Racial and Ethnic Differences in Low-Value Care Among Older Adults in a Large Statewide Health System

Abstract

Background: As value-based payment models incorporate both measures of health equity and low-value care (LVC), understanding how LVC varies by race is vital for interventions. Therefore, we measured racial differences in LVC in a contemporary sample.

Methods: We conducted a cross-sectional analysis of claims from adults ≥ 55 years receiving care at five academic medical centers in California from 2019 to 2021. Our sample included patients who received a service that could be classified as LVC. The primary outcome was whether a service was classified as LVC. Secondary outcomes included clinical categories of LVC (preventive screening, diagnostic testing, prescription drugs, and preoperative testing). We examined associations between race/ethnicity with outcomes using multivariable regression models adjusted for patient characteristics and medical center.

Results: Among 15,720 members who received potentially LVC, non-Hispanic White older adults comprised 59% of the sample, followed by Asian (17%), unknown race (8%), Latino (8%), non-Hispanic Black (5%), other race (2%). In adjusted models, Asian (-4.9 percentage points [pp]; 95% CI -5.9, -3.8 pp), Black (-5.4 pp; 95% CI -8.0, -2.7 pp), and Latino (-2.5 pp; 95% CI -4.6, -0.4 pp) older adults were less likely to receive LVC compared to White older adults, specifically preventive and preoperative services. Asian, Black, and Latino older adults, however, were more likely to receive low-value prescriptions.

Conclusions: These diverging racial patterns in LVC across different measures likely reflect differential mechanisms, underscoring the need to use clinically specific measures rather than composite measures, which obscure underlying heterogeneity and could lead to potentially harmful and inequity-producing interventions.

Keywords: low‐value care; quality of care; racial disparities.

FIGURE 1

Adjusted probabilities and average marginal differences of receiving any low‐value care among adults 55 years and older in UC Self‐Funded Health Plans by race and ethnicity, 2019–2021. Each marker reports the average marginal difference in the adjusted probability of receiving any of the 38 services measured by the Milliman Health Waste Calculator for each racial or ethnic group (compared to White older adults). Results are derived from a generalized linear model with binomial family and logit link, adjusted for age, sex, plan type (HMO vs. PPO), time period (pre and post March 2020), outpatient utilization (number of visits), number of comorbidities (1, 2, 3, and 4+), and fixed effects for each UC medical center. Error bars indicate the 95% CI for each estimate. The red vertical line denotes an average marginal difference of 0 from the reference group, non‐Hispanic White older adults. p values are adjusted for multiple comparisons using the Benjamini–Hochberg procedure.

FIGURE 2

Proportions of low‐value care by category among adults 55 years and older in UC Self‐Funded Health Plans by race and ethnicity, 2019–2021. Probability of receiving low‐value preventive services and diagnostic tests (panel A) and prescriptions and preoperative services (panel B), adjusted for age, sex, plan type (HMO vs. PPO), time period (pre and post March 2020), outpatient utilization (number of visits), number of comorbidities (1, 2, 3, and 4+), and UC medical center. Error bars indicate the 95% CI for each estimate. * indicate adjusted p < 0.05. Results suppressed if sample n < 10. NH = non‐Hispanic.