An analysis of multilevel factor contributions to breast cancer screening inequities in an academic health system

Abstract

Background

Breast cancer screening (BCS) inequities are evident at national and local levels, and many health systems want to address these inequities, but may lack data about contributing factors. The objective of this study was to inform health system interventions through an exploratory analysis of potential multilevel contributors to BCS inequities using health system data.

Methods

The authors conducted a cross-sectional analysis within a large academic health system including 19,774 individuals who identified as Black (n = 1445) or White (n = 18,329) race and were eligible for BCS. They evaluated individual-level, provider-level, and clinic-level factors. They conducted logistic regression and Blinder-Oaxaca (BO) decomposition analyses to quantitatively estimate the contribution of factors to the mean difference in BCS outcomes between the two racialized groups. They calculated average marginal effects (AME) for the logistic regression models representing the estimated additive probability of receiving BCS in the Black versus White group.

Results

BCS was completed in 63.7% of Black and 71.7% of White individuals (AME, –0.08; 95% confidence interval (CI), –0.10 to –0.04; p < .001). In the BO decomposition, observed factors explained 13.3% difference in BCS. Lower patient portal use among Black versus White patients had the greatest estimated contribution to the BCS inequity (4.6 percentage points; 95% CI, 3.0–6.2).

Conclusion

Racialized group differences in patient portal use had the greatest estimated contribution to the explained difference in BCS between Black and White individuals. Patient portal use promotion could be considered as a part of multifaceted health system efforts to address BCS inequities.