In this issue of Pediatrics, Goyal et al1 report race- and ethnicity-based differences in the receipt of antibiotics for acute respiratory viral illnesses over a 1-year period from sites participating in the Pediatric Emergency Care Applied Research Network. In their report, Goyal et al1 demonstrate that, regardless of race, non-Hispanic white children were given or prescribed antibiotics at a higher rate than non-Hispanic black patients or Hispanic children. Unfortunately, the findings in this study do not come as a surprise. Not only have researchers in other studies found differences in prescribing antibiotics in various pediatric practice settings but differences in all aspects of care, such as use of diagnostic testing, pain management, and even hospitalization rates.2–5
One of the challenges in assessing racial and ethnic disparities is in understanding the degree to which they are attributable to confounding factors such as socioeconomic status and access to care. In their study, Goyal et al1 control for insurance status but no other measures of socioeconomic status. Nonetheless the overall weight of scientific evidence supports that significant disparities exist on the basis of race and ethnicity. There was a time when results like this would have shocked us. That time is gone. We must recognize that we are biased in our treatment of patients. This does not mean we accept this fact. We should strive to provide care in our clinics and emergency departments that is of high quality, evidence-based, consistent across providers, and free from bias, implicit or explicit. It is unlikely that any given prescriber in this study demonstrated explicit bias; that is, it is unlikely that they considered themselves to have bias or consciously decided whether to use antibiotics based on race or ethnicity. It is likely these differences occurred despite the best intentions of these clinicians to provide equal treatment to all.
Goyal et al1 acknowledge that determining the reasons for the differences in antibiotic use was beyond the scope of their study but offer the following potential reasons for these differences: “race- and ethnicity-specific differences in parental expectations, differential parental pressures perceived by clinicians for treatment of viral infections with antibiotics, and implicit bias of clinicians.”
Parental expectations can likely be addressed by providing education regarding the risks and benefits of additional testing and/or treatment, but this requires time and communication skills, and the effectiveness is not well studied.6,7 The other 2 reasons for the disparities reported (the differential parental pressure perceived by the clinicians and how they respond and the problem of clinician implicit bias) seem to be tightly linked concepts.8 Although few data exist regarding methods to reduce the effects of these factors in the emergency department care of children, methods to reduce parental pressures and implicit biases deserve careful consideration.
Devine9 developed the notion of implicit bias, which comes from common cultural experiences that create awareness of stereotypes within our subconscious minds that can be automatically activated in ways that bypass deliberate thought and influence one’s judgment in unintended and unacknowledged ways.
Implicit bias in decision-making translated into action is not limited to the practice of medicine, but we need to develop strategies in our medical care to deal with this problem. Frameworks to approach the issue have been developed.10 The first step is for all of us who care for children to recognize the problem. Articles like the one in this issue by Goyal et al1 help identify differences in care possibly because of provider bias. Tests to identify implicit bias within individuals have been developed. One such test, the Implicit Association Test,11 has been used and tested with physicians. Despite no explicitly reported preferences for white individuals over black individuals, the testing reveals a preference for white individuals over black individuals among physicians in emergency medicine, pediatrics, and other medical fields.4,12–14 So, the hypothesis that implicit bias may play a role in the identified race and ethnicity differences in care such as reported by Goyal et al1 seems plausible. The second step is to determine and implement a proven method to reduce the identified disparities. Step 2 is much more difficult. Our society would benefit from the introduction of methods to reduce implicit bias-related disparities within and outside of medical care. Despite articles in the literature from an array of fields seeking to identify methods by which we can reduce disparities related to implicit bias, we do not yet have well-validated tools for doing so.
I am reminded of the early days in the development of infection control departments. The first step was education. It was common to be able to ask a hospital administrator, surgeon, or other clinician to demonstrate an understanding of nosocomial infections. Most could define and demonstrate an understanding of the term. However, when asked whether they had nosocomial infections in their own hospital or clinical setting, the answer was commonly no. It was not until the use of systematic measurement and reporting, with the associated cycles of quality improvement with which we have become familiar, that slow but steady progress was made. If you asked your colleagues to define socioeconomic, racial, and ethnic inequality in health care, they will likely be able to do so. Ask them if they might be providing care with related disparities and they will likely say no. It will require local, regional, and national initiatives to support innovative methods for medicine to slowly reduce 1 measured disparity at a time while working toward achieving the goal of eliminating all disparities.
A much less-developed approach for medical care would be to research methods of how implicit biases or their activation and subsequent control in decision-making can be altered at the provider level. It is possible that with such an approach, multiple disparities could be reduced simultaneously. Methods like “individuating” (consciously moving from stereotypes and focusing on the individual) and “perspective-taking” (consciously attempting to envision the viewpoint of your patient or their guardian) may be helpful, but these require conscious effort on the part of the clinician to overcome implicit bias. These methods require time for reflection and may be particularly difficult in time-pressured environments as is common in the delivery of health care, especially emergency health care. It is not known how these methods could or should be applied in health care and, if successful, how long lasting those changes might be.15
I applaud the efforts of Goyal et al1 and others to point out disparities, but encourage the next step of investigation be focused on finding methods that will be successful in eliminating these disparities. Doing so will be critical in ensuring that our patients receive consistent high-quality, evidence-based care directed at their individual preferences and not our often-biased impression of what care they should receive.
- Accepted July 21, 2017.
- Address correspondence to Marvin B. Harper, MD, Division of Emergency Medicine, Boston Children’s Hospital, 300 Longwood Ave, Boston, MA 02115. E-mail:
Opinions expressed in these commentaries are those of the author and not necessarily those of the American Academy of Pediatrics or its Committees.
FINANCIAL DISCLOSURE: The author has indicated he has no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The author has indicated he has no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2017-0203.
- Goyal M,
- Johnson T,
- Chamberlain J, et al
- Agency for Healthcare Research and Quality
- Warner AS,
- Shah N,
- Morse A, et al
- Forscher PS,
- Devine PG
- Copyright © 2017 by the American Academy of Pediatrics