1070 — Examining Facility-Specific Disparities in Electronic Quality Measures to Guide Equity-Focused Priorities

Lead/Presenter: Leslie Hausmann,  COIN - Pittsburgh/Philadelphia
All Authors: Hausmann LR (Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System), Lovelace, E (Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System, Pittsburgh, PA) Cashy, J (Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System, Pittsburgh, PA) Moy, E (Office of Health Equity, Veterans Health Administration, Washington, DC)

Objectives:
Quality measures used to guide VHA priorities do not identify inequities across populations. New methods to isolate and report facility-specific gaps in quality for patients with different social risks have been developed. This study used one such method to calculate national and facility-specific racial and ethnic disparities in diabetes quality measures from the VHA Electronic Quality Measurement (eQM) Program.

Methods:
Outcomes included three quality measures for optimal diabetes management calculated by the VHA eQM Program, including HbA1c < 9%, HbA1c < 8%, and blood pressure (BP) < 140/90. The eQM measures are calculated for the entire VHA enrolled population on a monthly basis using a point-in-time estimate that determines the number of people who were eligible for (denominator) and achieved the target goal (numerator) over the last 12 months. We analyzed eQM measures calculated on September 30, 2021. Using race and ethnicity data from the Corporate Data Warehouse, Veterans in the eQM cohort were categorized as follows for analysis: Hispanic (H), Non-Hispanic Black (B), Non-Hispanic white (W), and a combined (C) category of smaller racial/ethnic groups including Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, and multiple races. We used mixed effect random slope models to estimate a fixed effect for overall B-W, H-W, and C-W differences in passing rates for each measure and variation in facility-specific differences. Overall differences were reported as odds ratios (OR). Modeling results were used to calculate absolute rate differences (ARD) for each measure at each facility for all comparisons and median and interquartile range (IQR) of ARDs were reported.

Results:
Analyses included 740,557 Veterans (466,493 W; 175,633 B; 58,337 H; and 40,094 C) with diabetes. The percentage Veterans who met the target goal for each quality measure was higher among W Veterans than among B, H, and C Veterans, respectively: HbA1c < 9: 79.7%, 74.6%, 75.5%, 75.9%; HbA1c < 8: 71.0%, 66.6%, 66.8%, 67.2%; BP < 140/90: 66.2%, 60.3%, 63.1%, 62.6%. ORs were statistically significant (p < .0001) for all B-W, H-W, and C-W comparisons, ranging from OR = 0.75 for H-W differences in HbA1c < 9 to OR = 0.94 for C-W differences in BP < 140/90. Variation in facility-specific effects was also statistically significant for all comparisons except H-W variation in BP < 140/90.

Implications:
There are significant racial and ethnic differences in quality of diabetes management in the Veteran population overall. The magnitude of disparities also varies significantly across facilities.

Impacts:
This study demonstrates the feasibility of generating overall and facility-specific disparities in eQM measures. This method could be leveraged to set national goals for reducing disparities and engage facilities in equity-guided improvement initiatives.