[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.159.197.114. Please contact the publisher to request reinstatement.
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Download PDF
Figure.
Implicit and Explicit Race and Social Class Preferences Among Practitioners
Implicit and Explicit Race and Social Class Preferences Among Practitioners

Correlations are reported as Spearman rank correlation coefficients.

Table 1.  
Participant Demographic Characteristicsa
Participant Demographic Characteristicsa
Table 2.  
Scores for Single-Facility Practitioners
Scores for Single-Facility Practitioners
Table 3.  
Stratified Examination of Implicit Bias for Vignettes With Discordant Responses Based on Patient Race or Social Classa
Stratified Examination of Implicit Bias for Vignettes With Discordant Responses Based on Patient Race or Social Classa
1.
Haider  AH, Scott  VK, Rehman  KA,  et al.  Racial disparities in surgical care and outcomes in the United States: a comprehensive review of patient, provider, and systemic factors. J Am Coll Surg. 2013;216(3):482-92.e12.
PubMedArticle
2.
Haider  AH, Weygandt  PL, Bentley  JM,  et al.  Disparities in trauma care and outcomes in the United States: a systematic review and meta-analysis. J Trauma Acute Care Surg. 2013;74(5):1195-1205.
PubMedArticle
3.
Hicks  CW, Hashmi  ZG, Velopulos  C,  et al.  Association between race and age in survival after trauma. JAMA Surg. 2014;149(7):642-647.
PubMedArticle
4.
Hernandez  RA, Hevelone  ND, Lopez  L, Finlayson  SR, Chittenden  E, Cooper  Z.  Racial variation in the use of life-sustaining treatments who die after major elective surgery [published online October 13, 2014]. Am J Surg. doi:10.1016/j.amjsurg.2014.08.025.
PubMed
5.
Sukumar  S, Ravi  P, Sood  A,  et al.  Racial disparities in operative outcomes after major cancer surgery in the United States. World J Surg. 2015;39(3):634-643.
PubMedArticle
6.
Artinyan  A, Mailey  B, Sanchez-Luege  N,  et al.  Race, ethnicity, and socioeconomic status influence the survival of patients with hepatocellular carcinoma in the United States. Cancer. 2010;116(5):1367-1377.
PubMedArticle
7.
Nguyen  LL, Hevelone  N, Rogers  SO,  et al.  Disparity in outcomes of surgical revascularization for limb salvage: race and gender are synergistic determinants of vein graft failure and limb loss. Circulation. 2009;119(1):123-130.
PubMedArticle
8.
Egbert  LD, Rothman  IL.  Relation between the race and economic status of patients and who performs their surgery. N Engl J Med. 1977;297(2):90-91.
PubMedArticle
9.
Centers for Disease Control and Prevention.  Report of the Secretary’s Task Force on Black and Minority Health. MMWR Morb Mortal Wkly Rep. 1986;35(8):109-112.
PubMed
10.
Haider  AH, Chang  DC, Efron  DT, Haut  ER, Crandall  M, Cornwell  EE  III.  Race and insurance status as risk factors for trauma mortality. Arch Surg. 2008;143(10):945-949.
PubMedArticle
11.
Haider  AH, Ong’uti  S, Efron  DT,  et al.  Association between hospitals caring for a disproportionately high percentage of minority trauma patients and increased mortality: a nationwide analysis of 434 hospitals. Arch Surg. 2012;147(1):63-70.
PubMedArticle
12.
American College of Surgeons. Statement on health care disparities.https://www.facs.org/about-acs/statements/67-health-care-disparities. Accessed January 18, 2014.
13.
Cooper  LA, Roter  DL, Carson  KA,  et al.  The associations of clinicians’ implicit attitudes about race with medical visit communication and patient ratings of interpersonal care. Am J Public Health. 2012;102(5):979-987.
PubMedArticle
14.
Haider  AH, Sexton  J, Sriram  N,  et al.  Association of unconscious race and social class bias with vignette-based clinical assessments by medical students. JAMA. 2011;306(9):942-951.
PubMedArticle
15.
Nam  CB, Boyd  M.  Occupational status in 2000: over a century of census-based measurement. Popul Res Policy Rev. 2004;23:327-358.Article
16.
Nosek  BA, Greenwald  AG, Banaji  MR. The implicit association test at age 7: a methodological and conceptual review. In: Bargh  JA, ed. Automatic Processes in Social Thinking and Behavior. London, England: Psychology Press; 2007:265-292.
17.
Greenwald  AG, McGhee  DE, Schwartz  JL.  Measuring individual differences in implicit cognition: the Implicit Association Test. J Pers Soc Psychol. 1998;74(6):1464-1480.
PubMedArticle
18.
Nosek  BA, Smyth  FL, Hansen  JJ,  et al.  Pervasiveness and correlates of implicit attitudes and stereotypes. Eur Rev Soc Psychol. 2007;18:36-88.Article
19.
Centers for Disease Control and Prevention. STDs in women and infants. http://www.cdc.gov/std/stats10/womenandinf.htm#pid. Accessed June 15, 2014.
20.
Sutton  MY, Sternberg  M, Zaidi  A, St Louis  ME, Markowitz  LE.  Trends in pelvic inflammatory disease hospital discharges and ambulatory visits, United States, 1985-2001. Sex Transm Dis. 2005;32(12):778-784.
PubMedArticle
21.
Balsa  AI, McGuire  TG, Meredith  LS.  Testing for statistical discrimination in health care. Health Serv Res. 2005;40(1):227-252.
PubMedArticle
22.
Nathens  AB, Maier  RV, Copass  MK, Jurkovich  GJ.  Payer status: the unspoken triage criterion. J Trauma. 2001;50(5):776-783.
PubMedArticle
23.
Weissman  J, Epstein  AM.  Case mix and resource utilization by uninsured hospital patients in the Boston metropolitan area. JAMA. 1989;261(24):3572-3576.
PubMedArticle
24.
Moser  JW, Applegate  KE.  Imaging and insurance: do the uninsured get less imaging in emergency departments? J Am Coll Radiol. 2012;9(1):50-57.
PubMedArticle
25.
Burgess  DJ, van Ryn  M, Crowley-Matoka  M, Malat  J.  Understanding the provider contribution to race/ethnicity disparities in pain treatment: insights from dual process models of stereotyping. Pain Med. 2006;7(2):119-134.
PubMedArticle
26.
Chen  EH, Shofer  FS, Dean  AJ,  et al.  Gender disparity in analgesic treatment of emergency department patients with acute abdominal pain. Acad Emerg Med. 2008;15(5):414-418.
PubMedArticle
27.
Schulman  KA, Berlin  JA, Harless  W,  et al.  The effect of race and sex on physicians’ recommendations for cardiac catheterization. N Engl J Med. 1999;340(8):618-626.
PubMedArticle
28.
Todd  KH, Samaroo  N, Hoffman  JR.  Ethnicity as a risk factor for inadequate emergency department analgesia. JAMA. 1993;269(12):1537-1539.
PubMedArticle
29.
Todd  KH, Deaton  C, D’Adamo  AP, Goe  L.  Ethnicity and analgesic practice. Ann Emerg Med. 2000;35(1):11-16.
PubMedArticle
30.
Tamayo-Sarver  JH, Hinze  SW, Cydulka  RK, Baker  DW.  Racial and ethnic disparities in emergency department analgesic prescription. Am J Public Health. 2003;93(12):2067-2073.
PubMedArticle
31.
Pletcher  MJ, Kertesz  SG, Kohn  MA, Gonzales  R.  Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299(1):70-78.
PubMedArticle
32.
Tamayo-Sarver  JH, Dawson  NV, Hinze  SW,  et al.  The effect of race/ethnicity and desirable social characteristics on physicians’ decisions to prescribe opioid analgesics. Acad Emerg Med. 2003;10(11):1239-1248.
PubMedArticle
33.
Green  AR, Carney  DR, Pallin  DJ,  et al.  Implicit bias among physicians and its prediction of thrombolysis decisions for black and white patients. J Gen Intern Med. 2007;22(9):1231-1238.
PubMedArticle
34.
Gauri  AJ, Davis  A, Hong  T, Burke  MC, Knight  BP.  Disparities in the use of primary prevention and defibrillator therapy among blacks and women. Am J Med. 2006;119(2):167.e17-167.e21. doi:10.1016/j.amjmed.2005.08.021.
PubMedArticle
35.
McCarney  R, Warner  J, Iliffe  S, van Haselen  R, Griffin  M, Fisher  P.  The Hawthorne Effect: a randomised, controlled trial. BMC Med Res Methodol. 2007;7:30.
PubMedArticle
36.
Street  RL  Jr, O’Malley  KJ, Cooper  LA, Haidet  P.  Understanding concordance in patient-physician relationships: personal and ethnic dimensions of shared identity. Ann Fam Med. 2008;6(3):198-205.
PubMedArticle
37.
Zolnierek  KB, Dimatteo  MR.  Physician communication and patient adherence to treatment: a meta-analysis. Med Care. 2009;47(8):826-834.
PubMedArticle
38.
Fiscella  K, Meldrum  S, Franks  P,  et al.  Patient trust: is it related to patient-centered behavior of primary care physicians? Med Care. 2004;42(11):1049-1055.
PubMedArticle
39.
Schneider  J, Kaplan  SH, Greenfield  S, Li  W, Wilson  IB.  Better physician-patient relationships are associated with higher reported adherence to antiretroviral therapy in patients with HIV infection. J Gen Intern Med. 2004;19(11):1096-1103.
PubMedArticle
40.
McGinnis  B, Olson  KL, Magid  D,  et al.  Factors related to adherence to statin therapy. Ann Pharmacother. 2007;41(11):1805-1811.
PubMedArticle
41.
Safran  DG, Montgomery  JE, Chang  H, Murphy  J, Rogers  WH.  Switching doctors: predictors of voluntary disenrollment from a primary physician’s practice. J Fam Pract. 2001;50(2):130-136.
PubMed
42.
Haider  AH, Schneider  ED, Sriram  N,  et al.  Unconscious race and class bias among nurses? vignette-based study using implicit-association testing. J Am Coll Surg. In press.
43.
Cooper  LA, Roter  DL, Johnson  RL, Ford  DE, Steinwachs  DM, Powe  NR.  Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139(11):907-915.
PubMedArticle
Original Investigation
May 2015

Unconscious Race and Social Class Bias Among Acute Care Surgical Clinicians and Clinical Treatment Decisions

Author Affiliations
  • 1Center for Surgical Trials and Outcomes Research, Division of Acute Care Surgery, Department of Surgery, Johns Hopkins University School of Medicine, Baltimore, Maryland
  • 2now with Center for Surgery and Public Health, Department of Surgery, Brigham and Women’s Hospital, and Harvard Medical School and Harvard School of Public Health, Boston, Massachusetts
  • 3Department of Psychology, University of Virginia, Charlottesville
  • 4Department of Surgery, Mayo Clinic, Scottsdale, Arizona
  • 5Armstrong Institute of Patient Safety, Johns Hopkins University School of Medicine, Baltimore, Maryland
  • 6Department of Surgery, Howard University College of Medicine, Washington, DC
  • 7Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland
  • 8Center to Eliminate Cardiovascular Health Disparities, Johns Hopkins University School of Medicine, Baltimore, Maryland
  • 9Office of the Dean, University of California, Davis, School of Medicine, Sacramento
JAMA Surg. 2015;150(5):457-464. doi:10.1001/jamasurg.2014.4038
Abstract

Importance  Significant health inequities persist among minority and socially disadvantaged patients. Better understanding of how unconscious biases affect clinical decision making may help to illuminate clinicians’ roles in propagating disparities.

Objective  To determine whether clinicians’ unconscious race and/or social class biases correlate with patient management decisions.

Design, Setting, and Participants  We conducted a web-based survey among 230 physicians from surgery and related specialties at an academic, level I trauma center from December 1, 2011, through January 31, 2012.

Interventions  We administered clinical vignettes, each with 3 management questions. Eight vignettes assessed the relationship between unconscious bias and clinical decision making. We performed ordered logistic regression analysis on the Implicit Association Test (IAT) scores and used multivariable analysis to determine whether implicit bias was associated with the vignette responses.

Main Outcomes and Measures  Differential response times (D scores) on the IAT as a surrogate for unconscious bias. Patient management vignettes varied by patient race or social class. Resulting D scores were calculated for each management decision.

Results  In total, 215 clinicians were included and consisted of 74 attending surgeons, 32 fellows, 86 residents, 19 interns, and 4 physicians with an undetermined level of education. Specialties included surgery (32.1%), anesthesia (18.1%), emergency medicine (18.1%), orthopedics (7.9%), otolaryngology (7.0%), neurosurgery (7.0%), critical care (6.0%), and urology (2.8%); 1.9% did not report a departmental affiliation. Implicit race and social class biases were present in most respondents. Among all clinicians, mean IAT D scores for race and social class were 0.42 (95% CI, 0.37-0.48) and 0.71 (95% CI, 0.65-0.78), respectively. Race and class scores were similar across departments (general surgery, orthopedics, urology, etc), race, or age. Women demonstrated less bias concerning race (mean IAT D score, 0.39 [95% CI, 0.29-0.49]) and social class (mean IAT D score, 0.66 [95% CI, 0.57-0.75]) relative to men (mean IAT D scores, 0.44 [95% CI, 0.37-0.52] and 0.82 [95% CI, 0.75-0.89], respectively). In univariate analyses, we found an association between race/social class bias and 3 of 27 possible patient-care decisions. Multivariable analyses revealed no association between the IAT D scores and vignette-based clinical assessments.

Conclusions and Relevance  Unconscious social class and race biases were not significantly associated with clinical decision making among acute care surgical clinicians. Further studies involving real physician-patient interactions may be warranted.

Introduction

Disparities in the quality of care received by minority patients have been reported for decades across multiple medical conditions, types of care, and institutions, including our own.17 As recently as 1977, black surgical patients at the Johns Hopkins Hospital, Baltimore, Maryland, were 2 to 4 times more likely than white patients to undergo a cholecystectomy or hernia repair by a surgeon in training rather than by an attending surgeon.8 Although overt displays of racism have diminished, racial disparities in the quality of care persist. These disparities, in turn, may contribute to significant racial differences in mortality and morbidity across a range of health conditions.9 A recent review on race and surgical outcomes1 demonstrated a lower likelihood of receiving surgery and higher rates of surgical mortality among minority vs white patients. A significant body of work shows that black patients experience higher odds of death after trauma compared with white patients.2,10

Myriad risk factors for health outcome disparities exist. At the patient level, these factors include race/ethnicity, age, sex, socioeconomic class, and insurance status.1012 Significant risk factors also likely exist at the provider level. Potential disparities in health care quality may, in turn, result in health outcome disparities. For instance, implicit bias among clinicians may affect the provision of care to minority patients; however, this potential relationship is not well understood. Using the previously validated Implicit Association Test (IAT) for race13 and social class,14 which measures the strength of a person’s automatic association between mental representations of concepts (eg, race and positive or negative emotions), we sought to determine whether implicit race and social class biases of clinicians are present and whether they translate into differences in vignette-based clinical decision making for surgical patients seeking trauma care and acute care according to their race or social class.

Methods

This study was a single-center, cross-sectional survey of medical practitioners at the Johns Hopkins Hospital. The institutional review board of Johns Hopkins Medicine approved the study. Written informed consent was obtained during the study, with the true intent of the study revealed to participants after completion. Initial use of deception was deemed appropriate by the institutional review board to facilitate the needed study design. Data were deidentified during the analysis process. Clinicians at diverse stages of their careers, including attending physicians, fellows, and residents, were invited to participate. Physician assistants and nurse practitioners were not included. Clinicians from the following specialties that frequently participate in the treatment of surgical patients seeking trauma care and acute care were included: general surgery, neurosurgery, otolaryngology, orthopedic surgery, anesthesia, critical care, urology, and emergency medicine. A study sample-size calculation was performed before the study initiation, using as many as 10 variables with power estimates (1 − β) set at 0.80 and 0.95. With standard significance (α = .05), the required sample size was calculated to be 180 to 225 participants. The survey was closed once sufficient enrollment was reached (maximum of 230 participants).

To mitigate response bias, participants were not informed of the study’s intent; the survey was portrayed as a quality improvement and patient safety study. Clinicians at the Johns Hopkins Hospital were invited to participate in the study via email. Participants completed the study through the website http://www.traumastudy.com for a 6-week period from December 1, 2011, through January 31, 2012. The survey took approximately 20 minutes, and a $25 gift card was provided as compensation for participation. Any participants who did not supply complete demographic data or who correctly identified the intent of the study before study initiation were excluded from the analysis. They still received compensation.

After completing a consent page, participants were directed to review 9 clinical vignettes, 8 of which were intended to assess the relationship between unconscious bias and clinical decision making or the clinician’s perceptions of the vignette patient. The vignettes were designed and pilot tested by an expert panel of trauma/acute care surgeons, surgical residents, nurse educators, and medical students. The vignettes were intended to cover common scenarios in acute/trauma care and surgery. The ninth vignette focused on patient safety and was not intended to measure implicit bias. Because the answer was considered obvious, it was used to determine whether participants were reading the vignettes carefully. All individuals who incorrectly answered the patient-safety vignette were excluded from the study. In total, we included 4 race vignettes and 4 social class vignettes. Patient race and social class randomly alternated in the vignettes, with each study participant viewing 2 black and 2 white patients and 2 patients each of upper and lower social classes. Each of the vignettes was followed by 3 associated questions concerning clinical management.

Answer choices to each of the 27 clinical management questions were presented on a 7-point Likert scale. The direction of the scale was made inconsistent so that participants would not assume that the higher numeric values represented favored answer choices. The race vignettes showed a picture of a black or a white patient balanced for age and attractiveness. The social class vignettes used patient occupation as a surrogate for social class and randomly alternated between higher- and lower-class occupations. Occupations were chosen using the Nam-Powers occupational prestige scale, which ranks occupations from 1 to 100 in order of perceived prestige.15

Unconscious preferences for black vs white individuals and upper-class vs lower-class individuals were assessed using the race IAT and the social class IAT, respectively.1618 Participants who take the IAT classify items presented to them on a computer by rapidly pressing one of 2 computer keys. Unconscious attitudes are determined by the speed with which keys are pressed as a proxy for the ease with which study participants sort mental concepts. When associated concepts share the same response key, performance is faster. When associated concepts have different response keys, performance is slower. The implicit preference for one group over another is derived from differences in reaction times across blocks of trials.

We used D scores for IAT analysis. The D scores adjust for variations between a participant’s innate reaction time by calculating the difference in reaction time between 2 blocks of the IAT: for race or social class. Standard classifications of D scores were used to create a 7-point implicit preference scale (a D score of −0.15 to 0.15 indicated no preference; 0.16 to 0.34, minor preference; 0.35 to 0.64, moderate preference; and ≥0.65, strong preference) with the midpoint equivalent to no preference. In addition, explicit biases for race and social class were measured using a 10-point thermometer scale to assess the participant’s self-reported relative warmth toward black vs white and upper- vs lower-class individuals. Implicit preferences account for unconscious and often unrecognized differences; in contrast, explicit preferences are directly acknowledged and stated. Implicit D scores were compared with a 7-point composite scale of explicit preferences using the Spearman correlation coefficients for race and social class IAT responses.

Because the vignettes’ management decisions were ordinal, we initially conducted unadjusted ordered logistic regression to determine the relationship between vignette patient race or class with differences in management decisions. After analysis of the individual vignettes, participant responses for each vignette were aligned so that higher scores were associated with more clinically appropriate decisions. The most clinically appropriate vignette management decisions were decided a priori during the study design phase. Each vignette question was given the same weight, with a maximum score per vignette of 21 points. Scores for each vignette were then summed, and mean scores were calculated for each participant. We used analysis of variance to determine whether participant IAT D scores were associated with quartiles of composite vignette scores. Interaction terms between vignette patient race or social class and the study participant’s race or social class IAT score were calculated. After univariate analysis of D scores, multivariable analysis controlling for sex, explicit bias, and age was performed. We used commercially available software (STATA, version 12; StataCorp) to perform all calculations; a 2-sided P < .05 was considered statistically significant.

Results

A total of 230 Johns Hopkins Hospital physicians completed the online survey. Of these, 223 participants provided complete IAT and demographic information. Eight respondents were able to identify the purpose of the study before study initiation and were excluded. In total, 215 physicians were included in the study, exceeding our a priori sample size requirements. Most of the participants were white (139 [64.7%]), male (129 [60.0%]), and 30 years or older (159 [74.0%]). Residents accounted for most of the participants (86 [40.0%]), followed by attending surgeons (74 [34.4%]), fellows (32 [14.9%]), and interns (19 [8.8%]); 4 physicians (1.9%) did not provide their educational level. Nearly one-third of participants were associated with the Department of Surgery (32.1%), followed by 18.1% in anesthesia, 18.1% in emergency medicine, 7.9% in orthopedics, 7.0% in otolaryngology, 7.0% in neurosurgery, 6.0% in critical care, and 2.8% in urology; 1.9% did not report a departmental affiliation. Table 1 shows the complete demographic information of participants.

Quiz Ref IDFemale respondents demonstrated less unconscious bias for race (mean IAT D score, 0.39 [95% CI, 0.29-0.49]) and social class (mean IAT D score, 0.66 [95% CI, 0.57-0.75]) relative to male respondents (mean IAT D scores, 0.44 [95% CI, 0.37-0.52] and 0.82 [95% CI, 0.75-0.89], respectively). The difference was not statistically significant. Corresponding race and social class IAT D scores by participant sex, age, training level, and specialty are given in Table 2. Quiz Ref IDSome differences in management decisions by level of training were found. For instance, interns were more likely than residents in at least their second year to favor the possibility that a patient presenting to the emergency department with abdominal pain 4 weeks after cholecystectomy was abusing an analgesic consisting of acetaminophen and oxycodone hydrochloride (Percocet) (β = 1.49 [95% CI, 0.51-2.47]).Quiz Ref ID Race and class IAT D scores did not differ across training levels. Ultimately, we found no relationship between the level of training (resident or fellow vs attending physician), practitioner departmental affiliation, race, or age and IAT D score (eTables 1-3 in the Supplement).

The full list of vignettes can be found in the eAppendix in the Supplement. On univariate analysis, IAT D scores showed evidence of implicit biases in the following 3 of 27 possible management decisions:

  1. Respondents who saw black patients favored responses suggesting a hidden history of alcohol abuse in a patient who underwent splenectomy for trauma compared with those who saw a white patient (β = 0.65 [95% CI, 0.16-1.15]). Race IAT D score was not a significant predictor in this model (β = −0.06 [95% CI, −0.65 to 0.52]). However, respondents reporting explicit bias toward white patients were significantly more likely overall to favor responses that rejected the possibility of a hidden history of alcohol abuse (β = −1.89 [95% CI, −3.40 to −0.38]). Race was not associated with further questioning of the family regarding alcohol abuse or with consideration of pharmacologic prophylaxis for alcohol withdrawal.

  2. Respondents who saw a lower-class patient favored responses suggesting a reduced likelihood of obtaining magnetic resonance imaging (MRI) of the cervical spine compared with those who saw an upper-class patient (β = −0.61 [95% CI, −1.11 to −0.12]). Social class IAT D scores were not associated with the likelihood of an MRI of the cervical spine (β = 0.01 [95% CI, −0.53 to 0.55]).

  3. Respondents who saw a young black mother with right lower quadrant pain were more likely to favor the differential diagnosis of pelvic inflammatory disease (PID) vs appendicitis compared with those who saw a white patient (β = 0.71 [95% CI, 0.23-1.20]). No implicit or explicit biases were associated with the reported differential diagnosis. Despite the fact that respondents who were presented with a black patient were more likely to diagnose PID, we found no difference in the choice of diagnostic tool (computed tomography vs ultrasonography; P = .32) or the choice of treatment (pelvic ultrasonography vs diagnostic laparoscopy/appendectomy; P = .63).

A stratified examination of the IAT D scores for the 3 vignettes with discordant responses is given in Table 3. Quiz Ref IDMultivariable analysis controlling for sex, explicit bias, and age showed no statistically significant correlation between management decisions and the IAT D score. The interaction terms calculated between vignette patient race or social class and between the race or social class IAT D score from the regression models are shown in the Figure. Quiz Ref IDComparison of implicit and explicit preferences ultimately revealed no statistically significant correlation for race (Spearman rank correlation coefficient, r = 0.03; P = .69) or social class (Spearman rank correlation coefficient, r = 0.03; P = .64). None of the interaction terms were statistically significant. The full list of D scores for each vignette question is presented in eTables 1 through 3 in the Supplement.

Discussion

Although implicit biases of race and social class were present among most of the trauma and acute care clinician respondents, these biases were not associated with clinical decision making. Race and social class IAT D scores did not differ significantly by practitioner specialty, race, or age. Subtle sex differences found between the mean IAT scores for men and women also were not significant on stratified analysis. Across all clinicians, univariate analysis showed a correlation between implicit race or social class bias with decision making in only 3 of 27 vignettes. However, on adjusted multivariable analysis, these differences were not significant. Overall, we found no differential patient treatment related to race or social class bias.

However, this study suggests that patient race and social class are associated with patient care in certain clinical scenarios. For instance, clinicians were more likely to diagnose a young black woman with lower right quadrant pain as having PID rather than appendicitis when compared with a young white woman with the same symptoms and history. In the United States, rates of PID are 2 to 3 times higher among black women than white women.19,20 According to Balsa et al,21 statistical discrimination occurs when physicians who are uncertain about a clinical diagnosis or treatment behave differently based on patient race or ethnicity, particularly when an illness is associated with a certain racial or ethnic group. In this vignette, the clinicians were faced with clinical uncertainty and an illness more prevalent among 1 race group, perhaps leading them to use statistical discrimination in their assessment of the most likely cause of lower right quadrant pain. Future studies assessing practitioner tendencies in clinically uncertain situations would be wise to consider and control for physiological differences by race, such as the likelihood of experiencing PID.

Clinicians in this study were also less likely to order an MRI of the cervical spine for patients with neck tenderness after a motor vehicle crash if they were of low rather than high socioeconomic status. This finding is consistent with several studies showing that insurance status—often reflective of socioeconomic status—is associated with decreases in the use of health care resources.22,23 In particular, uninsured patients in the emergency department are less likely than non–Medicaid-insured patients to receive any imaging services.24 Given that implicit class bias was not associated with clinician-level decision making in this study, the clinicians’ reduced likelihood of ordering an MRI for patients of lower socioeconomic status may be reflective of prevalent reimbursement concerns rather than implicit bias. Alternatively, given the patient’s relatively mild presentation and negative findings on computed tomography, we can argue whether an MRI is indicated at all. Given the discrepancy observed, an increased tendency to order an MRI among patients of higher socioeconomic status may be indicative of a parallel tendency to cautiously overtreat affluent patients in situations where health care practitioners perceive a higher risk for litigation owing to a missed injury.

Implicit biases may be related to care decisions as divergent as the management of acute myocardial infarction to analgesia prescribing habits.2534 However, a salient finding of this study was the lack of correlation between implicit bias and statistically significant differences in decision making based on training level. A previous study using the IAT with first-year medical students also found that implicit bias toward white individuals did not translate into differences in clinical decision making.14 Whether the lack of association found between implicit bias and decision making in this study represents a true lack of association or the failure of clinical vignettes to capture the nuances of how implicit biases translate into management decisions remains unclear. Although vignettes have been used in multiple studies to investigate clinical decision making and have the advantage of minimizing the Hawthorne (observer) effect35 wherein practitioners would be expected to modify or improve their behavior knowing that they are being observed,32,33 they may not accurately reflect real-world clinical encounters.

Another possibility is that the algorithmic nature of decision making in trauma and acute care surgery mitigates the impact of implicit bias. Trauma, unlike outpatient primary care, focuses on urgent immediate management decisions. In a previous study, Cooper et al13 used race IAT to determine the association between implicit race bias and primary care physician communication style. Their study found that implicit race bias of the physician was associated with more clinical verbal dominance and poorer interpersonal interactions with patients. The authors suggest that poor communication skills and a lack of focus on the patient may affect health outcomes by undermining patient trust and, in turn, patient adherence to therapy.34,3641 In primary care, participatory decision making is clearly important to achieving the best long-term outpatient care. However, the algorithm-driven nature of trauma and acute care surgery does not require long-term participatory decision making. Deciding whether to offer an operation or to order an MRI of the cervical spine in the acute care setting, for instance, may be less affected by physician-patient interpersonal interactions than are less acute management decisions. In addition, the greater exposure of trauma and acute care surgical clinicians to patients from diverse racial and economic backgrounds may help to mitigate the effects of implicit bias on their decision making.

Generational differences across practitioners and within specific specialties may further mitigate the effect. Similar work among a group of registered nurses within the same institution found that although the oldest demographic group (≥35 years of age) had higher mean race and social class IAT D scores than younger groups, age differences were not associated with disparities in clinical decision making and the treatment of patients.42 Assuming a comparable pattern among clinicians, as our data suggest, training and exposure should be expected to elicit more salient effects. Alternatively, the low percentage of black and Hispanic practitioners (Table 1) may have affected the practitioners’ perceptions of vignette scenarios. Among clinicians, several practitioner-level factors, including implicit racial biases, have been implicated in racial disparities through their effect on patient-physician communication.13 Work by Cooper and colleagues13,36,43 suggests that, particularly in race-discordant relationships, patient-physician communication tends to be less patient centered. Lack of sufficient representation by minority health care practitioners, while expected, should be further explored in future studies.

Conclusions

Although this study of clinicians from surgical and other related specialties did not demonstrate any association between implicit race or social class bias and clinical decision making, existing biases might influence the quality of care received by minority patients and those of lower socioeconomic status in real-life clinical encounters. Further research incorporating patient outcomes and data from actual clinical interactions is warranted to clarify the effect of clinician implicit bias on the provision of health care and outcomes.

Back to top
Article Information

Accepted for Publication: August 29, 2014.

Corresponding Author: Adil H. Haider, MD, MPH, Center for Surgery and Public Health, Department of Surgery, Brigham and Women's Hospital, 1620 Tremont St, One Brigham Circle, Fourth Floor, Ste 4-020, Boston, MA 02120 (ahhaider@partners.org).

Published Online: March 18, 2015. doi:10.1001/jamasurg.2014.4038.

Author Contributions: Dr Haider had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Haider, Schneider, Swoboda, Losonczy, Haut, Efron, Pronovost, Lipsett, Cornwell, MacKenzie, Cooper, Freischlag.

Acquisition, analysis, or interpretation of data: Haider, Schneider, Sriram, Dossick, Scott, Swoboda, Losonczy, Lipsett, MacKenzie.

Drafting of the manuscript: Haider, Schneider, Dossick, Scott.

Critical revision of the manuscript for important intellectual content: Haider, Schneider, Sriram, Swoboda, Losonczy, Haut, Efron, Pronovost, Lipsett , Cornwell, MacKenzie, Cooper, Freischlag.

Statistical analysis: Haider, Schneider, Sriram, Dossick, Losonczy, Pronovost, MacKenzie.

Obtained funding: Haider, MacKenzie.

Administrative, technical, or material support: Haider, Scott, Swoboda, Losonczy, Pronovost, Lipsett, Cornwell, Cooper, Freischlag.

Study supervision: Haider, Swoboda, Efron, Pronovost, Lipsett, Cornwell.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by grant K23GM093112-01 from the National Institutes of Health/National Institute of General Medical Sciences (Dr Haider), by the American College of Surgeons C. James Carrico fellowship for the study of trauma and critical care (Dr Haider), and by grant K24 HL83113 from the National Heart, Lung, and Blood Institute (Dr Cooper).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr Freischlag was the editor of JAMA Surgery when this article was accepted for publication but was not involved in the editorial review or decision process.

Additional Contributions: Cheryl K. Zogg, MSPH, MHS, Center for Surgery and Public Health, Harvard Medical School and Harvard School of Public Health, and Department of Surgery, Brigham and Women’s Hospital, Boston, Massachusetts, assisted with manuscript preparation, content, and editing in addition to her handling of manuscript revisions. She received no compensation for this role.

References
1.
Haider  AH, Scott  VK, Rehman  KA,  et al.  Racial disparities in surgical care and outcomes in the United States: a comprehensive review of patient, provider, and systemic factors. J Am Coll Surg. 2013;216(3):482-92.e12.
PubMedArticle
2.
Haider  AH, Weygandt  PL, Bentley  JM,  et al.  Disparities in trauma care and outcomes in the United States: a systematic review and meta-analysis. J Trauma Acute Care Surg. 2013;74(5):1195-1205.
PubMedArticle
3.
Hicks  CW, Hashmi  ZG, Velopulos  C,  et al.  Association between race and age in survival after trauma. JAMA Surg. 2014;149(7):642-647.
PubMedArticle
4.
Hernandez  RA, Hevelone  ND, Lopez  L, Finlayson  SR, Chittenden  E, Cooper  Z.  Racial variation in the use of life-sustaining treatments who die after major elective surgery [published online October 13, 2014]. Am J Surg. doi:10.1016/j.amjsurg.2014.08.025.
PubMed
5.
Sukumar  S, Ravi  P, Sood  A,  et al.  Racial disparities in operative outcomes after major cancer surgery in the United States. World J Surg. 2015;39(3):634-643.
PubMedArticle
6.
Artinyan  A, Mailey  B, Sanchez-Luege  N,  et al.  Race, ethnicity, and socioeconomic status influence the survival of patients with hepatocellular carcinoma in the United States. Cancer. 2010;116(5):1367-1377.
PubMedArticle
7.
Nguyen  LL, Hevelone  N, Rogers  SO,  et al.  Disparity in outcomes of surgical revascularization for limb salvage: race and gender are synergistic determinants of vein graft failure and limb loss. Circulation. 2009;119(1):123-130.
PubMedArticle
8.
Egbert  LD, Rothman  IL.  Relation between the race and economic status of patients and who performs their surgery. N Engl J Med. 1977;297(2):90-91.
PubMedArticle
9.
Centers for Disease Control and Prevention.  Report of the Secretary’s Task Force on Black and Minority Health. MMWR Morb Mortal Wkly Rep. 1986;35(8):109-112.
PubMed
10.
Haider  AH, Chang  DC, Efron  DT, Haut  ER, Crandall  M, Cornwell  EE  III.  Race and insurance status as risk factors for trauma mortality. Arch Surg. 2008;143(10):945-949.
PubMedArticle
11.
Haider  AH, Ong’uti  S, Efron  DT,  et al.  Association between hospitals caring for a disproportionately high percentage of minority trauma patients and increased mortality: a nationwide analysis of 434 hospitals. Arch Surg. 2012;147(1):63-70.
PubMedArticle
12.
American College of Surgeons. Statement on health care disparities.https://www.facs.org/about-acs/statements/67-health-care-disparities. Accessed January 18, 2014.
13.
Cooper  LA, Roter  DL, Carson  KA,  et al.  The associations of clinicians’ implicit attitudes about race with medical visit communication and patient ratings of interpersonal care. Am J Public Health. 2012;102(5):979-987.
PubMedArticle
14.
Haider  AH, Sexton  J, Sriram  N,  et al.  Association of unconscious race and social class bias with vignette-based clinical assessments by medical students. JAMA. 2011;306(9):942-951.
PubMedArticle
15.
Nam  CB, Boyd  M.  Occupational status in 2000: over a century of census-based measurement. Popul Res Policy Rev. 2004;23:327-358.Article
16.
Nosek  BA, Greenwald  AG, Banaji  MR. The implicit association test at age 7: a methodological and conceptual review. In: Bargh  JA, ed. Automatic Processes in Social Thinking and Behavior. London, England: Psychology Press; 2007:265-292.
17.
Greenwald  AG, McGhee  DE, Schwartz  JL.  Measuring individual differences in implicit cognition: the Implicit Association Test. J Pers Soc Psychol. 1998;74(6):1464-1480.
PubMedArticle
18.
Nosek  BA, Smyth  FL, Hansen  JJ,  et al.  Pervasiveness and correlates of implicit attitudes and stereotypes. Eur Rev Soc Psychol. 2007;18:36-88.Article
19.
Centers for Disease Control and Prevention. STDs in women and infants. http://www.cdc.gov/std/stats10/womenandinf.htm#pid. Accessed June 15, 2014.
20.
Sutton  MY, Sternberg  M, Zaidi  A, St Louis  ME, Markowitz  LE.  Trends in pelvic inflammatory disease hospital discharges and ambulatory visits, United States, 1985-2001. Sex Transm Dis. 2005;32(12):778-784.
PubMedArticle
21.
Balsa  AI, McGuire  TG, Meredith  LS.  Testing for statistical discrimination in health care. Health Serv Res. 2005;40(1):227-252.
PubMedArticle
22.
Nathens  AB, Maier  RV, Copass  MK, Jurkovich  GJ.  Payer status: the unspoken triage criterion. J Trauma. 2001;50(5):776-783.
PubMedArticle
23.
Weissman  J, Epstein  AM.  Case mix and resource utilization by uninsured hospital patients in the Boston metropolitan area. JAMA. 1989;261(24):3572-3576.
PubMedArticle
24.
Moser  JW, Applegate  KE.  Imaging and insurance: do the uninsured get less imaging in emergency departments? J Am Coll Radiol. 2012;9(1):50-57.
PubMedArticle
25.
Burgess  DJ, van Ryn  M, Crowley-Matoka  M, Malat  J.  Understanding the provider contribution to race/ethnicity disparities in pain treatment: insights from dual process models of stereotyping. Pain Med. 2006;7(2):119-134.
PubMedArticle
26.
Chen  EH, Shofer  FS, Dean  AJ,  et al.  Gender disparity in analgesic treatment of emergency department patients with acute abdominal pain. Acad Emerg Med. 2008;15(5):414-418.
PubMedArticle
27.
Schulman  KA, Berlin  JA, Harless  W,  et al.  The effect of race and sex on physicians’ recommendations for cardiac catheterization. N Engl J Med. 1999;340(8):618-626.
PubMedArticle
28.
Todd  KH, Samaroo  N, Hoffman  JR.  Ethnicity as a risk factor for inadequate emergency department analgesia. JAMA. 1993;269(12):1537-1539.
PubMedArticle
29.
Todd  KH, Deaton  C, D’Adamo  AP, Goe  L.  Ethnicity and analgesic practice. Ann Emerg Med. 2000;35(1):11-16.
PubMedArticle
30.
Tamayo-Sarver  JH, Hinze  SW, Cydulka  RK, Baker  DW.  Racial and ethnic disparities in emergency department analgesic prescription. Am J Public Health. 2003;93(12):2067-2073.
PubMedArticle
31.
Pletcher  MJ, Kertesz  SG, Kohn  MA, Gonzales  R.  Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299(1):70-78.
PubMedArticle
32.
Tamayo-Sarver  JH, Dawson  NV, Hinze  SW,  et al.  The effect of race/ethnicity and desirable social characteristics on physicians’ decisions to prescribe opioid analgesics. Acad Emerg Med. 2003;10(11):1239-1248.
PubMedArticle
33.
Green  AR, Carney  DR, Pallin  DJ,  et al.  Implicit bias among physicians and its prediction of thrombolysis decisions for black and white patients. J Gen Intern Med. 2007;22(9):1231-1238.
PubMedArticle
34.
Gauri  AJ, Davis  A, Hong  T, Burke  MC, Knight  BP.  Disparities in the use of primary prevention and defibrillator therapy among blacks and women. Am J Med. 2006;119(2):167.e17-167.e21. doi:10.1016/j.amjmed.2005.08.021.
PubMedArticle
35.
McCarney  R, Warner  J, Iliffe  S, van Haselen  R, Griffin  M, Fisher  P.  The Hawthorne Effect: a randomised, controlled trial. BMC Med Res Methodol. 2007;7:30.
PubMedArticle
36.
Street  RL  Jr, O’Malley  KJ, Cooper  LA, Haidet  P.  Understanding concordance in patient-physician relationships: personal and ethnic dimensions of shared identity. Ann Fam Med. 2008;6(3):198-205.
PubMedArticle
37.
Zolnierek  KB, Dimatteo  MR.  Physician communication and patient adherence to treatment: a meta-analysis. Med Care. 2009;47(8):826-834.
PubMedArticle
38.
Fiscella  K, Meldrum  S, Franks  P,  et al.  Patient trust: is it related to patient-centered behavior of primary care physicians? Med Care. 2004;42(11):1049-1055.
PubMedArticle
39.
Schneider  J, Kaplan  SH, Greenfield  S, Li  W, Wilson  IB.  Better physician-patient relationships are associated with higher reported adherence to antiretroviral therapy in patients with HIV infection. J Gen Intern Med. 2004;19(11):1096-1103.
PubMedArticle
40.
McGinnis  B, Olson  KL, Magid  D,  et al.  Factors related to adherence to statin therapy. Ann Pharmacother. 2007;41(11):1805-1811.
PubMedArticle
41.
Safran  DG, Montgomery  JE, Chang  H, Murphy  J, Rogers  WH.  Switching doctors: predictors of voluntary disenrollment from a primary physician’s practice. J Fam Pract. 2001;50(2):130-136.
PubMed
42.
Haider  AH, Schneider  ED, Sriram  N,  et al.  Unconscious race and class bias among nurses? vignette-based study using implicit-association testing. J Am Coll Surg. In press.
43.
Cooper  LA, Roter  DL, Johnson  RL, Ford  DE, Steinwachs  DM, Powe  NR.  Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139(11):907-915.
PubMedArticle
×