ֱ

Implicit Bias Trainings Are Increasingly Common. But Do They Work?

— New research identifies several areas for improvement and redesign

MedpageToday
A photo of the word BIAS as seen through a magnifying glass.

While implicit bias trainings for healthcare providers have exploded in prevalence and popularity, most have flawed methodology and multiple translational gaps that may compromise their impact, a systematic review of 77 studies found.

All the reviewed studies had at least one translational gap, reported Nao Hagiwara, PhD, of the University of Virginia in Charlottesville, and colleagues in .

"What we found is that there is no scientific evidence to support the effectiveness of current implicit bias training for two major reasons," Hagiwara told ֱ. The first being that most training programs have multiple translational gaps, and secondly, more than two-thirds of the studies were deemed "moderate to high" in risk of bias in their methodologies. Just 22 studies were deemed to have low bias risk, while 55 had moderate to high bias risk.

"In other words, these studies had several methodological flaws in how they tested the effectiveness of the training," she said.

Moreover, there was no evidence that implicit bias trainings have an impact on long-term behavioral change, the authors noted.

"The potential for implicit bias training to improve the quality of patient care and promote health equity is grossly undermined by the way the current implicit bias training has been developed, evaluated, and implemented," Hagiwara concluded.

In 2020, Hagiwara published an arguing that effective implicit bias training should be grounded in research and executed in incremental stages, per the Clinical and Translational Science framework. This framework consists of five stages: T0 (basic science research), T1 (translation to humans), T2 (translation to patients), T3 (translation to practice), and T4 (translation to communities).

Translational gaps between T1 and T2 reflect a disconnect between the scientific evidence surrounding implicit bias training and the ways these trainings are designed; T2 and T3 gaps are between how effectiveness was assessed and how it was tested; and T3 and T4 gaps indicate a lack of rigorous testing of external validity.

"Recognizing where and what translational gaps exist is essential for improving implicit bias training and ultimately achieving health care equity," the authors wrote. Moreover, translational gaps "likely compromise [implicit bias trainings'] potential impacts," they added.

In this review, the most common T1-T2 gap was a failure to identify what component of implicit bias was being addressed in the training. For T2-T3 gaps, it was that few studies used best research practices to assess the efficacy of training. All included studies shared the major T3-T4 gap of not testing the external validity of the program.

The authors noted that there have been other systematic reviews of implicit bias research, but this study "differs from prior reviews in that it addresses the question of why implicit bias training may or may not be effective."

Lisa Cooper, MD, director of the Johns Hopkins Center for Health Equity in Baltimore, told ֱ that she was glad a research team conducted a thorough systematic review with a focus on why trainings did or didn't work.

"There's been this explosion of implicit bias trainings over the past 20 years since the first called for that training to be developed and implemented," Cooper said. "If ... we're not clear what the training is focused on, or if the focus of the training isn't clearly linked to the outcome, that's a problem, and we saw that in several of the studies they reviewed."

Hagiwara said this review "provides people who are involved in the development and implementation of implicit bias training clear and structured guidance on what common translational gaps are and how to identify them, so they can proactively address them in their own work."

For this systematic review, the team included 77 studies focused on implicit bias training for healthcare trainees and providers, the earliest of which was published in 2008. Half were quantitative studies, 36% were mixed methods, and 14% were qualitative; sample sizes ranged from just 12 to 1,250 and ran the gamut of professional status and specialty. For 35 studies, the implicit bias training was part of the curriculum, and in 42 studies it was an addition to the regular curriculum. In 41 studies, implicit bias was the primary training goal, and in 35 it was part of larger goals. The primary target was race and ethnicity in 49.3% of studies.

Hagiwara and colleagues pointed out that few studies were "rigorously grounded in theoretical evidence" and that "even among studies that specified components of bias, the majority failed to tailor the material appropriately." One positive was that most trainings were a mix of hands-on and didactic learning. However, most lasted an average of 6 hours and were delivered in a single day, meaning that "attendees were unlikely to have enough opportunities to practice newly learned strategies to mitigate their implicit bias."

Pointing to limitations of the study, Hagiwara and team said that many of the studies they reviewed didn't have the necessary information to code for the items in the review, and the studies included little information about the instruction method and content of the trainings. In addition, the review process was based on a level of subjective interpretation.

  • author['full_name']

    Rachael Robertson is a writer on the ֱ enterprise and investigative team, also covering OB/GYN news. Her print, data, and audio stories have appeared in Everyday Health, Gizmodo, the Bronx Times, and multiple podcasts.

Disclosures

This research was supported by grants from the NIH.

The study authors declared no conflicts of interest.

Cooper also had no conflicts of interest.

Primary Source

Science Advances

Hagiwara N, et al "The nature and validity of implicit bias training for health care providers and trainees: a systematic review" Sci Adv 2024; DOI: 10.1126/sciadv.ado5957.