Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

VA Health Systems Research

Go to the VA ORD website
Go to the QUERI website

PPO 09-247 – HSR Study

PPO 09-247
Does Cognitive Load Affect Provider Racial Bias in Decision-Making?
Diana J. Burgess, PhD
Minneapolis VA Health Care System, Minneapolis, MN
Minneapolis, MN
Funding Period: September 2010 - February 2012
Systematic reviews of healthcare disparities suggest that clinicians' diagnostic and therapeutic decision-making may be inappropriately influenced by racial stereotyping and that racial stereotyping may contribute to healthcare disparities. However, there is little understanding of the particular features of the healthcare setting under which clinicians are most likely to be influenced by racial stereotypes. Numerous experiments conducted in social psychology have shown that stereotyping is more likely when individuals are under high levels of "cognitive load"-the amount of mental activity imposed on working memory, which may come from competing mental tasks, environmental factors (e.g., stressful conditions, time pressure), and our own psychological or physiological state (e.g., burnout, fatigue). However, there have been no studies that have directly tested the hypothesis that racial stereotypes are more likely to bias healthcare providers' decisions when they are under greater cognitive load.

The overall objective of this project was to pilot an innovative experiment, to be administered to a national sample of VA primary care providers (PCPs) over the intranet, to test the hypothesis that patient race will be more likely to influence providers' clinical decisions about whether to provide opioids for patients with chronic pain under high levels of cognitive load. Based on studies showing that Blacks were less likely than Whites to receive opioids for pain, we expected that providers in our study would be less likely to state that they would prescribe opioids for a Black versus a White chronic pain patient when they were under greater cognitive load. Specific Aims of this pilot study were: 1) to determine the likely response rate for a larger-scale study, 2) to characterize the degree and type of likely selection bias, 3) to examine the effectiveness of our experimental manipulation of cognitive load, and 4) to gather information on effect sizes of our experimental manipulation.

A sample of VA primary care physicians was identified using the Primary Care Management Module (PCMM) database, with email addresses obtained from the Personnel and Accounting Integrated Data (PAID) database. Providers were invited by email to participate in a study of the treatment of chronic non-cancer pain. We recruited participants in phases until we met our recruitment goals of 40 providers in each of the 4 randomly assigned experimental conditions (160 in total). We employed a 2 (patient race: White vs. Black) X 2 (cognitive load: low vs. high) between-subjects factorial design. Participants logged into a secure website and read a clinical vignette of a pain patient (White vs. Black) with severe, chronic low back pain under conditions of low vs. high cognitive load. To induce high levels of cognitive load, we used a dual-task methodology, in which providers simultaneously performed the clinical decision task and a secondary task, which consisted of memorizing a series of dot patterns. Participants completed: 1) closed-ended measures of the clinical actions they would take, including decisions about prescribing opioid analgesics (our primary dependent measure); 2) concerns about opioid use; 3) the Mental Effort scale (which tested the effectiveness of our cognitive load manipulation); 4) pain-related beliefs and experience (covariates); and 5) questions designed to capture any difficulties they may have had completing the study. Primary analyses calculated: (1) the response rate; (2) the effect sizes of race, cognitive load, and the race X cognitive load interaction on decisions to provide opioids, and (3) the effect of cognitive load and race on providers' scores on the Mental Effort scale.

1)Response rate: 943 individuals were invited to participate. Of those 943, 42 were excluded due to the inability to contact them (i.e. bad email address, email process failure). We first calculated the rate of individuals who clicked on the link to the questionnaire, regardless of whether they completed the survey or not. Of the 901 individuals who received an email; 226 clicked on the link, yielding a response rate of 25.1%. We also calculated the completion rate among those who clicked on the survey link. Of those 226 individuals, 198 completed the survey; yielding a completion rate of 86.8%.

2)Selection bias: We were not able to obtain the demographic information we required from the PAID database that we needed in order to look at selection bias, due to information security concerns that arose.

3)Effect sizes of race, cognitive load, and the race X cognitive load interaction on decision to prescribe opioids. Neither the race X cognitive load interaction (p = .52), nor the main effects of cognitive load (p = .52) on decisions to prescribe opioids were significant. There was a marginal effect of race on decisions to prescribe opioids (p = .09), though in the opposite direction of what was predicted, in which 28.8% of providers prescribed opioids when the patient was Black, while 18.4% of prescribed opioids when the patient was White.

4)Did we successfully manipulate cognitive load using our "dual task paradigm"? In order to test our experimental manipulation of cognitive load, we examined whether providers in the high cognitive load condition reported expending higher levels of mental effort, on a "Mental Effort" scale, than patients in the low cognitive load condition. Counter to predictions, providers in the low cognitive load condition reported greater mental effort than providers in the high cognitive load condition (p < .0009). Neither patient race (p = .20) nor the race X cognitive load interaction (p = .82) had a significant effect on ratings of mental effort.

The results of this pilot study will provide useful information for other researchers studying the effect of racial stereotyping on provider decision-making. In this study, our experimental manipulation of cognitive load, which had been used in prior research, did not moderate the effect of race on providers' decisions to prescribe opioids for chronic pain. Moreover, counter to expectations, providers were more likely to prescribe opioids for Black rather than White patients. It is unclear whether these results were affected by social desirability biases, raising questions about the validity of this research paradigm for studying the effects of racial stereotyping on clinical decision-making.

External Links for this Project

NIH Reporter

Grant Number: I01HX000317-01

Dimensions for VA

Dimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.

Learn more about Dimensions for VA.

VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address.
    Search Dimensions for this project


Journal Articles

  1. Burgess DJ, Phelan S, Workman M, Hagel E, Nelson DB, Fu SS, Widome R, van Ryn M. The effect of cognitive load and patient race on physicians' decisions to prescribe opioids for chronic low back pain: a randomized trial. Pain medicine (Malden, Mass.). 2014 Jun 1; 15(6):965-74. [view]

DRA: Other Conditions
DRE: Etiology
Keywords: none
MeSH Terms: none

Questions about the HSR website? Email the Web Team

Any health information on this website is strictly for informational purposes and is not intended as medical advice. It should not be used to diagnose or treat any condition.