Emergency Physicians’ Fear of Malpractice in Evaluating Patients With Possible Acute Cardiac Ischemia

Abstract

Study objective:

We evaluate the association between emergency physicians’ fear of malpractice and the triage and evaluation patterns of patients with symptoms suggestive of acute coronary syndrome.

Methods:

We surveyed 33 emergency physicians of 2 university hospitals during the preintervention phase of an implementation trial of the Agency for Health Care Policy and Research Unstable Angina guideline in 1,134 study patients. The survey included a 6-item instrument that addressed concerns about malpractice and a measure of general risk aversion. We used hierarchical logistic regression to model emergency department (ED) triage decisions and diagnostic testing as a function of fear of malpractice, with adjustment for patient characteristics, Agency for Health Care Policy and Research guideline risk group, study site, and clustering by emergency physician.

Results:

Overall, emergency physicians in the upper tertile of malpractice fear were less likely to discharge low-risk patients compared with emergency physicians in the lower tertile (adjusted odds ratio [OR] 0.34; 95% confidence interval [CI] 0.12 to 0.99; P=.05). Patients treated by emergency physicians in this group were also more likely to be admitted to an ICU or telemetry bed (adjusted OR 1.7; 95% CI 1.2 to 2.4). In addition, emergency physicians in the upper tertile of malpractice fear were more likely to order chest radiography, as well as cardiac troponin. Malpractice fear accounted for a similar amount of variance after controlling for emergency physicians’ risk aversion.

Conclusion:

Malpractice fear accounts for significant variability in ED decisionmaking and is associated with increased hospitalization of low-risk patients and increased use of diagnostic tests. [Ann Emerg Med. 2005;46:525–533.]

INTRODUCTION

Background

Of the approximately 7 million people who present to US emergency departments (EDs) annually with chest pain or other symptoms of possible acute coronary syndrome (which includes unstable angina and acute myocardial infarction),¹ approximately 50% are hospitalized or admitted to chest pain observation units, and the majority of these patients are subsequently shown not to have acute coronary syndrome. Conversely, approximately 2% of ED patients with confirmed acute myocardial infarction and a similar proportion of patients with a diagnosis of unstable angina are inappropriately sent home.² The quest for diagnostic certainty in evaluating patients with possible acute coronary syndrome is driven by the fact that missed acute myocardial infarction has immediate life-threatening consequences and that missed acute myocardial infarction is among the most frequent causes of medical malpractice in adult medicine.³

The extent to which defensive medicine (ie, the use of diagnostic and treatment measures primarily for the purpose of averting malpractice lawsuits) drives emergency physicians’ triage decisions in patients with possible acute coronary syndrome is unclear. Some physician surveys have suggested that the practice of defensive medicine is not extensive and is probably not a major contributory cause of escalating health care costs.^4–6 In a survey of 162 cardiologists, only 4.4% cited malpractice concerns as the primary reason for choosing to admit a hypothetical patient with atypical chest pain.⁷ One possible explanation for this finding is the tendency of physicians to underestimate the influence of malpractice concerns on their use of diagnostic tests; in effect, they fail to acknowledge the degree to which medical standards have been influenced by fear of liability.^7,8 Another survey, however, suggested that physicians with a higher perceived risk of a lawsuit were significantly more likely to order tests and procedures.⁹

Practices aimed at avoiding malpractice liability may vary considerably among physicians, based on training, experience, and ability to tolerate risk and uncertainty. For example, some physicians have risk-seeking personalities (eg, adventurous, unconcerned with danger) and others have risk-avoiding personalities (eg, cautious, security minded). In the case of acute coronary syndrome, some of the variation in ED triage decisions can be attributed to differences in risk aversion.⁵ There are few empiric data from clinical practice, however, that specifically relate fear of malpractice to observed practice behavior in acute coronary syndrome (or other clinical syndromes commonly observed in ED). Thus, the purpose of this study is to evaluate the association between fear of malpractice and emergency physician triage and evaluation decisions for patients with symptoms of possible acute coronary syndrome.

MATERIALS AND METHODS

Study Design

We surveyed emergency physicians who participated in an implementation trial of the Agency for Healthcare Policy and Research Unstable Angina guideline, conducted between January 2000 and May 2001. Because this trial showed no significant impact of the guideline intervention on emergency physician triage decisions,¹⁰ we pooled data from the pre- and postintervention periods for analysis.

Setting

The study sites included 1 university hospital ED (17 beds, 33,000 visits per year) and 1 university-affiliated community hospital ED (23 beds, 56,000 visits per year). Both sites provided primary and tertiary care. Internal medicine residents rotated in both EDs; 1 hospital also trained emergency medicine residents.

Selection of Participants

The study sample included 1,134 consecutive ED patients who presented with symptoms of possible acute coronary syndrome (chest pain or left arm discomfort, shortness of breath, nausea or vomiting, epigastric pain, dizziness).¹⁰ These symptoms have been shown to capture more than 90% of patients with acute coronary syndrome in community-based studies.¹¹ This project was approved by the institutional research board at each study site.

Data Collection and Processing

After the emergency physician’s initial evaluation (and after obtaining informed consent), a research technician interviewed study patients to assess symptoms leading to ED presentation and pertinent cardiovascular history and abstracted data on physical findings and test results (creatine-kinase MB [CK-MB], troponin-I) from ED records.¹² We also calculated the Acute Cardiac Ischemia-Time Insensitive Predictive Instrument (ACI-TPI) predicted probability of cardiac ischemia based on clinical and electrocardiographic data.¹³ Each patient was classified according to Agency for Health Care Policy and Research guideline risk category (low, intermediate, or high risk of adverse outcomes), using the criteria shown in Appendix 1.¹⁴ The guideline recommends that emergency physicians consider outpatient management for low-risk patients and admission to a monitored (ICU or telemetry) bed for intermediate- and high-risk patients.¹⁵ The prognostic validity of the Agency for Health Care Policy and Research risk groups has been previously demonstrated.^14,16,17

A written survey was completed by emergency physicians before onset of the study intervention. The survey, which took approximately 5 minutes to complete, included instruments that addressed overall risk aversion and fear of malpractice. Data on emergency physicians’ demographics and training were obtained from departmental records. Each survey was reviewed to ensure that all items were answered. Details about the survey are provided below.

The malpractice fear scale was developed by one of the authors (GCW) and includes 6 items, which were selected based on expert opinion and factor analysis (Appendix 2).¹⁸ Validity of the scale is supported by the finding that patients treated by primary care physicians with higher levels of malpractice fear were more likely to be referred to specialists.¹⁹ Cronbach’s α, which measures the internal consistency reliability of a summative rating scale, was 0.88 in the current study. Items were rated on a 5-point Likert scale, and the sum of responses was calculated. Possible scores ranged from 6 to 30, with higher scores corresponding to increased malpractice fear (Appendix 3 includes item-total and interitem correlations). Emergency physicians were split into tertiles (corresponding to high, moderate, and low malpractice fear).

The 6-item risk-taking scale measure was derived from the risk-taking subscale of the Jackson Personality Index; Cronbach’s α was reported to be 0.71.⁵ Each item was rated on a 6-point Likert scale, and the sum of responses was calculated (possible scores ranged from 6 to 36). In one ED study that used this measure, risk-avoiding emergency physicians were significantly more likely to admit patients with chest pain than risk-tolerant emergency physicians.⁵

To assess emergency physicians’ attitudes toward practice guidelines in general, we used an 8-item measure developed by the American College of Physicians Clinical Efficacy Assessment Subcommittee.²⁰ Each item of this measure was rated on a 5-point Likert scale, and the sum of responses was calculated.

Outcome Measures

Emergency physicians’ triage decisions were abstracted from ED records. We examined 2 measures of diagnostic evaluation in the ED: cardiac troponin and chest radiography. Admission to a cardiology service was determined from hospital billing data.

A final diagnosis of acute coronary syndrome was determined by physician reviewers (blinded to study hypotheses) based on review of all available data, including ED and hospital records, outpatient visits and procedures, and a structured telephone interview at 30-day follow-up; vital status for patients who were lost to follow-up was determined by searching state death records. The diagnosis of acute myocardial infarction was based on current European Society of Cardiology/American College of Cardiology criteria.²¹ The diagnosis of unstable angina was based on clinical history plus objective evidence of ischemia on diagnostic testing. Evidence of ischemia included the following: dynamic ST-T-wave changes noted during acute ischemic symptoms or a positive noninvasive test result (eg, reversible ischemia on exercise or pharmacologic stress testing).¹² We also considered the presence of a 70% or greater occlusion in at least 1 vessel (or left main disease with >50% occlusion) on coronary angiography to be suggestive of an ischemic etiology.¹⁵ A 20% subsample of records was reviewed independently by a second reviewer. Interrater agreement of physician reviewers in the diagnosis of confirmed acute coronary syndrome was 87% (κ=0.73).

Primary Data Analysis

We report differences in triage and diagnostic testing decisions between tertiles of malpractice fear; 95% CIs the differences are based on the normal distribution (using 2 sample tests of proportions).

In our primary analysis, we used hierarchical logistic regression to model the decision to discharge ED patients as a function of the malpractice fear score in the entire study sample and in the subgroup of Agency for Health Care Policy and Research low-risk patients (Appendix 4). We similarly modeled the decision to admit patients to a monitored bed in the entire study sample and in the subgroup of patients with Agency for Health Care Policy and Research intermediate- to high-risk characteristics. In secondary analyses, we examined the relationship between malpractice fear score and the decision to obtain cardiac troponin (a recommended test) and the decision to obtain chest radiography (which is ordered if believed to be clinically indicated)²² in the entire study sample. We report the odds ratio (OR) and 95% confidence interval (CI) associated with membership in the middle and upper tertiles of malpractice fear score (using the lower tertile as the reference category).

We adjusted for patient covariates (age, sex, race, insurance type; chief complaint of chest pain, history of angina, acute myocardial infarction, stroke, or previous revascularization; history of hypertension, diabetes, high cholesterol, current smoking, family history of premature coronary artery disease; guideline risk group, and cardiac troponin), study site, study period (intervention versus control), and clustering by emergency physician. To determine whether risk aversion accounted for the association between malpractice fear score and triage or evaluation decisions, we repeated the above analyses with addition of the risk-taking score to the model.

To assess diagnostic accuracy across tertiles of malpractice fear, we compared the proportion of patients with confirmed acute coronary syndrome whom the emergency physician suspected of having probable acute coronary syndrome (based on an ED diagnosis of acute myocardial infarction, unstable angina, rule-out myocardial infarction, or unspecified angina). We also assessed whether less fearful emergency physicians were more likely to discharge patients with confirmed acute coronary syndrome from the ED (although statistical power to detect a difference in this measure was low).

Statistical analyses were performed using Stata, version 7.0 (Stata Corporation, College Station, TX) and MLwiN software (Centre for MultiLevel Modeling, London, UK) for hierarchical modeling.²³

RESULTS

Characteristics of Study Subjects

All 33 emergency physicians at both study sites completed the survey. Emergency physician characteristics were similar across tertiles of malpractice fear score (Table 1). Patients treated by emergency physicians in the 3 fear-score groups were comparable, except that a chief complaint of chest pain was found less often among patients treated by emergency physicians in the high-fear group compared with the middle tertile (difference −9%; 95% CI −15 to −3).

Table 1.

Emergency physician and patient characteristics, grouped by tertile of malpractice fear score.^*

	Malpractice Fear Score
Emergency Physician Characteristics	Lower (N=10)	Middle (N=11)	Upper (N=12)
Age, mean (SD)	43.7 (8)	37.7 (6)	39.7 (6)
Sex, male, %	90	91	92
Board-certified (emergency medicine), %	100	73	83
Years in clinical practice, mean (SD)	16 (8)	10 (5)	13 (6)
Full time (clinical), %	40	82	58
Attitude toward guidelines, mean (SD) †	22 (6)	22 (3)	21 (4)
Risk-taking scale, mean (SD) ‡	20 (7)	21 (6)	21 (4)
Patient characteristics	(N=257)	(N=371)	(N=506)
Chief complaint: chest pain, %	63	67	58
Prolonged ischemic symptoms at rest, %§	37	36	32
History of myocardial infarction, %	19	16	17
Abnormal baseline ECG, %ǁ	21	17	22
Abnormal cardiac enzyme levels, %ǁ	10	12	13
Intermediate- to high-risk characteristics, %¶	61	58	64
Probability of acute cardiac ischemia (%), mean (SD)¶	20 (14)	20 (15)	22 (17)

^*Tertiles of malpractice fear score were defined as follows: lower (<15), middle (≥15 to <20), and upper (≥20). The number of emergency physicians (and the number of study patients) in each tertile is shown in parentheses.

^†On a scale of 8–40, with higher scores indicating a more favorable attitude.

^‡On a scale of 6–36, with higher scores indicating a more risk-taking attitude.

^§Ongoing symptoms at ED presentation (>20 minutes in duration).

^ǁAbnormal baseline ECG indicates 1 or more of the following: pathologic q-waves, ST segment depression ≥0.5 mm, ST elevation ≥1 mm, or T-wave inversion ≥1 mm in at least 2 contiguous leads, bundle-branch block, or left ventricular hypertrophy.

^¶Abnormal values for troponin-I and CK-MB index were defined by each hospital’s clinical laboratory.

Main Results

To verify that our grouping strategy identified emergency physicians with distinctly different attitudes toward malpractice, we examined the pattern of responses to individual items on the malpractice fear scale (Table 2). As expected, emergency physicians in the upper tertile were more likely than those in the lower tertile to respond affirmatively to each item on the scale. There was no significant correlation between the malpractice fear score and general risk aversion (Spearman’s ρ=.01).

Table 2.

Emergency physician responses to items on the malpractice fear scale, grouped by tertile of summary score.^*

Item	Lower (N=10)	Middle (N=11)	Upper (N=12)
Significant changes in practice pattern for medicolegal reasons	0 (1.7)	9 (2.2)	42 (3.1)
Concern about malpractice case in the next 10 years	30 (2.7)	64 (3.5)	100 (4.6)
Pressure in daily practice from threat of malpractice	0 (1.8)	9 (2.5)	83 (4.1)
Ordering tests/consultations to avoid malpractice	0 (1.9)	45 (3.2)	58 (3.7)
Ask for consultant opinions to reduce risk of lawsuit	0 (1.7)	27 (2.7)	75 (3.9)
Relying on clinical judgment rather than technology is risky	20 (2.5)	45 (3.4)	100 (4.5)

^*Tertiles of malpractice fear score were defined as follows: lower (<15), middle (≥15 to <20), and upper (≥20). The percentage of emergency physicians who responded affirmatively (agree or strongly agree) is shown; the mean value for each item is shown in parentheses (on a scale of 1–5, where 1=strongly disagree and 5=strongly agree).

Our primary hypothesis was confirmed by the finding that patients treated by emergency physicians in the upper tertile were significantly less likely to be discharged than patients treated by emergency physicians in the lower tertile of malpractice fear (42% versus 52%; adjusted OR 0.56; 95% CI 0.4 to 0.9) (Tables 3 and 4). When this analysis was restricted to low-risk patients (for whom the Agency for Health Care Policy and Research guideline recommends outpatient management), only 64% of patients treated by emergency physicians in the upper tertile were discharged from the ED compared with 72% and 81% in the middle and lower tertiles (adjusted OR 0.34; 95% CI 0.12 to 0.99 for the contrast between upper and lower tertiles). The adjusted odds ratio for malpractice fear was virtually identical with addition of general risk aversion to the model (Table 4).

Table 3.

Unadjusted patient outcomes, grouped by tertile of emergency physician’s malpractice fear score.^*

Variable	Lower (n=257)	Middle (n=371)	Upper (n=506)	Difference, % (95% CI)
				Middle-Lower	Upper-Lower
Triage decisions
Discharged from ED, %	51.8	53.9	42.2	2 (−6 to 10)	−9 (−17 to −2)
AHCPR low-risk patients only, % (No.)	80.5 (41)	72.0 (50)	64.3 (84)	−9 (−26 to 9)	−16 (−32 to 0)
Admitted to monitored bed, %	42.0	43.7	51.2	2 (−6 to 10)	9 (2–17)
AHCPR intermediate-high risk patients only, % (No.)	48.6 (216)	47.0 (321)	55.0 (422)	−2 (−10 to 7)	6 (−2 to 15)
Admission to cardiology service, % (No.)†	59.5 (121)	70.6 (163)	64.6 (277)	11 (0–22)	5 (−5 to 16)
Diagnostic tests ordered in ED, %
Troponin-I	74.3	79.3	80.4	5 (−2 to 12)	6 (0–12)
Chest radiograph	73.5	80.0	85.8	7 (0–13)	12 (6–18)

AHCPR, Agency for Health Care Policy and Research.

^*Tertiles of malpractice fear score were defined as follows: lower (≤15), middle (≥15 to <20), and upper (≥20). For all subgroup analyses, the number of patients within each tertile of malpractice fear score is shown in parentheses.

^†Applies to hospitalized patients only.

Table 4.

Multivariable-adjusted odds ratios (with 95% CIs) for middle and upper tertiles of malpractice fear score. Emergency physicians in the lower tertile are the reference category.

Outcome	Base Model*		Base Model+Risk-Taking Score†
	Middle Tertile	Upper Tertile	Middle Tertile	Upper Tertile
Discharge from ED	0.67 (0.44–1.03)	0.56 (0.36–0.89)	0.65 (0.42–1.01)	0.56 (0.36–0.88)
AHCPR low-risk patients only‡	0.20 (0.06–0.61)	0.34 (0.12–0.99)	0.24 (0.08–0.78)	0.35 (0.12–1.05)
Admission to monitored bed	1.6 (1.2–2.3)	1.7 (1.2–2.4)	1.6 (1.2–2.3)	1.7 (1.2–2.4)
AHCPR intermediate- to high-risk patients only§	1.4 (1.0–2.1)	1.4 (0.96–2.1)	1.5 (1.0–2.1)	1.4 (0.96–2.1)
Admission to cardiology service	0.92 (0.56–1.5)	1.4 (0.8–2.3)	0.84 (0.52–1.4)	1.4 (0.8–2.3)
Troponin-I ordered in ED	1.8 (1.2–2.7)	1.9 (1.2–2.9)	1.7 (1.1–2.6)	1.9 (1.2–2.8)
Chest radiograph ordered in ED	1.6 (1.0–2.5)	2.0 (1.3–3.3)	1.6 (1.0–2.5)	2.0 (1.3–3.3)

^*Hierarchical logistic regression models with tertile of malpractice fear as the independent variable (adjusted for patient age, sex, race, insurance type, chest pain as chief complaint, history of myocardial infarction, history of angina, previous revascularization, history of stroke, hypertension, smoking status, diabetes, hypercholesterolemia, family history of premature coronary artery disease, abnormal troponin, AHCPR Unstable Angina guideline risk group, site, and study period). Abnormal troponin was dropped from the model predicting “Troponin-I ordered in ED.”

^†Risk-taking score was added to the base model as a physician-level covariate.

^‡N=175 for this subgroup analysis. Abnormal troponin was dropped from this analysis because of a zero cell (failure of model to converge).

^§N=955 for this subgroup analysis.

We also found that patients treated by emergency physicians in the high-fear group were more likely to be admitted to a monitored bed than patients treated by emergency physicians in the low-fear group (51% versus 42%; adjusted OR 1.7; 95% CI 1.2to2.4).When this analysis was restricted to intermediate to high-risk patients, this association was attenuated (adjusted OR 1.4; 95% CI 0.96 to 2.1 for the contrast between upper and lower tertiles). Hierarchical models showed no significant differences in the proportion of patients admitted to a cardiology service across tertiles of malpractice fear.

In analyses of test-ordering decisions, 80% of patients treated by emergency physicians in the upper tertile received troponin testing compared with 74% of patients treated by emergency physicians in the lower tertile of malpractice fear (adjusted OR 1.9; 95% CI 1.2 to 2.9) (Tables 3 and 4). These patients were also more likely to receive chest radiography (adjusted OR 2.0; 95% CI 1.3 to 3.3). These results were virtually identical after adjustment for general risk aversion (Table 4).

Of patients with confirmed acute coronary syndrome, the proportion of patients with ED diagnoses indicating low suspicion of acute coronary syndrome was similar across tertiles of malpractice fear (35%, 35%, and 31% in the lower, middle, and upper tertiles, respectively). Only 7 patients in the entire study sample had “missed acute coronary syndrome”; however, there was no tendency for less fearful emergency physicians to discharge these patients from the ED (3%, 9%, and 3% for lower, middle, and upper tertiles, respectively).

LIMITATIONS

Limitations of this study deserve comment. First, the study sample was composed of a relatively small number of emergency physicians at 2 teaching hospitals. Of note, however, this study sample included emergency physicians across the spectrum of malpractice fear and patients evaluated by these emergency physicians in actual practice. Second, we did not collect information on the emergency physicians’ history of being sued for malpractice. This factor, however, was not associated with increased ordering of tests and procedures in another study.⁹ Third, the observed associations between malpractice fear and ED decisions could be attributable to differences in the frequency of chest pain because a greater proportion of atypical symptoms in the high-fear tertile might have prompted emergency physicians to order additional diagnostic tests (eg, chest radiography) and to admit patients for observation. Our findings remained robust after statistical adjustment for chest pain and other clinical covariates. Fourth, it is possible that other unmeasured emergency physician attitudes or motivations that covary with fear of malpractice (such as discomfort with uncertainty)²⁴ could account for these differences. Finally, this study focuses only on the influence of malpractice fear on physician behavior but does not address whether a more defensive practice style prevents negligent injuries (eg, missed acute myocardial infarction).

DISCUSSION

Our findings indicate that fear of malpractice is associated with a defensive practice style in the ED and accounts for significant variability in triage and testing decisions. In particular, low-risk patients treated by emergency physicians in the upper tertile of malpractice fear were significantly less likely to be discharged from the ED than patients treated by emergency physicians in the lower tertile, although these patients can generally be safely treated as outpatients.¹⁵ One possible explanation for this finding is that emergency physicians with increased fear of malpractice overestimate the probability of acute ischemia, particularly in low-risk patients. This explanation is consistent with “value-induced” bias, which occurs when a physician assigns a diagnosis with serious consequences a greater subjective probability than it deserves because it is critical to exclude that diagnosis.^25,26 On the other hand, these same emergency physicians tended to be more diligent in admitting intermediate-high risk patients to monitored beds and were more likely to order cardiac troponin to guide risk stratification, as supported by current evidence.^27,28

Our findings are consistent with work that has shown a positive association between fear of malpractice and physician behavior.^6,19,29 With regard to diagnostic testing, physicians respond to the perceived threat of litigation by ordering more referrals and more tests, some of which may be highly indicated and beneficial, but others might be wasteful or even harmful.⁷ In surveys, between 20% and 81% of physicians reported having ordered more tests or procedures in response to increasing risk of professional liability.⁶ In a national study of neurologists, greater malpractice concern was associated with significantly greater test ordering in 2 of 4 scenarios presented.²⁹

Although some physicians justify defensive testing on the grounds that such testing correctly diagnoses additional patients with disease, others argue that such testing leads to potential harm when nondiseased patients with false-positive diagnoses receive unnecessary or inappropriate treatment. Using a decision-analytic framework, DeKay and Asch³⁰ suggest that defensive testing may reduce overall quality of care because it leads to decreased expected utility compared to what one would attain based solely on consideration of the patient’s interests. The problem is likely exacerbated by physicians’ tendency to overestimate the risk of being sued.^9,29

In summary, malpractice fear accounts for significant variability in ED decisionmaking that is not explained by general risk aversion and is associated with increased hospitalization and use of diagnostic tests in patients with possible acute coronary syndrome. Interventions to lower emergency physicians’ actual and perceived risk of lawsuits could be expected to reduce the practice of defensive medicine and its associated health care costs. Use of computerized decision-support tools, such as the ACI-TPI, may reduce the likelihood of a malpractice claim coming to litigation in cases of missed acute coronary syndrome.³¹ Quality-improvement initiatives that incorporate standardized protocols, preprinted orders, physician education, and interactive feedback also have potential to improve initial management of ED patients with ACS.^32,33 and subsequent risk of liability. For quality improvement to reach its full potential, however, health care organizations need to foster an atmosphere that is conducive to open disclosure of mistakes to shift the emphasis from clinicians’ negligence to the faulty systems in which clinicians work.³⁴

In Retrospect

Future research should aim to replicate our findings in other clinical problems and in other practice settings (eg, nonteaching community and rural hospitals) and to explore the relationship between malpractice fear and other important aspects of physician behavior (including attitudes toward uncertainty and job satisfaction). Asking emergency physicians to estimate the probability of acute coronary syndrome at triage would also help to clarify whether fear of malpractice is associated with systematic overestimation of risk (and value-induced bias). Future studies should prospectively evaluate strategies to reduce the practice of defensive medicine for their effects on physician behavior and patient outcomes. The malpractice fear scale captures a unique dimension of emergency physician decisionmaking that is not explained by the physician’s attitude toward risk in general and should be considered in intervention studies that aim to change practice behavior in the ED.

Editor’s Capsule Summary.

What is already known on this topic

Some of the variation in decisions made by emergency physicians when evaluating patients with chest pain can be attributed to differences in the physicians’ aversion to taking risks. Surveys also suggest that physicians with a higher perceived risk of lawsuit are more likely to order tests and procedures.

What question this study addressed

Is there an association between fear of malpractice (as distinct from general risk aversion) and emergency physician assessment and disposition of emergency department (ED) patients with possible acute coronary syndrome?

What this study adds to our knowledge

In 33 physicians treating 1,134 patients, those with a higher malpractice fear were more likely to order troponin-I measurements (6% greater) and chest radiographs (12% greater) and had a higher hospital admission rate (9% greater) than physicians with a lower malpractice fear, even after correction for patient presentation and physician personal risk-avoidance traits.

How this might change clinical practice

Identification and modification of physician malpractice fear should help reduce variations in care, which is believed to lead to better patient outcomes.

Research we’d like to see

Do the differences in patient assessment and disposition between low- versus high-malpractice-fear physicians result in any difference in patient outcomes? Can these fears be addressed in a way that helps standardize practice?

Appendix 1.

Agency for Health Care Policy and Research unstable angina guideline risk groups.

Risk of Complications	Criteria
High	At least 1 of the following features must be present:
	Prolonged ongoing (>20 minutes) rest pain
	Pulmonary edema, most likely related to ischemia
	Angina at rest with dynamic ST changes ≥1 mm
	Angina with new or worsening mitral regurgitation murmur, S3, new or worsening rales, or hypotension
Intermediate	No high-risk feature but must have any of the following:
	Prolonged ongoing (>20 minutes) rest angina, now resolved, with moderate-to-high likelihood of coronary artery disease
	Rest angina (>20 minutes or relieved with rest or sublingual nitroglycerin)
	Nocturnal angina
	Angina with dynamic T-wave changes
	New-onset Canadian Cardiovascular Society Classification grade III or IV angina in the past 2 weeks
	Pathologic Q waves or resting ST depression ≤1 mm in multiple-lead groups (anterior, inferior, lateral)
	Age >65 years
Low	No high- or intermediate-risk features but may have any of the following features:
	Increased angina frequency, severity, or duration
	Angina provoked at a lower threshold
	New-onset angina with onset 2 weeks to 2 months before presentation
	Normal or unchanged ECG

Appendix 2.

Fear of malpractice scale.

Please complete the remaining section from the perspective of an emergency physician treating a full range of patients, not just patients with coronary artery disease. Indicate the extent to which you agree or disagree with the following statements.

Item	Strongly Disagree	Disagree	Not Sure	Agree	Strongly Agree
1. I have had to make significant changes in my practice pattern because of recent legal developments concerning medical delivery.	1	2	3	4	5
2. I am concerned that I will be involved in a malpractice case sometime in the next 10 years.	1	2	3	4	5
3. I feel pressured in my day-to-day practice by the threat of malpractice litigation.	1	2	3	4	5
4. I order some tests or consultations simply to avoid the appearance of malpractice.	1	2	3	4	5
5. Sometimes I ask for consultant opinions primarily to reduce my risk of being sued.	1	2	3	4	5
6. Relying on clinical judgment rather than on technology to make a diagnosis is becoming riskier from a medicolegal perspective.	1	2	3	4	5

Appendix 3.

Interitem correlation matrix and item-total correlations for the fear of malpractice scale (n=33).^*

Item	1	2	3	4	5	6
1. Change in practice because of legal factors	1.0000
2. Concern about future malpractice suit	0.4213	1.0000
3. Practice pressure related to malpractice	0.6926	0.6195	1.0000
4. Ordering tests/consultations to avoid malpractice	0.3553	0.4270	0.5479	1.0000
5. Asking for consultant opinions to reduce risk	0.3240	0.4756	0.5543	0.7749	1.0000
6. Relying on clinical judgment for diagnosis	0.5365	0.5453	0.7224	0.5324	0.6337	1.0000
Item			Item-Total Correlation			Average Interitem Correlation
1. Change in practice because of legal factors		0.7047			0.5833
2. Concern about future malpractice suit		0.7383			0.5674
3. Practice pressure related to malpractice		0.8755			0.5026
4. Ordering tests/consultations to avoid malpractice		0.7698			0.5525
5. Asking for consultant opinions to reduce risk		0.7963			0.5400
6. Relying on clinical judgment for diagnosis		0.8403			0.5192

^*Cronbach’s α = 0.88. Full text of questionnaire items is shown in Appendix 2.

Appendix 4. Hierarchical model.

Because individual patients are naturally grouped under individual emergency physicians, we constructed 2-level hierarchical logistic regression models of triage and test-ordering decisions.³⁵ In a model of ED discharge, for example, y_ij denotes the dependent variable for the ith patient under the care of the jth emergency physician and has a value of 1 for patients who were discharged from the ED and 0 for all others. Because we are analyzing a 2-level model, we denote the fixed part of the model by f_ij and a random error variable by r_j (emergency physician level). If π_ij is the probability of patients being discharged from the ED, our 2-level logistic model may be defined as

$$\mathrm{logit}{\pi }_{ij}=\ln \frac{{\pi }_{ij}}{\left(1-{\pi }_{ij}\right)}=f_{\mathrm{ij}}+r_j$$ (1)

1. This can be rewritten and incorporated into a full model, including residual random error terms at the patient level (e_ij), as follows:

   $$y_{ij}={\pi }_{ij}+e_{ij}=\frac{\exp \left(f_{ij}+r_j\right)}{1+\exp \left(f_{ij}+r_j\right)}+e_{ij}$$ (2)

2. The result of the functional relationship in model (2) is the probability of the patient being discharged from the ED conditional on the fixed and random variables from all levels of information. Based on this model, we estimated the odds ratio (and 95% confidence interval) for the middle and upper tertiles of malpractice fear at the emergency physician level (with the low tertile as the reference category), patient characteristics known to affect triage decisions (see Materials and Methods), and study design variables (study site, intervention versus baseline period).