Skip to main content

Study tracks clinical team engagement with health records by patient race/ethnicity

Oct. 10, 2023, 3:01 PM

The study team from VUMC included, from left, Xinmeng Zhang, You Chen, PhD, Bradley Malin, PhD, and Chao Yan, PhD. On the computers are Northwestern Medicine colleagues Abel Kho, MD, and Yuyang Yang. (photo by Donn Jones)
The study team from VUMC included, from left, Xinmeng Zhang, You Chen, PhD, Bradley Malin, PhD, and Chao Yan, PhD. On the computers are Northwestern Medicine colleagues Abel Kho, MD, and Yuyang Yang. (photo by Donn Jones)

by Paul Govern

At two large academic medical centers a review of electronic health record (EHR) user access logs found that EHRs of adult inpatients from minority racial and ethnic populations on average received lower engagement from health care teams than the records of white adult inpatients.

The finding, reported Oct. 9 in JAMA Network Open by researchers at Vanderbilt University Medical Center in Nashville, Tennessee, and Northwestern Medicine in Chicago, measures EHR engagement in terms of health care team interactions per hour with the EHR during the adult inpatient stay, reflecting user activities such as reviewing test results, updating clinical notes, and maintaining active medications.

The study’s chief results are stated as odds ratios for receiving EHR engagement on a par with that afforded to white patients (after adjustment for potentially confounding factors). With a fair coin toss, the odds ratio of coming up heads is 50/50, or 1. Consider a coin that’s somewhat biased in favor of tails, with odds for tossing heads being, say, 40/50 instead of 50/50; with this biased coin, the odds ratio for heads comes to 4/5, or 0.8 instead of 1. In the study, with average engagement in view and white patients as the reference group, for patients of any non-white racial/ethnic group, an adjusted odds ratio of 1 would indicate EHR engagement precisely on par with the engagement afforded to white patients, while a ratio less than 1 would indicate a pattern of lower engagement.

For non-white patients overall, with white patients as the reference group, the adjusted odds ratio for health care team EHR engagement at Vanderbilt University Hospital was 0.86, while across adult hospitals at Northwestern it was 0.90.

“Our premise in undertaking this analysis is that EHR engagement, as reflected in access logs, is an overlooked but important measure of clinical effort and attention,” said the report’s lead author, Chao Yan, PhD, a research fellow in Biomedical Informatics at VUMC. “Our findings, as well as the analysis framework itself, provide an opportunity to detect potentially unknown disparities in health care.”

The analysis was based on all adult hospital admissions from 2018 through 2020 at both centers. (The report does not extend to user access by clinical role, such as physician, nurse, etc.) The analysis grouped non-white patients into Black, Hispanic, and, to preserve statistical power, “other race/ethnicity.”

With white patients as the reference group, the adjusted odds ratio brea­­­kdown for EHR engagement at VUMC was 0.93 for Black patients, 0.77 for Hispanic patients, and 0.67 for “other race/ethnicity.”

At Northwestern Medicine, the breakdown was 1.03 for Black patients (indicating slightly higher EHR engagement compared to white patients), 0.84 for Hispanic patients, and 0.81 for “other race/ethnicity.”

From VUMC, the r­eport’s co-first author is PhD student Xinmeng Zhang and the senior authors are You Chen, PhD, assistant professor Biomedical Informatics, and Bradley Malin, PhD, Accenture professor of Biomedical Informatics. From Northwestern University, the co-first author is PhD student Yuyang Yang and the senior author is Abel Kho, MD, professor of Medicine.

“It is unknown if the same findings would be observed in other health care institutions,” Chen said. “We invite researchers elsewhere to adopt our methodology so that this issue, which could amount to a lingering structural defect in our system of care, can be assessed and addressed head-on. More systematic research is needed to determine whether and how the observed differences in EHR engagement have an impact on patients’ outcomes.”

The report’s discussion section mentions four factors that could contribute to lower overall EHR engagement having been afforded to non-white patients.

  • Minority populations historically have had less access to health care, leaving these patients with less complex, more readily grasped health records.
  • Hispanics are more apt to face an English language barrier, which could translate as less EHR activity.
  • Clinical research documentation by the health care team increasingly figures in the EHR and minorities are less prone to participate in this research.
  • The U.S. health care system might not be immune to the remnants of American racism that haunt the broader society.

With racial discrimination in U.S. health care having been well documented, this last point receives the brunt of the authors’ discussion as they adduce findings from some of the “numerous studies [that] have provided evidence of implicit racial bias in health professionals.”

The analysis also looks at EHR engagement by patient exposure to socio-economic deprivation — this based on home address and multidimensional deprivation scores across the two metropolitan areas, reported by the U.S. Census Bureau. With patients from areas with below-median deprivation scores serving as the reference group — on this scale, areas with lower scores are faring better — patients from areas more exposed to socio-economic deprivation received higher EHR engagement, the adjusted odds ratios being 1.04 at VUMC, 1.07 at Northwestern Medicine.

With publicly insured patients as the reference group, patients without insurance received greater EHR engagement from the health care team: adjusted odds ratios were 1.91 at VUMC, 1.51 at Northwestern Medicine.

Finally, compared to patients with public insurance (such as Medicare or Medicaid), patients with private insurance received greater EHR engagement at VUMC, where the adjusted odds ratio was 1.11, and less engagement at Northwestern Medicine, where the figure was 0.86.

Others on the study from VUMC include Kaidi Kang, MS, Martin Were, MD, MS, Peter Embí, MD, MS, and Mayur Patel, MD, MPH. The study was supported by the National Institutes of Health (grants LM012854, HG012510, AG058639, GM120484).

Recent Stories from VUMC News and Communications Publications

Vanderbilt Medicine
VUMC Voice