Evaluating Bias within State Laws and an Association with Premature Mortality Rate

Introduction

This study seeks to evaluate whether policies and laws that were classified as either protective or harmful are associated with premature mortality rates. In essence, the analysis is a new method for measuring structural racism and estimating the joint effects of multiple laws on racial health equity.

In the study, laws are deemed as protective or harmful depending on whether the law possessed anti-Black sentiments and/or discriminated against ethnically minoritized peoples. The study found that there are associations between  laws that were deemed discriminatory or  structurally racist and premature mortality rates for minority groups. Overall, the study also found that premature mortality rates were lowest in states that had more protective laws and highest in states which had harmful laws. The authors included [1] Jaquelyn L. Jahn, Ph.D., MPH, Assistant Professor at the University of Drexel; Dougie Zubizarreta, MS, PhD student at the FXB Center for Human Rights at Harvard University; Jarvis T. Chen, Associate Director for the Population Health Sciences Ph.D. Program at the Harvard T.H. Chan School of Public Health; Belinda Needham, PhD, MA, social epidemiologist and the Chair of Epidemiology at the University of Michigan’s School of Public Health; Goleen Samari, Ph.D, population health demographer at Columbia University’s Mailman School of Public Health; Alecia J. McGregor, PhD, an Assistant Professor of Health Policy and Politics at the Harvard T.H. Chan School of Public Health; Megan Daugherty Douglas, J.D.,      Assistant Professor at the Department of Community Health and Preventive Medicine at the Morehouse School of Medicine; S. Bryn Austin, ScD, Professor of Social and Behavioral Health Sciences at Harvard T.H. Chan’s School of Public Health and Professor of Pediatrics at Harvard Medical School; Madina Agénor, ScD, MPH, Associate Professor in the Departments of Behavioral and Social Sciences and Epidemiology and Center for Health Promotion and Health Equity at Brown University School of Public Health.

Methods and Findings

The study utilizes a latent class analysis to group states statistically based on laws present within a respective state; the analysis aims to provide an understanding on how states group based on protective or harmful laws, which were categorized based on if they carried anti-Black sentiment. This analysis facilitates the evaluation of states, particularly states with a history of utilizing Jim Crow Laws and/or laws that promote restriction, or to better understand the association between premature mortality and state laws (protective and harmful). The study uses the CDC Wonder’s Database for premature mortality and couples this data with state-level and age-adjusted data that comes from the 2013 National Center for Health Statistics (NCHS) dataset. The premature mortality cutoff was indicated to be before age seventy-five in this study. To estimate the premature mortality, the authors utilized a “three-step” estimating method which centered on:

• a weighted linear regression model;
• an association with the latent class analyses for the states; and
• a regression for the non-Hispanic Black and White adults and the ratios for Black and White premature mortality.

Because the study aims to address socioeconomic factors, the research team did not adjust for socioeconomic factors and political contexts, such as the political party in place at the time. However, to ensure that the latent class analyses are effective and that the findings still stand true, socioeconomic factors and political contexts were controlled for after the initial regression.

The findings of the study showcase three different state classes,

• Class 1 is the class of states with strictly harmful laws. All southern states were a part of class 1.
•  Class 2 is the class most prone to have states with protective laws. Many West Coast, mid-Atlantic, and southwestern states were class 2
• Class 3 states possess  protective laws but are also most likely to have harmful laws that related to welfare, state Earned Income Tax Credit, tuition, and renter protections. Class 3 states were scattered but had some clustering in the Northeast.

For associations between premature mortality and the aforementioned classes, the study shows that the more protective a state was classified, it was also more likely to have less premature mortality rates.

A limitation to this study is the research team’s inability to empirically evaluate much of the informal racist practices that may permeate the state as laws are not the only way to perpetuate structural racism. In addition, the study only analyzes laws in their written context and does not evaluate or assess if a law was implemented in a discriminatory fashion. Other confounding factors for the study include no controls for residential mobility and individual-level factors, such as behavior. In addition, state laws that are implemented at local levels also cannot be analyzed because of the state-level approximation used in the study.

Conclusions

This study shows how differences in racist and discriminatory laws are associated with an increase in premature mortality rates. The historical context and racism behind these laws also signifies that health systems and structures have been shaped by discriminatory ideologies that result in lower health outcomes. The authors conclude, “Dismantling structural racism to address racialized health inequities will require macro-level interventions in the form of sets of laws and policies across social systems and institutions, as well as a deep understanding of the historical underpinnings of structural racism and how it has shaped and continues to shape social norms, institutional practices, ideology, and politics.”

RRAPP Summary Title: Reducing Bias in the Clinical Space: A New Evaluation Tool to Reduce Bias Against Resident Ratings

Introduction

This study empirically evaluates how the implementation of a new program rating residents, termed “Milestone,” reduced      bias against Asian, Latino, and Black residents. The Milestone program utilizes new forms of performance evaluation, including competency trainings and knowledge-based questions for residents to evaluate their clinical excellence. The study used linear regression models and robustness checks to ascertain whether there was a reduction in racial and gender bias towards residents by resident directors in their performance evaluations because of this program.

Bradley M. Gray, Ph.D., Senior Health Services Researcher at the American Board of Internal Medicine; Rebecca S. Lipner, head of the Division of Assessment and Research at the American Board of Internal Medicine; Robert O. Roswell, MD, FACP, FACC, Associate Dean for Diversity, Equity, and Inclusion and Associate Professor of Cardiology and Science Education at the Zucker School of Medicine; Alicia Fernandez, MD, Associate Dean of Population Health and Health Equity at the University of California San Francisco (UCSF) School of Management, Professor of Medicine at UCSF, and internal medicine provider at Zuckerberg San Francisco General Hospital; Jonathan L. Vandergrift, MS, Senior Research Associate at the American Board of Internal Medicine; Marcella Alsan, MD, PhD, Professor at the Harvard Kennedy School.    

Methods and Findings

The methodology includes cross-sectional and time-series comparisons [1] for the adoption of the program and reduction in bias for the residents that are of interest; residents of interest for this study were only those who identified as either Asian, Black, or Latino. The study, firstly, divided participants into U.S.-born and non-U.S.-born residents to minimize the discrimination that non-U.S.-born residents might experience within the context of the study. In addition, residents of interest were only those who identified as either Asian, Black, or Latino. The study compared “knowledge ratings,” which residents were assigned using examinations to test their clinical excellence.

To determine the difference between knowledge ratings, the first certification examination (ABIM), in which residents were rated prior to the Milestone system, was used as the base-case. Then, the Milestone system of resident rating was seen as the treatment variable.

The findings of the study indicate that after the Milestone system was in use, gender bias against female residents was relatively insignificant with how they were rated in their knowledge. However, a note was that there was a deviation from this trend of diminished gender bias for Black female residents, as they experienced significantly more discrimination than both their Black male and other female residents. With the new resident system of rating, there was a significant reduction in bias for residents of color, except for Black residents, where bias persisted. With the small population of residents that are Black (6%) in comparison to the population of Black U.S.-born Black residents in the U.S. (34%), the perpetuation of bias in how residents were rated proved to be especially concerning.

The study’s limitations include possible confounding variables influencing the study that correspond to ratings bias and Milestone bias for how residents are rated on performance. Another limitation of the study, particularly for study participants who identify as mixed race, is participants’ ability to change their race and/or ethnicity, which could alter study results.

Can you put this sentence in more everyday language? The audience for the article is not researchers, but rather practitioners (people who might be making performance evaluations in their jobs). They won’t know what this is.

Conclusions

This study highlights how Black, Asian, and Latino residents were evaluated under new rating system. The Milestone system introduced several measures meant to reduce bias in evaluation systems for residents by resident directors. These measures included several more competency ratings and more comprehensive clinical and diagnostic knowledge questions on clinical comprehension. The Milestone system proved effective in reducing bias in performance rating by resident directors against Asian and Latino residents.
However, for U.S.-born Black residents, the bias remained.