Contributions of Geography and Nongeographic Factors to the White-Black Gap in Hospital Quality for Coronary Heart Disease: A Decomposition Analysis

Implications of Bias in Artificial Intelligence: Considerations for Cardiovascular Imaging

PURPOSE OF REVIEW

Bias in artificial intelligence (AI) models can result in unintended consequences. In cardiovascular imaging, biased AI models used in clinical practice can negatively affect patient outcomes. Biased AI models result from decisions made when training and evaluating a model. This paper is a comprehensive guide for AI development teams to understand assumptions in datasets and chosen metrics for outcome/ground truth, and how this translates to real-world performance for cardiovascular disease (CVD).

RECENT FINDINGS

CVDs are the number one cause of mortality worldwide; however, the prevalence, burden, and outcomes of CVD vary across gender and race. Several biomarkers are also shown to vary among different populations and ethnic/racial groups. Inequalities in clinical trial inclusion, clinical presentation, diagnosis, and treatment are preserved in health data that is ultimately used to train AI algorithms, leading to potential biases in model performance. Despite the notion that AI models themselves are biased, AI can also help to mitigate bias (e.g., bias auditing tools). In this review paper, we describe in detail implicit and explicit biases in the care of cardiovascular disease that may be present in existing datasets but are not obvious to model developers. We review disparities in CVD outcomes across different genders and race groups, differences in treatment of historically marginalized groups, and disparities in clinical trials for various cardiovascular diseases and outcomes. Thereafter, we summarize some CVD AI literature that shows bias in CVD AI as well as approaches that AI is being used to mitigate CVD bias.

The segregation of physician networks providing care to black and white patients with heart disease: Concepts, measures, and empirical evaluation

Black-White disparities in cardiac care may be related to physician referral network segregation. We developed and tested new geographic physician network segregation measures. We used Medicare claims to identify Black and White Medicare heart disease patients and map physician networks for 169 hospital referral regions (HRRs) with over 1000 Black patients. We constructed two network segregation indexes ranging from 0 (integration) to 100 (total segregation): Dissimilarity (the unevenness of Black and White patient distribution across physicians [Dn]) and Absolute Clustering (the propensity of Black patients' physicians to have closer ties with each other than with other physicians [ACLn]). We employed conditional logit models to estimate the probability of using the best (lowest mortality) geographically available hospital for Black and White patients undergoing coronary artery bypass grafting (CABG) surgery in 126 markets with sufficient sample size at increasing levels of network segregation and for low vs. high HRR Black patient population. Physician network segregation was lower than residential segregation (Dissimilarity 21.9 vs. 48.7, and Absolute Clustering 4.8 vs. 32.4) and positively correlated with residential segregation (p < .001). Network segregation effects differed by race and HRR Black patient population. For White patients, higher network segregation was associated with a higher probability of using the best available hospitals in HRRs with few black patients but unchanged (ACLn) or lower (Dn) probability of best hospital use in HRRs with many Black patients. For Black patients, higher network segregation was not associated with a substantial change in the probability of best hospital use regardless of the HRR Black patient population size. Measuring physician network segregation is feasible and associated with nuanced effects on Black-White differences in high-quality hospital use for heart disease. Further work is needed to understand underlying mechanisms and potential uses in health equity policy.

The geography of Medicare's hospital value-based purchasing in relation to market demographics

OBJECTIVE

To illustrate the association between the sociodemographic characteristics of hospital markets and the geographic patterns of Medicare hospital value-based purchasing (HVBP) scores.

DATA SOURCES AND STUDY SETTING

This is a secondary analysis of United States hospitals with a HVBP Total Performance Score (TPS) for 2019 in the Centers for Medicare and Medicaid Services (CMS) Hospital Compare database (4/2021 release) and American Community Survey (ACS) data for 2015-2019.

STUDY DESIGN

This is a cross-sectional study using spatial multivariable autoregressive models with HVBP TPS and component domain scores as dependent variables and hospital market demographics as the independent variables.

DATA COLLECTION/EXTRACTION METHODS

We calculated hospital market demographics using ZIP code level data from the ACS, weighted the 2019 CMS inpatient Hospital Service Area file.

PRINCIPAL FINDINGS

Spatial autoregressive models using eight nearest neighbors with diversity index, race and ethnicity distribution, families in poverty, unemployment, and lack of health insurance among residents ages 19-64 years provided the best model fit. Diversity index had the highest statistically significant contribution to lower TPS (ß = -12.79, p < 0.0001), followed by the percent of the population coded to "non-Hispanic, some other race" (ß = -2.59, p < 0.0023), and the percent of families in poverty (ß = -0.26, p < 0.0001). Percent of the population was non-Hispanic American Indian/Alaskan Native (ß = 0.35, p < 0.0001) and percent non-Hispanic Asian (ß = 0.12, p < 0.02071) were associated with higher TPS. Lower predicted TPS was observed in large urban cities throughout the US as well as in states throughout the Southeastern US. Similar geographic patterns were observed for the predicted Patient Safety, Person and Community Engagement, and Efficiency and Cost Reduction domain scores but are not for predicted Clinical Outcomes scores.

CONCLUSIONS

The lower predicted scores seen in cities and in the Southeastern region potentially reflect an inherent-that is, structural-association between market sociodemographics and HVBP scores.

Health Inequities in Coronary Artery Bypass Grafting Literature: A Scoping Review

Although life saving, health inequities exist regarding access and patient outcomes in Coronary artery bypass grafting (CABG), especially among marginalized groups. This scoping review's goal is to outline existing literature and highlight gaps for future research. Researchers followed guidance from the Joanna Briggs Institute and PRISMA extension for scoping reviews. We conducted a search to identify articles published between 2016 and 2022 regarding CABG and inequity groups, defined by the National Institutes of Health. Fifty-seven articles were included in our final sample. Race/Ethnicity was examined in 39 incidences, Sex or Gender 29 times, Income 17 instances, Geography 10 instances, and Education Level 3 instances. Occupation Status 2 instances, and LGBTQ+ 0 times. Important disparities exist regarding CABG access and outcomes, especially involving members of the LGBTQ+, Native American, and Black communities. Further research is needed to address health disparities and their root causes for focused action and improved health of minoritized groups.

Patient Characteristics Associated With Admission to Low-Safety Inpatient Psychiatric Facilities: Evidence for Racial Inequities

OBJECTIVE

The author examined patient demographic, clinical, payment, and geographic factors associated with admission to low-safety inpatient psychiatric facilities.

METHODS

Massachusetts all-payer 2017 discharge data (N=39,128 psychiatric patients) were linked to facility-level indicators of safety (N=38 facilities). A composite of safety was created by averaging standardized measures of restraint and seclusion as well as 5-year averages of overall, substantiated, and abuse-related (i.e., verbal, physical, or sexual) complaints per 1,000 discharges (α=0.73). This composite informed quintile groups of safety performance. A series of multinomial regression models were fit, with payment and geography added separately.

RESULTS

Notable factors independently associated with admission to low-safety facilities were belonging to a racial or ethnic minority group compared with being a White patient (for non-Hispanic Black, relative risk ratio [RRR]=1.71, p<0.01; for non-Hispanic Asian, RRR=5.60, p<0.01; for non-Hispanic "other" race, RRR=2.17, p<0.01; and for Hispanic-Latinx, RRR=1.29, p<0.01) and not having private insurance (for self-pay or uninsured, RRR=2.40, p<0.01; for Medicaid, RRR=1.80, p<0.01; and for Medicare, RRR=1.31, p<0.01).

CONCLUSIONS

To the best of the author's knowledge, this is the first study to examine differences in admission to low-safety inpatient psychiatric facilities. Even after accounting for potential clinical, geographic, and insurance mediators of structural racism, stark racial and ethnic inequities were found in admission to low-safety inpatient psychiatric facilities. In addition to addressing safety performance, policy makers should invest in gaining a better understanding of how differences in community-based referrals, mode of transport (e.g., police or self), and deliberate or unintentional steering and selection affect admissions and outcomes.

Racial Diversity Among American Cardiologists: Implications for the Past, Present, and Future

In the United States, race-based disparities in cardiovascular disease care have proven to be pervasive, deadly, and expensive. African American/Black, Hispanic/Latinx, and Native/Indigenous American individuals are at an increased risk of cardiovascular disease and are less likely to receive high-quality, evidence-based medical care as compared with their White American counterparts. Although the United States population is diverse, the cardiovascular workforce that provides its much-needed care lacks diversity. The available data show that care provided by physicians from racially diverse backgrounds is associated with better quality, both for minoritized patients and for majority patients. Not only is cardiovascular workforce diversity associated with improvements in health care quality, but racial diversity among academic teams and research scientists is linked with research quality. We outline documented barriers to achieving workforce diversity and suggest evidence-based strategies to overcome these barriers. Key strategies to enhance racial diversity in cardiology include improving recruitment and retention of racially diverse members of the cardiology workforce and focusing on cardiovascular health equity for patients. This review draws attention to academic institutions, but the implications should be considered relevant for nonacademic and community settings as well.

Predictors and In-Hospital Outcomes Among Patients Using a Single Versus Bilateral Mammary Arteries in Coronary Artery Bypass Grafting

The benefit of bilateral mammary artery (BIMA) use during coronary artery bypass grafting (CABG) continues to be debated. This study examined nationwide trends in BIMA use and factors influencing its utilization. Using the National Inpatient Sample, adults undergoing isolated multivessel CABG between 2005 and 2015 were identified and stratified based on the use of a single mammary artery or BIMA. Regression models were fit to identify patient and hospital level predictors of BIMA use and characterize the association of BIMA on outcomes including sternal infection, mortality, and resource utilization. An estimated 4.5% (n = 60,698) of patients underwent CABG with BIMA, with a steady increase from 3.8% to 5.0% over time (p<0.001). Younger age, male gender, and elective admission, were significant predictors of BIMA use. Moreover, private insurance was associated with higher odds of BIMA use (adjusted odds ratio 1.24) compared with Medicare. BIMA use was not a predictor of postoperative sternal infection, in-hospital mortality, or hospitalization costs. Overall, BIMA use remains uncommon in the United States despite no significant differences in acute postoperative outcomes. Several patient, hospital, and socioeconomic factors appear to be associated with BIMA utilization.

Contributions of Geography and Nongeographic Factors to the White-Black Gap in Hospital Quality for Coronary Heart Disease: A Decomposition Analysis

Background

Differences in hospital proximity and nongeographic factors affect disparities in hospital quality for heart disease, but their relative contributions are unknown. The current study quantifies the influences of these factors on the white‐black gap in high‐ and low‐quality hospital use for acute myocardial infarction (AMI) and coronary artery bypass grafting (CABG) surgery.

Methods and Results

We used Medicare claims to identify fee‐for‐service Medicare beneficiaries aged 65 and older hospitalized during 2009–2011 with AMI (n=384 443) and CABG (n=71 411). Hospital quality was measured using publicly available AMI mortality rates. In national and regional analyses, we used conditional multinomial logit models to estimate the white‐black gap in high‐ and low‐quality hospital use and decompose the gap into geographic and nongeographic contributions. Overall, more whites used high‐quality hospitals for both conditions (34.8% versus 32.4% for AMI; 39.0% versus 29.9% for CABG; P<0.001), but after accounting for distance to hospitals, the white‐black gap was significant only for CABG (9.1%; P<0.001). The nongeographic component was significant for both conditions (3.4% for AMI and 7.7% for CABG; P<0.001) and accounted for nearly the entire gap for CABG. In contrast, hospital geographic proximity was not significant. In regional analyses, white beneficiaries had higher rates of high‐quality hospital use in the Northeast (CABG) and South (AMI and CABG), whereas black had higher rates of high‐quality hospital use in the Midwest (AMI).

Conclusions

White‐black differences in high‐quality hospital use were significant for CABG and related to nongeographic factors. Interventions should consider health system and contextual reasons for these disparities.