Skip to main content

Table 1 Studies included in the systematic review

From: Implicit bias in healthcare professionals: a systematic review

Year First author Country Method Population Recruitment Response rate Main findings (relevant to systematic review)
Age
 2010 Protière [25] France Assumption Method 388 oncologists and radiotherapists Mail-out countrywide. 69% Significant negative differences in treatment choice for older patients.
AIDS patients
 2007 Li [61] China Assumption Method 1101 (including just over 50% doctors, 40% nurses) Random selection of institutions and individuals. Less than 8% refusal rate (includes 10% lab technicians) Attitudes as measured across subjects were more negative towards AIDS patients than towards hepatitis B patients.
Brain injured patients who have contributed to their injury
 2011 Linden [58] UK Assumption Method 69 nurses Email. 24% More negative attitudes as measured across subjects were found against individuals seen as contributing towards their injury.
 2010 Redpath [59] UK Assumption Method 155 (94 qualified nurses, 61 qualified doctors) Not specified. Not specified. More negative attitudes as measured across subjects were found against individuals seen as contributing towards their injury. Attitudes significantly related to intended helping behaviour.
Disability
 2012 Aaberg [62] US IAT (prejudice) 132 nurse educators Email. 21% Negative implicit bias against the disabled, stronger than that of the average population.
Gender
 2005 Abuful [63] Israel Assumption Method 172 physicians (internists, cardiologists, family physicians, general practitioners) Continuing medical education. Not specified. Negative bias against women in diagnosis of risk and prescription of lipid-lowering medications and aspirin.
Intravenous Drug Users (IDUs)
 2007 Brener [49] Australia IAT (prejudice) 60 health care workers (HCW) from drug and alcohol facilities and liver clinics: 21 physicians, 37 nurses, two medical students Different facilities and GPs identified through networking. Not specified. HCW had positive explicit attitudes and negative implicit attitudes towards HCV positive IDUs. Contact (as estimated by HCW) predicted explicit (positive) and implicit attitudes (negative) towards IDU beyond the effect of conservatism.
 2008 von Hippel [48] Australia IAT (prejudice) 44 Drug and Alcohol (D&A) nurses Selection of Drug & Alcohol treatment facilities, needle and syringe exchange programs, and primary-care facilities. Not specified. Challenging behaviours by IDUs predicted self-reported stress of nurses, which, in turn, predicted intention to change jobs. The relation between stress and intention to change jobs significantly mediated by the nurses’ implicit prejudice, not explicit prejudice.
Mentally ill
 2007 Chow [64] Hong Kong Assumption Method 433 (107 physicians, 322 nurses and four who didn't state) Random distribution via ward managers (nurses) and email (physicians). 36.1% More negative attitudes as measured across subjects found towards psychiatric patients than to non-psychiatric patients.
 2005 Mackay [26] UK Assumption Method 89 A&E medical and nursing staff 4 A&E Departments in Greater Manchester. 49% The greater the attributions of controllability to self-harming patients, the greater the negative affect of staff towards the patient as measured across subjects, and the less the propensity to help.
 2012 Neauport [65] France Assumption Method 322 medical residents of all specialties in one hospital Email. 47.4% Those assigned the vignette that included the psychiatric illness label said that they were less likely to want to treat the individual and be involved with her/him in various ways.
 2008 Peris [42] Worldwide IAT (prejudice and stereotype) 682 mental health professionals (clinical psychologists, social workers, counsellors, psychiatrists and others) and clinical graduate students. Project Implicit website and 110 clinicians and graduate students recruited directly through list serves. 81% (after random assignment to study via Project Implicit and including 747 non-mental health professionals) Overall, explicit and implicit views were not negative towards individuals with mental illness. Those with mental health training displayed less implicit and explicit prejudices. Their explicit (but not implicit) biases predicted more negative patient prognoses, but implicit (and not explicit) biases predicted over-diagnosis.
Multiple biases
 2004 Arbera [27] US/UK Assumption Method Biases: Age, Gender, Racial/ethnic and Socio-Economic Status (SES) 256 primary care physicians in the US and the UK Screening telephone calls. 65% in the US and 60% in the UK. Gender and age influenced the doctors’ questioning of patients presenting with coronary heart disease (CHD) in both countries. Men were asked more questions overall and middle-aged patients were asked more lifestyle questions.
 2006 Arbera [32] UK and US Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 256 primary care physicians in the US and the UK Screening telephone calls. 65% in the US and 60% in the UK. The gender of the patient significantly influenced doctors’ diagnostic and management activities. Midlife women were asked fewest questions and prescribed least medication appropriate for CHD.
 2006 Barnhart [66] US Assumption Method Biases: Racial/ethnic, Gender, and Social Circumstances. 544 family medicine physicians, internists, cardiologists, and cardiothoracic surgeons Mail-out. 70% The patient’s race and gender did not significantly affect the physicians’ treatment preferences. However, significant differences were found according to social circumstance.
 2008 Böntea [31] US, UK and Germany Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 384 physicians (internists or family practitioners in the US and Germany or general practitioners in the UK) Screening telephone calls. 64.9% in the US, 59.6% in the UK, and 65% in Germany. Results showed gender differences in the diagnostic strategies of the doctors.
 2010 Dehlendorf [39] US Assumption Method Biases: Racial/ethnic and SES. 524 health care providers (96% MD/DO, 4% Nurse Practitioner/Physician Assistant) Convenience sample from meetings of professional societies. Not specified. Low SES whites were less likely to have intrauterine contraception recommended than high SES whites. By race/ethnicity, low SES Latinas and blacks were more likely to have intrauterine contraception recommended than low SES whites, with no effect of race/ethnicity for high SES patients. Low SES patients were judged to be significantly more likely than high SES patients to have an STI and an unintended pregnancy, and were also judged to be less knowledgeable.
 2005a Kales [37] US Assumption Method Biases: Racial/ethnic and Gender. 321 psychiatrists Attendees at the 2002 annual meeting of the American Psychiatric Association. Not specified. Patients’ race and gender was not associated with significant differences in the diagnoses of major depression. However, white patients were rated as being of significantly higher SES than black patients. A significant relationship was found between rating of SES and estimates of patient demeanour (lower SES associated with more hostile demeanour).
 2005b Kales [38] US Assumption Method Biases: Racial/ethnic and Gender. 178 Primary Care Physicians (PCPs) Attendees at the 2002 American Academy of Family Physicians Annual Meeting. Not specified. Patients’ race and gender was not associated with significant differences in the diagnoses of major depression. However, white patients were rated as being of significantly higher SES than black patients.
 2009a Lutfeya [29] US, UK and Germany Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 384 physicians (internists or family practitioners in the US and Germany and general practitioners in the UK) Screening telephone calls. 64.9% in the US, 59.6% in the UK, and 65.0% in Germany Physicians were least certain of CHD diagnoses when patients were younger and female. Certainty was positively correlated with several clinical actions, including test ordering, prescriptions, referrals to specialists, and time to follow-up.
 2009b Lutfeya [35] US Assumption Method Biases: Age, Gender, Racial/ethnic and SES 128 generalist physicians Screening telephone calls. 64.9% Physicians were least certain of their CHD diagnoses for black patients and for younger women. Physicians responded differentially to diagnostic certainty in terms of their clinical therapeutic actions such as test ordering and writing prescriptions.
 2010 Lutfey [33] US Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 256 internists or family/general practitioners Screening telephone calls. Not specified. Physicians primed with CHD were more likely to order CHD-related tests and prescriptions. Main effects for patient gender and age remained, suggesting that physicians treated these demographic variables as diagnostic features indicating lower risk of CHD for these patients.
 2009a Maserejiana [34] US Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 128 physicians (internist or family practitioner) Screening telephone calls. Not specified. Physicians were significantly less certain of the underlying cause of symptoms among female patients regardless of age, but only among middle-aged women were they significantly less certain of the CHD diagnosis.
 2009b Maserejian [36] US Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 256 internists or family/general practitioners Screening telephone calls. Not specified. 48% of physicians were inconsistent in their population-level and individual-level CHD assessments. Physicians’ assessments of CHD prevalence did not attenuate the observed gender effect in diagnostic certainty for the individual patient.
 2006 McKinlaya [28] US/UK Assumption Method Biases: Age, Gender, Race and SES. 256 primary care physicians in the US and the UK Screening telephone calls. 64.9% in the US and 59.6% in the UK. Age, race, gender, but not SES, influenced decision-making for the conditions studied in both countries. Differences were also found between treatment decisions in the UK and the US.
 2007 McKinlaya [30] US Assumption Method Biases: Age, Gender, Racial/ethnic and SES. 128 internists and family physicians Screening telephone calls. 64.9% Female patients were less likely than males to receive 4 of 5 types of physical examination; older patients were less likely to be advised to stop smoking. Race and SES of patients had no effect on provider adherence to guidelines.
 2003 Tamayo-Sarver [40] US Assumption Method Biases: Racial/ethnic 2872 emergency physicians Mail-out. 53% The race/ethnicity of patients in the vignettes had no effect on physician prescription of opioids. Making socially desirable information explicit increased the prescribing rates by 4% for the migraine vignette and 6% for the back pain vignette.
Racial/ethnic
 2011 Barnato [41] US Assumption Method 33 hospital-based physicians (emergency physicians, hospitalists, intensivists) Probability sampling (15) and convenience sampling (18). Reported as ‘low’. Physicians did not make different treatment decisions for black and white patients, despite believing that black patients were more likely to prefer intensive, life-sustaining treatment.
 2013a Blair [44] US IAT (prejudice) and interpersonal interaction measures 134 clinicians Data for clinicians collected in the Blair [44] study. Primary data from patients in a broader study on hypertension. 60% Clinicians with greater implicit bias against blacks were rated lower in patient-centred care by their black patients as compared with a reference group of white patients.
 2013b Blair [22] US IAT (prejudice) 210 experienced primary care providers (PCPs) Invitation launched with presentations. 60% Substantial implicit bias against Latinos and African Americans in PCPs
 2008 Burgess [43] US Assumption Method 382 general internal medicine physicians Mail-out. 40% There was no significant effect of patient race alone. Among black patients, physicians were significantly more likely to state that they would switch to a higher dose or stronger opioid for patients exhibiting “challenging” behaviours compared to those exhibiting “non-challenging” behaviours.
 2012 Cooper [23] US IAT (prejudice and stereotype), audiotape measures of visit communication and patient ratings. 40 primary care clinicians (36 physicians, four nurses) in urban community-based practices. Secondary data from two previous studies, where patients and providers participated in randomised clinical trials of interventions to enhance communication. 63% Clinician implicit race bias and race and compliance stereotyping were associated with markers of poor visit communication and poor ratings of care, particularly among black patients.
 2007 Green [46] US IAT (prejudice and stereotype) and vignettes 287 physicians Email. 50.6% All 3 IATs showed significant race bias. Physicians were more likely to diagnose black patients than white patients with CAD. However, physicians were equally likely to give thrombolysis for black and white patients. There was thus a racial disparity in thrombolysis relative to CAD diagnosis.
 2012 Moskowitz [20] US Subliminal priming (faces). Study 1: 16 physicians. Study 2: 11 physicians Convenience sample. Not specified. When primed with a black face, physicians reacted more quickly for stereotypical diseases, indicating an implicit association of certain diseases with black patients. These comprised not only diseases that black patients are genetically predisposed to, but also conditions and social behaviours with no biological association (e.g. obesity, drug use).
 2010 Penner [24] US IAT (prejudice) and interaction measures 15 resident physicians (3 white and 12 self-identified as Indian, Pakistani or Asian) Patients recruited consecutively and physicians invited. 83% physicians Overall, physicians did not display implicit race bias. However, black patients had less positive reactions to medical interactions with physicians relatively low in explicit but relatively high in implicit bias than to interactions with physicians who were either (a) low in both explicit and implicit bias, or (b) high in both explicit and implicit bias.
 2008 Sabinb [47] US IAT (prejudice and stereotype) and vignettes 86 academic paediatricians from one department Invited all faculty, residents and fellows at a large, urban research university to participate. 58% Paediatricians held implicit race bias, but it was weaker than that of other MDs and others in society.
 2009 Sabin [67] Worldwide IAT (prejudice) 2535 MDs Project Implicit website. Not applicable Medical doctors, like the rest of the sample, showed a strong implicit preference for whites over blacks.
 2012a Sabinb [45] US IAT (prejudice and stereotype) and vignettes 86 academic paediatricians from one department. Invited all faculty, residents and fellows at a large, urban research university to participate. 58% Paediatricians’ implicit bias was associated with treatment recommendations. As paediatricians’ implicit pro-white bias increased, prescribing narcotic medication decreased for black patients, but not for white patients.
 2012 Stepanikova [21] US Subliminal priming (words) and vignettes 81 family physicians and general internists Email. 2% Under higher time pressure, but not lower, implicit biases against blacks and Hispanics led to less serious diagnosis. Under higher time pressure, implicit bias against blacks led to lower rate of referral to specialist.
Weight
 2012b Sabin [60] Worldwide IAT (prejudice) 2284 medical doctors (MDs) Project Implicit website. Not applicable. MDs, like the wider population tested, had a strong implicit anti-fat bias and a strong explicit anti-fat bias.
 2003 Schwartz [50] Canada IAT (prejudice and stereotype) 389 (122 physicians, 12 psychologists, five nurses, 18 other obesity clinicians) Attendees of the Annual Meeting of the North American Association for the Study of Obesity. Not specified. There was a significant implicit anti-fat prejudice and stereotype found.
 2007 Vallis [68] Canada IAT (prejudice and stereotype) 78 (14.3% physicians, 15.4% nurses) Attendees of on obesity conference. 86% of total attendees. Strong evidence for anti-fat prejudice and stereotype.
  1. aWhat appear to be the same data from the US, the UK and Germany are selectively analysed in different ways in these eight studies
  2. bWhat appear to be the same data are analysed in different ways in these two studies