Adequate sleep is important for health, yet the hospital environment commonly disrupts sleep.13 Sleep improves after several days in the hospital.3, 4 Sleep deprivation increases cortisol levels5 and sleep loss of greater than 4 hours may be hyperalgesic.6 Even a few days' suppression of slow‐wave sleep worsens glucose tolerance.7 Sleep disruption may cause irritability and aggressiveness,8 impaired memory consolidation, and delirium.2

Noise may disrupt sleep. The World Health Organization recommends a maximum of 30 to 40 dBA in patients' rooms at night.9, 10 Normal conversation occurs at 60 dBA. Medical equipment alarms are about 80 dBA.

Sedative use is common in the hospital.3 Sedatives typically shorten sleep latency and suppress rapid eye movement (REM) sleep. However, some sedatives cause delirium, falls, amnesia, and confusion, particularly in the elderly.1113

Most research on sleep in hospitalized patients has been done in the critical care setting, often in sedated ventilated patients, where sleep disruption is well‐described.1416 Only a few small studies have assessed the sleep of hospitalized patients outside critical care.17, 18

A single blinded interventional trial assessed sedative use, but was a nonrandomized study.19, 20 As‐needed sedative use was measured among hospitalized elderly patients as a secondary endpoint. The intervention, known as the Hospital Elder Life Program (HELP), included a protocol with noise reduction, massage, music, and warm drinks, as well as rescheduling of medications and procedures; it resulted in a 24% reduction in as‐needed sedative use. Another trial decreased noise and reduced overnight X‐rays on a surgical unit, then measured staff and patient attitudes.21 Two interventional studies in nursing homes reduced noise and light, and/or increased daytime activity and found no effect on most objective measures of sleep.22, 23 One descriptive study found most sleep disturbances in medical‐surgical patients came from noise and sleeping in an unfamiliar bed.4

We hypothesized that an intervention designed to improve patient sleep through changes in staff behavior would decrease sedative use among unselected patients in a medical‐surgical unit. We measured sedative use as our primary endpoint as a marker for effective sleep, and because decreased sedative use is desirable. We also hypothesized that the intervention would lead to improved sleep experiences, as measured by a questionnaire and Verran Snyder‐Halpern (VSH) sleep scores as secondary endpoints.24

Materials And Methods

Study Protocol

A single investigator surveyed patients in the morning about the prior night's sleep experience. The surveys consisted of the VSH sleep scale, as well as an 8‐item questionnaire developed from informal pilot interviews with about 18 patients conducted by 1 of the investigators (M.B.) (Supporting Information Figure 1). The VSH scale is a visual analog scale using a 100‐cm line,24 which we modified with a 100‐mm line to make it easier to collect data. The questionnaire and VSH scores of patients with cognitive impairment were not included in the final analysis. Cognitive impairment was determined by diagnoses present in chart review. Surveys and consent forms were available in 4 languages and trained interpreters were used as needed. Nurses, providers, and patients were blinded to the measurement of as‐needed sedative use, and staff were unaware of which patients were study subjects.

Figure 1

Results

During the preintervention phase, 334 patients were screened, 294 were eligible, and 54.7% of eligible subjects were enrolled (n = 161). During the intervention phase, 211 patients were screened, 188 were eligible, and 56.3% of eligible patients were enrolled (n = 106). The mean patient age was 60.6 years. The preintervention and intervention groups did not differ significantly in enrollment rate, age, gender, cognitive impairment, surgical status, or hearing deficiencies (Table 1). Over 93% of patients were nonsurgical.

Sedative Use

Preintervention, 31.7% of patients received nighttime as‐needed sedatives, versus 16.0% of the intervention group, a 49.4% reduction (P = 0.0041; 95% confidence interval [CI]: 0.056‐0.26) (Figure 2). In patients aged 65 years or older, 38.2% received nighttime as‐needed sedatives preintervention, and 14.6% did postintervention, a 61.2% reduction (P = 0.0054; 95% CI: 0.084‐0.39).

Figure 2

Questionnaire Results

Preintervention, hospital staff was by far the biggest factor keeping patients awake, with 42.4% of patients reporting it (Figure 3). This dropped to only 25.7% with the intervention, a 39.3% decrease (P = 0.009; 95% CI: 0.0452‐0.2765). Preintervention, 19.2% of patients selected voices as the noise most likely to bother them at night, and this dropped to 9.9% with the intervention, a 48% decrease (P = 0.045; 95% CI: 0.0074‐0.1787). No other significant differences were found.

Figure 3

Discussion

Our trial found that hospital staff was the factor most responsible for patient sleep disruption, and that behavioral interventions on hospital staff can reduce use of as‐needed sedatives. The only previously reported intervention to reduce sedative use, the HELP strategy, involved a complex intervention requiring extra staff, with adherence ranging from 10% to 75%.19, 20, 25 In contrast, our protocol can be easily replicated at minimal cost.

Our results are consistent with those of Freedman et al.,26 who found that noise was not the primary factor responsible for sleep disruption in ICU patients, and that staff activities were at least as important a factor. The study is also consistent with the nursing home studies in which decreases in noise and light did not improve sleep.22, 23 It refutes the study that showed that most sleep disturbance in medical‐surgical patients comes from noise and sleeping in an unfamiliar bed.4 Our results call into question the use of the VSH scale in hospitalized patients, which was designed for use in healthy subjects.

Limitations of this study were as follows: moderate size, lack of refined measures of disease severity, and, as in previous studies,19, 2123 the lack of randomized concurrent controls. Evaluation of secondary endpoints was limited by lack of validation of the questionnaire with objective observations, and inability to use the modified VSH scale. Self‐reports of sleep may correlate imperfectly with objective measures, such as polysomnography.27

A larger concurrent trial randomizing similar units at multiple hospitals would be ideal. Future research is needed to determine whether improving sleep in the hospital improves other outcomes, such as recovery times, delirium, falls, or cost.

The need to reduce as‐needed sedatives is an important safety issue and similar interventions in other hospitals may be helpful. Simple changes in staff routines and provider prescribing habits can yield significant reductions in sedative use.

Adequate sleep is important for health, yet the hospital environment commonly disrupts sleep.13 Sleep improves after several days in the hospital.3, 4 Sleep deprivation increases cortisol levels5 and sleep loss of greater than 4 hours may be hyperalgesic.6 Even a few days' suppression of slow‐wave sleep worsens glucose tolerance.7 Sleep disruption may cause irritability and aggressiveness,8 impaired memory consolidation, and delirium.2

Noise may disrupt sleep. The World Health Organization recommends a maximum of 30 to 40 dBA in patients' rooms at night.9, 10 Normal conversation occurs at 60 dBA. Medical equipment alarms are about 80 dBA.

Sedative use is common in the hospital.3 Sedatives typically shorten sleep latency and suppress rapid eye movement (REM) sleep. However, some sedatives cause delirium, falls, amnesia, and confusion, particularly in the elderly.1113

Most research on sleep in hospitalized patients has been done in the critical care setting, often in sedated ventilated patients, where sleep disruption is well‐described.1416 Only a few small studies have assessed the sleep of hospitalized patients outside critical care.17, 18

A single blinded interventional trial assessed sedative use, but was a nonrandomized study.19, 20 As‐needed sedative use was measured among hospitalized elderly patients as a secondary endpoint. The intervention, known as the Hospital Elder Life Program (HELP), included a protocol with noise reduction, massage, music, and warm drinks, as well as rescheduling of medications and procedures; it resulted in a 24% reduction in as‐needed sedative use. Another trial decreased noise and reduced overnight X‐rays on a surgical unit, then measured staff and patient attitudes.21 Two interventional studies in nursing homes reduced noise and light, and/or increased daytime activity and found no effect on most objective measures of sleep.22, 23 One descriptive study found most sleep disturbances in medical‐surgical patients came from noise and sleeping in an unfamiliar bed.4

We hypothesized that an intervention designed to improve patient sleep through changes in staff behavior would decrease sedative use among unselected patients in a medical‐surgical unit. We measured sedative use as our primary endpoint as a marker for effective sleep, and because decreased sedative use is desirable. We also hypothesized that the intervention would lead to improved sleep experiences, as measured by a questionnaire and Verran Snyder‐Halpern (VSH) sleep scores as secondary endpoints.24

Materials And Methods

Study Protocol

A single investigator surveyed patients in the morning about the prior night's sleep experience. The surveys consisted of the VSH sleep scale, as well as an 8‐item questionnaire developed from informal pilot interviews with about 18 patients conducted by 1 of the investigators (M.B.) (Supporting Information Figure 1). The VSH scale is a visual analog scale using a 100‐cm line,24 which we modified with a 100‐mm line to make it easier to collect data. The questionnaire and VSH scores of patients with cognitive impairment were not included in the final analysis. Cognitive impairment was determined by diagnoses present in chart review. Surveys and consent forms were available in 4 languages and trained interpreters were used as needed. Nurses, providers, and patients were blinded to the measurement of as‐needed sedative use, and staff were unaware of which patients were study subjects.

Figure 1

Results

During the preintervention phase, 334 patients were screened, 294 were eligible, and 54.7% of eligible subjects were enrolled (n = 161). During the intervention phase, 211 patients were screened, 188 were eligible, and 56.3% of eligible patients were enrolled (n = 106). The mean patient age was 60.6 years. The preintervention and intervention groups did not differ significantly in enrollment rate, age, gender, cognitive impairment, surgical status, or hearing deficiencies (Table 1). Over 93% of patients were nonsurgical.

Sedative Use

Preintervention, 31.7% of patients received nighttime as‐needed sedatives, versus 16.0% of the intervention group, a 49.4% reduction (P = 0.0041; 95% confidence interval [CI]: 0.056‐0.26) (Figure 2). In patients aged 65 years or older, 38.2% received nighttime as‐needed sedatives preintervention, and 14.6% did postintervention, a 61.2% reduction (P = 0.0054; 95% CI: 0.084‐0.39).

Figure 2

Questionnaire Results

Preintervention, hospital staff was by far the biggest factor keeping patients awake, with 42.4% of patients reporting it (Figure 3). This dropped to only 25.7% with the intervention, a 39.3% decrease (P = 0.009; 95% CI: 0.0452‐0.2765). Preintervention, 19.2% of patients selected voices as the noise most likely to bother them at night, and this dropped to 9.9% with the intervention, a 48% decrease (P = 0.045; 95% CI: 0.0074‐0.1787). No other significant differences were found.

Figure 3

Discussion

Our trial found that hospital staff was the factor most responsible for patient sleep disruption, and that behavioral interventions on hospital staff can reduce use of as‐needed sedatives. The only previously reported intervention to reduce sedative use, the HELP strategy, involved a complex intervention requiring extra staff, with adherence ranging from 10% to 75%.19, 20, 25 In contrast, our protocol can be easily replicated at minimal cost.

Our results are consistent with those of Freedman et al.,26 who found that noise was not the primary factor responsible for sleep disruption in ICU patients, and that staff activities were at least as important a factor. The study is also consistent with the nursing home studies in which decreases in noise and light did not improve sleep.22, 23 It refutes the study that showed that most sleep disturbance in medical‐surgical patients comes from noise and sleeping in an unfamiliar bed.4 Our results call into question the use of the VSH scale in hospitalized patients, which was designed for use in healthy subjects.

Limitations of this study were as follows: moderate size, lack of refined measures of disease severity, and, as in previous studies,19, 2123 the lack of randomized concurrent controls. Evaluation of secondary endpoints was limited by lack of validation of the questionnaire with objective observations, and inability to use the modified VSH scale. Self‐reports of sleep may correlate imperfectly with objective measures, such as polysomnography.27

A larger concurrent trial randomizing similar units at multiple hospitals would be ideal. Future research is needed to determine whether improving sleep in the hospital improves other outcomes, such as recovery times, delirium, falls, or cost.

The need to reduce as‐needed sedatives is an important safety issue and similar interventions in other hospitals may be helpful. Simple changes in staff routines and provider prescribing habits can yield significant reductions in sedative use.

Adequate sleep is important for health, yet the hospital environment commonly disrupts sleep.13 Sleep improves after several days in the hospital.3, 4 Sleep deprivation increases cortisol levels5 and sleep loss of greater than 4 hours may be hyperalgesic.6 Even a few days' suppression of slow‐wave sleep worsens glucose tolerance.7 Sleep disruption may cause irritability and aggressiveness,8 impaired memory consolidation, and delirium.2

Noise may disrupt sleep. The World Health Organization recommends a maximum of 30 to 40 dBA in patients' rooms at night.9, 10 Normal conversation occurs at 60 dBA. Medical equipment alarms are about 80 dBA.

Sedative use is common in the hospital.3 Sedatives typically shorten sleep latency and suppress rapid eye movement (REM) sleep. However, some sedatives cause delirium, falls, amnesia, and confusion, particularly in the elderly.1113

Most research on sleep in hospitalized patients has been done in the critical care setting, often in sedated ventilated patients, where sleep disruption is well‐described.1416 Only a few small studies have assessed the sleep of hospitalized patients outside critical care.17, 18

A single blinded interventional trial assessed sedative use, but was a nonrandomized study.19, 20 As‐needed sedative use was measured among hospitalized elderly patients as a secondary endpoint. The intervention, known as the Hospital Elder Life Program (HELP), included a protocol with noise reduction, massage, music, and warm drinks, as well as rescheduling of medications and procedures; it resulted in a 24% reduction in as‐needed sedative use. Another trial decreased noise and reduced overnight X‐rays on a surgical unit, then measured staff and patient attitudes.21 Two interventional studies in nursing homes reduced noise and light, and/or increased daytime activity and found no effect on most objective measures of sleep.22, 23 One descriptive study found most sleep disturbances in medical‐surgical patients came from noise and sleeping in an unfamiliar bed.4

We hypothesized that an intervention designed to improve patient sleep through changes in staff behavior would decrease sedative use among unselected patients in a medical‐surgical unit. We measured sedative use as our primary endpoint as a marker for effective sleep, and because decreased sedative use is desirable. We also hypothesized that the intervention would lead to improved sleep experiences, as measured by a questionnaire and Verran Snyder‐Halpern (VSH) sleep scores as secondary endpoints.24

Materials And Methods

Study Protocol

A single investigator surveyed patients in the morning about the prior night's sleep experience. The surveys consisted of the VSH sleep scale, as well as an 8‐item questionnaire developed from informal pilot interviews with about 18 patients conducted by 1 of the investigators (M.B.) (Supporting Information Figure 1). The VSH scale is a visual analog scale using a 100‐cm line,24 which we modified with a 100‐mm line to make it easier to collect data. The questionnaire and VSH scores of patients with cognitive impairment were not included in the final analysis. Cognitive impairment was determined by diagnoses present in chart review. Surveys and consent forms were available in 4 languages and trained interpreters were used as needed. Nurses, providers, and patients were blinded to the measurement of as‐needed sedative use, and staff were unaware of which patients were study subjects.

Figure 1

Results

During the preintervention phase, 334 patients were screened, 294 were eligible, and 54.7% of eligible subjects were enrolled (n = 161). During the intervention phase, 211 patients were screened, 188 were eligible, and 56.3% of eligible patients were enrolled (n = 106). The mean patient age was 60.6 years. The preintervention and intervention groups did not differ significantly in enrollment rate, age, gender, cognitive impairment, surgical status, or hearing deficiencies (Table 1). Over 93% of patients were nonsurgical.

Sedative Use

Preintervention, 31.7% of patients received nighttime as‐needed sedatives, versus 16.0% of the intervention group, a 49.4% reduction (P = 0.0041; 95% confidence interval [CI]: 0.056‐0.26) (Figure 2). In patients aged 65 years or older, 38.2% received nighttime as‐needed sedatives preintervention, and 14.6% did postintervention, a 61.2% reduction (P = 0.0054; 95% CI: 0.084‐0.39).

Figure 2

Questionnaire Results

Preintervention, hospital staff was by far the biggest factor keeping patients awake, with 42.4% of patients reporting it (Figure 3). This dropped to only 25.7% with the intervention, a 39.3% decrease (P = 0.009; 95% CI: 0.0452‐0.2765). Preintervention, 19.2% of patients selected voices as the noise most likely to bother them at night, and this dropped to 9.9% with the intervention, a 48% decrease (P = 0.045; 95% CI: 0.0074‐0.1787). No other significant differences were found.

Figure 3

Discussion

Our trial found that hospital staff was the factor most responsible for patient sleep disruption, and that behavioral interventions on hospital staff can reduce use of as‐needed sedatives. The only previously reported intervention to reduce sedative use, the HELP strategy, involved a complex intervention requiring extra staff, with adherence ranging from 10% to 75%.19, 20, 25 In contrast, our protocol can be easily replicated at minimal cost.

Our results are consistent with those of Freedman et al.,26 who found that noise was not the primary factor responsible for sleep disruption in ICU patients, and that staff activities were at least as important a factor. The study is also consistent with the nursing home studies in which decreases in noise and light did not improve sleep.22, 23 It refutes the study that showed that most sleep disturbance in medical‐surgical patients comes from noise and sleeping in an unfamiliar bed.4 Our results call into question the use of the VSH scale in hospitalized patients, which was designed for use in healthy subjects.

Limitations of this study were as follows: moderate size, lack of refined measures of disease severity, and, as in previous studies,19, 2123 the lack of randomized concurrent controls. Evaluation of secondary endpoints was limited by lack of validation of the questionnaire with objective observations, and inability to use the modified VSH scale. Self‐reports of sleep may correlate imperfectly with objective measures, such as polysomnography.27

A larger concurrent trial randomizing similar units at multiple hospitals would be ideal. Future research is needed to determine whether improving sleep in the hospital improves other outcomes, such as recovery times, delirium, falls, or cost.

The need to reduce as‐needed sedatives is an important safety issue and similar interventions in other hospitals may be helpful. Simple changes in staff routines and provider prescribing habits can yield significant reductions in sedative use.

Historically, diagnosis has been central to patient care. Making the correct diagnosis serves as a guide to the choice of treatment, permits assessment of prognosis, and indicates what complications to expect. Arriving at the correct diagnosis has been a major goalthe Holy Grail, as it were. Accurate diagnosis continues to be a major focus of medical practice, and accurate diagnoses are routinely made every day. Still, many experienced clinicians have the impression that in recent years the primacy of diagnosis has been coming under attack from several sources.

A decade ago, Thomas Szasz pointed out that disease is a fact of nature, while diagnosis is man made.1 The noun diagnosis is derived from the Greek verb diagignoskeinindicating knowledge attained through analysis. As defined in Merriam‐Webster's Collegiate Dictionary, the diagnosis essentially means the conclusion arrived at by the art of identifying a disease. It is the product of an intellectual effort of a particular analytic type. The response to the question What is the diagnosis? has been the name of the specific disease entity with which the patient is afflicted.

Disease entities represent coherent, organizing concepts.2 A specific disease is a condition with characteristic manifestationsclinical, histologic, or pathophysiologic. If untreated, it results in dysfunction or, in some cases, death. Differentiation of one disease from another is enhanced when there is some sort of understanding, even if incomplete, of the specific pathophysiology at play. Admittedly, concepts of what constitute specific disease entities are not fixed; they evolve with time. Not all diseases have been identified. The underlying etiology may or may not be known. Nonetheless, diseases are recognized as specific entities, distinct from other diseases. Thus, anemia is not regarded as a disease, while pernicious anemia and iron deficiency anemia are diseases. Fever is not a disease, while typhoid fever is. Arthritis is not a disease, while gonococcal arthritis is.

Billable Terms Are Replacing Traditional Medical Diagnoses

The term diagnosis has been redefined to comply with the need to enter a diagnosis for billing purposes. Use of this term for this purpose has confused the issue. Diagnoses entered for such purposes are largely derived from International Classification of Diseases (ICD) lists.3 However, the ICD was not intended to definitively identify underlying diseases, nor to serve as a guide to management and prognostication. The 6th revision of the ICD in 1948, the first revision to be widely employed, was designed for epidemiologic purposes and achieved widespread use to obtain mortality and morbidity statistics.4 It was subsequently also used as a tool to index hospital medical records.

Significantly, it was also employed for billing purposes, with far‐reaching pernicious consequences. Although the ICD purports to be a list of diseases, it actually includes symptoms and signs. Consequently, in the billing context, diagnosis no longer necessarily refers to specific disease states; it now refers to billable termsoften the manifestation that was responsible for the patient seeking medical assistance. Far from being the product of an intellectual effort, it is often merely a justification for submitting a bill. Examples of such diagnoses are shown in Table 1. Many of them represent symptoms, signs, or laboratory abnormalities. The importance of accurate medical diagnosis has been cheapened by this change. The effect is to devalue diagnosisto lessen its status as the Holy Grail.

The effect of this on trainees is invidious, and predictable. The traditional meaning of diagnosis is being replaced in our minds. Physicians in training are tempted to deceive themselves into believing that they have arrived at an understanding of what they are dealing with when they enter such a diagnosis. After all, have they not responded to the question: what is the diagnosis?

We do not mean to imply that physicians are doing anything wrong by entering ICD terms for billing purposes. What must be done for billing purposes must be done. It is important to be aware, however, and to continually remind ourselves, that what has been entered for this purpose is often not a true medical diagnosis.

Further, when the diagnosis is not yet known, it is not possible to enter a true diagnosis. There is no way to say I don't know. It would be preferable to simply admit that the diagnosis is not yet established, as a medical resident has recently emphasized.5

Diagnosis Often Gets Short Shrift Because of the Perceived Urgency of Discharge

The emphasis on diagnosis several generations ago may have resulted, at least in part, from the relative paucity of effective therapeutic interventions before the 1930s. Things have changed; therapeutic capabilities are much more powerful now. Making the correct diagnosis seems to have lost its urgency. Instead of the major question being what is the diagnosis? it now is often what do we do now? The diagnosis is often an afterthought. Indeed, it is sometimes not even mentioned in discharge summaries, where, not uncommonly, one sees nondiagnoses such as blood in stool or polyarthritis.

In addition, we are under pressure to shorten the inpatient stay of hospitalized patients. At least a portion of the public is aware of this; thus, it has been noted in the New York Times that: The pressure to get patients out of the hospital rapidly can focus medical attention on treatment rather than diagnosis.6 We commonly seek to ameliorate the patients' status to permit discharge before (or often without) learning what we are dealing with. Sometimes one senses that the primary question has become how soon can we discharge this patient?

A price is paid for this. In the absence of a valid diagnosis, patients may be subjected to a broad array of nonessential investigations and therapeutic interventions, each with its own possible complications. Patients are often discharged without a diagnosis having been made, presenting a serious challenge to outpatient physicians who are left to manage them without a clear idea of what they are dealing with. It often falls to the outpatient physicians to make the diagnosis. This is somewhat problematic, since they themselves are under harsh time pressure. Patients often require rehospitalization for the same as‐yet‐undiagnosed condition.

The Problem‐Oriented Record Poses Problems

The widespread use of the problem‐oriented record, originated by Lawrence Weed,7 has led to problems of its own.8 It has evolved, away from its original intent. In practice, its major emphasis often seems to be on identification of problems and tracking their progress, rather than on synthesis. This often leads to muddy rather than clear diagnostic thinking. Assessments and progress notes frequently consist of lists of symptoms, organs, abnormal laboratory findings, or even medical specialties. The net effect is often fragmented thinkingas Weed7 put it, failure to integrate findings into a single entity. Synthesizing diverse findings into a single entity, when possible, is necessary to define a diagnosis. Failure to do so may have serious consequences. In a recent study of diagnostic errors in internal medicine, cognitive errors were frequently found to contribute to such errors.9 The most common cognitive problem was faulty synthesis. How much worse than faulty synthesis is failure to synthesize at all!

Presumptive Diagnoses, Even if Incorrect, Metamorphose into Established Diagnoses

We must often treat empirically. When there is no firm diagnosis, presumptive diagnoses must be made and acted upon. Unfortunately, there are not always mechanisms for the physician to make it clear that his or her diagnosis is only presumptive. (A common example is acute viral syndrome, generally an educated guess.) All too often, presumptive diagnoses are entered, without qualification, as definitive diagnoses, and then achieve immortality. Thus, if a patient is incorrectly diagnosed as having rheumatoid arthritis, all subsequent presentations will start: A so‐and‐so year‐old woman with rheumatoid arthritis for many years Presumptive diagnoses are frequently not questioned. It is easier to assume that they were arrived at after due consideration. Once entered in the medical record, they may be difficult to remove.

It is true that the need to arrive at a precise diagnosis is less pressing for some medical specialties than for others. Emergency physicians, critical care physicians, and frequently, surgeons, must commonly act on the basis of presumptive diagnoses. In contrast, internists, family physicians, psychiatrists, and indeed all physicians who care for patients with chronic illnesses can, with time, be expected to sort out accurate from inaccurate presumptive diagnoses.

A specific example of the problem of presumptive diagnosis is of interest. It is not uncommon, following a first encounter, for a diagnosis to be entered based on the patient's history alone. While such diagnoses are frequently correct, they are not invariably correct. The patient may have arrived at the conclusion herself; she may have misunderstood what she was told by her physician, or her physician may have been in error. Such inaccurate diagnoses also often achieve immortality in the medical record.

Apparent Disparaging of the Importance of Diagnosis

Further trivialization has come from a number of publications expressing concerns about the importance of diagnosis. Thus we read that there are negative consequences of emphasis on diagnosis. When we know what is wrong, we focus less on the individual and more on the disease.10 In his recent book Our Present Complaint. American Medicine, Then and Now, the scholar C.E. Rosenberg11 includes a chapter with the provocative title The Tyranny of Diagnosis. He points out that even a century ago the fear was expressed that burgeoning scientific medicine would lead to denigration of physicians' holistic and intuitive skills.11 Other authors maintain that firm diagnoses may be misleading, since many diseases are a matter of degree in a continuuma spectrumthat are best defined employing a statistical model of risk prediction.12 The suggestion is made that the usefulness of diagnostic tests should not be related to the presence or absence of a disease, but rather to whether they influence outcome.13

Scientific medicine is focused on diagnosis. Denigration of diagnosis has often come, as a philosophical posture, from opponents of reductionist thinking. As Rosenberg11 points out: It has become fashionable among humanistic and social science‐oriented commentators to dwell on the distinction between illness and disease, between the patient's felt experience and the constructions placed on that experience by the world of medicine. Their opposition, he feels, reflects the value‐laden mutual incompatibility (real or apparent) of art and science, of holism and reductionism.2

It is true that medicine is more than just biology. There is a great deal to be said for the view that scientific medicine tends to deemphasize the humanistic, holistic aspects of medical practice. However, despite all these concerns, most physiciansand, to be fair, most criticsagree that making an accurate diagnosis is important. Thus, though the title of his relevant chapter is The Tyranny of Diagnosis, Rosenberg11 states: I might just as well have used the term indispensability. Indeed, the opening words of that chapter are: Diagnosis has always played a pivotal role in medicine.11 Other authors cited above issue this disclaimer: We are not against diagnosis. Diagnosis does and always will play a central role in clinical medicine.12

The importance of diagnosis is underscored by the vigorous debate about how to assess diagnostic tests;14 apparently, diagnosis does indeed matter. While it is true that diagnoses are not always precise, objective, and quantifiable,10 abundant evidence points to the unavoidable conclusion that identifying the patient's disease is heuristically useful; that is, it works.2 The track record of modern scientific medicine in improving mortality and morbidity speaks for itself. It hardly seems necessary to defend it. In addition to representing a valuable intellectual challenge in its own right, diagnosis is pivotal to the scientific mission of medicine.

What Can Be Done?

The net effect of all these forces: the use of billable terms as diagnoses, the pressures of managed care, fragmented problem lists, persistence of incorrect presumptive diagnoses in medical records, and antireductionist criticisms is to encourage sloppy diagnostic thinking in some physicians. What can be done to emphasize the proper use of differential diagnosis in arriving at a definitive diagnosis? What can be done to underscore the importance of differentiating between presumptive and definitive diagnoses? Most importantly, how can we instill the respect for the intellectual honesty necessary to acquire and retain these skills?

Above all, we should relentlessly impress on our students and trainees the importance of arriving at an accurate definitive diagnosis. They should be aware that the job is only half done if the diagnosis has not been made. We should do this repeatedly, both by word and by example. We ourselves must display intellectual honesty.

In addition, we ought to be able to enter diagnosis uncertain, so coded, or to append the phrase cause unknown after the manifestation of concern, when we don't really know what is going on. We should routinely indicate when a diagnosis is merely presumptive. Perhaps we need a way to indicate: This diagnosis is definitive or to indicate the specific evidence that led to the diagnosis (eg, biopsy, laboratory result, radiographic finding). Finally, we need to correct the current confusion between diseases and billable terms, to differentiate the disease from the symptom, perhaps by labeling ICD‐9‐CM codes simply as billing codes, with a separate entry for actual medical diagnoses.

Although powerful historical forces have brought us to this state, we believe that arriving at the correct diagnosis is at least as important now as it has been in the past, and that its primacy should be recognized, celebrated, and fought for. We owe our patients no less.

Historically, diagnosis has been central to patient care. Making the correct diagnosis serves as a guide to the choice of treatment, permits assessment of prognosis, and indicates what complications to expect. Arriving at the correct diagnosis has been a major goalthe Holy Grail, as it were. Accurate diagnosis continues to be a major focus of medical practice, and accurate diagnoses are routinely made every day. Still, many experienced clinicians have the impression that in recent years the primacy of diagnosis has been coming under attack from several sources.

A decade ago, Thomas Szasz pointed out that disease is a fact of nature, while diagnosis is man made.1 The noun diagnosis is derived from the Greek verb diagignoskeinindicating knowledge attained through analysis. As defined in Merriam‐Webster's Collegiate Dictionary, the diagnosis essentially means the conclusion arrived at by the art of identifying a disease. It is the product of an intellectual effort of a particular analytic type. The response to the question What is the diagnosis? has been the name of the specific disease entity with which the patient is afflicted.

Disease entities represent coherent, organizing concepts.2 A specific disease is a condition with characteristic manifestationsclinical, histologic, or pathophysiologic. If untreated, it results in dysfunction or, in some cases, death. Differentiation of one disease from another is enhanced when there is some sort of understanding, even if incomplete, of the specific pathophysiology at play. Admittedly, concepts of what constitute specific disease entities are not fixed; they evolve with time. Not all diseases have been identified. The underlying etiology may or may not be known. Nonetheless, diseases are recognized as specific entities, distinct from other diseases. Thus, anemia is not regarded as a disease, while pernicious anemia and iron deficiency anemia are diseases. Fever is not a disease, while typhoid fever is. Arthritis is not a disease, while gonococcal arthritis is.

Billable Terms Are Replacing Traditional Medical Diagnoses

The term diagnosis has been redefined to comply with the need to enter a diagnosis for billing purposes. Use of this term for this purpose has confused the issue. Diagnoses entered for such purposes are largely derived from International Classification of Diseases (ICD) lists.3 However, the ICD was not intended to definitively identify underlying diseases, nor to serve as a guide to management and prognostication. The 6th revision of the ICD in 1948, the first revision to be widely employed, was designed for epidemiologic purposes and achieved widespread use to obtain mortality and morbidity statistics.4 It was subsequently also used as a tool to index hospital medical records.

Significantly, it was also employed for billing purposes, with far‐reaching pernicious consequences. Although the ICD purports to be a list of diseases, it actually includes symptoms and signs. Consequently, in the billing context, diagnosis no longer necessarily refers to specific disease states; it now refers to billable termsoften the manifestation that was responsible for the patient seeking medical assistance. Far from being the product of an intellectual effort, it is often merely a justification for submitting a bill. Examples of such diagnoses are shown in Table 1. Many of them represent symptoms, signs, or laboratory abnormalities. The importance of accurate medical diagnosis has been cheapened by this change. The effect is to devalue diagnosisto lessen its status as the Holy Grail.

The effect of this on trainees is invidious, and predictable. The traditional meaning of diagnosis is being replaced in our minds. Physicians in training are tempted to deceive themselves into believing that they have arrived at an understanding of what they are dealing with when they enter such a diagnosis. After all, have they not responded to the question: what is the diagnosis?

We do not mean to imply that physicians are doing anything wrong by entering ICD terms for billing purposes. What must be done for billing purposes must be done. It is important to be aware, however, and to continually remind ourselves, that what has been entered for this purpose is often not a true medical diagnosis.

Further, when the diagnosis is not yet known, it is not possible to enter a true diagnosis. There is no way to say I don't know. It would be preferable to simply admit that the diagnosis is not yet established, as a medical resident has recently emphasized.5

Diagnosis Often Gets Short Shrift Because of the Perceived Urgency of Discharge

The emphasis on diagnosis several generations ago may have resulted, at least in part, from the relative paucity of effective therapeutic interventions before the 1930s. Things have changed; therapeutic capabilities are much more powerful now. Making the correct diagnosis seems to have lost its urgency. Instead of the major question being what is the diagnosis? it now is often what do we do now? The diagnosis is often an afterthought. Indeed, it is sometimes not even mentioned in discharge summaries, where, not uncommonly, one sees nondiagnoses such as blood in stool or polyarthritis.

In addition, we are under pressure to shorten the inpatient stay of hospitalized patients. At least a portion of the public is aware of this; thus, it has been noted in the New York Times that: The pressure to get patients out of the hospital rapidly can focus medical attention on treatment rather than diagnosis.6 We commonly seek to ameliorate the patients' status to permit discharge before (or often without) learning what we are dealing with. Sometimes one senses that the primary question has become how soon can we discharge this patient?

A price is paid for this. In the absence of a valid diagnosis, patients may be subjected to a broad array of nonessential investigations and therapeutic interventions, each with its own possible complications. Patients are often discharged without a diagnosis having been made, presenting a serious challenge to outpatient physicians who are left to manage them without a clear idea of what they are dealing with. It often falls to the outpatient physicians to make the diagnosis. This is somewhat problematic, since they themselves are under harsh time pressure. Patients often require rehospitalization for the same as‐yet‐undiagnosed condition.

The Problem‐Oriented Record Poses Problems

The widespread use of the problem‐oriented record, originated by Lawrence Weed,7 has led to problems of its own.8 It has evolved, away from its original intent. In practice, its major emphasis often seems to be on identification of problems and tracking their progress, rather than on synthesis. This often leads to muddy rather than clear diagnostic thinking. Assessments and progress notes frequently consist of lists of symptoms, organs, abnormal laboratory findings, or even medical specialties. The net effect is often fragmented thinkingas Weed7 put it, failure to integrate findings into a single entity. Synthesizing diverse findings into a single entity, when possible, is necessary to define a diagnosis. Failure to do so may have serious consequences. In a recent study of diagnostic errors in internal medicine, cognitive errors were frequently found to contribute to such errors.9 The most common cognitive problem was faulty synthesis. How much worse than faulty synthesis is failure to synthesize at all!

Presumptive Diagnoses, Even if Incorrect, Metamorphose into Established Diagnoses

We must often treat empirically. When there is no firm diagnosis, presumptive diagnoses must be made and acted upon. Unfortunately, there are not always mechanisms for the physician to make it clear that his or her diagnosis is only presumptive. (A common example is acute viral syndrome, generally an educated guess.) All too often, presumptive diagnoses are entered, without qualification, as definitive diagnoses, and then achieve immortality. Thus, if a patient is incorrectly diagnosed as having rheumatoid arthritis, all subsequent presentations will start: A so‐and‐so year‐old woman with rheumatoid arthritis for many years Presumptive diagnoses are frequently not questioned. It is easier to assume that they were arrived at after due consideration. Once entered in the medical record, they may be difficult to remove.

It is true that the need to arrive at a precise diagnosis is less pressing for some medical specialties than for others. Emergency physicians, critical care physicians, and frequently, surgeons, must commonly act on the basis of presumptive diagnoses. In contrast, internists, family physicians, psychiatrists, and indeed all physicians who care for patients with chronic illnesses can, with time, be expected to sort out accurate from inaccurate presumptive diagnoses.

A specific example of the problem of presumptive diagnosis is of interest. It is not uncommon, following a first encounter, for a diagnosis to be entered based on the patient's history alone. While such diagnoses are frequently correct, they are not invariably correct. The patient may have arrived at the conclusion herself; she may have misunderstood what she was told by her physician, or her physician may have been in error. Such inaccurate diagnoses also often achieve immortality in the medical record.

Apparent Disparaging of the Importance of Diagnosis

Further trivialization has come from a number of publications expressing concerns about the importance of diagnosis. Thus we read that there are negative consequences of emphasis on diagnosis. When we know what is wrong, we focus less on the individual and more on the disease.10 In his recent book Our Present Complaint. American Medicine, Then and Now, the scholar C.E. Rosenberg11 includes a chapter with the provocative title The Tyranny of Diagnosis. He points out that even a century ago the fear was expressed that burgeoning scientific medicine would lead to denigration of physicians' holistic and intuitive skills.11 Other authors maintain that firm diagnoses may be misleading, since many diseases are a matter of degree in a continuuma spectrumthat are best defined employing a statistical model of risk prediction.12 The suggestion is made that the usefulness of diagnostic tests should not be related to the presence or absence of a disease, but rather to whether they influence outcome.13

Scientific medicine is focused on diagnosis. Denigration of diagnosis has often come, as a philosophical posture, from opponents of reductionist thinking. As Rosenberg11 points out: It has become fashionable among humanistic and social science‐oriented commentators to dwell on the distinction between illness and disease, between the patient's felt experience and the constructions placed on that experience by the world of medicine. Their opposition, he feels, reflects the value‐laden mutual incompatibility (real or apparent) of art and science, of holism and reductionism.2

It is true that medicine is more than just biology. There is a great deal to be said for the view that scientific medicine tends to deemphasize the humanistic, holistic aspects of medical practice. However, despite all these concerns, most physiciansand, to be fair, most criticsagree that making an accurate diagnosis is important. Thus, though the title of his relevant chapter is The Tyranny of Diagnosis, Rosenberg11 states: I might just as well have used the term indispensability. Indeed, the opening words of that chapter are: Diagnosis has always played a pivotal role in medicine.11 Other authors cited above issue this disclaimer: We are not against diagnosis. Diagnosis does and always will play a central role in clinical medicine.12

The importance of diagnosis is underscored by the vigorous debate about how to assess diagnostic tests;14 apparently, diagnosis does indeed matter. While it is true that diagnoses are not always precise, objective, and quantifiable,10 abundant evidence points to the unavoidable conclusion that identifying the patient's disease is heuristically useful; that is, it works.2 The track record of modern scientific medicine in improving mortality and morbidity speaks for itself. It hardly seems necessary to defend it. In addition to representing a valuable intellectual challenge in its own right, diagnosis is pivotal to the scientific mission of medicine.

What Can Be Done?

The net effect of all these forces: the use of billable terms as diagnoses, the pressures of managed care, fragmented problem lists, persistence of incorrect presumptive diagnoses in medical records, and antireductionist criticisms is to encourage sloppy diagnostic thinking in some physicians. What can be done to emphasize the proper use of differential diagnosis in arriving at a definitive diagnosis? What can be done to underscore the importance of differentiating between presumptive and definitive diagnoses? Most importantly, how can we instill the respect for the intellectual honesty necessary to acquire and retain these skills?

Above all, we should relentlessly impress on our students and trainees the importance of arriving at an accurate definitive diagnosis. They should be aware that the job is only half done if the diagnosis has not been made. We should do this repeatedly, both by word and by example. We ourselves must display intellectual honesty.

In addition, we ought to be able to enter diagnosis uncertain, so coded, or to append the phrase cause unknown after the manifestation of concern, when we don't really know what is going on. We should routinely indicate when a diagnosis is merely presumptive. Perhaps we need a way to indicate: This diagnosis is definitive or to indicate the specific evidence that led to the diagnosis (eg, biopsy, laboratory result, radiographic finding). Finally, we need to correct the current confusion between diseases and billable terms, to differentiate the disease from the symptom, perhaps by labeling ICD‐9‐CM codes simply as billing codes, with a separate entry for actual medical diagnoses.

Although powerful historical forces have brought us to this state, we believe that arriving at the correct diagnosis is at least as important now as it has been in the past, and that its primacy should be recognized, celebrated, and fought for. We owe our patients no less.

Historically, diagnosis has been central to patient care. Making the correct diagnosis serves as a guide to the choice of treatment, permits assessment of prognosis, and indicates what complications to expect. Arriving at the correct diagnosis has been a major goalthe Holy Grail, as it were. Accurate diagnosis continues to be a major focus of medical practice, and accurate diagnoses are routinely made every day. Still, many experienced clinicians have the impression that in recent years the primacy of diagnosis has been coming under attack from several sources.

A decade ago, Thomas Szasz pointed out that disease is a fact of nature, while diagnosis is man made.1 The noun diagnosis is derived from the Greek verb diagignoskeinindicating knowledge attained through analysis. As defined in Merriam‐Webster's Collegiate Dictionary, the diagnosis essentially means the conclusion arrived at by the art of identifying a disease. It is the product of an intellectual effort of a particular analytic type. The response to the question What is the diagnosis? has been the name of the specific disease entity with which the patient is afflicted.

Disease entities represent coherent, organizing concepts.2 A specific disease is a condition with characteristic manifestationsclinical, histologic, or pathophysiologic. If untreated, it results in dysfunction or, in some cases, death. Differentiation of one disease from another is enhanced when there is some sort of understanding, even if incomplete, of the specific pathophysiology at play. Admittedly, concepts of what constitute specific disease entities are not fixed; they evolve with time. Not all diseases have been identified. The underlying etiology may or may not be known. Nonetheless, diseases are recognized as specific entities, distinct from other diseases. Thus, anemia is not regarded as a disease, while pernicious anemia and iron deficiency anemia are diseases. Fever is not a disease, while typhoid fever is. Arthritis is not a disease, while gonococcal arthritis is.

Billable Terms Are Replacing Traditional Medical Diagnoses

The term diagnosis has been redefined to comply with the need to enter a diagnosis for billing purposes. Use of this term for this purpose has confused the issue. Diagnoses entered for such purposes are largely derived from International Classification of Diseases (ICD) lists.3 However, the ICD was not intended to definitively identify underlying diseases, nor to serve as a guide to management and prognostication. The 6th revision of the ICD in 1948, the first revision to be widely employed, was designed for epidemiologic purposes and achieved widespread use to obtain mortality and morbidity statistics.4 It was subsequently also used as a tool to index hospital medical records.

Significantly, it was also employed for billing purposes, with far‐reaching pernicious consequences. Although the ICD purports to be a list of diseases, it actually includes symptoms and signs. Consequently, in the billing context, diagnosis no longer necessarily refers to specific disease states; it now refers to billable termsoften the manifestation that was responsible for the patient seeking medical assistance. Far from being the product of an intellectual effort, it is often merely a justification for submitting a bill. Examples of such diagnoses are shown in Table 1. Many of them represent symptoms, signs, or laboratory abnormalities. The importance of accurate medical diagnosis has been cheapened by this change. The effect is to devalue diagnosisto lessen its status as the Holy Grail.

The effect of this on trainees is invidious, and predictable. The traditional meaning of diagnosis is being replaced in our minds. Physicians in training are tempted to deceive themselves into believing that they have arrived at an understanding of what they are dealing with when they enter such a diagnosis. After all, have they not responded to the question: what is the diagnosis?

We do not mean to imply that physicians are doing anything wrong by entering ICD terms for billing purposes. What must be done for billing purposes must be done. It is important to be aware, however, and to continually remind ourselves, that what has been entered for this purpose is often not a true medical diagnosis.

Further, when the diagnosis is not yet known, it is not possible to enter a true diagnosis. There is no way to say I don't know. It would be preferable to simply admit that the diagnosis is not yet established, as a medical resident has recently emphasized.5

Diagnosis Often Gets Short Shrift Because of the Perceived Urgency of Discharge

The emphasis on diagnosis several generations ago may have resulted, at least in part, from the relative paucity of effective therapeutic interventions before the 1930s. Things have changed; therapeutic capabilities are much more powerful now. Making the correct diagnosis seems to have lost its urgency. Instead of the major question being what is the diagnosis? it now is often what do we do now? The diagnosis is often an afterthought. Indeed, it is sometimes not even mentioned in discharge summaries, where, not uncommonly, one sees nondiagnoses such as blood in stool or polyarthritis.

In addition, we are under pressure to shorten the inpatient stay of hospitalized patients. At least a portion of the public is aware of this; thus, it has been noted in the New York Times that: The pressure to get patients out of the hospital rapidly can focus medical attention on treatment rather than diagnosis.6 We commonly seek to ameliorate the patients' status to permit discharge before (or often without) learning what we are dealing with. Sometimes one senses that the primary question has become how soon can we discharge this patient?

A price is paid for this. In the absence of a valid diagnosis, patients may be subjected to a broad array of nonessential investigations and therapeutic interventions, each with its own possible complications. Patients are often discharged without a diagnosis having been made, presenting a serious challenge to outpatient physicians who are left to manage them without a clear idea of what they are dealing with. It often falls to the outpatient physicians to make the diagnosis. This is somewhat problematic, since they themselves are under harsh time pressure. Patients often require rehospitalization for the same as‐yet‐undiagnosed condition.

The Problem‐Oriented Record Poses Problems

The widespread use of the problem‐oriented record, originated by Lawrence Weed,7 has led to problems of its own.8 It has evolved, away from its original intent. In practice, its major emphasis often seems to be on identification of problems and tracking their progress, rather than on synthesis. This often leads to muddy rather than clear diagnostic thinking. Assessments and progress notes frequently consist of lists of symptoms, organs, abnormal laboratory findings, or even medical specialties. The net effect is often fragmented thinkingas Weed7 put it, failure to integrate findings into a single entity. Synthesizing diverse findings into a single entity, when possible, is necessary to define a diagnosis. Failure to do so may have serious consequences. In a recent study of diagnostic errors in internal medicine, cognitive errors were frequently found to contribute to such errors.9 The most common cognitive problem was faulty synthesis. How much worse than faulty synthesis is failure to synthesize at all!

Presumptive Diagnoses, Even if Incorrect, Metamorphose into Established Diagnoses

We must often treat empirically. When there is no firm diagnosis, presumptive diagnoses must be made and acted upon. Unfortunately, there are not always mechanisms for the physician to make it clear that his or her diagnosis is only presumptive. (A common example is acute viral syndrome, generally an educated guess.) All too often, presumptive diagnoses are entered, without qualification, as definitive diagnoses, and then achieve immortality. Thus, if a patient is incorrectly diagnosed as having rheumatoid arthritis, all subsequent presentations will start: A so‐and‐so year‐old woman with rheumatoid arthritis for many years Presumptive diagnoses are frequently not questioned. It is easier to assume that they were arrived at after due consideration. Once entered in the medical record, they may be difficult to remove.

It is true that the need to arrive at a precise diagnosis is less pressing for some medical specialties than for others. Emergency physicians, critical care physicians, and frequently, surgeons, must commonly act on the basis of presumptive diagnoses. In contrast, internists, family physicians, psychiatrists, and indeed all physicians who care for patients with chronic illnesses can, with time, be expected to sort out accurate from inaccurate presumptive diagnoses.

A specific example of the problem of presumptive diagnosis is of interest. It is not uncommon, following a first encounter, for a diagnosis to be entered based on the patient's history alone. While such diagnoses are frequently correct, they are not invariably correct. The patient may have arrived at the conclusion herself; she may have misunderstood what she was told by her physician, or her physician may have been in error. Such inaccurate diagnoses also often achieve immortality in the medical record.

Apparent Disparaging of the Importance of Diagnosis

Further trivialization has come from a number of publications expressing concerns about the importance of diagnosis. Thus we read that there are negative consequences of emphasis on diagnosis. When we know what is wrong, we focus less on the individual and more on the disease.10 In his recent book Our Present Complaint. American Medicine, Then and Now, the scholar C.E. Rosenberg11 includes a chapter with the provocative title The Tyranny of Diagnosis. He points out that even a century ago the fear was expressed that burgeoning scientific medicine would lead to denigration of physicians' holistic and intuitive skills.11 Other authors maintain that firm diagnoses may be misleading, since many diseases are a matter of degree in a continuuma spectrumthat are best defined employing a statistical model of risk prediction.12 The suggestion is made that the usefulness of diagnostic tests should not be related to the presence or absence of a disease, but rather to whether they influence outcome.13

Scientific medicine is focused on diagnosis. Denigration of diagnosis has often come, as a philosophical posture, from opponents of reductionist thinking. As Rosenberg11 points out: It has become fashionable among humanistic and social science‐oriented commentators to dwell on the distinction between illness and disease, between the patient's felt experience and the constructions placed on that experience by the world of medicine. Their opposition, he feels, reflects the value‐laden mutual incompatibility (real or apparent) of art and science, of holism and reductionism.2

It is true that medicine is more than just biology. There is a great deal to be said for the view that scientific medicine tends to deemphasize the humanistic, holistic aspects of medical practice. However, despite all these concerns, most physiciansand, to be fair, most criticsagree that making an accurate diagnosis is important. Thus, though the title of his relevant chapter is The Tyranny of Diagnosis, Rosenberg11 states: I might just as well have used the term indispensability. Indeed, the opening words of that chapter are: Diagnosis has always played a pivotal role in medicine.11 Other authors cited above issue this disclaimer: We are not against diagnosis. Diagnosis does and always will play a central role in clinical medicine.12

The importance of diagnosis is underscored by the vigorous debate about how to assess diagnostic tests;14 apparently, diagnosis does indeed matter. While it is true that diagnoses are not always precise, objective, and quantifiable,10 abundant evidence points to the unavoidable conclusion that identifying the patient's disease is heuristically useful; that is, it works.2 The track record of modern scientific medicine in improving mortality and morbidity speaks for itself. It hardly seems necessary to defend it. In addition to representing a valuable intellectual challenge in its own right, diagnosis is pivotal to the scientific mission of medicine.

What Can Be Done?

The net effect of all these forces: the use of billable terms as diagnoses, the pressures of managed care, fragmented problem lists, persistence of incorrect presumptive diagnoses in medical records, and antireductionist criticisms is to encourage sloppy diagnostic thinking in some physicians. What can be done to emphasize the proper use of differential diagnosis in arriving at a definitive diagnosis? What can be done to underscore the importance of differentiating between presumptive and definitive diagnoses? Most importantly, how can we instill the respect for the intellectual honesty necessary to acquire and retain these skills?

Above all, we should relentlessly impress on our students and trainees the importance of arriving at an accurate definitive diagnosis. They should be aware that the job is only half done if the diagnosis has not been made. We should do this repeatedly, both by word and by example. We ourselves must display intellectual honesty.

In addition, we ought to be able to enter diagnosis uncertain, so coded, or to append the phrase cause unknown after the manifestation of concern, when we don't really know what is going on. We should routinely indicate when a diagnosis is merely presumptive. Perhaps we need a way to indicate: This diagnosis is definitive or to indicate the specific evidence that led to the diagnosis (eg, biopsy, laboratory result, radiographic finding). Finally, we need to correct the current confusion between diseases and billable terms, to differentiate the disease from the symptom, perhaps by labeling ICD‐9‐CM codes simply as billing codes, with a separate entry for actual medical diagnoses.

Although powerful historical forces have brought us to this state, we believe that arriving at the correct diagnosis is at least as important now as it has been in the past, and that its primacy should be recognized, celebrated, and fought for. We owe our patients no less.

Central venous catheter (CVC) insertions are commonly performed at the bedside in medical intensive care unit (MICU) settings. Internal medicine residents are required to demonstrate knowledge regarding CVC indications, complications, and sterile technique,1 and often perform the procedure during training. Education in CVC insertion is needed because many internal medicine residents are uncomfortable performing this procedure.2 CVC insertion also carries the risk of potentially life‐threatening complications including infection, pneumothorax, arterial puncture, deep vein thrombosis, and bleeding. Education and training may also contribute to improved patient care because increased physician experience with CVC insertion reduces complication risk.3, 4 Similarly, a higher number of needle passes or attempts during CVC insertion correlates with mechanical complications such as pneumothorax or arterial punctures.48 Pneumothorax rates for internal jugular (IJ) CVCs have been reported to range from 0% to 0.2% and for subclavian (SC) CVCs from 1.5% to 3.1%.4, 5 The arterial puncture rate for IJ CVCs ranges from 5.0% to 9.4% and for SC CVCs from 3.1% to 4.9%.4, 5 Proper use of ultrasound to assist with IJ CVC insertion has been shown to decrease these mechanical complications.4, 5 However, studies of ultrasound use with SC CVC insertion have mixed results.4

Simulation‐based training has been used in medical education to increase knowledge, provide opportunities for deliberate and safe practice, and shape the development of clinical skills.9, 10 We previously used simulation‐based mastery learning to improve the thoracentesis and advanced cardiac life support (ACLS) skills of internal medicine residents.11, 12 Although a few small studies have linked simulation‐based interventions to improved quality of care,1319 more work is needed to show that results from a simulated environment transfer to actual patient care.

This study had 2 aims. The first was to expand our simulation‐based mastery learning to CVC insertion using a CVC simulator and ultrasound device. The second was to assess quality indicators (number of needle passes, pneumothorax, arterial punctures, and need for catheter adjustment) and resident confidence related to actual CVC insertions in the MICU before and after an educational intervention.

Materials and Methods

Design

This was a cohort study20 of IJ and SC CVC insertions by 41 second‐ and third‐year internal medicine residents rotating through the MICU in a university‐affiliated program from October 2006 to February 2007. The Northwestern University Institutional Review Board approved the study. All study participants were required to give informed consent prior to participation.

Thirteen residents rotated through the MICU during a 6‐week preintervention phase. These residents served as a traditionally trained group that did not receive CVC insertion simulator training. Simultaneously, 28 residents who rotated through the MICU later in the study period received simulation‐based training in CVC insertion and served as the simulator‐trained group (Figure 1). Demographic data were obtained from the participants including age, gender, ethnicity, year of training, and scores on the United States Medical Licensing Examination (USMLE) Steps 1 and 2.

Figure 1

Simulator‐trained residents underwent baseline skill assessment (pretest) using a 27‐item checklist in IJ and SC CVC insertions (see Appendix). Checklists were developed by one author (J.H.B.) using appropriate references4, 5 and a step‐by‐step process,21 and reviewed for completeness by another author with expertise in checklist development (D.B.W.). Each skill or other action was listed in order and given equal weight. A dichotomous scoring scale of 1 = done correctly and 0 = done incorrectly/not done was imposed for each item. Assessments were performed using Simulab's CentralLineMan. This model features realistic tissue with ultrasound compatibility, an arterial pulse, and self‐sealing veins and skins. Needles, dilators, and guidewires can be inserted and realistic venous and arterial pressures demonstrated (Figure 2).

Figure 2

Residents in the simulator‐trained group received two, 2‐hour education sessions featuring a lecture, ultrasound training, deliberate practice with the CVC simulator, and feedback.22 Education sessions contained standardized didactic material on CVC indications and complications, as well as a stepwise demonstration of IJ and SC CVC insertions using ultrasound and landmark techniques. These sessions were supervised by a senior hospitalist faculty member with expertise in CVC insertions (J.H.B.). Residents were expected to use the ultrasound device for all IJ CVC insertions. However, its use was optional for SC CVC insertion. After training, residents were retested (posttest) and required to meet or exceed a minimum passing score (MPS) set by an expert panel for both IJ and SC procedures.23 This 11 member expert panel provided item‐based (Angoff) and group‐based (Hofstee) judgments on the 27‐item checklists as described previously.23

Residents who did not achieve the MPS had more deliberate practice and were retested until the MPS was reached; the key feature of mastery learning.24 After completing simulation‐based mastery learning in CVC insertion, the 28 simulator‐trained residents rotated through the MICU.

Results

All eligible residents participated in the study and completed the entire protocol. There was no significant difference in age, gender, ethnicity, year of training, or USMLE Step 1 and 2 scores between the traditionally‐trained and simulator‐trained groups.

Interrater reliability measured by the mean kappa coefficient was very high (_n = 0.94) across the 27 IJ and SC checklist items. No resident met the MPS (79.1%) for CVC insertion at baseline testing. In the simulator‐trained group, 25 of 28 (89%) residents achieved SC skill mastery and 27 of 28 (96%) achieved IJ skill mastery within the standard four hour curriculum. All residents subsequently reached the MPS with less than one hour of additional practice time. A graphic portrait of the residents' pretest and posttest performance on the simulated CVC clinical skills examination with descriptive statistics is shown in Figure 3. After the educational intervention, posttest scores significantly improved (p < 0.001), to meet or exceed the MPS.

Figure 3

Traditionally trained and simulator‐trained residents independently inserted 46 CVCs during the study period. Simulator‐trained residents required significantly fewer needle passes to insert all actual CVCs in the MICU compared to traditionally trained residents: mean (M) = 1.79, standard deviation (SD) = 1.03 versus M = 2.78, SD = 1.77 (p = 0.04). As shown in Table 1, the groups did not differ in pneumothorax, arterial puncture, or mean number of CVC adjustments. In addition, the groups did not differ in use of ultrasound for IJ or SC CVC insertions. One IJ CVC was inserted without ultrasound in the traditionally‐trained group; 2 were inserted without ultrasound in the simulator‐trained group. Ultrasound was not used during any SC CVC insertions in the traditionally‐trained group and was used for 1 SC CVC insertion in the simulator‐trained group.

Simulator‐trained residents displayed more self‐confidence about their procedural skills than traditionally‐trained residents (M = 81, SD = 11 versus M = 68, SD = 20, p = 0.02). Spearman correlations showed no practical association between resident self‐confidence and performance on CVC insertion quality indicators.

Discussion

This study demonstrates the use of a mastery learning model to develop CVC insertion skills to a high achievement level among internal medicine residents. Our data support prior work showing that procedural skills that are poor at baseline can be increased significantly using simulation‐based training and deliberate practice.1118, 28 This report on CVC insertion adds to the growing body of literature showing that simulation training complements standard medical education,1119, 28 and expands the clinical application of the mastery model beyond thoracentesis and ACLS.11, 12 Use of the mastery model described in this study also has important implications for patients. In our training program, residents are required to demonstrate procedural mastery in a simulated environment before independently performing a CVC insertion on an actual patient. This is in sharp contrast to the traditional clinical model of procedural training at the bedside, and may be used in other training programs and with other invasive procedures.

The second aim of our study was to determine the impact of simulation‐based training on actual clinical practice by residents in the MICU. To our knowledge, no prior study has demonstrated that simulation‐based training in CVC insertion improves patient outcomes. We believe our results advance what is known about the impact of simulation‐based training because simulator‐trained residents in this study performed actual CVC insertions in the MICU using significantly fewer needle passes. Needle passes have been used by other investigators as a surrogate measure for reduced CVC‐associated complications because mechanical complications rise exponentially with more than two insertion attempts.47, 29 We believe this finding demonstrates transfer of skill acquired from simulation‐based training to the actual clinical environment. It is possible that ultrasound training accounts for the improvement in the simulator‐trained group. However, we do not believe that ultrasound training is entirely responsible as prior work has shown that deliberate practice using mastery learning without ultrasound significantly improved resident performance of thoracentesis11 and ACLS12, 19 procedures. We did not show a significant reduction in complications such as pneumothorax or arterial puncture. This is likely due to the small sample size and the low number of procedures and complications during the study period.

Our results also show that resident self‐confidence regarding actual CVC insertions improved after simulation training. These findings are similar to prior reports linking improved confidence among trainees after simulation‐based training in CVC insertion.29, 30 Our results did not reveal a correlation between improved self‐confidence and clinical skill acquisition. Linking improved self‐confidence to improved clinical skill is important because self‐assessment does not always correlate with performance ability.31, 32

More study is needed to evaluate the impact of simulation‐based training on the quality of CVC insertions by trainees. Mechanisms shown to decrease complications of CVC placement include use of ultrasound,4, 7, 3336 full sterile barrier technique,3739 chlorhexidine skin preparations,4042 and nurse‐physician education.43 Our simulation‐training program incorporates each of these elements. We plan to expand our simulation‐based training intervention to a larger sample size to determine its impact on mechanical and infectious complication rates linked to CVC insertion.

This study has several limitations. It was performed at a single institution over a short time period. However, demonstration of significantly fewer needle passes and improved resident self‐confidence after simulator training are important findings that warrant further study. It was impossible to blind raters during the skills assessment examination about whether the resident was performing a pretest or posttest. This was accounted for by using a second rater, who was blind to the pretest and posttest status of the examinee. The arterial puncture rate of 7% among simulator‐trained residents was higher than expected, although it remains within published ranges.4, 5 Also, a low total number of CVCs were evaluated during the study. This is likely due to strict exclusion criteria employed in order to study the impact of simulation training. For example, CVC insertions were only evaluated if they were actually performed by study residents (supervised insertions were excluded) and femoral catheters were not evaluated. We did not track clinical experience with CVC insertion by residents before the study. Residents who were simulator‐trained may have had more clinical experience with CVC insertion and this may have impacted their performance. However, residents did not differ in year of training or clinical rotations, and there is clear evidence that clinical training is not a proxy for skill acquisition.44 Finally, outcome data were measured via resident questionnaires that relied on resident recall about CVC insertion rather than observer ratings. This method was selected because observer ratings could not be standardized given the large number of clinical supervisors in the MICU over the study period. Information about needle passes and arterial puncture also may not be documented in procedural notes and could not be obtained by medical record review. We attempted to minimize recall bias by surveying residents within 24 hours of CVC placement.

In conclusion, this study demonstrates that simulation‐based training and deliberate practice in a mastery learning setting improves performance of both simulated and actual CVC insertions by internal medicine residents. Procedural training remains an important component of internal medicine training, although internists are performing fewer invasive procedures now than in years past.45, 46 Use of a mastery model of CVC insertion requires that trainees demonstrate skill in a simulated environment before independently performing this invasive procedure on patients. Further study is needed to assess clinical outcomes such as reduced CVC‐related infections and mechanical complications after simulation‐based training.

Central venous catheter (CVC) insertions are commonly performed at the bedside in medical intensive care unit (MICU) settings. Internal medicine residents are required to demonstrate knowledge regarding CVC indications, complications, and sterile technique,1 and often perform the procedure during training. Education in CVC insertion is needed because many internal medicine residents are uncomfortable performing this procedure.2 CVC insertion also carries the risk of potentially life‐threatening complications including infection, pneumothorax, arterial puncture, deep vein thrombosis, and bleeding. Education and training may also contribute to improved patient care because increased physician experience with CVC insertion reduces complication risk.3, 4 Similarly, a higher number of needle passes or attempts during CVC insertion correlates with mechanical complications such as pneumothorax or arterial punctures.48 Pneumothorax rates for internal jugular (IJ) CVCs have been reported to range from 0% to 0.2% and for subclavian (SC) CVCs from 1.5% to 3.1%.4, 5 The arterial puncture rate for IJ CVCs ranges from 5.0% to 9.4% and for SC CVCs from 3.1% to 4.9%.4, 5 Proper use of ultrasound to assist with IJ CVC insertion has been shown to decrease these mechanical complications.4, 5 However, studies of ultrasound use with SC CVC insertion have mixed results.4

Simulation‐based training has been used in medical education to increase knowledge, provide opportunities for deliberate and safe practice, and shape the development of clinical skills.9, 10 We previously used simulation‐based mastery learning to improve the thoracentesis and advanced cardiac life support (ACLS) skills of internal medicine residents.11, 12 Although a few small studies have linked simulation‐based interventions to improved quality of care,1319 more work is needed to show that results from a simulated environment transfer to actual patient care.

This study had 2 aims. The first was to expand our simulation‐based mastery learning to CVC insertion using a CVC simulator and ultrasound device. The second was to assess quality indicators (number of needle passes, pneumothorax, arterial punctures, and need for catheter adjustment) and resident confidence related to actual CVC insertions in the MICU before and after an educational intervention.

Materials and Methods

Design

This was a cohort study20 of IJ and SC CVC insertions by 41 second‐ and third‐year internal medicine residents rotating through the MICU in a university‐affiliated program from October 2006 to February 2007. The Northwestern University Institutional Review Board approved the study. All study participants were required to give informed consent prior to participation.

Thirteen residents rotated through the MICU during a 6‐week preintervention phase. These residents served as a traditionally trained group that did not receive CVC insertion simulator training. Simultaneously, 28 residents who rotated through the MICU later in the study period received simulation‐based training in CVC insertion and served as the simulator‐trained group (Figure 1). Demographic data were obtained from the participants including age, gender, ethnicity, year of training, and scores on the United States Medical Licensing Examination (USMLE) Steps 1 and 2.

Figure 1

Simulator‐trained residents underwent baseline skill assessment (pretest) using a 27‐item checklist in IJ and SC CVC insertions (see Appendix). Checklists were developed by one author (J.H.B.) using appropriate references4, 5 and a step‐by‐step process,21 and reviewed for completeness by another author with expertise in checklist development (D.B.W.). Each skill or other action was listed in order and given equal weight. A dichotomous scoring scale of 1 = done correctly and 0 = done incorrectly/not done was imposed for each item. Assessments were performed using Simulab's CentralLineMan. This model features realistic tissue with ultrasound compatibility, an arterial pulse, and self‐sealing veins and skins. Needles, dilators, and guidewires can be inserted and realistic venous and arterial pressures demonstrated (Figure 2).

Figure 2

Residents in the simulator‐trained group received two, 2‐hour education sessions featuring a lecture, ultrasound training, deliberate practice with the CVC simulator, and feedback.22 Education sessions contained standardized didactic material on CVC indications and complications, as well as a stepwise demonstration of IJ and SC CVC insertions using ultrasound and landmark techniques. These sessions were supervised by a senior hospitalist faculty member with expertise in CVC insertions (J.H.B.). Residents were expected to use the ultrasound device for all IJ CVC insertions. However, its use was optional for SC CVC insertion. After training, residents were retested (posttest) and required to meet or exceed a minimum passing score (MPS) set by an expert panel for both IJ and SC procedures.23 This 11 member expert panel provided item‐based (Angoff) and group‐based (Hofstee) judgments on the 27‐item checklists as described previously.23

Residents who did not achieve the MPS had more deliberate practice and were retested until the MPS was reached; the key feature of mastery learning.24 After completing simulation‐based mastery learning in CVC insertion, the 28 simulator‐trained residents rotated through the MICU.

Results

All eligible residents participated in the study and completed the entire protocol. There was no significant difference in age, gender, ethnicity, year of training, or USMLE Step 1 and 2 scores between the traditionally‐trained and simulator‐trained groups.

Interrater reliability measured by the mean kappa coefficient was very high (_n = 0.94) across the 27 IJ and SC checklist items. No resident met the MPS (79.1%) for CVC insertion at baseline testing. In the simulator‐trained group, 25 of 28 (89%) residents achieved SC skill mastery and 27 of 28 (96%) achieved IJ skill mastery within the standard four hour curriculum. All residents subsequently reached the MPS with less than one hour of additional practice time. A graphic portrait of the residents' pretest and posttest performance on the simulated CVC clinical skills examination with descriptive statistics is shown in Figure 3. After the educational intervention, posttest scores significantly improved (p < 0.001), to meet or exceed the MPS.

Figure 3

Traditionally trained and simulator‐trained residents independently inserted 46 CVCs during the study period. Simulator‐trained residents required significantly fewer needle passes to insert all actual CVCs in the MICU compared to traditionally trained residents: mean (M) = 1.79, standard deviation (SD) = 1.03 versus M = 2.78, SD = 1.77 (p = 0.04). As shown in Table 1, the groups did not differ in pneumothorax, arterial puncture, or mean number of CVC adjustments. In addition, the groups did not differ in use of ultrasound for IJ or SC CVC insertions. One IJ CVC was inserted without ultrasound in the traditionally‐trained group; 2 were inserted without ultrasound in the simulator‐trained group. Ultrasound was not used during any SC CVC insertions in the traditionally‐trained group and was used for 1 SC CVC insertion in the simulator‐trained group.

Simulator‐trained residents displayed more self‐confidence about their procedural skills than traditionally‐trained residents (M = 81, SD = 11 versus M = 68, SD = 20, p = 0.02). Spearman correlations showed no practical association between resident self‐confidence and performance on CVC insertion quality indicators.

Discussion

This study demonstrates the use of a mastery learning model to develop CVC insertion skills to a high achievement level among internal medicine residents. Our data support prior work showing that procedural skills that are poor at baseline can be increased significantly using simulation‐based training and deliberate practice.1118, 28 This report on CVC insertion adds to the growing body of literature showing that simulation training complements standard medical education,1119, 28 and expands the clinical application of the mastery model beyond thoracentesis and ACLS.11, 12 Use of the mastery model described in this study also has important implications for patients. In our training program, residents are required to demonstrate procedural mastery in a simulated environment before independently performing a CVC insertion on an actual patient. This is in sharp contrast to the traditional clinical model of procedural training at the bedside, and may be used in other training programs and with other invasive procedures.

The second aim of our study was to determine the impact of simulation‐based training on actual clinical practice by residents in the MICU. To our knowledge, no prior study has demonstrated that simulation‐based training in CVC insertion improves patient outcomes. We believe our results advance what is known about the impact of simulation‐based training because simulator‐trained residents in this study performed actual CVC insertions in the MICU using significantly fewer needle passes. Needle passes have been used by other investigators as a surrogate measure for reduced CVC‐associated complications because mechanical complications rise exponentially with more than two insertion attempts.47, 29 We believe this finding demonstrates transfer of skill acquired from simulation‐based training to the actual clinical environment. It is possible that ultrasound training accounts for the improvement in the simulator‐trained group. However, we do not believe that ultrasound training is entirely responsible as prior work has shown that deliberate practice using mastery learning without ultrasound significantly improved resident performance of thoracentesis11 and ACLS12, 19 procedures. We did not show a significant reduction in complications such as pneumothorax or arterial puncture. This is likely due to the small sample size and the low number of procedures and complications during the study period.

Our results also show that resident self‐confidence regarding actual CVC insertions improved after simulation training. These findings are similar to prior reports linking improved confidence among trainees after simulation‐based training in CVC insertion.29, 30 Our results did not reveal a correlation between improved self‐confidence and clinical skill acquisition. Linking improved self‐confidence to improved clinical skill is important because self‐assessment does not always correlate with performance ability.31, 32

More study is needed to evaluate the impact of simulation‐based training on the quality of CVC insertions by trainees. Mechanisms shown to decrease complications of CVC placement include use of ultrasound,4, 7, 3336 full sterile barrier technique,3739 chlorhexidine skin preparations,4042 and nurse‐physician education.43 Our simulation‐training program incorporates each of these elements. We plan to expand our simulation‐based training intervention to a larger sample size to determine its impact on mechanical and infectious complication rates linked to CVC insertion.

This study has several limitations. It was performed at a single institution over a short time period. However, demonstration of significantly fewer needle passes and improved resident self‐confidence after simulator training are important findings that warrant further study. It was impossible to blind raters during the skills assessment examination about whether the resident was performing a pretest or posttest. This was accounted for by using a second rater, who was blind to the pretest and posttest status of the examinee. The arterial puncture rate of 7% among simulator‐trained residents was higher than expected, although it remains within published ranges.4, 5 Also, a low total number of CVCs were evaluated during the study. This is likely due to strict exclusion criteria employed in order to study the impact of simulation training. For example, CVC insertions were only evaluated if they were actually performed by study residents (supervised insertions were excluded) and femoral catheters were not evaluated. We did not track clinical experience with CVC insertion by residents before the study. Residents who were simulator‐trained may have had more clinical experience with CVC insertion and this may have impacted their performance. However, residents did not differ in year of training or clinical rotations, and there is clear evidence that clinical training is not a proxy for skill acquisition.44 Finally, outcome data were measured via resident questionnaires that relied on resident recall about CVC insertion rather than observer ratings. This method was selected because observer ratings could not be standardized given the large number of clinical supervisors in the MICU over the study period. Information about needle passes and arterial puncture also may not be documented in procedural notes and could not be obtained by medical record review. We attempted to minimize recall bias by surveying residents within 24 hours of CVC placement.

In conclusion, this study demonstrates that simulation‐based training and deliberate practice in a mastery learning setting improves performance of both simulated and actual CVC insertions by internal medicine residents. Procedural training remains an important component of internal medicine training, although internists are performing fewer invasive procedures now than in years past.45, 46 Use of a mastery model of CVC insertion requires that trainees demonstrate skill in a simulated environment before independently performing this invasive procedure on patients. Further study is needed to assess clinical outcomes such as reduced CVC‐related infections and mechanical complications after simulation‐based training.

Central venous catheter (CVC) insertions are commonly performed at the bedside in medical intensive care unit (MICU) settings. Internal medicine residents are required to demonstrate knowledge regarding CVC indications, complications, and sterile technique,1 and often perform the procedure during training. Education in CVC insertion is needed because many internal medicine residents are uncomfortable performing this procedure.2 CVC insertion also carries the risk of potentially life‐threatening complications including infection, pneumothorax, arterial puncture, deep vein thrombosis, and bleeding. Education and training may also contribute to improved patient care because increased physician experience with CVC insertion reduces complication risk.3, 4 Similarly, a higher number of needle passes or attempts during CVC insertion correlates with mechanical complications such as pneumothorax or arterial punctures.48 Pneumothorax rates for internal jugular (IJ) CVCs have been reported to range from 0% to 0.2% and for subclavian (SC) CVCs from 1.5% to 3.1%.4, 5 The arterial puncture rate for IJ CVCs ranges from 5.0% to 9.4% and for SC CVCs from 3.1% to 4.9%.4, 5 Proper use of ultrasound to assist with IJ CVC insertion has been shown to decrease these mechanical complications.4, 5 However, studies of ultrasound use with SC CVC insertion have mixed results.4

Simulation‐based training has been used in medical education to increase knowledge, provide opportunities for deliberate and safe practice, and shape the development of clinical skills.9, 10 We previously used simulation‐based mastery learning to improve the thoracentesis and advanced cardiac life support (ACLS) skills of internal medicine residents.11, 12 Although a few small studies have linked simulation‐based interventions to improved quality of care,1319 more work is needed to show that results from a simulated environment transfer to actual patient care.

This study had 2 aims. The first was to expand our simulation‐based mastery learning to CVC insertion using a CVC simulator and ultrasound device. The second was to assess quality indicators (number of needle passes, pneumothorax, arterial punctures, and need for catheter adjustment) and resident confidence related to actual CVC insertions in the MICU before and after an educational intervention.

Materials and Methods

Design

This was a cohort study20 of IJ and SC CVC insertions by 41 second‐ and third‐year internal medicine residents rotating through the MICU in a university‐affiliated program from October 2006 to February 2007. The Northwestern University Institutional Review Board approved the study. All study participants were required to give informed consent prior to participation.

Thirteen residents rotated through the MICU during a 6‐week preintervention phase. These residents served as a traditionally trained group that did not receive CVC insertion simulator training. Simultaneously, 28 residents who rotated through the MICU later in the study period received simulation‐based training in CVC insertion and served as the simulator‐trained group (Figure 1). Demographic data were obtained from the participants including age, gender, ethnicity, year of training, and scores on the United States Medical Licensing Examination (USMLE) Steps 1 and 2.

Figure 1

Simulator‐trained residents underwent baseline skill assessment (pretest) using a 27‐item checklist in IJ and SC CVC insertions (see Appendix). Checklists were developed by one author (J.H.B.) using appropriate references4, 5 and a step‐by‐step process,21 and reviewed for completeness by another author with expertise in checklist development (D.B.W.). Each skill or other action was listed in order and given equal weight. A dichotomous scoring scale of 1 = done correctly and 0 = done incorrectly/not done was imposed for each item. Assessments were performed using Simulab's CentralLineMan. This model features realistic tissue with ultrasound compatibility, an arterial pulse, and self‐sealing veins and skins. Needles, dilators, and guidewires can be inserted and realistic venous and arterial pressures demonstrated (Figure 2).

Figure 2

Residents in the simulator‐trained group received two, 2‐hour education sessions featuring a lecture, ultrasound training, deliberate practice with the CVC simulator, and feedback.22 Education sessions contained standardized didactic material on CVC indications and complications, as well as a stepwise demonstration of IJ and SC CVC insertions using ultrasound and landmark techniques. These sessions were supervised by a senior hospitalist faculty member with expertise in CVC insertions (J.H.B.). Residents were expected to use the ultrasound device for all IJ CVC insertions. However, its use was optional for SC CVC insertion. After training, residents were retested (posttest) and required to meet or exceed a minimum passing score (MPS) set by an expert panel for both IJ and SC procedures.23 This 11 member expert panel provided item‐based (Angoff) and group‐based (Hofstee) judgments on the 27‐item checklists as described previously.23

Residents who did not achieve the MPS had more deliberate practice and were retested until the MPS was reached; the key feature of mastery learning.24 After completing simulation‐based mastery learning in CVC insertion, the 28 simulator‐trained residents rotated through the MICU.

Results

All eligible residents participated in the study and completed the entire protocol. There was no significant difference in age, gender, ethnicity, year of training, or USMLE Step 1 and 2 scores between the traditionally‐trained and simulator‐trained groups.

Interrater reliability measured by the mean kappa coefficient was very high (_n = 0.94) across the 27 IJ and SC checklist items. No resident met the MPS (79.1%) for CVC insertion at baseline testing. In the simulator‐trained group, 25 of 28 (89%) residents achieved SC skill mastery and 27 of 28 (96%) achieved IJ skill mastery within the standard four hour curriculum. All residents subsequently reached the MPS with less than one hour of additional practice time. A graphic portrait of the residents' pretest and posttest performance on the simulated CVC clinical skills examination with descriptive statistics is shown in Figure 3. After the educational intervention, posttest scores significantly improved (p < 0.001), to meet or exceed the MPS.

Figure 3

Traditionally trained and simulator‐trained residents independently inserted 46 CVCs during the study period. Simulator‐trained residents required significantly fewer needle passes to insert all actual CVCs in the MICU compared to traditionally trained residents: mean (M) = 1.79, standard deviation (SD) = 1.03 versus M = 2.78, SD = 1.77 (p = 0.04). As shown in Table 1, the groups did not differ in pneumothorax, arterial puncture, or mean number of CVC adjustments. In addition, the groups did not differ in use of ultrasound for IJ or SC CVC insertions. One IJ CVC was inserted without ultrasound in the traditionally‐trained group; 2 were inserted without ultrasound in the simulator‐trained group. Ultrasound was not used during any SC CVC insertions in the traditionally‐trained group and was used for 1 SC CVC insertion in the simulator‐trained group.

Simulator‐trained residents displayed more self‐confidence about their procedural skills than traditionally‐trained residents (M = 81, SD = 11 versus M = 68, SD = 20, p = 0.02). Spearman correlations showed no practical association between resident self‐confidence and performance on CVC insertion quality indicators.

Discussion

This study demonstrates the use of a mastery learning model to develop CVC insertion skills to a high achievement level among internal medicine residents. Our data support prior work showing that procedural skills that are poor at baseline can be increased significantly using simulation‐based training and deliberate practice.1118, 28 This report on CVC insertion adds to the growing body of literature showing that simulation training complements standard medical education,1119, 28 and expands the clinical application of the mastery model beyond thoracentesis and ACLS.11, 12 Use of the mastery model described in this study also has important implications for patients. In our training program, residents are required to demonstrate procedural mastery in a simulated environment before independently performing a CVC insertion on an actual patient. This is in sharp contrast to the traditional clinical model of procedural training at the bedside, and may be used in other training programs and with other invasive procedures.

The second aim of our study was to determine the impact of simulation‐based training on actual clinical practice by residents in the MICU. To our knowledge, no prior study has demonstrated that simulation‐based training in CVC insertion improves patient outcomes. We believe our results advance what is known about the impact of simulation‐based training because simulator‐trained residents in this study performed actual CVC insertions in the MICU using significantly fewer needle passes. Needle passes have been used by other investigators as a surrogate measure for reduced CVC‐associated complications because mechanical complications rise exponentially with more than two insertion attempts.47, 29 We believe this finding demonstrates transfer of skill acquired from simulation‐based training to the actual clinical environment. It is possible that ultrasound training accounts for the improvement in the simulator‐trained group. However, we do not believe that ultrasound training is entirely responsible as prior work has shown that deliberate practice using mastery learning without ultrasound significantly improved resident performance of thoracentesis11 and ACLS12, 19 procedures. We did not show a significant reduction in complications such as pneumothorax or arterial puncture. This is likely due to the small sample size and the low number of procedures and complications during the study period.

Our results also show that resident self‐confidence regarding actual CVC insertions improved after simulation training. These findings are similar to prior reports linking improved confidence among trainees after simulation‐based training in CVC insertion.29, 30 Our results did not reveal a correlation between improved self‐confidence and clinical skill acquisition. Linking improved self‐confidence to improved clinical skill is important because self‐assessment does not always correlate with performance ability.31, 32

More study is needed to evaluate the impact of simulation‐based training on the quality of CVC insertions by trainees. Mechanisms shown to decrease complications of CVC placement include use of ultrasound,4, 7, 3336 full sterile barrier technique,3739 chlorhexidine skin preparations,4042 and nurse‐physician education.43 Our simulation‐training program incorporates each of these elements. We plan to expand our simulation‐based training intervention to a larger sample size to determine its impact on mechanical and infectious complication rates linked to CVC insertion.

This study has several limitations. It was performed at a single institution over a short time period. However, demonstration of significantly fewer needle passes and improved resident self‐confidence after simulator training are important findings that warrant further study. It was impossible to blind raters during the skills assessment examination about whether the resident was performing a pretest or posttest. This was accounted for by using a second rater, who was blind to the pretest and posttest status of the examinee. The arterial puncture rate of 7% among simulator‐trained residents was higher than expected, although it remains within published ranges.4, 5 Also, a low total number of CVCs were evaluated during the study. This is likely due to strict exclusion criteria employed in order to study the impact of simulation training. For example, CVC insertions were only evaluated if they were actually performed by study residents (supervised insertions were excluded) and femoral catheters were not evaluated. We did not track clinical experience with CVC insertion by residents before the study. Residents who were simulator‐trained may have had more clinical experience with CVC insertion and this may have impacted their performance. However, residents did not differ in year of training or clinical rotations, and there is clear evidence that clinical training is not a proxy for skill acquisition.44 Finally, outcome data were measured via resident questionnaires that relied on resident recall about CVC insertion rather than observer ratings. This method was selected because observer ratings could not be standardized given the large number of clinical supervisors in the MICU over the study period. Information about needle passes and arterial puncture also may not be documented in procedural notes and could not be obtained by medical record review. We attempted to minimize recall bias by surveying residents within 24 hours of CVC placement.

In conclusion, this study demonstrates that simulation‐based training and deliberate practice in a mastery learning setting improves performance of both simulated and actual CVC insertions by internal medicine residents. Procedural training remains an important component of internal medicine training, although internists are performing fewer invasive procedures now than in years past.45, 46 Use of a mastery model of CVC insertion requires that trainees demonstrate skill in a simulated environment before independently performing this invasive procedure on patients. Further study is needed to assess clinical outcomes such as reduced CVC‐related infections and mechanical complications after simulation‐based training.

An 80‐year‐old man with coronary artery disease and chronic obstructive pulmonary disease (COPD) was admitted to an outside hospital after a mechanical fall. On presentation to the emergency room his systolic blood pressure was found to be 86/62 mm Hg. He complained of right flank, groin, and thigh pain. On physical exam, a hematoma extending from his right groin down to his right knee was found, as well as scattered ecchymoses involving his trunk and all 4 extremities. His hemoglobin was low, at 5.6 g/dL (14‐17 g/dL). A computed tomography (CT) scan revealed a right‐sided retroperitoneal bleed extending from the iliopsoas into his right thigh. The patient received 13 transfusions of packed red blood cells over the course of 9 days as he continued to bleed. Transfer to our facility for further workup and management ensued.

On serial testing at our institution his activated partial thromboplastin time (aPTT) was elevated at >160 seconds (normal range, 24‐36 seconds). Further coagulation parameters were found as follows: platelets 182.000/L; prothrombin time 17.6 seconds; international normalized ratio (INR) 1.4; thrombin 18 seconds; fibrinogen 778 mg/dL; and D‐dimer 3866 ng/mL. Of note, the patient had not received any medications known to potentially interfere with the measured aPTT. Because the source of his bleeding was not apparent at this point, disorders of primary hemostasis, including hereditary disease states (eg, von Willebrand disease), iatrogenic disorders (eg, drug‐induced), or acquired disorders, such as immune thrombocytopenia, were considered and ruled out. At this point the differential diagnoses had to be expanded, and secondary disorders of hemostasis were considered. A deficiency or decreased activity of coagulation factors was suspected. Whereas factor IX and XI were found to be normal, the factor VIII level was significantly decreased at 3% (50%‐150% being normal). This prompted an assay to check for the presence of a factor VIII inhibitor. It proved to be significantly elevated at 25.6 Bethesda Units (BU) (normal, 0.00‐0.04 BU). On that basis we arrived at the diagnosis of acquired factor VIII deficiency, but the etiology of such remained unclear to this point. Hence a search for the specific etiology of acquired factor IIII deficiency was launched, and connective tissue disease, as well as malignancy, was ruled out. While inflammatory bowel disease is a known potential cause for this condition, the clinical picture was not consistent with such and this diagnosis was not considered further. The patient received immunosuppressive therapy with prednisone 1 mg/kg orally per day. Rituximab and cyclophosphamide were considered, but due to bacteremia from bilateral parotitis, this was deferred. Of note, his bleeding abnormality was apparent prior to initiation of antibiotic therapy.

The bleeding stopped 2 days after initiation of treatment. At the time of discharge, 2 weeks after presentation, factor VIII inhibitor levels had decreased to 13 BU and his partial thromboplastin time (PTT) was 100 seconds.

Discussion

Acquired factor VIII inhibitor, also called acquired hemophilia A, is a rare, potentially life‐threatening bleeding disorder. It is caused by autoantibodies directed against coagulation factor VIII.1

The estimated incidence in the general population is 1 in 4 million/year. Risk factors include advanced age, pregnancy and the postpartum period, rheumatoid disease/connective tissue disease, inflammatory bowel disease, medications (especially antibiotics and psychiatric drugs), and malignancy. Both solid tumors as well as hematologic malignancies have been associated with acquired hemophilia A.2

Patients older than 85 years are more frequently affected. The annual incidence is 14.7 in 1 million in this age group. Hence, it is found rarely in young patients, but pregnancy and the postpartum period represent the exception.

Patients with acquired factor VIII inhibitor tend to bleed into the skin, soft tissue, muscle, brain, and mucous membranes. Most of the time, they present with epistaxis, retroperitoneal hematomas, or gastrointestinal bleeds, while patients with congenital factor VIII deficiency3 are more likely to bleed into the large joints. Acquired factor VIII inhibitor is associated with a high morbidity and mortality.

In the presence of an isolated elevated aPTT, once heparin has been ruled out, specific factor deficiencies and/or inhibitors need to be considered. The inhibitor assay helps to establish the diagnosis of acquired factor VIII deficiency and allows the quantification of factor VIII inhibitor. A search for specific etiologies of acquired factor VIII inhibitors should be undertaken; however, in 50% of cases no concomitant condition is found. The differential diagnoses should be expanded within the appropriate framework and tailored to the individual patient.

Control of bleeding might be achieved by factor VIII concentrate if the bleeding is mild. However, if the hemorrhage is life‐threatening, recombinant factor VII is frequently required to stop the bleeding.4 One has to be aware that recombinant factor VII may precipitate thromboembolic events and as such might pose a dilemma, as the degree of bleeding has to be balanced with the risk of unintended side effects. Therapy to eliminate factor VIII inhibitor is the combination of prednisone and cyclophosphamide, though monoclonal CD20 antibody (Rituximab) has become the first‐line agent in the appropriate setting.5 Risk and benefit of therapy have to be balanced with the severity of the bleed and potential unintended side effects of immunosuppression, especially in the presence of infection.

As hospitalists, we are challenged daily by a high degree of complexities in inpatient care. Hospitalists are well trained to manage a wide variety of conditions, and coagulopathies are no exception. They are so common in the inpatient setting that every hospitalist should be familiar with the basic principles of diagnosing and managing bleeding disorders. Because of the hospitalist's ability to promptly react, the consulting role of the hematologist can be reserved for the more unusual blood dyscrasias.

This article is intended to raise physician awareness for the discussed condition because early recognition and treatment are of paramount importance in patient outcome.

An 80‐year‐old man with coronary artery disease and chronic obstructive pulmonary disease (COPD) was admitted to an outside hospital after a mechanical fall. On presentation to the emergency room his systolic blood pressure was found to be 86/62 mm Hg. He complained of right flank, groin, and thigh pain. On physical exam, a hematoma extending from his right groin down to his right knee was found, as well as scattered ecchymoses involving his trunk and all 4 extremities. His hemoglobin was low, at 5.6 g/dL (14‐17 g/dL). A computed tomography (CT) scan revealed a right‐sided retroperitoneal bleed extending from the iliopsoas into his right thigh. The patient received 13 transfusions of packed red blood cells over the course of 9 days as he continued to bleed. Transfer to our facility for further workup and management ensued.

On serial testing at our institution his activated partial thromboplastin time (aPTT) was elevated at >160 seconds (normal range, 24‐36 seconds). Further coagulation parameters were found as follows: platelets 182.000/L; prothrombin time 17.6 seconds; international normalized ratio (INR) 1.4; thrombin 18 seconds; fibrinogen 778 mg/dL; and D‐dimer 3866 ng/mL. Of note, the patient had not received any medications known to potentially interfere with the measured aPTT. Because the source of his bleeding was not apparent at this point, disorders of primary hemostasis, including hereditary disease states (eg, von Willebrand disease), iatrogenic disorders (eg, drug‐induced), or acquired disorders, such as immune thrombocytopenia, were considered and ruled out. At this point the differential diagnoses had to be expanded, and secondary disorders of hemostasis were considered. A deficiency or decreased activity of coagulation factors was suspected. Whereas factor IX and XI were found to be normal, the factor VIII level was significantly decreased at 3% (50%‐150% being normal). This prompted an assay to check for the presence of a factor VIII inhibitor. It proved to be significantly elevated at 25.6 Bethesda Units (BU) (normal, 0.00‐0.04 BU). On that basis we arrived at the diagnosis of acquired factor VIII deficiency, but the etiology of such remained unclear to this point. Hence a search for the specific etiology of acquired factor IIII deficiency was launched, and connective tissue disease, as well as malignancy, was ruled out. While inflammatory bowel disease is a known potential cause for this condition, the clinical picture was not consistent with such and this diagnosis was not considered further. The patient received immunosuppressive therapy with prednisone 1 mg/kg orally per day. Rituximab and cyclophosphamide were considered, but due to bacteremia from bilateral parotitis, this was deferred. Of note, his bleeding abnormality was apparent prior to initiation of antibiotic therapy.

The bleeding stopped 2 days after initiation of treatment. At the time of discharge, 2 weeks after presentation, factor VIII inhibitor levels had decreased to 13 BU and his partial thromboplastin time (PTT) was 100 seconds.

Discussion

Acquired factor VIII inhibitor, also called acquired hemophilia A, is a rare, potentially life‐threatening bleeding disorder. It is caused by autoantibodies directed against coagulation factor VIII.1

The estimated incidence in the general population is 1 in 4 million/year. Risk factors include advanced age, pregnancy and the postpartum period, rheumatoid disease/connective tissue disease, inflammatory bowel disease, medications (especially antibiotics and psychiatric drugs), and malignancy. Both solid tumors as well as hematologic malignancies have been associated with acquired hemophilia A.2

Patients older than 85 years are more frequently affected. The annual incidence is 14.7 in 1 million in this age group. Hence, it is found rarely in young patients, but pregnancy and the postpartum period represent the exception.

Patients with acquired factor VIII inhibitor tend to bleed into the skin, soft tissue, muscle, brain, and mucous membranes. Most of the time, they present with epistaxis, retroperitoneal hematomas, or gastrointestinal bleeds, while patients with congenital factor VIII deficiency3 are more likely to bleed into the large joints. Acquired factor VIII inhibitor is associated with a high morbidity and mortality.

In the presence of an isolated elevated aPTT, once heparin has been ruled out, specific factor deficiencies and/or inhibitors need to be considered. The inhibitor assay helps to establish the diagnosis of acquired factor VIII deficiency and allows the quantification of factor VIII inhibitor. A search for specific etiologies of acquired factor VIII inhibitors should be undertaken; however, in 50% of cases no concomitant condition is found. The differential diagnoses should be expanded within the appropriate framework and tailored to the individual patient.

Control of bleeding might be achieved by factor VIII concentrate if the bleeding is mild. However, if the hemorrhage is life‐threatening, recombinant factor VII is frequently required to stop the bleeding.4 One has to be aware that recombinant factor VII may precipitate thromboembolic events and as such might pose a dilemma, as the degree of bleeding has to be balanced with the risk of unintended side effects. Therapy to eliminate factor VIII inhibitor is the combination of prednisone and cyclophosphamide, though monoclonal CD20 antibody (Rituximab) has become the first‐line agent in the appropriate setting.5 Risk and benefit of therapy have to be balanced with the severity of the bleed and potential unintended side effects of immunosuppression, especially in the presence of infection.

As hospitalists, we are challenged daily by a high degree of complexities in inpatient care. Hospitalists are well trained to manage a wide variety of conditions, and coagulopathies are no exception. They are so common in the inpatient setting that every hospitalist should be familiar with the basic principles of diagnosing and managing bleeding disorders. Because of the hospitalist's ability to promptly react, the consulting role of the hematologist can be reserved for the more unusual blood dyscrasias.

This article is intended to raise physician awareness for the discussed condition because early recognition and treatment are of paramount importance in patient outcome.

An 80‐year‐old man with coronary artery disease and chronic obstructive pulmonary disease (COPD) was admitted to an outside hospital after a mechanical fall. On presentation to the emergency room his systolic blood pressure was found to be 86/62 mm Hg. He complained of right flank, groin, and thigh pain. On physical exam, a hematoma extending from his right groin down to his right knee was found, as well as scattered ecchymoses involving his trunk and all 4 extremities. His hemoglobin was low, at 5.6 g/dL (14‐17 g/dL). A computed tomography (CT) scan revealed a right‐sided retroperitoneal bleed extending from the iliopsoas into his right thigh. The patient received 13 transfusions of packed red blood cells over the course of 9 days as he continued to bleed. Transfer to our facility for further workup and management ensued.

On serial testing at our institution his activated partial thromboplastin time (aPTT) was elevated at >160 seconds (normal range, 24‐36 seconds). Further coagulation parameters were found as follows: platelets 182.000/L; prothrombin time 17.6 seconds; international normalized ratio (INR) 1.4; thrombin 18 seconds; fibrinogen 778 mg/dL; and D‐dimer 3866 ng/mL. Of note, the patient had not received any medications known to potentially interfere with the measured aPTT. Because the source of his bleeding was not apparent at this point, disorders of primary hemostasis, including hereditary disease states (eg, von Willebrand disease), iatrogenic disorders (eg, drug‐induced), or acquired disorders, such as immune thrombocytopenia, were considered and ruled out. At this point the differential diagnoses had to be expanded, and secondary disorders of hemostasis were considered. A deficiency or decreased activity of coagulation factors was suspected. Whereas factor IX and XI were found to be normal, the factor VIII level was significantly decreased at 3% (50%‐150% being normal). This prompted an assay to check for the presence of a factor VIII inhibitor. It proved to be significantly elevated at 25.6 Bethesda Units (BU) (normal, 0.00‐0.04 BU). On that basis we arrived at the diagnosis of acquired factor VIII deficiency, but the etiology of such remained unclear to this point. Hence a search for the specific etiology of acquired factor IIII deficiency was launched, and connective tissue disease, as well as malignancy, was ruled out. While inflammatory bowel disease is a known potential cause for this condition, the clinical picture was not consistent with such and this diagnosis was not considered further. The patient received immunosuppressive therapy with prednisone 1 mg/kg orally per day. Rituximab and cyclophosphamide were considered, but due to bacteremia from bilateral parotitis, this was deferred. Of note, his bleeding abnormality was apparent prior to initiation of antibiotic therapy.

The bleeding stopped 2 days after initiation of treatment. At the time of discharge, 2 weeks after presentation, factor VIII inhibitor levels had decreased to 13 BU and his partial thromboplastin time (PTT) was 100 seconds.

Discussion

Acquired factor VIII inhibitor, also called acquired hemophilia A, is a rare, potentially life‐threatening bleeding disorder. It is caused by autoantibodies directed against coagulation factor VIII.1

The estimated incidence in the general population is 1 in 4 million/year. Risk factors include advanced age, pregnancy and the postpartum period, rheumatoid disease/connective tissue disease, inflammatory bowel disease, medications (especially antibiotics and psychiatric drugs), and malignancy. Both solid tumors as well as hematologic malignancies have been associated with acquired hemophilia A.2

Patients older than 85 years are more frequently affected. The annual incidence is 14.7 in 1 million in this age group. Hence, it is found rarely in young patients, but pregnancy and the postpartum period represent the exception.

Patients with acquired factor VIII inhibitor tend to bleed into the skin, soft tissue, muscle, brain, and mucous membranes. Most of the time, they present with epistaxis, retroperitoneal hematomas, or gastrointestinal bleeds, while patients with congenital factor VIII deficiency3 are more likely to bleed into the large joints. Acquired factor VIII inhibitor is associated with a high morbidity and mortality.

In the presence of an isolated elevated aPTT, once heparin has been ruled out, specific factor deficiencies and/or inhibitors need to be considered. The inhibitor assay helps to establish the diagnosis of acquired factor VIII deficiency and allows the quantification of factor VIII inhibitor. A search for specific etiologies of acquired factor VIII inhibitors should be undertaken; however, in 50% of cases no concomitant condition is found. The differential diagnoses should be expanded within the appropriate framework and tailored to the individual patient.

Control of bleeding might be achieved by factor VIII concentrate if the bleeding is mild. However, if the hemorrhage is life‐threatening, recombinant factor VII is frequently required to stop the bleeding.4 One has to be aware that recombinant factor VII may precipitate thromboembolic events and as such might pose a dilemma, as the degree of bleeding has to be balanced with the risk of unintended side effects. Therapy to eliminate factor VIII inhibitor is the combination of prednisone and cyclophosphamide, though monoclonal CD20 antibody (Rituximab) has become the first‐line agent in the appropriate setting.5 Risk and benefit of therapy have to be balanced with the severity of the bleed and potential unintended side effects of immunosuppression, especially in the presence of infection.

As hospitalists, we are challenged daily by a high degree of complexities in inpatient care. Hospitalists are well trained to manage a wide variety of conditions, and coagulopathies are no exception. They are so common in the inpatient setting that every hospitalist should be familiar with the basic principles of diagnosing and managing bleeding disorders. Because of the hospitalist's ability to promptly react, the consulting role of the hematologist can be reserved for the more unusual blood dyscrasias.

This article is intended to raise physician awareness for the discussed condition because early recognition and treatment are of paramount importance in patient outcome.