).

The total score is broken down into 5 classes: I (< 65 points), II (65–85), III (86–105), IV (106–125), and V (> 126). Classes I and II are low risk, and the higher ones are high risk.

The simplified PESI score⁸ was developed to more rapidly risk-stratify patients and has been found to be similar to the PESI score in prognostic accuracy. Patients get 1 point for each of the following:

• Age over 80
• Cancer
• Chronic cardiopulmonary disease (heart failure or chronic lung disease)
• Heart rate 110 per minute or greater
• Systolic blood pressure less than 100 mm Hg
• Arterial oxygen saturation less than 90%.

A total score of 0 is low risk; anything higher is high risk.

Back to our patient

Our patient had proximal and calf deep vein thrombosis of the right leg, bilateral submassive pulmonary emboli with associated biomarker elevation and right ventricular dysfunction, and left renal artery thrombosis with infarction. Using the PESI score, his risk of death in the next 30 days was 13.7% and his 30-day risk of a complicated course was 27%. Using the Bova score, his 30-day risk of death was 15.5% and his 30-day risk of a complicated course was 29.2%.^6,7

Notably, the patient’s right ventricular function had also been impaired on the echocardiogram performed during his admission 1 month previously. On transthoracic echocardiography during the current admission, the patient was found to have a similar degree of right ventricular dysfunction. This finding, along with the oxygen requirement that developed during the earlier admission, suggested that his pulmonary embolism may have been subacute and that the diagnosis may have been missed during the earlier hospital stay.

The patient was treated with unfractionated heparin. After the hospital’s multidisciplinary pulmonary embolism response team discussed and weighed the above factors, they recommended to not pursue thrombolytic therapy or inferior vena cava filter placement.

Of note, the patient’s pulses in the left lower extremity continued to be weak but palpable, and the left leg was cooler to touch than the right leg.

ASSESSING PERIPHERAL ARTERY DISEASE

4. How should the finding of weak pulses in this patient’s left leg be initially investigated?

• Computed tomographic angiography with runoff
• Ankle-brachial indices with pulse-volume recordings
• Arterial duplex ultrasonography
• Magnetic resonance angiography of the lower extremities

The ankle-brachial index is the initial diagnostic test for assessment of pulse abnormalities and for diagnosis of lower-extremity peripheral artery disease. It is calculated by dividing the higher of the ankle systolic pressures (posterior tibial or dorsalis pedis) by the higher of the 2 brachial pressures (left or right).⁹ Normal values are between 1.00 and 1.40.

Ankle-brachial indices in our patient

Our patient underwent measurement of his brachial, dorsalis pedis, and posterior tibial artery systolic pressures using blood pressure cuffs and continuous-wave Doppler. Ankle pulse-volume recordings were also obtained.

Given the patient’s poor renal function and concern for acute renal infarction, we thought it best to avoid iodinated or gadolinium contrast, such as with magnetic resonance or computed tomographic angiography.

Segmental leg pressures and pulse-volume recordings can be performed to help localize the level of arterial disease in the extremities, but were not done in this case because of the extensive deep vein thrombosis in the right leg.^10,11

Arterial ultrasonography in our patient

Arterial duplex ultrasonography was performed to help determine the location of arterial disease. It showed patent arteries in the right leg. In the left lower extremity there was slow, monophasic blood flow in the distal superficial femoral artery. The popliteal artery was occluded. The posterior tibial artery was occluded at the origin, with reconstitution distally. The peroneal artery was occluded throughout. The anterior tibial artery was patent throughout. The ultrasonographic findings were thought to be suspicious for arterial thromboembolism.

5. Given that the patient had both arterial thrombosis (renal artery, lower-extremity arteries) and venous thromboembolism (deep vein thrombosis and pulmonary embolism), which of the following would be included in the differential diagnosis?

• Antiphospholipid antibody syndrome
• Protein C or protein S deficiency
• Malignancy
• Paradoxical embolization
• Factor V Leiden mutation

Correct answers include antiphospholipid antibody syndrome, malignancy, and paradoxical embolization.

The differential diagnosis for concomitant venous and arterial thrombosis is broad,¹² and includes the following:

• Structural factors: patent foramen ovale, popliteal artery aneurysm
• Malignancy
• Inflammatory diseases: Behçet disease, Buerger disease, inflammatory bowel disease, antiphospholipid antibody syndrome, elevated lipoprotein(a), elevated homocysteine
• Hematologic diseases: myelodysplastic syndrome, disseminated intravascular coagulation, paroxysmal nocturnal hemoglobinuria, heparin-induced thrombocytopenia.

Traditional risk factors for venous thromboembolism include protein C deficiency, protein S deficiency, factor V Leiden mutation, the prothrombin G20210A gene mutation, and others. These are relatively minor risk factors for venous thrombosis and do not pose a risk for arterial thrombosis.¹² In contrast, antiphospholipid antibody syndrome and malignancy pose a risk for both venous and arterial thrombosis. Paradoxical embolism is a mechanism by which arterial thrombosis (emboli) can develop in the setting of existing venous thrombosis.¹²

Our patient underwent testing for antiphospholipid antibodies and lupus anticoagulant, and he was encouraged to undergo age-appropriate cancer screening as an outpatient.¹²

ANTIPHOSPHOLIPID ANTIBODY SYNDROME

Antiphospholipid antibody syndrome is defined by both clinical and laboratory criteria. Clinical symptoms include vascular thrombosis (arterial, venous, or both) and pregnancy-related complications.¹³

Laboratory criteria require the presence of antiphospholipid antibodies or lupus anticoagulant. These must be confirmed with repeat testing in 12 weeks. Antiphospholipid antibodies are detected by an enzyme-linked immunosorbent assay; laboratory assessment for the presence of lupus anticoagulant is a stepwise process and relies on 4 criteria:

• There should be prolongation of a phospholipid-dependent clotting test (eg, activated partial thromboplastin time, dilute Russell viper venom time test).
• There must be evidence of an inhibitory activity with mixing study.
• The inhibitor must exhibit phospholipid dependence; that is, with more phospholipid there is shortening of clotting time.
• Specific inhibitors must be excluded, including factor VIII and anticoagulant drugs such as heparin.^14–17

From Houghton DE, Moll S. Antiphospholipid antibodies. Vasc Med 2017; 22:545–550.

Clinically, one should consider antiphospholipid antibody syndrome in patients who have arterial thrombosis, a history of pregnancy morbidity, or unexplained prolongation of activated partial thromboplastin time.¹³

Antiphospholipid antibodies may be present in up to a quarter of patients with venous thromboembolism, but it is persistent positivity of antibody assays that is associated with increased future risk of venous thromboembolism.¹⁹ Of note, the risk of venous thromboembolism in patients with confirmed antiphospholipid antibody syndrome is 10 times higher than in the general population.²⁰

ANTIPHOSPHOLIPID ANTIBODIES ARE NOT ALL THE SAME

6. Which of the following antiphospholipid antibodies have not been associated with an increased thrombotic risk?

• Anti-beta-2-glycoprotein I IgG
• Lupus anticoagulant
• Antiphosphatidylserine
• Anticardiolipin IgM
• Anticardiolipin IgG

The correct answer is antiphosphatidylserine.¹⁵

Antiphospholipid antibodies are directed against a portion of select plasma proteins that are uncovered upon phospholipid binding. While lupus anticoagulant, anti-beta-2-glycoprotein I, and anticardiolipin antibodies are associated with thrombosis, antiprothrombin antibodies (including antiprothrombin and antiphosphatidylserine antibodies) are not.^15,21

Patent foramen ovale, a communication between the right and left atrium in the interatrial septum, is associated with an increased risk of paradoxical embolization. The prevalence of patent foramen ovale is estimated to be 27% to 29% in the general population.²² Noncerebral systemic paradoxical embolism occurs less frequently than cerebral embolism, accounting for approximately 5% to 10% of paradoxical emboli.²²

To evaluate for patent foramen ovale, transthoracic echocardiography is performed with a bubble (agitated saline contrast) study to assess for interatrial shunting. Transesophageal echocardiography or transcranial Doppler bubble studies may also be performed.

Although patent foramen ovale is most commonly associated with cerebral embolism, peripheral emboli can occur. Some research suggests that this may be a more common cause of arterial thromboembolism in younger patients. There have also been reports of other sites of systemic embolization, including the renal artery.¹²

Back to our patient

Initial antiphospholipid antibody testing was positive for lupus anticoagulant. Anticardiolipin and anti-beta-2-glycoprotein I antibodies were not detected.

Transesophageal echocardiography revealed a patent foramen ovale with a highly mobile atrial septum (atrial septal aneurysm).

The patient was treated with intravenous unfractionated heparin with bridging to warfarin with a target international normalized ratio (INR) of 2 to 3. His renal artery infarction and his lower-extremity arterial thromboembolic event were conservatively managed. His respiratory status improved, and he no longer required supplemental oxygen. His creatinine peaked at 1.7 mg/dL during his admission and improved to 1.2 mg/dL before he was discharged.

At follow-up, repeat echocardiography showed that his right ventricular systolic pressure had improved (decreased) to 37 mm Hg from 54 mm Hg. Repeat confirmatory testing was positive for lupus anticoagulant 12 weeks later. He has been maintained on warfarin with an INR goal of 2 to 3 as well as low-dose aspirin with plans for long-term anticoagulation. We decided to keep the patient on anticoagulation indefinitely with warfarin; he was not a candidate for a direct oral anticoagulant, given limited data on the use of these agents in the setting of lupus anticoagulant and antiphospholipid antibody syndrome.

SUMMARY OF CASE

In summary, this patient was a 75-year-old man with COPD who presented with abdominal pain. He was noted to have a left renal infarction, extensive unprovoked lower-extremity deep vein thrombosis with pulmonary emboli, and lower limb arterial thromboembolism.

He also had an underlying hypercoagulable state—antiphospholipid antibody syndrome—that predisposed him to both arterial and venous thrombosis. He was ultimately found to have a patent foramen ovale, which further increased the risk of arterial thrombosis by facilitating paradoxical embolization of venous thrombi. It is not certain whether the renal infarction and leg artery thrombi were due to paradoxical embolism or to in situ thrombosis, but we believe that it was most likely paradoxical embolization.        

Case

A 29-year-old man presented to the ED several hours after ingesting what he described as a “hefty” bottle of hand sanitizer. The patient stated that he ingested such a considerable quantity of liquid hand sanitizer because he was unable to obtain beer or liquor. He further admitted to drinking two 40-ounce beers daily for the past several years, noting that he last consumed drinking alcohol the preceding day.

The patient denied any other coingestants. He also denied nausea, vomiting, abdominal pain, or other somatic complaints. The patient’s medical history was significant for hypertension and hepatitis C, and his social history was significant for daily alcohol consumption, tobacco abuse, and former benzodiazepine, marijuana, and intravenous heroin abuse. His psychiatric history was significant for borderline personality disorder, major depression, and bulimia. The patient’s home medications included a daily multivitamin, folate, thiamine, sertraline, mirtazapine, and prazosin.

Initial vital signs at presentation were: blood pressure, 124/77 mm Hg; heart rate, 86 beats/min; respiratory rate, 15 breaths/min; and temperature, 98.0°F. On physical examination, he was noted to have slurred speech and nystagmus. His pupils were equal and reactive, without scleral icterus. The abdomen was nontender and nondistended, with regular bowel sounds, and without ascites or varicosities visualized. The rest of the examination was unremarkable. The patient did express thoughts of suicidality, but denied any homicidal ideation.

Laboratory studies revealed a serum ethanol concentration of 446 mg/dL. The patient’s basic metabolic panel was unremarkable, and liver function test results showed mildly elevated enzymes. The coagulation panel was within normal limits.

Is alcohol-based hand sanitizer consumption an emerging public health concern?

Excessive alcohol consumption is a recognized public health problem in the United States and is associated with an average of 88,000 deaths per year.¹ In a select population of patients, an untoward effect has developed from another public health target—that of hand hygiene.

Alcohol-based liquid hand sanitizers have become ubiquitous as a weapon in the antimicrobial arsenal with recommendations for its use as an alternative to soap and water in certain clinical settings. Liquid hand sanitizers are ideal for hospital or community use as they are faster, more effective, and less irritating to the skin than traditional hand-washing techniques.²

The downside to the widespread availability of hand sanitizers is that they offer easy access to individuals in search of clandestine sources of alcohol. Prior case reports have discussed the practice of consuming alcohol-based hand sanitizers for the purpose of intoxication in institutionalized persons, such as prisoners or patients in psychiatric facilities who are restricted to conventional sources of alcohol.

Children and confused elderly patients are also at risk for unintentional ingestions.^3,4 An article reviewed exposures reported to the American Association of Poison Control Center’s National Poison Data System over a 5-year period from 2005 to 2009.³ Of the 68,712 reported cases in this cohort, 80.5% were in children younger than 6 years of age. The investigators also noted an increased incidence of exposure over this period with an average of 1,894 additional cases per year.³There were 17,154 children aged 12 years and younger reported in 2017 to poison centers with exposures to hand sanitizers. Young children may be enticed by the bright colorful packaging and similarity to food and candy smells.⁵

What are the clinical manifestations of alcohol-based hand sanitizer ingestion?

Significant hazards exist from ingesting liquid hand sanitizer, including the high alcohol content, which varies from 40% to 85%.² Because isopropanol is commonly one of the components (if not the sole component) of many hand-sanitizer preparations, isopropanol toxicity may occur when ingested. The effects of isopropanol are similar to those of ethanol, with clinical effects reported after ingestion of as little as 100 mL of 70% isopropanol solution.⁴

Hand sanitizer formulations vary by manufacturer and contain different concentrations of ethanol and/or isopropanol, as well as additional potential inactive ingredients such as acetone, 1-propanol, 2-propanol, benzyl alcohol, hydrogen peroxide, glycerin, water, and different perfumes.^3,4

Persons who consume hand sanitizers recreationally are often unaware of the large alcohol content by volume that they are consuming. Recreational ingestion of hand sanitizer is believed to be the cause of at least one case of lethal ethanol intoxication. An articlereported a case of a male patient who suffered respiratory arrest after consuming an ethanol-based hand sanitizer.⁶ This patient was noted to have a serum ethanol of 536 mg/dL after consuming an unknown quantity of a 354 mL container of a 62% ethanol by volume hand sanitizer.⁶

Institutionalized individuals seeking alcohol through this source have discovered novel ways to yield a stronger product. Through the use of table salt and a cotton sock, it is possible to extract a liquid from a gel hand sanitizer preparation, yielding an alcohol context 30% higher by volume than the parent mixture.⁷

Alcohol intoxication poses a host of health effects. In nonhabituated individuals, a lethal load of alcohol can be achieved by consuming a volume of as little as 400 mL of an 80% alcohol-based solution.⁴ Symptoms from ingestion of an alcohol-based liquid hand sanitizer typically appear 1 to 2 hours after ingestion and mirror that of the alcohol toxidrome. Most commonly, this includes nausea, vomiting, epigastric pain, and varying degrees of central nervous system (CNS) depression.⁴ The life-threatening clinical manifestation of alcohol intoxication includes severe CNS and respiratory depression resulting in respiratory arrest, hypothermia, cardiac dysrhythmias with possible cardiac arrest, hypoglycemia, ketoacidosis, and hypotension.³

How is alcohol-based hand sanitizer ingestion managed?

The management of patients with alcohol-based hand sanitizer ingestion is the same as the management of alcohol ingestion from more socially acceptable sources and is mainly supportive.^3,4 These measures are directed at managing the patient’s airway with intubation and mechanical ventilation when appropriate, as well as supportive measures to address any underlying metabolic derangement or hypotension.² While hemodialysis has been used in some patients who had severe organ dysfunction and did not respond to supportive measures, it is usually not necessary.^1,3

Case Conclusion

The patient in this case was subsequently admitted to an intermediate level of care. He did not require intubation or further hemodynamic support during his initial acute intoxication. Later in the patient’s hospital course, he was noted to be in alcohol withdrawal, and proper management was initiated. He also required therapeutic one-to-one supervision after members of the nursing staff observed the patient consuming the hand sanitizer gel present in patient-care areas. He was later seen by psychiatry services. The psychiatrist recommended transfer to an inpatient psychiatric facility upon medical clearance for treatment of his psychiatric illness as well as alcohol dependence.

Case

A 29-year-old man presented to the ED several hours after ingesting what he described as a “hefty” bottle of hand sanitizer. The patient stated that he ingested such a considerable quantity of liquid hand sanitizer because he was unable to obtain beer or liquor. He further admitted to drinking two 40-ounce beers daily for the past several years, noting that he last consumed drinking alcohol the preceding day.

The patient denied any other coingestants. He also denied nausea, vomiting, abdominal pain, or other somatic complaints. The patient’s medical history was significant for hypertension and hepatitis C, and his social history was significant for daily alcohol consumption, tobacco abuse, and former benzodiazepine, marijuana, and intravenous heroin abuse. His psychiatric history was significant for borderline personality disorder, major depression, and bulimia. The patient’s home medications included a daily multivitamin, folate, thiamine, sertraline, mirtazapine, and prazosin.

Initial vital signs at presentation were: blood pressure, 124/77 mm Hg; heart rate, 86 beats/min; respiratory rate, 15 breaths/min; and temperature, 98.0°F. On physical examination, he was noted to have slurred speech and nystagmus. His pupils were equal and reactive, without scleral icterus. The abdomen was nontender and nondistended, with regular bowel sounds, and without ascites or varicosities visualized. The rest of the examination was unremarkable. The patient did express thoughts of suicidality, but denied any homicidal ideation.

Laboratory studies revealed a serum ethanol concentration of 446 mg/dL. The patient’s basic metabolic panel was unremarkable, and liver function test results showed mildly elevated enzymes. The coagulation panel was within normal limits.

Is alcohol-based hand sanitizer consumption an emerging public health concern?

Excessive alcohol consumption is a recognized public health problem in the United States and is associated with an average of 88,000 deaths per year.¹ In a select population of patients, an untoward effect has developed from another public health target—that of hand hygiene.

Alcohol-based liquid hand sanitizers have become ubiquitous as a weapon in the antimicrobial arsenal with recommendations for its use as an alternative to soap and water in certain clinical settings. Liquid hand sanitizers are ideal for hospital or community use as they are faster, more effective, and less irritating to the skin than traditional hand-washing techniques.²

The downside to the widespread availability of hand sanitizers is that they offer easy access to individuals in search of clandestine sources of alcohol. Prior case reports have discussed the practice of consuming alcohol-based hand sanitizers for the purpose of intoxication in institutionalized persons, such as prisoners or patients in psychiatric facilities who are restricted to conventional sources of alcohol.

Children and confused elderly patients are also at risk for unintentional ingestions.^3,4 An article reviewed exposures reported to the American Association of Poison Control Center’s National Poison Data System over a 5-year period from 2005 to 2009.³ Of the 68,712 reported cases in this cohort, 80.5% were in children younger than 6 years of age. The investigators also noted an increased incidence of exposure over this period with an average of 1,894 additional cases per year.³There were 17,154 children aged 12 years and younger reported in 2017 to poison centers with exposures to hand sanitizers. Young children may be enticed by the bright colorful packaging and similarity to food and candy smells.⁵

What are the clinical manifestations of alcohol-based hand sanitizer ingestion?

Significant hazards exist from ingesting liquid hand sanitizer, including the high alcohol content, which varies from 40% to 85%.² Because isopropanol is commonly one of the components (if not the sole component) of many hand-sanitizer preparations, isopropanol toxicity may occur when ingested. The effects of isopropanol are similar to those of ethanol, with clinical effects reported after ingestion of as little as 100 mL of 70% isopropanol solution.⁴

Hand sanitizer formulations vary by manufacturer and contain different concentrations of ethanol and/or isopropanol, as well as additional potential inactive ingredients such as acetone, 1-propanol, 2-propanol, benzyl alcohol, hydrogen peroxide, glycerin, water, and different perfumes.^3,4

Persons who consume hand sanitizers recreationally are often unaware of the large alcohol content by volume that they are consuming. Recreational ingestion of hand sanitizer is believed to be the cause of at least one case of lethal ethanol intoxication. An articlereported a case of a male patient who suffered respiratory arrest after consuming an ethanol-based hand sanitizer.⁶ This patient was noted to have a serum ethanol of 536 mg/dL after consuming an unknown quantity of a 354 mL container of a 62% ethanol by volume hand sanitizer.⁶

Institutionalized individuals seeking alcohol through this source have discovered novel ways to yield a stronger product. Through the use of table salt and a cotton sock, it is possible to extract a liquid from a gel hand sanitizer preparation, yielding an alcohol context 30% higher by volume than the parent mixture.⁷

Alcohol intoxication poses a host of health effects. In nonhabituated individuals, a lethal load of alcohol can be achieved by consuming a volume of as little as 400 mL of an 80% alcohol-based solution.⁴ Symptoms from ingestion of an alcohol-based liquid hand sanitizer typically appear 1 to 2 hours after ingestion and mirror that of the alcohol toxidrome. Most commonly, this includes nausea, vomiting, epigastric pain, and varying degrees of central nervous system (CNS) depression.⁴ The life-threatening clinical manifestation of alcohol intoxication includes severe CNS and respiratory depression resulting in respiratory arrest, hypothermia, cardiac dysrhythmias with possible cardiac arrest, hypoglycemia, ketoacidosis, and hypotension.³

How is alcohol-based hand sanitizer ingestion managed?

The management of patients with alcohol-based hand sanitizer ingestion is the same as the management of alcohol ingestion from more socially acceptable sources and is mainly supportive.^3,4 These measures are directed at managing the patient’s airway with intubation and mechanical ventilation when appropriate, as well as supportive measures to address any underlying metabolic derangement or hypotension.² While hemodialysis has been used in some patients who had severe organ dysfunction and did not respond to supportive measures, it is usually not necessary.^1,3

Case Conclusion

The patient in this case was subsequently admitted to an intermediate level of care. He did not require intubation or further hemodynamic support during his initial acute intoxication. Later in the patient’s hospital course, he was noted to be in alcohol withdrawal, and proper management was initiated. He also required therapeutic one-to-one supervision after members of the nursing staff observed the patient consuming the hand sanitizer gel present in patient-care areas. He was later seen by psychiatry services. The psychiatrist recommended transfer to an inpatient psychiatric facility upon medical clearance for treatment of his psychiatric illness as well as alcohol dependence.

Case

A 29-year-old man presented to the ED several hours after ingesting what he described as a “hefty” bottle of hand sanitizer. The patient stated that he ingested such a considerable quantity of liquid hand sanitizer because he was unable to obtain beer or liquor. He further admitted to drinking two 40-ounce beers daily for the past several years, noting that he last consumed drinking alcohol the preceding day.

The patient denied any other coingestants. He also denied nausea, vomiting, abdominal pain, or other somatic complaints. The patient’s medical history was significant for hypertension and hepatitis C, and his social history was significant for daily alcohol consumption, tobacco abuse, and former benzodiazepine, marijuana, and intravenous heroin abuse. His psychiatric history was significant for borderline personality disorder, major depression, and bulimia. The patient’s home medications included a daily multivitamin, folate, thiamine, sertraline, mirtazapine, and prazosin.

Initial vital signs at presentation were: blood pressure, 124/77 mm Hg; heart rate, 86 beats/min; respiratory rate, 15 breaths/min; and temperature, 98.0°F. On physical examination, he was noted to have slurred speech and nystagmus. His pupils were equal and reactive, without scleral icterus. The abdomen was nontender and nondistended, with regular bowel sounds, and without ascites or varicosities visualized. The rest of the examination was unremarkable. The patient did express thoughts of suicidality, but denied any homicidal ideation.

Laboratory studies revealed a serum ethanol concentration of 446 mg/dL. The patient’s basic metabolic panel was unremarkable, and liver function test results showed mildly elevated enzymes. The coagulation panel was within normal limits.

Is alcohol-based hand sanitizer consumption an emerging public health concern?

Excessive alcohol consumption is a recognized public health problem in the United States and is associated with an average of 88,000 deaths per year.¹ In a select population of patients, an untoward effect has developed from another public health target—that of hand hygiene.

Alcohol-based liquid hand sanitizers have become ubiquitous as a weapon in the antimicrobial arsenal with recommendations for its use as an alternative to soap and water in certain clinical settings. Liquid hand sanitizers are ideal for hospital or community use as they are faster, more effective, and less irritating to the skin than traditional hand-washing techniques.²

The downside to the widespread availability of hand sanitizers is that they offer easy access to individuals in search of clandestine sources of alcohol. Prior case reports have discussed the practice of consuming alcohol-based hand sanitizers for the purpose of intoxication in institutionalized persons, such as prisoners or patients in psychiatric facilities who are restricted to conventional sources of alcohol.

Children and confused elderly patients are also at risk for unintentional ingestions.^3,4 An article reviewed exposures reported to the American Association of Poison Control Center’s National Poison Data System over a 5-year period from 2005 to 2009.³ Of the 68,712 reported cases in this cohort, 80.5% were in children younger than 6 years of age. The investigators also noted an increased incidence of exposure over this period with an average of 1,894 additional cases per year.³There were 17,154 children aged 12 years and younger reported in 2017 to poison centers with exposures to hand sanitizers. Young children may be enticed by the bright colorful packaging and similarity to food and candy smells.⁵

What are the clinical manifestations of alcohol-based hand sanitizer ingestion?

Significant hazards exist from ingesting liquid hand sanitizer, including the high alcohol content, which varies from 40% to 85%.² Because isopropanol is commonly one of the components (if not the sole component) of many hand-sanitizer preparations, isopropanol toxicity may occur when ingested. The effects of isopropanol are similar to those of ethanol, with clinical effects reported after ingestion of as little as 100 mL of 70% isopropanol solution.⁴

Hand sanitizer formulations vary by manufacturer and contain different concentrations of ethanol and/or isopropanol, as well as additional potential inactive ingredients such as acetone, 1-propanol, 2-propanol, benzyl alcohol, hydrogen peroxide, glycerin, water, and different perfumes.^3,4

Persons who consume hand sanitizers recreationally are often unaware of the large alcohol content by volume that they are consuming. Recreational ingestion of hand sanitizer is believed to be the cause of at least one case of lethal ethanol intoxication. An articlereported a case of a male patient who suffered respiratory arrest after consuming an ethanol-based hand sanitizer.⁶ This patient was noted to have a serum ethanol of 536 mg/dL after consuming an unknown quantity of a 354 mL container of a 62% ethanol by volume hand sanitizer.⁶

Institutionalized individuals seeking alcohol through this source have discovered novel ways to yield a stronger product. Through the use of table salt and a cotton sock, it is possible to extract a liquid from a gel hand sanitizer preparation, yielding an alcohol context 30% higher by volume than the parent mixture.⁷

Alcohol intoxication poses a host of health effects. In nonhabituated individuals, a lethal load of alcohol can be achieved by consuming a volume of as little as 400 mL of an 80% alcohol-based solution.⁴ Symptoms from ingestion of an alcohol-based liquid hand sanitizer typically appear 1 to 2 hours after ingestion and mirror that of the alcohol toxidrome. Most commonly, this includes nausea, vomiting, epigastric pain, and varying degrees of central nervous system (CNS) depression.⁴ The life-threatening clinical manifestation of alcohol intoxication includes severe CNS and respiratory depression resulting in respiratory arrest, hypothermia, cardiac dysrhythmias with possible cardiac arrest, hypoglycemia, ketoacidosis, and hypotension.³

How is alcohol-based hand sanitizer ingestion managed?

The management of patients with alcohol-based hand sanitizer ingestion is the same as the management of alcohol ingestion from more socially acceptable sources and is mainly supportive.^3,4 These measures are directed at managing the patient’s airway with intubation and mechanical ventilation when appropriate, as well as supportive measures to address any underlying metabolic derangement or hypotension.² While hemodialysis has been used in some patients who had severe organ dysfunction and did not respond to supportive measures, it is usually not necessary.^1,3

Case Conclusion

The patient in this case was subsequently admitted to an intermediate level of care. He did not require intubation or further hemodynamic support during his initial acute intoxication. Later in the patient’s hospital course, he was noted to be in alcohol withdrawal, and proper management was initiated. He also required therapeutic one-to-one supervision after members of the nursing staff observed the patient consuming the hand sanitizer gel present in patient-care areas. He was later seen by psychiatry services. The psychiatrist recommended transfer to an inpatient psychiatric facility upon medical clearance for treatment of his psychiatric illness as well as alcohol dependence.

A 96-year-old woman with hypertension, diabetes,    and dementia was found unresponsive in her nursing home and was transferred to the hospital.

At presentation to the hospital, her blood pressure was 76/43 mm Hg, heart rate 42 beats per minute, rectal temperature 31.6°C (88.8°F), and blood glucose 36 mg/dL.

Causes of secondary hypothermia were sought. Blood and urine cultures were negative. Computed tomography of the head showed no acute intracranial abnormalities. Tests for adrenal insufficiency and hypothyroidism were negative.

HYPOTHERMIA AND THE ECG

Hypothermia can produce a number of changes on the ECG. At the start of hypothermia, a stress reaction is induced, resulting in sinus tachycardia. But when the temperature goes below 32°C, sinus bradycardia ensues,¹ resulting in various degrees of heart block.² In our patient, a severely prolonged PR interval resulted in first-degree heart block.

Other findings on ECG associated with hypothermia include atrial fibrillation, widening of the P and T waves, prolonging of the QT interval, and widening of the QRS interval. Progressive widening of the QRS interval can predispose to ventricular fibrillation.^1,3

An Osborn or J wave is a wave found between the end of the QRS and the beginning of the ST segment and is usually seen on the inferior and lateral precordial leads. It is found in as many as 80% of patients when the body temperature is below 30°C.^1,3,4

Although Osborn waves are a common finding in hypothermia, they are also seen in electrolyte imbalances such as hypercalcemia and in central nervous system diseases.^5,6 Hypothermia-associated changes on ECG are usually readily reversible with rewarming.¹

TAKE-HOME MESSAGES

The ECG should always be interpreted in the proper clinical context and, whenever possible, compared with a previous ECG. It is prudent to always consider potentially reversible triggers of hypothermia other than environmental exposure such as hypothyroidism, infection, adrenal insufficiency, ketoacidosis, medication side effects, and alcohol use.

Hypothermia, especially in elderly patients with multiple comorbidities, can lead to bradycardia and varying degrees of heart block.

A 96-year-old woman with hypertension, diabetes,    and dementia was found unresponsive in her nursing home and was transferred to the hospital.

At presentation to the hospital, her blood pressure was 76/43 mm Hg, heart rate 42 beats per minute, rectal temperature 31.6°C (88.8°F), and blood glucose 36 mg/dL.

Causes of secondary hypothermia were sought. Blood and urine cultures were negative. Computed tomography of the head showed no acute intracranial abnormalities. Tests for adrenal insufficiency and hypothyroidism were negative.

HYPOTHERMIA AND THE ECG

Hypothermia can produce a number of changes on the ECG. At the start of hypothermia, a stress reaction is induced, resulting in sinus tachycardia. But when the temperature goes below 32°C, sinus bradycardia ensues,¹ resulting in various degrees of heart block.² In our patient, a severely prolonged PR interval resulted in first-degree heart block.

Other findings on ECG associated with hypothermia include atrial fibrillation, widening of the P and T waves, prolonging of the QT interval, and widening of the QRS interval. Progressive widening of the QRS interval can predispose to ventricular fibrillation.^1,3

An Osborn or J wave is a wave found between the end of the QRS and the beginning of the ST segment and is usually seen on the inferior and lateral precordial leads. It is found in as many as 80% of patients when the body temperature is below 30°C.^1,3,4

Although Osborn waves are a common finding in hypothermia, they are also seen in electrolyte imbalances such as hypercalcemia and in central nervous system diseases.^5,6 Hypothermia-associated changes on ECG are usually readily reversible with rewarming.¹

TAKE-HOME MESSAGES

The ECG should always be interpreted in the proper clinical context and, whenever possible, compared with a previous ECG. It is prudent to always consider potentially reversible triggers of hypothermia other than environmental exposure such as hypothyroidism, infection, adrenal insufficiency, ketoacidosis, medication side effects, and alcohol use.

Hypothermia, especially in elderly patients with multiple comorbidities, can lead to bradycardia and varying degrees of heart block.

A 96-year-old woman with hypertension, diabetes,    and dementia was found unresponsive in her nursing home and was transferred to the hospital.

At presentation to the hospital, her blood pressure was 76/43 mm Hg, heart rate 42 beats per minute, rectal temperature 31.6°C (88.8°F), and blood glucose 36 mg/dL.

Causes of secondary hypothermia were sought. Blood and urine cultures were negative. Computed tomography of the head showed no acute intracranial abnormalities. Tests for adrenal insufficiency and hypothyroidism were negative.

HYPOTHERMIA AND THE ECG

Hypothermia can produce a number of changes on the ECG. At the start of hypothermia, a stress reaction is induced, resulting in sinus tachycardia. But when the temperature goes below 32°C, sinus bradycardia ensues,¹ resulting in various degrees of heart block.² In our patient, a severely prolonged PR interval resulted in first-degree heart block.

Other findings on ECG associated with hypothermia include atrial fibrillation, widening of the P and T waves, prolonging of the QT interval, and widening of the QRS interval. Progressive widening of the QRS interval can predispose to ventricular fibrillation.^1,3

An Osborn or J wave is a wave found between the end of the QRS and the beginning of the ST segment and is usually seen on the inferior and lateral precordial leads. It is found in as many as 80% of patients when the body temperature is below 30°C.^1,3,4

Although Osborn waves are a common finding in hypothermia, they are also seen in electrolyte imbalances such as hypercalcemia and in central nervous system diseases.^5,6 Hypothermia-associated changes on ECG are usually readily reversible with rewarming.¹

TAKE-HOME MESSAGES

The ECG should always be interpreted in the proper clinical context and, whenever possible, compared with a previous ECG. It is prudent to always consider potentially reversible triggers of hypothermia other than environmental exposure such as hypothyroidism, infection, adrenal insufficiency, ketoacidosis, medication side effects, and alcohol use.

Hypothermia, especially in elderly patients with multiple comorbidities, can lead to bradycardia and varying degrees of heart block.

A 50-year-old woman presented to the emergency department after a witnessed loss of consciousness and seizurelike activity. She reported that she had been sitting outside her home, drinking coffee in the morning, but became very lightheaded when she went back into her house. At that time she felt could not focus and had a sense of impending doom. She sat down in a chair and her symptoms worsened.

According to her family, her eyes rolled back and she became rigid. The family helped her to the floor. Her body then made jerking movements that lasted for about 1 minute. She regained consciousness but was very confused for about 10 minutes until emergency medical services personnel arrived. She had no recollection of passing out. She said nothing like this had ever happened to her before.

On arrival in the emergency department, she complained of generalized headache and muscle soreness. She said the headache had been present for 1 week and was constant and dull. There were no aggravating or alleviating factors associated with the headache, and she denied fever, chills, nausea, numbness, tingling, incontinence, tongue biting, tremor, poor balance, ringing in ears, speech difficulty, or weakness.

Medical history: Multiple problems, medications

The patient’s medical history included depression, hypertension, anxiety, osteoarthritis, and asthma. She was allergic to penicillin. She had undergone carpal tunnel surgery on her right hand 5 years previously. She was perimenopausal with no children.

She denied using illicit drugs. She said she had smoked a half pack of cigarettes per day for more than 10 years and was a current smoker but was actively trying to quit. She said she occasionally used alcohol but had not consumed any alcohol in the last 2 weeks.

She had no history of central nervous system infection. She did report an episode of head trauma in grade school when a portable basketball hoop fell, striking her on the top of the head and causing her to briefly lose consciousness, but she did not seek medical attention.

She had no family history of seizure or neurologic disease.

Her current medications included atenolol, naproxen, gabapentin, venlafaxine, zolpidem, lorazepam, bupropion, and meloxicam. The bupropion and lorazepam had been prescribed recently for her anxiety. She reported that she had been given only 10 tablets of lorazepam and had taken the last tablet 48 hours previously. She had been taking the bupropion for 7 days. She reported an increase in stress lately and had been taking zolpidem due to an altered sleep pattern.

PHYSICAL EXAMINATION, INITIAL TESTS

On examination, the patient did not appear to be in acute distress. Her blood pressure was 107/22 mm Hg, pulse 100 beats per minute, respiratory rate 16 breaths per minute, temperature 37.1°C (98.8°F), and oxygen saturation 98% on room air.

Examination of her head, eyes, mouth, and neck were unremarkable. Cardiovascular, pulmonary, and abdominal examinations were normal. She had no neurologic deficits and was fully alert and oriented. She had no visible injuries.

Blood and urine samples were obtained about 15 minutes after her arrival to the emergency department. Results showed:

• Glucose 73 mg/dL (reference range 74–99)
• Sodium 142 mmol/L (136–144)
• Blood urea nitrogen 12 mg/dL (7–21)
• Creatinine 0.95 mg/dL (0.58–0.96)
• Chloride 97 mmol/L (97–105)
• Carbon dioxide (bicarbonate) 16 mmol/L (22–30)
• Prolactin 50.9 ng/mL (4.5–26.8)
• Anion gap 29 mmol/L (9–18)
• Ethanol undetectable
• White blood cell count 11.03 × 10⁹/L (3.70–11.00)
• Creatine kinase 89 U/L (30–220)
• Urinalysis normal, specific gravity 1.010 (1.005–1.030), no detectable ketones, and no crystals seen on microscopic evaluation.

Electrocardiography showed normal sinus rhythm with no ectopy and no ST-segment changes. Chest radiography was negative for any acute process.

The patient was transferred to the 23-hour observation unit in stable condition for further evaluation, monitoring, and management.

SIGNS AND SYMPTOMS OF SEIZURE

1. What findings are consistent with seizure?

• Jerking movements
• Confusion following the event
• Tongue-biting
• Focal motor weakness
• Urinary incontinence
• Aura before the event

All of the above findings are consistent with seizure.

The first consideration in evaluating a patient who presents with a possible seizure is whether the patient’s recollections of the event—and those of the witnesses—are consistent with the symptoms of seizure.¹

In generalized tonic-clonic or grand mal seizure, the patient may experience an aura or subjective sensations before the onset. These vary greatly among patients.² There may be an initial vocalization at the onset of the seizure, such as crying out or unintelligible speech. The patient’s eyes may roll back in the head. This is followed by loss of muscle tone, and if the patient is standing, he or she may fall to the ground. The patient becomes unresponsive and may go into respiratory arrest. There is tonic stiffening of the limbs and body, followed by clonic movements typically lasting 1 to 2 minutes, or sometimes longer.^1,3,4 The patient will then relax and experience a period of unconsciousness or confusion (postictal state).

Urinary incontinence and tongue-biting strongly suggest seizure activity, and turning the head to one side and posturing may also be seen.^3,5 After the event, the patient may report headache, generalized muscle soreness, exhaustion, or periods of transient focal weakness, also known as Todd paralysis.^2,5

Our patient had aura-like symptoms at the outset. She felt very lightheaded, had difficulty focusing, and felt a sense of impending doom. She did not make any vocalizations at the onset, but her eyes did roll backward and she became rigid (tonic). She then lost muscle tone and became unresponsive. Her family had to help her to the floor. Jerking (clonic) movements were witnessed.

She regained consciousness but was described as being confused (postictal) for 10 minutes. Additionally, she denied ever having had symptoms like this previously. On arrival in the emergency department, she reported generalized headache and muscle soreness, but no tongue-biting or urinary incontinence. Her event did not last for more than 1 to 2 minutes according to her family.

Her symptoms strongly suggest new-onset tonic-clonic or grand mal seizure, though this is not completely certain.

2. What laboratory results are consistent with seizure?

• Prolactin elevation
• Anion gap acidosis
• Leukocytosis

As noted above, the patient had an elevated prolactin level and elevated anion gap. Both of these findings can be used, with caution, in evaluating seizure activity.

Prolactin testing is controversial

Prolactin testing in diagnosing seizure activity is controversial. The exact mechanism of prolactin release in seizures is not fully understood. Generalized tonic-clonic seizures and complex partial seizures have both been shown to elevate prolactin. Prolactin levels after these types of seizures should rise within 30 minutes of the event and normalize 1 hour later.⁶

However, other events and conditions that mimic seizure have been shown to cause a rise in prolactin; these include syncope, transient ischemic attack, cardiac dysrhythmia, migraine, and other epilepsy-like variants. This effect has not been adequately studied. Therefore, an elevated prolactin level alone cannot diagnose or exclude seizure.⁷

For the prolactin level to be helpful, the blood sample must be drawn within 10 to 20 minutes after a possible seizure. Even if the prolactin level remains normal, it does not rule out seizure. Prolactin levels should therefore be used in combination with other testing to make a definitive diagnosis or exclusion of seizure.⁸

Anion gap and Denver Seizure Score

The anion gap has also been shown to rise after generalized seizure due to the metabolic acidosis that occurs. With a bicarbonate level of 16 mmol/L, an elevated anion gap, and normal breathing, our patient very likely had metabolic acidosis.

It is sometimes difficult to differentiate syncope from seizure, as they share several features.

The Denver Seizure Score can help differentiate these two conditions. It is based on the patient’s anion gap and bicarbonate level and is calculated as follows: 

(24 – bicarbonate) + [2 × (anion gap – 12)]

A score above 20 strongly indicates seizure activity. However, this is not a definitive tool for diagnosis. Like an elevated prolactin level, the Denver Seizure Score should be used in combination with other testing to move toward a definitive diagnosis.⁹

Our patient’s anion gap was 29 mmol/L and her bicarbonate level was 16 mmol/L. Her Denver Seizure Score was therefore 42, which supports this being an episode of generalized seizure activity.

Leukocytosis

The patient had a white blood cell count of 11.03 × 10⁹/L, which was mildly elevated. She had no history of fever and no source of infection by history.

Leukocytosis is common following generalized tonic-clonic seizure. A fever may lower the seizure threshold; however, our patient was not febrile and clinically had no factors that raised concern for an underlying infection.

ANION GAP ACIDOSIS AND SEIZURE

3. Which of the following can cause both anion gap acidosis and seizure?

• Ethylene glycol
• Salicylate overdose
• Ethanol withdrawal without ketosis
• Alcoholic ketoacidosis
• Methanol

All of the above except for ethanol withdrawal without ketosis can cause both anion gap acidosis and seizure.

Ethylene glycol can cause seizure and an elevated anion gap acidosis. However, this patient had no history of ingesting antifreeze (the most common source of ethylene glycol in the home) and no evidence of calcium oxalate crystals in the urine, which would be a sign of ethylene glycol toxicity. Additional testing for ethylene glycol may include serum ethylene glycol levels and ultraviolet light testing of the urine to detect fluorescein, which is commonly added to automotive antifreeze to help mechanics find fluid leaks in engines.

Salicylate overdose can cause seizure and an elevated anion gap acidosis. However, this patient has no history of aspirin ingestion, and a serum aspirin level was later ordered and found to be negative. In addition, the acid-base disorder in salicylate overdose may be respiratory alkalosis from direct stimulation of respiratory centers in conjunction with metabolic acidosis.

Ethanol withdrawal can cause seizure and may result in ketoacidosis, which would appear as anion gap acidosis. The undetectable ethanol level in this patient would be consistent with withdrawal from ethanol, which may also lead to ketoacidosis.

Alcoholic ketoacidosis is a late finding in patients who have been drinking ethanol and is thus a possible cause of an elevated anion gap in this patient. However, the absence of ketones in her urine speaks against this diagnosis.

Methanol can cause seizure and acidosis, but laboratory testing would reveal a normal anion gap and an elevated osmolar gap. This was not likely in this patient.

The presence of anion gap acidosis is important in forming a differential diagnosis. Several causes of anion gap acidosis may also cause seizure. These include salicylates, ethanol withdrawal with ketosis, methanol, and isoniazid. None of these appears to be a factor in this patient’s case.

DIFFERENTIAL DIAGNOSIS IN OUR PATIENT

4. What is the most likely cause of this patient’s seizure?

• Bupropion side effect
• Benzodiazepine withdrawal
• Ethanol withdrawal
• Brain lesion
• Central nervous system infection
• Unprovoked seizure (new-onset epilepsy)

Bupropion, an inhibitor of neuronal reuptake of norepinephrine and dopamine, has been used in the United States since 1989 to treat major depression.¹⁰ At therapeutic doses, it lowers the seizure threshold; in cases of acute overdose, seizures typically occur within hours of the dose, or up to 24 hours in patients taking extended-release formulations.¹¹

Bupropion should be used with caution or avoided in patients taking other medications that also lower the seizure threshold, or during withdrawal from alcohol, benzodiazepines, or barbiturates.¹⁰

Benzodiazepine withdrawal. Abrupt cessation of benzodiazepines also lowers the seizure threshold, and seizures are commonly seen in benzodiazepine withdrawal syndrome. The use of benzodiazepines is controversial in many situations, and discontinuing them may prove problematic for both the patient and physician, as the potential for abuse and addiction is significant.

Seizures have occurred during withdrawal from even short-term benzodiazepine use. Other factors, such as concomitant use of other medications that lower the seizure threshold, may play a more significant role in causing withdrawal seizures than the duration of benzodiazepine therapy.¹²

Medications shown to be useful in managing withdrawal from benzodiazepines include carbamazepine, imipramine, valproate, and trazodone. Paroxetine has also been shown to be helpful in patients with major depression who are being taken off a benzodiazepine.¹³

Ethanol withdrawal is common in patients presenting to emergency departments, and seizures are frequently seen in these patients. This patient reported social drinking but not drinking ethanol daily, although many patients are not forthcoming about alcohol or drug use when talking with a physician or other healthcare provider.

Alcohol withdrawal seizures may accompany delirium tremens or major withdrawal syndrome, but they are seen more often in the absence of major withdrawal or delirium tremens. Seizures are typically single or occur in a short grouping over a brief period of time and mostly occur in chronic alcoholism. The role of anticonvulsants in patients with alcohol withdrawal seizure has not been established.¹⁴

Brain lesion. A previously undiagnosed brain tumor is not a common cause of new-onset seizure, although it is not unusual for a brain tumor to cause new-onset seizure. In 1 study, 6% of patients with new-onset seizure had a clinically significant lesion on brain imaging.¹⁵ In addition, 15% to 30% of patients with a previously undiagnosed brain tumor present with seizure as the first symptom.¹⁶ Patients with abnormal findings on neurologic examination after the seizure activity are more likely to have a structural lesion that may be identified by computed tomography (CT) or magnetic resonance imaging. (MRI)¹⁵

Unprovoked seizure occurs without an identifiable precipitating factor, or from a central nervous system insult that occurred more than 7 days earlier. Patients who may have recurrent unprovoked seizure will likely be diagnosed with epilepsy.¹⁵ Patients with a first-time unprovoked seizure have a 30% or higher likelihood of having another unprovoked seizure within 5 years.¹⁷

It is most likely that bupropion is the key factor in lowering the seizure threshold in this patient. However, patients sometimes underreport the amount of alcohol they consume, and though less likely, our patient’s report of not drinking for 2 weeks may also be unreliable. Ethanol withdrawal, though unlikely, may also be a consideration with this case.

5. Which tests may be helpful in this patient’s workup?

• CT of the brain
• Lumbar puncture for spinal fluid analysis
• MRI of the brain
• Electroencephalography (EEG)

This patient had had a headache for 1 week before presenting to the emergency department. Indications for neuroimaging in a patient with headache include sudden onset of severe headache, neurologic deficits, human immunodeficiency virus infection, loss of consciousness, immunosuppression, pregnancy, malignancy, and age over 50 with a new type of headache.^18,19 Therefore, she should undergo some form of neuroimaging, either CT or MRI.

CT is the most readily available and fastest imaging study for the central nervous system available to emergency physicians. CT is limited, however, due to its decreased sensitivity in detecting some brain lesions. Therefore, many patients with first-time seizure may eventually require MRI.¹⁵ Furthermore, patients with focal onset of the seizure activity are more likely to have a structural lesion precipitating the seizure.  MRI may have a higher yield than CT in these cases.^15,20

Lumbar puncture for spinal fluid analysis is helpful in evaluating a patient with a suspected central nervous system infection such as meningitis or encephalitis, or subarachnoid hemorrhage.

This patient had a normal neurologic examination, no fever, and no meningeal signs, and central nervous system infection was very unlikely. Also, because she had had a headache for 1 week before the presentation with seizurelike activity, subarachnoid hemorrhage was very unlikely, and emergency lumbar puncture was not indicated.

MRI is less readily available than CT in a timely fashion in most emergency departments in the United States. It offers a higher yield than CT in diagnosing pathology such as acute stroke, brain tumor, and plaques seen in multiple sclerosis. CT is superior to MRI in diagnosing bony abnormalities and is very sensitive for detecting acute bleeding.

If MRI is performed, it should follow a specific protocol that includes high-resolution images for epilepsy evaluation rather than the more commonly ordered stroke protocol. The stroke protocol is more likely to be ordered in the emergency department.

EEG is well established in evaluating new-onset seizure in pediatric patients. Studies also demonstrate its utility in evaluating first-time seizure in adults, providing evidence that both epileptiform and nonepileptiform abnormalities seen on EEG are associated with a higher risk of recurrent seizure activity than in patients with normal findings on EEG.¹

EEG may be difficult to interpret in adults. According to Benbadis,⁵ as many as one-third of adult patients diagnosed with epilepsy on EEG did not have epilepsy. This is because of normal variants, simple fluctuations of background rhythms, or fragmented alpha activity that can have a similar appearance to epileptiform patterns. EEG must always be interpreted in the context of the patient’s history and symptoms.⁵

Though EEG has limitations, it remains a crucial tool for identifying epilepsy. Following a single seizure, the decision to prescribe antiepileptic drugs is highly influenced by patterns on EEG associated with a risk of recurrence. In fact, a patient experiencing a single, idiopathic seizure and exhibiting an EEG pattern of spike wave discharges is likely to have recurrent seizure activity.²¹ Also, the appropriate use of EEG after even a single unprovoked seizure can identify patients with epilepsy and a risk of recurrent seizure greater than 60%.^21,22

NO FURTHER SEIZURES

The patient was admitted to the observation unit from the emergency department after undergoing CT without intravenous contrast. While in observation, she had no additional episodes, and her vital signs remained within normal limits.

She underwent MRI and EEG as well as repeat laboratory studies and consultation by a neurologist. CT showed no structural abnormality, MRI results were read as normal, and EEG showed no epileptiform spikes or abnormal slow waves or other abnormality consistent with seizure. The repeat laboratory studies revealed normalization of the prolactin level at 11.3 ng/mL (reference range 2.0–17.4).

The final impression of the neurology consultant was that the patient had had a seizure that was most likely due to recently starting bupropion in combination with the withdrawal of the benzodiazepine, which lowered the seizure threshold. The neurologist also believed that our patient had no findings or symptoms other than the seizure that would suggest benzodiazepine withdrawal syndrome. According to the patient’s social history, it was unlikely that she had the pattern of alcohol consumption that would result in ethanol withdrawal seizure.

Seizures are common. In fact, every year, 180,000 US adults have their first seizure, and 10% of Americans will experience at least 1 seizure during their lifetime. However, only 20% to 25% of seizures are generalized tonic-clonic seizures as in our patient.²³

As this patient had an identifiable cause for the seizure, there was no need to initiate anticonvulsant therapy at the time of discharge. She was discharged to home without any anticonvulsant, the bupropion was discontinued, and the lorazepam was not restarted. When contacted by telephone at 1 month and 18 months after discharge, she reported she had not experienced any additional seizures and has not needed antiepileptic medications.

A 50-year-old woman presented to the emergency department after a witnessed loss of consciousness and seizurelike activity. She reported that she had been sitting outside her home, drinking coffee in the morning, but became very lightheaded when she went back into her house. At that time she felt could not focus and had a sense of impending doom. She sat down in a chair and her symptoms worsened.

According to her family, her eyes rolled back and she became rigid. The family helped her to the floor. Her body then made jerking movements that lasted for about 1 minute. She regained consciousness but was very confused for about 10 minutes until emergency medical services personnel arrived. She had no recollection of passing out. She said nothing like this had ever happened to her before.

On arrival in the emergency department, she complained of generalized headache and muscle soreness. She said the headache had been present for 1 week and was constant and dull. There were no aggravating or alleviating factors associated with the headache, and she denied fever, chills, nausea, numbness, tingling, incontinence, tongue biting, tremor, poor balance, ringing in ears, speech difficulty, or weakness.

Medical history: Multiple problems, medications

The patient’s medical history included depression, hypertension, anxiety, osteoarthritis, and asthma. She was allergic to penicillin. She had undergone carpal tunnel surgery on her right hand 5 years previously. She was perimenopausal with no children.

She denied using illicit drugs. She said she had smoked a half pack of cigarettes per day for more than 10 years and was a current smoker but was actively trying to quit. She said she occasionally used alcohol but had not consumed any alcohol in the last 2 weeks.

She had no history of central nervous system infection. She did report an episode of head trauma in grade school when a portable basketball hoop fell, striking her on the top of the head and causing her to briefly lose consciousness, but she did not seek medical attention.

She had no family history of seizure or neurologic disease.

Her current medications included atenolol, naproxen, gabapentin, venlafaxine, zolpidem, lorazepam, bupropion, and meloxicam. The bupropion and lorazepam had been prescribed recently for her anxiety. She reported that she had been given only 10 tablets of lorazepam and had taken the last tablet 48 hours previously. She had been taking the bupropion for 7 days. She reported an increase in stress lately and had been taking zolpidem due to an altered sleep pattern.

PHYSICAL EXAMINATION, INITIAL TESTS

On examination, the patient did not appear to be in acute distress. Her blood pressure was 107/22 mm Hg, pulse 100 beats per minute, respiratory rate 16 breaths per minute, temperature 37.1°C (98.8°F), and oxygen saturation 98% on room air.

Examination of her head, eyes, mouth, and neck were unremarkable. Cardiovascular, pulmonary, and abdominal examinations were normal. She had no neurologic deficits and was fully alert and oriented. She had no visible injuries.

Blood and urine samples were obtained about 15 minutes after her arrival to the emergency department. Results showed:

• Glucose 73 mg/dL (reference range 74–99)
• Sodium 142 mmol/L (136–144)
• Blood urea nitrogen 12 mg/dL (7–21)
• Creatinine 0.95 mg/dL (0.58–0.96)
• Chloride 97 mmol/L (97–105)
• Carbon dioxide (bicarbonate) 16 mmol/L (22–30)
• Prolactin 50.9 ng/mL (4.5–26.8)
• Anion gap 29 mmol/L (9–18)
• Ethanol undetectable
• White blood cell count 11.03 × 10⁹/L (3.70–11.00)
• Creatine kinase 89 U/L (30–220)
• Urinalysis normal, specific gravity 1.010 (1.005–1.030), no detectable ketones, and no crystals seen on microscopic evaluation.

Electrocardiography showed normal sinus rhythm with no ectopy and no ST-segment changes. Chest radiography was negative for any acute process.

The patient was transferred to the 23-hour observation unit in stable condition for further evaluation, monitoring, and management.

SIGNS AND SYMPTOMS OF SEIZURE

1. What findings are consistent with seizure?

• Jerking movements
• Confusion following the event
• Tongue-biting
• Focal motor weakness
• Urinary incontinence
• Aura before the event

All of the above findings are consistent with seizure.

The first consideration in evaluating a patient who presents with a possible seizure is whether the patient’s recollections of the event—and those of the witnesses—are consistent with the symptoms of seizure.¹

In generalized tonic-clonic or grand mal seizure, the patient may experience an aura or subjective sensations before the onset. These vary greatly among patients.² There may be an initial vocalization at the onset of the seizure, such as crying out or unintelligible speech. The patient’s eyes may roll back in the head. This is followed by loss of muscle tone, and if the patient is standing, he or she may fall to the ground. The patient becomes unresponsive and may go into respiratory arrest. There is tonic stiffening of the limbs and body, followed by clonic movements typically lasting 1 to 2 minutes, or sometimes longer.^1,3,4 The patient will then relax and experience a period of unconsciousness or confusion (postictal state).

Urinary incontinence and tongue-biting strongly suggest seizure activity, and turning the head to one side and posturing may also be seen.^3,5 After the event, the patient may report headache, generalized muscle soreness, exhaustion, or periods of transient focal weakness, also known as Todd paralysis.^2,5

Our patient had aura-like symptoms at the outset. She felt very lightheaded, had difficulty focusing, and felt a sense of impending doom. She did not make any vocalizations at the onset, but her eyes did roll backward and she became rigid (tonic). She then lost muscle tone and became unresponsive. Her family had to help her to the floor. Jerking (clonic) movements were witnessed.

She regained consciousness but was described as being confused (postictal) for 10 minutes. Additionally, she denied ever having had symptoms like this previously. On arrival in the emergency department, she reported generalized headache and muscle soreness, but no tongue-biting or urinary incontinence. Her event did not last for more than 1 to 2 minutes according to her family.

Her symptoms strongly suggest new-onset tonic-clonic or grand mal seizure, though this is not completely certain.

2. What laboratory results are consistent with seizure?

• Prolactin elevation
• Anion gap acidosis
• Leukocytosis

As noted above, the patient had an elevated prolactin level and elevated anion gap. Both of these findings can be used, with caution, in evaluating seizure activity.

Prolactin testing is controversial

Prolactin testing in diagnosing seizure activity is controversial. The exact mechanism of prolactin release in seizures is not fully understood. Generalized tonic-clonic seizures and complex partial seizures have both been shown to elevate prolactin. Prolactin levels after these types of seizures should rise within 30 minutes of the event and normalize 1 hour later.⁶

However, other events and conditions that mimic seizure have been shown to cause a rise in prolactin; these include syncope, transient ischemic attack, cardiac dysrhythmia, migraine, and other epilepsy-like variants. This effect has not been adequately studied. Therefore, an elevated prolactin level alone cannot diagnose or exclude seizure.⁷

For the prolactin level to be helpful, the blood sample must be drawn within 10 to 20 minutes after a possible seizure. Even if the prolactin level remains normal, it does not rule out seizure. Prolactin levels should therefore be used in combination with other testing to make a definitive diagnosis or exclusion of seizure.⁸

Anion gap and Denver Seizure Score

The anion gap has also been shown to rise after generalized seizure due to the metabolic acidosis that occurs. With a bicarbonate level of 16 mmol/L, an elevated anion gap, and normal breathing, our patient very likely had metabolic acidosis.

It is sometimes difficult to differentiate syncope from seizure, as they share several features.

The Denver Seizure Score can help differentiate these two conditions. It is based on the patient’s anion gap and bicarbonate level and is calculated as follows: 

(24 – bicarbonate) + [2 × (anion gap – 12)]

A score above 20 strongly indicates seizure activity. However, this is not a definitive tool for diagnosis. Like an elevated prolactin level, the Denver Seizure Score should be used in combination with other testing to move toward a definitive diagnosis.⁹

Our patient’s anion gap was 29 mmol/L and her bicarbonate level was 16 mmol/L. Her Denver Seizure Score was therefore 42, which supports this being an episode of generalized seizure activity.

Leukocytosis

The patient had a white blood cell count of 11.03 × 10⁹/L, which was mildly elevated. She had no history of fever and no source of infection by history.

Leukocytosis is common following generalized tonic-clonic seizure. A fever may lower the seizure threshold; however, our patient was not febrile and clinically had no factors that raised concern for an underlying infection.

ANION GAP ACIDOSIS AND SEIZURE

3. Which of the following can cause both anion gap acidosis and seizure?

• Ethylene glycol
• Salicylate overdose
• Ethanol withdrawal without ketosis
• Alcoholic ketoacidosis
• Methanol

All of the above except for ethanol withdrawal without ketosis can cause both anion gap acidosis and seizure.

Ethylene glycol can cause seizure and an elevated anion gap acidosis. However, this patient had no history of ingesting antifreeze (the most common source of ethylene glycol in the home) and no evidence of calcium oxalate crystals in the urine, which would be a sign of ethylene glycol toxicity. Additional testing for ethylene glycol may include serum ethylene glycol levels and ultraviolet light testing of the urine to detect fluorescein, which is commonly added to automotive antifreeze to help mechanics find fluid leaks in engines.

Salicylate overdose can cause seizure and an elevated anion gap acidosis. However, this patient has no history of aspirin ingestion, and a serum aspirin level was later ordered and found to be negative. In addition, the acid-base disorder in salicylate overdose may be respiratory alkalosis from direct stimulation of respiratory centers in conjunction with metabolic acidosis.

Ethanol withdrawal can cause seizure and may result in ketoacidosis, which would appear as anion gap acidosis. The undetectable ethanol level in this patient would be consistent with withdrawal from ethanol, which may also lead to ketoacidosis.

Alcoholic ketoacidosis is a late finding in patients who have been drinking ethanol and is thus a possible cause of an elevated anion gap in this patient. However, the absence of ketones in her urine speaks against this diagnosis.

Methanol can cause seizure and acidosis, but laboratory testing would reveal a normal anion gap and an elevated osmolar gap. This was not likely in this patient.

The presence of anion gap acidosis is important in forming a differential diagnosis. Several causes of anion gap acidosis may also cause seizure. These include salicylates, ethanol withdrawal with ketosis, methanol, and isoniazid. None of these appears to be a factor in this patient’s case.

DIFFERENTIAL DIAGNOSIS IN OUR PATIENT

4. What is the most likely cause of this patient’s seizure?

• Bupropion side effect
• Benzodiazepine withdrawal
• Ethanol withdrawal
• Brain lesion
• Central nervous system infection
• Unprovoked seizure (new-onset epilepsy)

Bupropion, an inhibitor of neuronal reuptake of norepinephrine and dopamine, has been used in the United States since 1989 to treat major depression.¹⁰ At therapeutic doses, it lowers the seizure threshold; in cases of acute overdose, seizures typically occur within hours of the dose, or up to 24 hours in patients taking extended-release formulations.¹¹

Bupropion should be used with caution or avoided in patients taking other medications that also lower the seizure threshold, or during withdrawal from alcohol, benzodiazepines, or barbiturates.¹⁰

Benzodiazepine withdrawal. Abrupt cessation of benzodiazepines also lowers the seizure threshold, and seizures are commonly seen in benzodiazepine withdrawal syndrome. The use of benzodiazepines is controversial in many situations, and discontinuing them may prove problematic for both the patient and physician, as the potential for abuse and addiction is significant.

Seizures have occurred during withdrawal from even short-term benzodiazepine use. Other factors, such as concomitant use of other medications that lower the seizure threshold, may play a more significant role in causing withdrawal seizures than the duration of benzodiazepine therapy.¹²

Medications shown to be useful in managing withdrawal from benzodiazepines include carbamazepine, imipramine, valproate, and trazodone. Paroxetine has also been shown to be helpful in patients with major depression who are being taken off a benzodiazepine.¹³

Ethanol withdrawal is common in patients presenting to emergency departments, and seizures are frequently seen in these patients. This patient reported social drinking but not drinking ethanol daily, although many patients are not forthcoming about alcohol or drug use when talking with a physician or other healthcare provider.

Alcohol withdrawal seizures may accompany delirium tremens or major withdrawal syndrome, but they are seen more often in the absence of major withdrawal or delirium tremens. Seizures are typically single or occur in a short grouping over a brief period of time and mostly occur in chronic alcoholism. The role of anticonvulsants in patients with alcohol withdrawal seizure has not been established.¹⁴

Brain lesion. A previously undiagnosed brain tumor is not a common cause of new-onset seizure, although it is not unusual for a brain tumor to cause new-onset seizure. In 1 study, 6% of patients with new-onset seizure had a clinically significant lesion on brain imaging.¹⁵ In addition, 15% to 30% of patients with a previously undiagnosed brain tumor present with seizure as the first symptom.¹⁶ Patients with abnormal findings on neurologic examination after the seizure activity are more likely to have a structural lesion that may be identified by computed tomography (CT) or magnetic resonance imaging. (MRI)¹⁵

Unprovoked seizure occurs without an identifiable precipitating factor, or from a central nervous system insult that occurred more than 7 days earlier. Patients who may have recurrent unprovoked seizure will likely be diagnosed with epilepsy.¹⁵ Patients with a first-time unprovoked seizure have a 30% or higher likelihood of having another unprovoked seizure within 5 years.¹⁷

It is most likely that bupropion is the key factor in lowering the seizure threshold in this patient. However, patients sometimes underreport the amount of alcohol they consume, and though less likely, our patient’s report of not drinking for 2 weeks may also be unreliable. Ethanol withdrawal, though unlikely, may also be a consideration with this case.

5. Which tests may be helpful in this patient’s workup?

• CT of the brain
• Lumbar puncture for spinal fluid analysis
• MRI of the brain
• Electroencephalography (EEG)

This patient had had a headache for 1 week before presenting to the emergency department. Indications for neuroimaging in a patient with headache include sudden onset of severe headache, neurologic deficits, human immunodeficiency virus infection, loss of consciousness, immunosuppression, pregnancy, malignancy, and age over 50 with a new type of headache.^18,19 Therefore, she should undergo some form of neuroimaging, either CT or MRI.

CT is the most readily available and fastest imaging study for the central nervous system available to emergency physicians. CT is limited, however, due to its decreased sensitivity in detecting some brain lesions. Therefore, many patients with first-time seizure may eventually require MRI.¹⁵ Furthermore, patients with focal onset of the seizure activity are more likely to have a structural lesion precipitating the seizure.  MRI may have a higher yield than CT in these cases.^15,20

Lumbar puncture for spinal fluid analysis is helpful in evaluating a patient with a suspected central nervous system infection such as meningitis or encephalitis, or subarachnoid hemorrhage.

This patient had a normal neurologic examination, no fever, and no meningeal signs, and central nervous system infection was very unlikely. Also, because she had had a headache for 1 week before the presentation with seizurelike activity, subarachnoid hemorrhage was very unlikely, and emergency lumbar puncture was not indicated.

MRI is less readily available than CT in a timely fashion in most emergency departments in the United States. It offers a higher yield than CT in diagnosing pathology such as acute stroke, brain tumor, and plaques seen in multiple sclerosis. CT is superior to MRI in diagnosing bony abnormalities and is very sensitive for detecting acute bleeding.

If MRI is performed, it should follow a specific protocol that includes high-resolution images for epilepsy evaluation rather than the more commonly ordered stroke protocol. The stroke protocol is more likely to be ordered in the emergency department.

EEG is well established in evaluating new-onset seizure in pediatric patients. Studies also demonstrate its utility in evaluating first-time seizure in adults, providing evidence that both epileptiform and nonepileptiform abnormalities seen on EEG are associated with a higher risk of recurrent seizure activity than in patients with normal findings on EEG.¹

EEG may be difficult to interpret in adults. According to Benbadis,⁵ as many as one-third of adult patients diagnosed with epilepsy on EEG did not have epilepsy. This is because of normal variants, simple fluctuations of background rhythms, or fragmented alpha activity that can have a similar appearance to epileptiform patterns. EEG must always be interpreted in the context of the patient’s history and symptoms.⁵

Though EEG has limitations, it remains a crucial tool for identifying epilepsy. Following a single seizure, the decision to prescribe antiepileptic drugs is highly influenced by patterns on EEG associated with a risk of recurrence. In fact, a patient experiencing a single, idiopathic seizure and exhibiting an EEG pattern of spike wave discharges is likely to have recurrent seizure activity.²¹ Also, the appropriate use of EEG after even a single unprovoked seizure can identify patients with epilepsy and a risk of recurrent seizure greater than 60%.^21,22

NO FURTHER SEIZURES

The patient was admitted to the observation unit from the emergency department after undergoing CT without intravenous contrast. While in observation, she had no additional episodes, and her vital signs remained within normal limits.

She underwent MRI and EEG as well as repeat laboratory studies and consultation by a neurologist. CT showed no structural abnormality, MRI results were read as normal, and EEG showed no epileptiform spikes or abnormal slow waves or other abnormality consistent with seizure. The repeat laboratory studies revealed normalization of the prolactin level at 11.3 ng/mL (reference range 2.0–17.4).

The final impression of the neurology consultant was that the patient had had a seizure that was most likely due to recently starting bupropion in combination with the withdrawal of the benzodiazepine, which lowered the seizure threshold. The neurologist also believed that our patient had no findings or symptoms other than the seizure that would suggest benzodiazepine withdrawal syndrome. According to the patient’s social history, it was unlikely that she had the pattern of alcohol consumption that would result in ethanol withdrawal seizure.

Seizures are common. In fact, every year, 180,000 US adults have their first seizure, and 10% of Americans will experience at least 1 seizure during their lifetime. However, only 20% to 25% of seizures are generalized tonic-clonic seizures as in our patient.²³

As this patient had an identifiable cause for the seizure, there was no need to initiate anticonvulsant therapy at the time of discharge. She was discharged to home without any anticonvulsant, the bupropion was discontinued, and the lorazepam was not restarted. When contacted by telephone at 1 month and 18 months after discharge, she reported she had not experienced any additional seizures and has not needed antiepileptic medications.

A 50-year-old woman presented to the emergency department after a witnessed loss of consciousness and seizurelike activity. She reported that she had been sitting outside her home, drinking coffee in the morning, but became very lightheaded when she went back into her house. At that time she felt could not focus and had a sense of impending doom. She sat down in a chair and her symptoms worsened.

According to her family, her eyes rolled back and she became rigid. The family helped her to the floor. Her body then made jerking movements that lasted for about 1 minute. She regained consciousness but was very confused for about 10 minutes until emergency medical services personnel arrived. She had no recollection of passing out. She said nothing like this had ever happened to her before.

On arrival in the emergency department, she complained of generalized headache and muscle soreness. She said the headache had been present for 1 week and was constant and dull. There were no aggravating or alleviating factors associated with the headache, and she denied fever, chills, nausea, numbness, tingling, incontinence, tongue biting, tremor, poor balance, ringing in ears, speech difficulty, or weakness.

Medical history: Multiple problems, medications

The patient’s medical history included depression, hypertension, anxiety, osteoarthritis, and asthma. She was allergic to penicillin. She had undergone carpal tunnel surgery on her right hand 5 years previously. She was perimenopausal with no children.

She denied using illicit drugs. She said she had smoked a half pack of cigarettes per day for more than 10 years and was a current smoker but was actively trying to quit. She said she occasionally used alcohol but had not consumed any alcohol in the last 2 weeks.

She had no history of central nervous system infection. She did report an episode of head trauma in grade school when a portable basketball hoop fell, striking her on the top of the head and causing her to briefly lose consciousness, but she did not seek medical attention.

She had no family history of seizure or neurologic disease.

Her current medications included atenolol, naproxen, gabapentin, venlafaxine, zolpidem, lorazepam, bupropion, and meloxicam. The bupropion and lorazepam had been prescribed recently for her anxiety. She reported that she had been given only 10 tablets of lorazepam and had taken the last tablet 48 hours previously. She had been taking the bupropion for 7 days. She reported an increase in stress lately and had been taking zolpidem due to an altered sleep pattern.

PHYSICAL EXAMINATION, INITIAL TESTS

On examination, the patient did not appear to be in acute distress. Her blood pressure was 107/22 mm Hg, pulse 100 beats per minute, respiratory rate 16 breaths per minute, temperature 37.1°C (98.8°F), and oxygen saturation 98% on room air.

Examination of her head, eyes, mouth, and neck were unremarkable. Cardiovascular, pulmonary, and abdominal examinations were normal. She had no neurologic deficits and was fully alert and oriented. She had no visible injuries.

Blood and urine samples were obtained about 15 minutes after her arrival to the emergency department. Results showed:

• Glucose 73 mg/dL (reference range 74–99)
• Sodium 142 mmol/L (136–144)
• Blood urea nitrogen 12 mg/dL (7–21)
• Creatinine 0.95 mg/dL (0.58–0.96)
• Chloride 97 mmol/L (97–105)
• Carbon dioxide (bicarbonate) 16 mmol/L (22–30)
• Prolactin 50.9 ng/mL (4.5–26.8)
• Anion gap 29 mmol/L (9–18)
• Ethanol undetectable
• White blood cell count 11.03 × 10⁹/L (3.70–11.00)
• Creatine kinase 89 U/L (30–220)
• Urinalysis normal, specific gravity 1.010 (1.005–1.030), no detectable ketones, and no crystals seen on microscopic evaluation.

Electrocardiography showed normal sinus rhythm with no ectopy and no ST-segment changes. Chest radiography was negative for any acute process.

The patient was transferred to the 23-hour observation unit in stable condition for further evaluation, monitoring, and management.

SIGNS AND SYMPTOMS OF SEIZURE

1. What findings are consistent with seizure?

• Jerking movements
• Confusion following the event
• Tongue-biting
• Focal motor weakness
• Urinary incontinence
• Aura before the event

All of the above findings are consistent with seizure.

The first consideration in evaluating a patient who presents with a possible seizure is whether the patient’s recollections of the event—and those of the witnesses—are consistent with the symptoms of seizure.¹

In generalized tonic-clonic or grand mal seizure, the patient may experience an aura or subjective sensations before the onset. These vary greatly among patients.² There may be an initial vocalization at the onset of the seizure, such as crying out or unintelligible speech. The patient’s eyes may roll back in the head. This is followed by loss of muscle tone, and if the patient is standing, he or she may fall to the ground. The patient becomes unresponsive and may go into respiratory arrest. There is tonic stiffening of the limbs and body, followed by clonic movements typically lasting 1 to 2 minutes, or sometimes longer.^1,3,4 The patient will then relax and experience a period of unconsciousness or confusion (postictal state).

Urinary incontinence and tongue-biting strongly suggest seizure activity, and turning the head to one side and posturing may also be seen.^3,5 After the event, the patient may report headache, generalized muscle soreness, exhaustion, or periods of transient focal weakness, also known as Todd paralysis.^2,5

Our patient had aura-like symptoms at the outset. She felt very lightheaded, had difficulty focusing, and felt a sense of impending doom. She did not make any vocalizations at the onset, but her eyes did roll backward and she became rigid (tonic). She then lost muscle tone and became unresponsive. Her family had to help her to the floor. Jerking (clonic) movements were witnessed.

She regained consciousness but was described as being confused (postictal) for 10 minutes. Additionally, she denied ever having had symptoms like this previously. On arrival in the emergency department, she reported generalized headache and muscle soreness, but no tongue-biting or urinary incontinence. Her event did not last for more than 1 to 2 minutes according to her family.

Her symptoms strongly suggest new-onset tonic-clonic or grand mal seizure, though this is not completely certain.

2. What laboratory results are consistent with seizure?

• Prolactin elevation
• Anion gap acidosis
• Leukocytosis

As noted above, the patient had an elevated prolactin level and elevated anion gap. Both of these findings can be used, with caution, in evaluating seizure activity.

Prolactin testing is controversial

Prolactin testing in diagnosing seizure activity is controversial. The exact mechanism of prolactin release in seizures is not fully understood. Generalized tonic-clonic seizures and complex partial seizures have both been shown to elevate prolactin. Prolactin levels after these types of seizures should rise within 30 minutes of the event and normalize 1 hour later.⁶

However, other events and conditions that mimic seizure have been shown to cause a rise in prolactin; these include syncope, transient ischemic attack, cardiac dysrhythmia, migraine, and other epilepsy-like variants. This effect has not been adequately studied. Therefore, an elevated prolactin level alone cannot diagnose or exclude seizure.⁷

For the prolactin level to be helpful, the blood sample must be drawn within 10 to 20 minutes after a possible seizure. Even if the prolactin level remains normal, it does not rule out seizure. Prolactin levels should therefore be used in combination with other testing to make a definitive diagnosis or exclusion of seizure.⁸

Anion gap and Denver Seizure Score

The anion gap has also been shown to rise after generalized seizure due to the metabolic acidosis that occurs. With a bicarbonate level of 16 mmol/L, an elevated anion gap, and normal breathing, our patient very likely had metabolic acidosis.

It is sometimes difficult to differentiate syncope from seizure, as they share several features.

The Denver Seizure Score can help differentiate these two conditions. It is based on the patient’s anion gap and bicarbonate level and is calculated as follows: 

(24 – bicarbonate) + [2 × (anion gap – 12)]

A score above 20 strongly indicates seizure activity. However, this is not a definitive tool for diagnosis. Like an elevated prolactin level, the Denver Seizure Score should be used in combination with other testing to move toward a definitive diagnosis.⁹

Our patient’s anion gap was 29 mmol/L and her bicarbonate level was 16 mmol/L. Her Denver Seizure Score was therefore 42, which supports this being an episode of generalized seizure activity.

Leukocytosis

The patient had a white blood cell count of 11.03 × 10⁹/L, which was mildly elevated. She had no history of fever and no source of infection by history.

Leukocytosis is common following generalized tonic-clonic seizure. A fever may lower the seizure threshold; however, our patient was not febrile and clinically had no factors that raised concern for an underlying infection.

ANION GAP ACIDOSIS AND SEIZURE

3. Which of the following can cause both anion gap acidosis and seizure?

• Ethylene glycol
• Salicylate overdose
• Ethanol withdrawal without ketosis
• Alcoholic ketoacidosis
• Methanol

All of the above except for ethanol withdrawal without ketosis can cause both anion gap acidosis and seizure.

Ethylene glycol can cause seizure and an elevated anion gap acidosis. However, this patient had no history of ingesting antifreeze (the most common source of ethylene glycol in the home) and no evidence of calcium oxalate crystals in the urine, which would be a sign of ethylene glycol toxicity. Additional testing for ethylene glycol may include serum ethylene glycol levels and ultraviolet light testing of the urine to detect fluorescein, which is commonly added to automotive antifreeze to help mechanics find fluid leaks in engines.

Salicylate overdose can cause seizure and an elevated anion gap acidosis. However, this patient has no history of aspirin ingestion, and a serum aspirin level was later ordered and found to be negative. In addition, the acid-base disorder in salicylate overdose may be respiratory alkalosis from direct stimulation of respiratory centers in conjunction with metabolic acidosis.

Ethanol withdrawal can cause seizure and may result in ketoacidosis, which would appear as anion gap acidosis. The undetectable ethanol level in this patient would be consistent with withdrawal from ethanol, which may also lead to ketoacidosis.

Alcoholic ketoacidosis is a late finding in patients who have been drinking ethanol and is thus a possible cause of an elevated anion gap in this patient. However, the absence of ketones in her urine speaks against this diagnosis.

Methanol can cause seizure and acidosis, but laboratory testing would reveal a normal anion gap and an elevated osmolar gap. This was not likely in this patient.

The presence of anion gap acidosis is important in forming a differential diagnosis. Several causes of anion gap acidosis may also cause seizure. These include salicylates, ethanol withdrawal with ketosis, methanol, and isoniazid. None of these appears to be a factor in this patient’s case.

DIFFERENTIAL DIAGNOSIS IN OUR PATIENT

4. What is the most likely cause of this patient’s seizure?

• Bupropion side effect
• Benzodiazepine withdrawal
• Ethanol withdrawal
• Brain lesion
• Central nervous system infection
• Unprovoked seizure (new-onset epilepsy)

Bupropion, an inhibitor of neuronal reuptake of norepinephrine and dopamine, has been used in the United States since 1989 to treat major depression.¹⁰ At therapeutic doses, it lowers the seizure threshold; in cases of acute overdose, seizures typically occur within hours of the dose, or up to 24 hours in patients taking extended-release formulations.¹¹

Bupropion should be used with caution or avoided in patients taking other medications that also lower the seizure threshold, or during withdrawal from alcohol, benzodiazepines, or barbiturates.¹⁰

Benzodiazepine withdrawal. Abrupt cessation of benzodiazepines also lowers the seizure threshold, and seizures are commonly seen in benzodiazepine withdrawal syndrome. The use of benzodiazepines is controversial in many situations, and discontinuing them may prove problematic for both the patient and physician, as the potential for abuse and addiction is significant.

Seizures have occurred during withdrawal from even short-term benzodiazepine use. Other factors, such as concomitant use of other medications that lower the seizure threshold, may play a more significant role in causing withdrawal seizures than the duration of benzodiazepine therapy.¹²

Medications shown to be useful in managing withdrawal from benzodiazepines include carbamazepine, imipramine, valproate, and trazodone. Paroxetine has also been shown to be helpful in patients with major depression who are being taken off a benzodiazepine.¹³

Ethanol withdrawal is common in patients presenting to emergency departments, and seizures are frequently seen in these patients. This patient reported social drinking but not drinking ethanol daily, although many patients are not forthcoming about alcohol or drug use when talking with a physician or other healthcare provider.

Alcohol withdrawal seizures may accompany delirium tremens or major withdrawal syndrome, but they are seen more often in the absence of major withdrawal or delirium tremens. Seizures are typically single or occur in a short grouping over a brief period of time and mostly occur in chronic alcoholism. The role of anticonvulsants in patients with alcohol withdrawal seizure has not been established.¹⁴

Brain lesion. A previously undiagnosed brain tumor is not a common cause of new-onset seizure, although it is not unusual for a brain tumor to cause new-onset seizure. In 1 study, 6% of patients with new-onset seizure had a clinically significant lesion on brain imaging.¹⁵ In addition, 15% to 30% of patients with a previously undiagnosed brain tumor present with seizure as the first symptom.¹⁶ Patients with abnormal findings on neurologic examination after the seizure activity are more likely to have a structural lesion that may be identified by computed tomography (CT) or magnetic resonance imaging. (MRI)¹⁵

Unprovoked seizure occurs without an identifiable precipitating factor, or from a central nervous system insult that occurred more than 7 days earlier. Patients who may have recurrent unprovoked seizure will likely be diagnosed with epilepsy.¹⁵ Patients with a first-time unprovoked seizure have a 30% or higher likelihood of having another unprovoked seizure within 5 years.¹⁷

It is most likely that bupropion is the key factor in lowering the seizure threshold in this patient. However, patients sometimes underreport the amount of alcohol they consume, and though less likely, our patient’s report of not drinking for 2 weeks may also be unreliable. Ethanol withdrawal, though unlikely, may also be a consideration with this case.

5. Which tests may be helpful in this patient’s workup?

• CT of the brain
• Lumbar puncture for spinal fluid analysis
• MRI of the brain
• Electroencephalography (EEG)

This patient had had a headache for 1 week before presenting to the emergency department. Indications for neuroimaging in a patient with headache include sudden onset of severe headache, neurologic deficits, human immunodeficiency virus infection, loss of consciousness, immunosuppression, pregnancy, malignancy, and age over 50 with a new type of headache.^18,19 Therefore, she should undergo some form of neuroimaging, either CT or MRI.

CT is the most readily available and fastest imaging study for the central nervous system available to emergency physicians. CT is limited, however, due to its decreased sensitivity in detecting some brain lesions. Therefore, many patients with first-time seizure may eventually require MRI.¹⁵ Furthermore, patients with focal onset of the seizure activity are more likely to have a structural lesion precipitating the seizure.  MRI may have a higher yield than CT in these cases.^15,20

Lumbar puncture for spinal fluid analysis is helpful in evaluating a patient with a suspected central nervous system infection such as meningitis or encephalitis, or subarachnoid hemorrhage.

This patient had a normal neurologic examination, no fever, and no meningeal signs, and central nervous system infection was very unlikely. Also, because she had had a headache for 1 week before the presentation with seizurelike activity, subarachnoid hemorrhage was very unlikely, and emergency lumbar puncture was not indicated.

MRI is less readily available than CT in a timely fashion in most emergency departments in the United States. It offers a higher yield than CT in diagnosing pathology such as acute stroke, brain tumor, and plaques seen in multiple sclerosis. CT is superior to MRI in diagnosing bony abnormalities and is very sensitive for detecting acute bleeding.

If MRI is performed, it should follow a specific protocol that includes high-resolution images for epilepsy evaluation rather than the more commonly ordered stroke protocol. The stroke protocol is more likely to be ordered in the emergency department.

EEG is well established in evaluating new-onset seizure in pediatric patients. Studies also demonstrate its utility in evaluating first-time seizure in adults, providing evidence that both epileptiform and nonepileptiform abnormalities seen on EEG are associated with a higher risk of recurrent seizure activity than in patients with normal findings on EEG.¹

EEG may be difficult to interpret in adults. According to Benbadis,⁵ as many as one-third of adult patients diagnosed with epilepsy on EEG did not have epilepsy. This is because of normal variants, simple fluctuations of background rhythms, or fragmented alpha activity that can have a similar appearance to epileptiform patterns. EEG must always be interpreted in the context of the patient’s history and symptoms.⁵

Though EEG has limitations, it remains a crucial tool for identifying epilepsy. Following a single seizure, the decision to prescribe antiepileptic drugs is highly influenced by patterns on EEG associated with a risk of recurrence. In fact, a patient experiencing a single, idiopathic seizure and exhibiting an EEG pattern of spike wave discharges is likely to have recurrent seizure activity.²¹ Also, the appropriate use of EEG after even a single unprovoked seizure can identify patients with epilepsy and a risk of recurrent seizure greater than 60%.^21,22

NO FURTHER SEIZURES

The patient was admitted to the observation unit from the emergency department after undergoing CT without intravenous contrast. While in observation, she had no additional episodes, and her vital signs remained within normal limits.

She underwent MRI and EEG as well as repeat laboratory studies and consultation by a neurologist. CT showed no structural abnormality, MRI results were read as normal, and EEG showed no epileptiform spikes or abnormal slow waves or other abnormality consistent with seizure. The repeat laboratory studies revealed normalization of the prolactin level at 11.3 ng/mL (reference range 2.0–17.4).

The final impression of the neurology consultant was that the patient had had a seizure that was most likely due to recently starting bupropion in combination with the withdrawal of the benzodiazepine, which lowered the seizure threshold. The neurologist also believed that our patient had no findings or symptoms other than the seizure that would suggest benzodiazepine withdrawal syndrome. According to the patient’s social history, it was unlikely that she had the pattern of alcohol consumption that would result in ethanol withdrawal seizure.

Seizures are common. In fact, every year, 180,000 US adults have their first seizure, and 10% of Americans will experience at least 1 seizure during their lifetime. However, only 20% to 25% of seizures are generalized tonic-clonic seizures as in our patient.²³

As this patient had an identifiable cause for the seizure, there was no need to initiate anticonvulsant therapy at the time of discharge. She was discharged to home without any anticonvulsant, the bupropion was discontinued, and the lorazepam was not restarted. When contacted by telephone at 1 month and 18 months after discharge, she reported she had not experienced any additional seizures and has not needed antiepileptic medications.

Emergency departments are not primary care clinics, but some patients use them that way. This relatively small group of patients consumes a disproportionate share of healthcare at great cost, earning them the label of “high users.” Mostly poor and often burdened with mental illness and addiction, they are not necessarily sicker than other patients, and they do not enjoy better outcomes from the extra money spent on them. (Another subset of high users, those with end-stage chronic disease, is outside the scope of this review.)

Herein lies an opportunity. If—and this is a big if—we could manage their care in a systematic way instead of haphazardly, proactively instead of reactively, with continuity of care instead of episodically, and in a way that is convenient for the patient, we might be able to improve quality and save money.

A DISPROPORTIONATE SHARE OF COSTS

In the United States in 2012, the 5% of the population who were the highest users were responsible for 50% of healthcare costs.¹ The mean cost per person in this group was more than $43,000 annually. The top 1% of users accounted for nearly 23% of all expenditures, averaging nearly $98,000 per patient per year—10 times more than the average yearly cost per patient.

CARE IS OFTEN INAPPROPRIATE AND UNNECESSARY

In addition to being disproportionately expensive, the care that these patients receive is often inappropriate and unnecessary for the severity of their disease.

A 2007–2009 study² of 1,969 patients who had visited the emergency department 10 or more times in a year found they received more than twice as many computed tomography (CT) scans as a control group of infrequent users (< 3 visits/year). This occurred even though they were not as sick as infrequent users, based on significantly lower hospital admission rates (11.1% vs 17.9%; P < .001) and mortality rates (0.7% vs 1.5%; P < .002).²

This inverse relationship between emergency department use and illness severity was even more exaggerated at the upper extreme of the use curve. The highest users (> 29 visits to the emergency department in a year) had the lowest triage acuity and hospital admission rates but the highest number of CT scans. Charges per visit were lower among frequent users, but total charges rose steadily with increasing emergency department use, accounting for significantly more costs per year.²

We believe that one reason these patients receive more medical care than necessary is because their medical records are too large and complex for the average physician to distill effectively in a 20-minute physician-patient encounter. Physicians therefore simply order more tests, procedures, and admissions, which are often medically unnecessary and redundant.

WHAT DRIVES HIGH COST?

Mental illness and chemical dependence

Drug addiction, mental illness, and poverty frequently accompany (and influence) high-use behavior, particularly in patients without end-stage diseases.

Szekendi et al,³ in a study of 28,291 patients who had been admitted at least 5 times in a year in a Chicago health system, found that these high users were 2 to 3 times more likely to suffer from comorbid depression (40% vs 13%), psychosis (18% vs 5%), recreational drug dependence (20% vs 7%), and alcohol abuse (16% vs 7%) than non-high-use hospitalized patients.³

Mercer et al⁴ conducted a study at Duke University Medical Center, Durham, NC, aimed at reducing emergency department visits and hospital admissions among 24 of its highest users. They found that 23 (96%) were either addicted to drugs or mentally ill, and 20 (83%) suffered from chronic pain.⁴

Drug abuse among high users is becoming even more relevant as the opioid epidemic worsens. Given that most patients requiring high levels of care suffer from chronic pain and many of them develop an opioid addiction while treating their pain, physicians have a moral imperative to reduce the prevalence of drug abuse in this population.

Low socioeconomic status

Low socioeconomic status is an important factor among high users, as it is highly associated with greater disease severity, which usually increases cost without any guarantee of an associated increase in quality. Data suggest that patients of low socioeconomic status are twice as likely to require urgent emergency department visits, 4 times as likely to require admission to the hospital, and, importantly, about half as likely to use ambulatory care compared with patients of higher socioeconomic status.⁵ While this pattern of low-quality, high-cost spending in acute care settings reflects spending in the healthcare system at large, the pattern is greatly exaggerated among high users.

Lost to follow-up

Low socioeconomic status also complicates communication and follow-up. In a 2013 study, physician researchers in St. Paul, MN, documented attempts to interview 64 recently discharged high users. They could not reach 47 (73%) of them, for reasons largely attributable to low socioeconomic status, such as disconnected phone lines and changes in address.⁶

Clearly, the usual contact methods for follow-up care after discharge, such as phone calls and mailings, are unlikely to be effective in coordinating the outpatient care of these individuals.

Additionally, we must find ways of making primary care more convenient, gaining our patients’ trust, and finding ways to engage patients in follow-up without relying on traditional means of communication.

Do high users have medical insurance?

Surprisingly, most high users of the emergency department have health insurance. The Chicago health system study³ found that most (72.4%) of their high users had either Medicare or private health insurance, while 27.6% had either Medicaid or no insurance (compared with 21.6% in the general population). Other studies also found that most of the frequent emergency department users are insured,⁷ although the overall percentage who rely on publicly paid insurance is greater than in the population at large.

Many prefer acute care over primary care

Although one might think that high users go to the emergency department because they have nowhere else to go for care, a report published in 2013 by Kangovi et al⁵ suggests another reason—they prefer the emergency department.⁵ They interviewed 40 urban patients of low socioeconomic status who consistently cited the 24-hour, no-appointment-necessary structure of the emergency department as an advantage over primary care. The flexibility of emergency access to healthcare makes sense if one reflects on how difficult it is for even high-functioning individuals to schedule and keep medical appointments.

Specific reasons for preferring the emergency department included the following:

Affordability. Even if their insurance fully paid for visits to their primary care physicians, the primary care physician was likely to refer them to specialists, whose visits required a copay, and which required taking another day off of work. The emergency department is cheaper for the patient and it is a “one-stop shop.” Patients appreciated the emergency department guarantee of seeing a physician regardless of proof of insurance, a policy not guaranteed in primary care and specialist offices.

Accessibility. For those without a car, public transportation and even patient transportation services are inconvenient and unreliable, whereas emergency medical services will take you to the emergency department.

Accommodations. Although medical centers may tout their same-day appointments, often same-day appointments are all that they have—and you have no choice about the time. You have to call first thing in the morning and stay on hold for a long time, and then when you finally get through, all the same-day appointments are gone.

Availability. Patients said they often had a hard time getting timely medical advice from their primary care physicians. When they could get through to their primary care physicians on the phone, they would be told to go to the emergency department.

Acceptability. Men, especially, feel they need to be very sick indeed to seek medical care, so going to the emergency department is more acceptable.

Trust in the provider. For reasons that were not entirely clear, patients felt that acute care providers were more trustworthy, competent, and compassionate than primary care physicians.⁵

None of these reasons for using the emergency department has anything to do with disease severity, which supports the findings that high users of the emergency department were not as sick as their normal-use peers.²

QUALITY IMPROVEMENT AND COST-REDUCTION STRATEGIES

Efforts are being made to reduce the cost of healthcare for high users while improving the quality of their care. Promising strategies focus on coordinating care management, creating individualized patient care plans, and improving the components and instructions of discharge summaries.

Care management organizations

A care management organization (CMO) model has emerged as a strategy for quality improvement and cost reduction in the high-use population. In this model, social workers, health coaches, nurses, mid-level providers, and physicians collaborate on designing individualized care plans to meet the specific needs of patients.

Teams typically work in stepwise fashion, first identifying and engaging patients at high risk of poor outcomes and unnecessary care, often using sophisticated quantitative, risk-prediction tools. Then, they perform health assessments and identify potential interventions aimed at preventing expensive acute-care medical interventions. Third, they work with patients to rapidly identify and effectively respond to changes in their conditions and direct them to the most appropriate medical setting, typically primary or urgent care.

Effective models

In 1998, the Camden (NJ) Coalition of Healthcare Providers established a model for CMO care plans. Starting with the first 36 patients enrolled in the program, hospital admissions and emergency department visits were cut by 47% (from 62 to 37 per month), and collective hospital costs were cut by 56% (from $1.2 million to about $500,000 per month).⁸ It should be noted that this was a small, nonrandomized study and these preliminary numbers did not take into account the cost of outpatient physician visits or new medications. Thus, how much money this program actually saves is not clear.

Similar programs have had similar results. A nurse-led care coordination program in Doylestown, PA, showed an impressive 25% reduction in annual mortality and a 36% reduction in overall costs during a 10-year period.⁹

A program in Atlantic City, NJ, combined the typical CMO model with a primary care clinic to provide high users with unlimited access, while paying its providers in a capitation model (as opposed to fee for service). It achieved a 40% reduction in yearly emergency department visits and hospital admissions.⁸

Patient care plans

Individualized patient care plans for high users are among the most promising tools for reducing costs and improving quality in this group. They are low-cost and relatively easy to implement. The goal of these care plans is to provide practitioners with a concise care summary to help them make rational and consistent medical decisions.

Typically, a care plan is written by an interdisciplinary committee composed of physicians, nurses, and social workers. It is based on the patient’s pertinent medical and psychiatric history, which may include recent imaging results or other relevant diagnostic tests. It provides suggestions for managing complex chronic issues, such as drug abuse, that lead to high use of healthcare resources.

These care plans provide a rational and prespecified approach to workup and management, typically including a narcotic prescription protocol, regardless of the setting or the number of providers who see the patient. Practitioners guided by effective care plans are much more likely to effectively navigate a complex patient encounter as opposed to looking through extensive medical notes and hoping to find relevant information.

Effective models

Data show these plans can be effective. For example, Regions Hospital in St. Paul, MN, implemented patient care plans in 2010. During the first 4 months, hospital admissions in the first 94 patients were reduced by 67%.¹⁰

A study of high users at Duke University Medical Center reported similar results. One year after starting care plans, inpatient admissions had decreased by 50.5%, readmissions had decreased by 51.5%, and variable direct costs per admission were reduced by 35.8%. Paradoxically, emergency department visits went up, but this anomaly was driven by 134 visits incurred by a single dialysis patient. After removing this patient from the data, emergency department visits were relatively stable.⁴

Better discharge summaries

Although improving discharge summaries is not a novel concept, changing the summary from a historical document to a proactive discharge plan has the potential to prevent readmissions and promote a durable de-escalation in care acuity.

For example, when moving a patient to a subacute care facility, providing a concise summary of which treatments worked and which did not, a list of comorbidities, and a list of medications and strategies to consider, can help the next providers to better target their plan of care. Studies have shown that nearly half of discharge statements lack important information on treatments and tests.¹¹

Improvement can be as simple as encouraging practitioners to construct their summaries in an “if-then” format. Instead of noting for instance that “Mr. Smith was treated for pneumonia with antibiotics and discharged to a rehab facility,” the following would be more useful: “Family would like to see if Mr. Smith can get back to his functional baseline after his acute pneumonia. If he clinically does not do well over the next 1 to 2 weeks and has a poor quality of life, then family would like to pursue hospice.”

In addition to shifting the philosophy, we believe that providing timely discharge summaries is a fundamental, high-yield aspect of ensuring their effectiveness. As an example, patients being discharged to a skilled nursing facility should have a discharge summary completed and in hand before leaving the hospital.

Evidence suggests that timely writing of discharge summaries improves their quality. In a retrospective cohort study published in 2012, discharge summaries created more than 24 hours after discharge were less likely to include important plan-of-care components.¹²

FUTURE NEEDS

Randomized trials

Although initial results have been promising for the strategies outlined above, much of the apparent cost reduction of these interventions may be at least partially related to the study design as opposed to the interventions themselves.

For example, Hong et al¹³ examined 18 of the more promising CMOs that had reported initial cost savings. Of these, only 4 had conducted randomized controlled trials. When broken down further, the initial cost reduction reported by most of these randomized controlled trials was generated primarily by small subgroups.¹⁴

These results, however, do not necessarily reflect an inherent failure in the system. We contend that they merely demonstrate that CMOs and care plan administrators need to be more selective about whom they enroll, either by targeting patients at the extremes of the usage curve or by identifying patient characteristics and usage parameters amenable to cost reduction and quality improvement strategies.

Better social infrastructure

Although patient care plans and CMOs have been effective in managing high users, we believe that the most promising quality improvement and cost-reduction strategy involves redirecting much of the expensive healthcare spending to the social determinants of health (eg, homelessness, mental illness, low socioeconomic status).

Organization for Economic Co-operation and Development. Health at a Glance 2009: OECD Indicators. Paris, France: OECD Publishing; 2009.

Among developed countries, the United States has the highest healthcare spending and the lowest social service spending as a percentage of its gross domestic product (Figure 1).¹⁵ Although seemingly discouraging, these data can actually be interpreted as hopeful, as they support the notion that the inefficiencies of our current system are not part of an inescapable reality, but rather reflect a system that has evolved uniquely in this country.

Using the available social programs

Exemplifying this medical and social services balance is a high user who visited her local emergency department 450 times in 1 year for reasons primarily related to homelessness.¹⁶ Each time, the medical system (as it is currently designed to do) applied a short-term medical solution to this patient’s problems and discharged her home, ie, back to the street.

But this patient’s high use was really a manifestation of a deeper social issue: homelessness. When the medical staff eventually noted how much this lack of stable shelter was contributing to her pattern of use, she was referred to appropriate social resources and provided with the housing she needed. Her hospital visits decreased from 450 to 12 in the subsequent year, amounting to a huge cost reduction and a clear improvement in her quality of life.

Similar encouraging results have resulted when available social programs are applied to the high-use population at large, which is particularly reassuring given this population’s preponderance of low socioeconomic status, mental illness, and homelessness. (The prevalence of homelessness is roughly 20%, depending on the definition of a high user).

New York Medicaid, for example, has a housing program that provides stable shelter outside of acute care medical settings for patients at a rate as low as $50 per day, compared with area hospital costs that often exceed $2,200 daily.¹⁷ A similar program in Westchester County, NY, reported a 45.9% reduction in inpatient costs and a 15.4% reduction in emergency department visits among 61 of its highest users after 2 years of enrollment.¹⁷

Need to reform privacy laws

Although legally daunting, reform of the Health Insurance Portability and Accountability Act (HIPAA) and other privacy laws in favor of a more open model of information sharing, particularly for high-risk patients, holds great opportunity for quality improvement. For patients who obtain their care from several healthcare facilities, the documentation is often inscrutable. If some of the HIPAA regulations and other patient privacy laws were exchanged for rules more akin to the current model of narcotic prescription tracking, for example, physicians would be better equipped to provide safe, organized, and efficient medical care for high-use patients.

Need to reform the system

A fundamental flaw in our healthcare system, which is largely based on a fee-for-service model, is that it was not designed for patients who use the system at the highest frequency and greatest cost. Also, it does not account for the psychosocial factors that beset many high-use patients. As such, it is imperative for the safety of our patients as well as the viability of the healthcare system that we change our historical way of thinking and reform this system that provides high users with care that is high-cost, low-quality, and not patient-centered.

IMPROVING QUALITY, REDUCING COST

High users of emergency services are a medically and socially complex group, predominantly characterized by low socioeconomic status and high rates of mental illness and drug dependency. Despite their increased healthcare use, they do not have better outcomes even though they are not sicker. Improving those outcomes requires both medical and social efforts.

Among the effective medical efforts are strategies aimed at creating individualized patient care plans, using coordinated care teams, and improving discharge summaries. Addressing patients’ social factors, such as homelessness, is more difficult, but healthcare systems can help patients navigate the available social programs. These strategies are part of a comprehensive care plan that can help reduce the cost and improve the quality of healthcare for high users.

Emergency departments are not primary care clinics, but some patients use them that way. This relatively small group of patients consumes a disproportionate share of healthcare at great cost, earning them the label of “high users.” Mostly poor and often burdened with mental illness and addiction, they are not necessarily sicker than other patients, and they do not enjoy better outcomes from the extra money spent on them. (Another subset of high users, those with end-stage chronic disease, is outside the scope of this review.)

Herein lies an opportunity. If—and this is a big if—we could manage their care in a systematic way instead of haphazardly, proactively instead of reactively, with continuity of care instead of episodically, and in a way that is convenient for the patient, we might be able to improve quality and save money.

A DISPROPORTIONATE SHARE OF COSTS

In the United States in 2012, the 5% of the population who were the highest users were responsible for 50% of healthcare costs.¹ The mean cost per person in this group was more than $43,000 annually. The top 1% of users accounted for nearly 23% of all expenditures, averaging nearly $98,000 per patient per year—10 times more than the average yearly cost per patient.

CARE IS OFTEN INAPPROPRIATE AND UNNECESSARY

In addition to being disproportionately expensive, the care that these patients receive is often inappropriate and unnecessary for the severity of their disease.

A 2007–2009 study² of 1,969 patients who had visited the emergency department 10 or more times in a year found they received more than twice as many computed tomography (CT) scans as a control group of infrequent users (< 3 visits/year). This occurred even though they were not as sick as infrequent users, based on significantly lower hospital admission rates (11.1% vs 17.9%; P < .001) and mortality rates (0.7% vs 1.5%; P < .002).²

This inverse relationship between emergency department use and illness severity was even more exaggerated at the upper extreme of the use curve. The highest users (> 29 visits to the emergency department in a year) had the lowest triage acuity and hospital admission rates but the highest number of CT scans. Charges per visit were lower among frequent users, but total charges rose steadily with increasing emergency department use, accounting for significantly more costs per year.²

We believe that one reason these patients receive more medical care than necessary is because their medical records are too large and complex for the average physician to distill effectively in a 20-minute physician-patient encounter. Physicians therefore simply order more tests, procedures, and admissions, which are often medically unnecessary and redundant.

WHAT DRIVES HIGH COST?

Mental illness and chemical dependence

Drug addiction, mental illness, and poverty frequently accompany (and influence) high-use behavior, particularly in patients without end-stage diseases.

Szekendi et al,³ in a study of 28,291 patients who had been admitted at least 5 times in a year in a Chicago health system, found that these high users were 2 to 3 times more likely to suffer from comorbid depression (40% vs 13%), psychosis (18% vs 5%), recreational drug dependence (20% vs 7%), and alcohol abuse (16% vs 7%) than non-high-use hospitalized patients.³

Mercer et al⁴ conducted a study at Duke University Medical Center, Durham, NC, aimed at reducing emergency department visits and hospital admissions among 24 of its highest users. They found that 23 (96%) were either addicted to drugs or mentally ill, and 20 (83%) suffered from chronic pain.⁴

Drug abuse among high users is becoming even more relevant as the opioid epidemic worsens. Given that most patients requiring high levels of care suffer from chronic pain and many of them develop an opioid addiction while treating their pain, physicians have a moral imperative to reduce the prevalence of drug abuse in this population.

Low socioeconomic status

Low socioeconomic status is an important factor among high users, as it is highly associated with greater disease severity, which usually increases cost without any guarantee of an associated increase in quality. Data suggest that patients of low socioeconomic status are twice as likely to require urgent emergency department visits, 4 times as likely to require admission to the hospital, and, importantly, about half as likely to use ambulatory care compared with patients of higher socioeconomic status.⁵ While this pattern of low-quality, high-cost spending in acute care settings reflects spending in the healthcare system at large, the pattern is greatly exaggerated among high users.

Lost to follow-up

Low socioeconomic status also complicates communication and follow-up. In a 2013 study, physician researchers in St. Paul, MN, documented attempts to interview 64 recently discharged high users. They could not reach 47 (73%) of them, for reasons largely attributable to low socioeconomic status, such as disconnected phone lines and changes in address.⁶

Clearly, the usual contact methods for follow-up care after discharge, such as phone calls and mailings, are unlikely to be effective in coordinating the outpatient care of these individuals.

Additionally, we must find ways of making primary care more convenient, gaining our patients’ trust, and finding ways to engage patients in follow-up without relying on traditional means of communication.

Do high users have medical insurance?

Surprisingly, most high users of the emergency department have health insurance. The Chicago health system study³ found that most (72.4%) of their high users had either Medicare or private health insurance, while 27.6% had either Medicaid or no insurance (compared with 21.6% in the general population). Other studies also found that most of the frequent emergency department users are insured,⁷ although the overall percentage who rely on publicly paid insurance is greater than in the population at large.

Many prefer acute care over primary care

Although one might think that high users go to the emergency department because they have nowhere else to go for care, a report published in 2013 by Kangovi et al⁵ suggests another reason—they prefer the emergency department.⁵ They interviewed 40 urban patients of low socioeconomic status who consistently cited the 24-hour, no-appointment-necessary structure of the emergency department as an advantage over primary care. The flexibility of emergency access to healthcare makes sense if one reflects on how difficult it is for even high-functioning individuals to schedule and keep medical appointments.

Specific reasons for preferring the emergency department included the following:

Affordability. Even if their insurance fully paid for visits to their primary care physicians, the primary care physician was likely to refer them to specialists, whose visits required a copay, and which required taking another day off of work. The emergency department is cheaper for the patient and it is a “one-stop shop.” Patients appreciated the emergency department guarantee of seeing a physician regardless of proof of insurance, a policy not guaranteed in primary care and specialist offices.

Accessibility. For those without a car, public transportation and even patient transportation services are inconvenient and unreliable, whereas emergency medical services will take you to the emergency department.

Accommodations. Although medical centers may tout their same-day appointments, often same-day appointments are all that they have—and you have no choice about the time. You have to call first thing in the morning and stay on hold for a long time, and then when you finally get through, all the same-day appointments are gone.

Availability. Patients said they often had a hard time getting timely medical advice from their primary care physicians. When they could get through to their primary care physicians on the phone, they would be told to go to the emergency department.

Acceptability. Men, especially, feel they need to be very sick indeed to seek medical care, so going to the emergency department is more acceptable.

Trust in the provider. For reasons that were not entirely clear, patients felt that acute care providers were more trustworthy, competent, and compassionate than primary care physicians.⁵

None of these reasons for using the emergency department has anything to do with disease severity, which supports the findings that high users of the emergency department were not as sick as their normal-use peers.²

QUALITY IMPROVEMENT AND COST-REDUCTION STRATEGIES

Efforts are being made to reduce the cost of healthcare for high users while improving the quality of their care. Promising strategies focus on coordinating care management, creating individualized patient care plans, and improving the components and instructions of discharge summaries.

Care management organizations

A care management organization (CMO) model has emerged as a strategy for quality improvement and cost reduction in the high-use population. In this model, social workers, health coaches, nurses, mid-level providers, and physicians collaborate on designing individualized care plans to meet the specific needs of patients.

Teams typically work in stepwise fashion, first identifying and engaging patients at high risk of poor outcomes and unnecessary care, often using sophisticated quantitative, risk-prediction tools. Then, they perform health assessments and identify potential interventions aimed at preventing expensive acute-care medical interventions. Third, they work with patients to rapidly identify and effectively respond to changes in their conditions and direct them to the most appropriate medical setting, typically primary or urgent care.

Effective models

In 1998, the Camden (NJ) Coalition of Healthcare Providers established a model for CMO care plans. Starting with the first 36 patients enrolled in the program, hospital admissions and emergency department visits were cut by 47% (from 62 to 37 per month), and collective hospital costs were cut by 56% (from $1.2 million to about $500,000 per month).⁸ It should be noted that this was a small, nonrandomized study and these preliminary numbers did not take into account the cost of outpatient physician visits or new medications. Thus, how much money this program actually saves is not clear.

Similar programs have had similar results. A nurse-led care coordination program in Doylestown, PA, showed an impressive 25% reduction in annual mortality and a 36% reduction in overall costs during a 10-year period.⁹

A program in Atlantic City, NJ, combined the typical CMO model with a primary care clinic to provide high users with unlimited access, while paying its providers in a capitation model (as opposed to fee for service). It achieved a 40% reduction in yearly emergency department visits and hospital admissions.⁸

Patient care plans

Individualized patient care plans for high users are among the most promising tools for reducing costs and improving quality in this group. They are low-cost and relatively easy to implement. The goal of these care plans is to provide practitioners with a concise care summary to help them make rational and consistent medical decisions.

Typically, a care plan is written by an interdisciplinary committee composed of physicians, nurses, and social workers. It is based on the patient’s pertinent medical and psychiatric history, which may include recent imaging results or other relevant diagnostic tests. It provides suggestions for managing complex chronic issues, such as drug abuse, that lead to high use of healthcare resources.

These care plans provide a rational and prespecified approach to workup and management, typically including a narcotic prescription protocol, regardless of the setting or the number of providers who see the patient. Practitioners guided by effective care plans are much more likely to effectively navigate a complex patient encounter as opposed to looking through extensive medical notes and hoping to find relevant information.

Effective models

Data show these plans can be effective. For example, Regions Hospital in St. Paul, MN, implemented patient care plans in 2010. During the first 4 months, hospital admissions in the first 94 patients were reduced by 67%.¹⁰

A study of high users at Duke University Medical Center reported similar results. One year after starting care plans, inpatient admissions had decreased by 50.5%, readmissions had decreased by 51.5%, and variable direct costs per admission were reduced by 35.8%. Paradoxically, emergency department visits went up, but this anomaly was driven by 134 visits incurred by a single dialysis patient. After removing this patient from the data, emergency department visits were relatively stable.⁴

Better discharge summaries

Although improving discharge summaries is not a novel concept, changing the summary from a historical document to a proactive discharge plan has the potential to prevent readmissions and promote a durable de-escalation in care acuity.

For example, when moving a patient to a subacute care facility, providing a concise summary of which treatments worked and which did not, a list of comorbidities, and a list of medications and strategies to consider, can help the next providers to better target their plan of care. Studies have shown that nearly half of discharge statements lack important information on treatments and tests.¹¹

Improvement can be as simple as encouraging practitioners to construct their summaries in an “if-then” format. Instead of noting for instance that “Mr. Smith was treated for pneumonia with antibiotics and discharged to a rehab facility,” the following would be more useful: “Family would like to see if Mr. Smith can get back to his functional baseline after his acute pneumonia. If he clinically does not do well over the next 1 to 2 weeks and has a poor quality of life, then family would like to pursue hospice.”

In addition to shifting the philosophy, we believe that providing timely discharge summaries is a fundamental, high-yield aspect of ensuring their effectiveness. As an example, patients being discharged to a skilled nursing facility should have a discharge summary completed and in hand before leaving the hospital.

Evidence suggests that timely writing of discharge summaries improves their quality. In a retrospective cohort study published in 2012, discharge summaries created more than 24 hours after discharge were less likely to include important plan-of-care components.¹²

FUTURE NEEDS

Randomized trials

Although initial results have been promising for the strategies outlined above, much of the apparent cost reduction of these interventions may be at least partially related to the study design as opposed to the interventions themselves.

For example, Hong et al¹³ examined 18 of the more promising CMOs that had reported initial cost savings. Of these, only 4 had conducted randomized controlled trials. When broken down further, the initial cost reduction reported by most of these randomized controlled trials was generated primarily by small subgroups.¹⁴

These results, however, do not necessarily reflect an inherent failure in the system. We contend that they merely demonstrate that CMOs and care plan administrators need to be more selective about whom they enroll, either by targeting patients at the extremes of the usage curve or by identifying patient characteristics and usage parameters amenable to cost reduction and quality improvement strategies.

Better social infrastructure

Although patient care plans and CMOs have been effective in managing high users, we believe that the most promising quality improvement and cost-reduction strategy involves redirecting much of the expensive healthcare spending to the social determinants of health (eg, homelessness, mental illness, low socioeconomic status).

Organization for Economic Co-operation and Development. Health at a Glance 2009: OECD Indicators. Paris, France: OECD Publishing; 2009.

Among developed countries, the United States has the highest healthcare spending and the lowest social service spending as a percentage of its gross domestic product (Figure 1).¹⁵ Although seemingly discouraging, these data can actually be interpreted as hopeful, as they support the notion that the inefficiencies of our current system are not part of an inescapable reality, but rather reflect a system that has evolved uniquely in this country.

Using the available social programs

Exemplifying this medical and social services balance is a high user who visited her local emergency department 450 times in 1 year for reasons primarily related to homelessness.¹⁶ Each time, the medical system (as it is currently designed to do) applied a short-term medical solution to this patient’s problems and discharged her home, ie, back to the street.

But this patient’s high use was really a manifestation of a deeper social issue: homelessness. When the medical staff eventually noted how much this lack of stable shelter was contributing to her pattern of use, she was referred to appropriate social resources and provided with the housing she needed. Her hospital visits decreased from 450 to 12 in the subsequent year, amounting to a huge cost reduction and a clear improvement in her quality of life.

Similar encouraging results have resulted when available social programs are applied to the high-use population at large, which is particularly reassuring given this population’s preponderance of low socioeconomic status, mental illness, and homelessness. (The prevalence of homelessness is roughly 20%, depending on the definition of a high user).

New York Medicaid, for example, has a housing program that provides stable shelter outside of acute care medical settings for patients at a rate as low as $50 per day, compared with area hospital costs that often exceed $2,200 daily.¹⁷ A similar program in Westchester County, NY, reported a 45.9% reduction in inpatient costs and a 15.4% reduction in emergency department visits among 61 of its highest users after 2 years of enrollment.¹⁷

Need to reform privacy laws

Although legally daunting, reform of the Health Insurance Portability and Accountability Act (HIPAA) and other privacy laws in favor of a more open model of information sharing, particularly for high-risk patients, holds great opportunity for quality improvement. For patients who obtain their care from several healthcare facilities, the documentation is often inscrutable. If some of the HIPAA regulations and other patient privacy laws were exchanged for rules more akin to the current model of narcotic prescription tracking, for example, physicians would be better equipped to provide safe, organized, and efficient medical care for high-use patients.

Need to reform the system

A fundamental flaw in our healthcare system, which is largely based on a fee-for-service model, is that it was not designed for patients who use the system at the highest frequency and greatest cost. Also, it does not account for the psychosocial factors that beset many high-use patients. As such, it is imperative for the safety of our patients as well as the viability of the healthcare system that we change our historical way of thinking and reform this system that provides high users with care that is high-cost, low-quality, and not patient-centered.

IMPROVING QUALITY, REDUCING COST

High users of emergency services are a medically and socially complex group, predominantly characterized by low socioeconomic status and high rates of mental illness and drug dependency. Despite their increased healthcare use, they do not have better outcomes even though they are not sicker. Improving those outcomes requires both medical and social efforts.

Among the effective medical efforts are strategies aimed at creating individualized patient care plans, using coordinated care teams, and improving discharge summaries. Addressing patients’ social factors, such as homelessness, is more difficult, but healthcare systems can help patients navigate the available social programs. These strategies are part of a comprehensive care plan that can help reduce the cost and improve the quality of healthcare for high users.

Emergency departments are not primary care clinics, but some patients use them that way. This relatively small group of patients consumes a disproportionate share of healthcare at great cost, earning them the label of “high users.” Mostly poor and often burdened with mental illness and addiction, they are not necessarily sicker than other patients, and they do not enjoy better outcomes from the extra money spent on them. (Another subset of high users, those with end-stage chronic disease, is outside the scope of this review.)

Herein lies an opportunity. If—and this is a big if—we could manage their care in a systematic way instead of haphazardly, proactively instead of reactively, with continuity of care instead of episodically, and in a way that is convenient for the patient, we might be able to improve quality and save money.

A DISPROPORTIONATE SHARE OF COSTS

In the United States in 2012, the 5% of the population who were the highest users were responsible for 50% of healthcare costs.¹ The mean cost per person in this group was more than $43,000 annually. The top 1% of users accounted for nearly 23% of all expenditures, averaging nearly $98,000 per patient per year—10 times more than the average yearly cost per patient.

CARE IS OFTEN INAPPROPRIATE AND UNNECESSARY

In addition to being disproportionately expensive, the care that these patients receive is often inappropriate and unnecessary for the severity of their disease.

A 2007–2009 study² of 1,969 patients who had visited the emergency department 10 or more times in a year found they received more than twice as many computed tomography (CT) scans as a control group of infrequent users (< 3 visits/year). This occurred even though they were not as sick as infrequent users, based on significantly lower hospital admission rates (11.1% vs 17.9%; P < .001) and mortality rates (0.7% vs 1.5%; P < .002).²

This inverse relationship between emergency department use and illness severity was even more exaggerated at the upper extreme of the use curve. The highest users (> 29 visits to the emergency department in a year) had the lowest triage acuity and hospital admission rates but the highest number of CT scans. Charges per visit were lower among frequent users, but total charges rose steadily with increasing emergency department use, accounting for significantly more costs per year.²

We believe that one reason these patients receive more medical care than necessary is because their medical records are too large and complex for the average physician to distill effectively in a 20-minute physician-patient encounter. Physicians therefore simply order more tests, procedures, and admissions, which are often medically unnecessary and redundant.

WHAT DRIVES HIGH COST?

Mental illness and chemical dependence

Drug addiction, mental illness, and poverty frequently accompany (and influence) high-use behavior, particularly in patients without end-stage diseases.

Szekendi et al,³ in a study of 28,291 patients who had been admitted at least 5 times in a year in a Chicago health system, found that these high users were 2 to 3 times more likely to suffer from comorbid depression (40% vs 13%), psychosis (18% vs 5%), recreational drug dependence (20% vs 7%), and alcohol abuse (16% vs 7%) than non-high-use hospitalized patients.³

Mercer et al⁴ conducted a study at Duke University Medical Center, Durham, NC, aimed at reducing emergency department visits and hospital admissions among 24 of its highest users. They found that 23 (96%) were either addicted to drugs or mentally ill, and 20 (83%) suffered from chronic pain.⁴

Drug abuse among high users is becoming even more relevant as the opioid epidemic worsens. Given that most patients requiring high levels of care suffer from chronic pain and many of them develop an opioid addiction while treating their pain, physicians have a moral imperative to reduce the prevalence of drug abuse in this population.

Low socioeconomic status

Low socioeconomic status is an important factor among high users, as it is highly associated with greater disease severity, which usually increases cost without any guarantee of an associated increase in quality. Data suggest that patients of low socioeconomic status are twice as likely to require urgent emergency department visits, 4 times as likely to require admission to the hospital, and, importantly, about half as likely to use ambulatory care compared with patients of higher socioeconomic status.⁵ While this pattern of low-quality, high-cost spending in acute care settings reflects spending in the healthcare system at large, the pattern is greatly exaggerated among high users.

Lost to follow-up

Low socioeconomic status also complicates communication and follow-up. In a 2013 study, physician researchers in St. Paul, MN, documented attempts to interview 64 recently discharged high users. They could not reach 47 (73%) of them, for reasons largely attributable to low socioeconomic status, such as disconnected phone lines and changes in address.⁶

Clearly, the usual contact methods for follow-up care after discharge, such as phone calls and mailings, are unlikely to be effective in coordinating the outpatient care of these individuals.

Additionally, we must find ways of making primary care more convenient, gaining our patients’ trust, and finding ways to engage patients in follow-up without relying on traditional means of communication.

Do high users have medical insurance?

Surprisingly, most high users of the emergency department have health insurance. The Chicago health system study³ found that most (72.4%) of their high users had either Medicare or private health insurance, while 27.6% had either Medicaid or no insurance (compared with 21.6% in the general population). Other studies also found that most of the frequent emergency department users are insured,⁷ although the overall percentage who rely on publicly paid insurance is greater than in the population at large.

Many prefer acute care over primary care

Although one might think that high users go to the emergency department because they have nowhere else to go for care, a report published in 2013 by Kangovi et al⁵ suggests another reason—they prefer the emergency department.⁵ They interviewed 40 urban patients of low socioeconomic status who consistently cited the 24-hour, no-appointment-necessary structure of the emergency department as an advantage over primary care. The flexibility of emergency access to healthcare makes sense if one reflects on how difficult it is for even high-functioning individuals to schedule and keep medical appointments.

Specific reasons for preferring the emergency department included the following:

Affordability. Even if their insurance fully paid for visits to their primary care physicians, the primary care physician was likely to refer them to specialists, whose visits required a copay, and which required taking another day off of work. The emergency department is cheaper for the patient and it is a “one-stop shop.” Patients appreciated the emergency department guarantee of seeing a physician regardless of proof of insurance, a policy not guaranteed in primary care and specialist offices.

Accessibility. For those without a car, public transportation and even patient transportation services are inconvenient and unreliable, whereas emergency medical services will take you to the emergency department.

Accommodations. Although medical centers may tout their same-day appointments, often same-day appointments are all that they have—and you have no choice about the time. You have to call first thing in the morning and stay on hold for a long time, and then when you finally get through, all the same-day appointments are gone.

Availability. Patients said they often had a hard time getting timely medical advice from their primary care physicians. When they could get through to their primary care physicians on the phone, they would be told to go to the emergency department.

Acceptability. Men, especially, feel they need to be very sick indeed to seek medical care, so going to the emergency department is more acceptable.

Trust in the provider. For reasons that were not entirely clear, patients felt that acute care providers were more trustworthy, competent, and compassionate than primary care physicians.⁵

None of these reasons for using the emergency department has anything to do with disease severity, which supports the findings that high users of the emergency department were not as sick as their normal-use peers.²

QUALITY IMPROVEMENT AND COST-REDUCTION STRATEGIES

Efforts are being made to reduce the cost of healthcare for high users while improving the quality of their care. Promising strategies focus on coordinating care management, creating individualized patient care plans, and improving the components and instructions of discharge summaries.

Care management organizations

A care management organization (CMO) model has emerged as a strategy for quality improvement and cost reduction in the high-use population. In this model, social workers, health coaches, nurses, mid-level providers, and physicians collaborate on designing individualized care plans to meet the specific needs of patients.

Teams typically work in stepwise fashion, first identifying and engaging patients at high risk of poor outcomes and unnecessary care, often using sophisticated quantitative, risk-prediction tools. Then, they perform health assessments and identify potential interventions aimed at preventing expensive acute-care medical interventions. Third, they work with patients to rapidly identify and effectively respond to changes in their conditions and direct them to the most appropriate medical setting, typically primary or urgent care.

Effective models

In 1998, the Camden (NJ) Coalition of Healthcare Providers established a model for CMO care plans. Starting with the first 36 patients enrolled in the program, hospital admissions and emergency department visits were cut by 47% (from 62 to 37 per month), and collective hospital costs were cut by 56% (from $1.2 million to about $500,000 per month).⁸ It should be noted that this was a small, nonrandomized study and these preliminary numbers did not take into account the cost of outpatient physician visits or new medications. Thus, how much money this program actually saves is not clear.

Similar programs have had similar results. A nurse-led care coordination program in Doylestown, PA, showed an impressive 25% reduction in annual mortality and a 36% reduction in overall costs during a 10-year period.⁹

A program in Atlantic City, NJ, combined the typical CMO model with a primary care clinic to provide high users with unlimited access, while paying its providers in a capitation model (as opposed to fee for service). It achieved a 40% reduction in yearly emergency department visits and hospital admissions.⁸

Patient care plans

Individualized patient care plans for high users are among the most promising tools for reducing costs and improving quality in this group. They are low-cost and relatively easy to implement. The goal of these care plans is to provide practitioners with a concise care summary to help them make rational and consistent medical decisions.

Typically, a care plan is written by an interdisciplinary committee composed of physicians, nurses, and social workers. It is based on the patient’s pertinent medical and psychiatric history, which may include recent imaging results or other relevant diagnostic tests. It provides suggestions for managing complex chronic issues, such as drug abuse, that lead to high use of healthcare resources.

These care plans provide a rational and prespecified approach to workup and management, typically including a narcotic prescription protocol, regardless of the setting or the number of providers who see the patient. Practitioners guided by effective care plans are much more likely to effectively navigate a complex patient encounter as opposed to looking through extensive medical notes and hoping to find relevant information.

Effective models

Data show these plans can be effective. For example, Regions Hospital in St. Paul, MN, implemented patient care plans in 2010. During the first 4 months, hospital admissions in the first 94 patients were reduced by 67%.¹⁰

A study of high users at Duke University Medical Center reported similar results. One year after starting care plans, inpatient admissions had decreased by 50.5%, readmissions had decreased by 51.5%, and variable direct costs per admission were reduced by 35.8%. Paradoxically, emergency department visits went up, but this anomaly was driven by 134 visits incurred by a single dialysis patient. After removing this patient from the data, emergency department visits were relatively stable.⁴

Better discharge summaries

Although improving discharge summaries is not a novel concept, changing the summary from a historical document to a proactive discharge plan has the potential to prevent readmissions and promote a durable de-escalation in care acuity.

For example, when moving a patient to a subacute care facility, providing a concise summary of which treatments worked and which did not, a list of comorbidities, and a list of medications and strategies to consider, can help the next providers to better target their plan of care. Studies have shown that nearly half of discharge statements lack important information on treatments and tests.¹¹

Improvement can be as simple as encouraging practitioners to construct their summaries in an “if-then” format. Instead of noting for instance that “Mr. Smith was treated for pneumonia with antibiotics and discharged to a rehab facility,” the following would be more useful: “Family would like to see if Mr. Smith can get back to his functional baseline after his acute pneumonia. If he clinically does not do well over the next 1 to 2 weeks and has a poor quality of life, then family would like to pursue hospice.”

In addition to shifting the philosophy, we believe that providing timely discharge summaries is a fundamental, high-yield aspect of ensuring their effectiveness. As an example, patients being discharged to a skilled nursing facility should have a discharge summary completed and in hand before leaving the hospital.

Evidence suggests that timely writing of discharge summaries improves their quality. In a retrospective cohort study published in 2012, discharge summaries created more than 24 hours after discharge were less likely to include important plan-of-care components.¹²

FUTURE NEEDS

Randomized trials

Although initial results have been promising for the strategies outlined above, much of the apparent cost reduction of these interventions may be at least partially related to the study design as opposed to the interventions themselves.

For example, Hong et al¹³ examined 18 of the more promising CMOs that had reported initial cost savings. Of these, only 4 had conducted randomized controlled trials. When broken down further, the initial cost reduction reported by most of these randomized controlled trials was generated primarily by small subgroups.¹⁴

These results, however, do not necessarily reflect an inherent failure in the system. We contend that they merely demonstrate that CMOs and care plan administrators need to be more selective about whom they enroll, either by targeting patients at the extremes of the usage curve or by identifying patient characteristics and usage parameters amenable to cost reduction and quality improvement strategies.

Better social infrastructure

Although patient care plans and CMOs have been effective in managing high users, we believe that the most promising quality improvement and cost-reduction strategy involves redirecting much of the expensive healthcare spending to the social determinants of health (eg, homelessness, mental illness, low socioeconomic status).

Organization for Economic Co-operation and Development. Health at a Glance 2009: OECD Indicators. Paris, France: OECD Publishing; 2009.

Among developed countries, the United States has the highest healthcare spending and the lowest social service spending as a percentage of its gross domestic product (Figure 1).¹⁵ Although seemingly discouraging, these data can actually be interpreted as hopeful, as they support the notion that the inefficiencies of our current system are not part of an inescapable reality, but rather reflect a system that has evolved uniquely in this country.

Using the available social programs

Exemplifying this medical and social services balance is a high user who visited her local emergency department 450 times in 1 year for reasons primarily related to homelessness.¹⁶ Each time, the medical system (as it is currently designed to do) applied a short-term medical solution to this patient’s problems and discharged her home, ie, back to the street.

But this patient’s high use was really a manifestation of a deeper social issue: homelessness. When the medical staff eventually noted how much this lack of stable shelter was contributing to her pattern of use, she was referred to appropriate social resources and provided with the housing she needed. Her hospital visits decreased from 450 to 12 in the subsequent year, amounting to a huge cost reduction and a clear improvement in her quality of life.

Similar encouraging results have resulted when available social programs are applied to the high-use population at large, which is particularly reassuring given this population’s preponderance of low socioeconomic status, mental illness, and homelessness. (The prevalence of homelessness is roughly 20%, depending on the definition of a high user).

New York Medicaid, for example, has a housing program that provides stable shelter outside of acute care medical settings for patients at a rate as low as $50 per day, compared with area hospital costs that often exceed $2,200 daily.¹⁷ A similar program in Westchester County, NY, reported a 45.9% reduction in inpatient costs and a 15.4% reduction in emergency department visits among 61 of its highest users after 2 years of enrollment.¹⁷

Need to reform privacy laws

Although legally daunting, reform of the Health Insurance Portability and Accountability Act (HIPAA) and other privacy laws in favor of a more open model of information sharing, particularly for high-risk patients, holds great opportunity for quality improvement. For patients who obtain their care from several healthcare facilities, the documentation is often inscrutable. If some of the HIPAA regulations and other patient privacy laws were exchanged for rules more akin to the current model of narcotic prescription tracking, for example, physicians would be better equipped to provide safe, organized, and efficient medical care for high-use patients.

Need to reform the system

A fundamental flaw in our healthcare system, which is largely based on a fee-for-service model, is that it was not designed for patients who use the system at the highest frequency and greatest cost. Also, it does not account for the psychosocial factors that beset many high-use patients. As such, it is imperative for the safety of our patients as well as the viability of the healthcare system that we change our historical way of thinking and reform this system that provides high users with care that is high-cost, low-quality, and not patient-centered.

IMPROVING QUALITY, REDUCING COST

High users of emergency services are a medically and socially complex group, predominantly characterized by low socioeconomic status and high rates of mental illness and drug dependency. Despite their increased healthcare use, they do not have better outcomes even though they are not sicker. Improving those outcomes requires both medical and social efforts.

Among the effective medical efforts are strategies aimed at creating individualized patient care plans, using coordinated care teams, and improving discharge summaries. Addressing patients’ social factors, such as homelessness, is more difficult, but healthcare systems can help patients navigate the available social programs. These strategies are part of a comprehensive care plan that can help reduce the cost and improve the quality of healthcare for high users.

ANSWER

The radiograph shows an oblique fracture through the radial styloid process. The patient was placed in a splint and referred to outpatient orthopedics for follow-up.

ANSWER

The radiograph shows an oblique fracture through the radial styloid process. The patient was placed in a splint and referred to outpatient orthopedics for follow-up.

ANSWER

The radiograph shows an oblique fracture through the radial styloid process. The patient was placed in a splint and referred to outpatient orthopedics for follow-up.