Updated Nonresponse Adjustment for the Department of Veterans Affairs Bereaved Family Survey Performance Measure.

Abstract

Context: Declining response rates pose a threat to the validity of large surveys designed to measure quality of care. For over a decade, the Department of Veterans Affairs has evaluated the quality of end-of-life care delivered to Veterans using the Bereaved Family Survey (BFS). The BFS Performance Measure (BFS-PM), derived from the BFS, is a widely used metric across the VA healthcare system that informs facility benchmarking and quality improvement efforts.

Objectives: The objective of this analysis was to develop and test an updated nonresponse adjustment model for the BFS-PM.

Methods: This cross-sectional analysis used BFS data linked with socio-demographic and clinical data available in VA's Corporate Data Warehouse for October 1, 2021-September 30, 2022. A direct inverse propensity weighting approach was applied to adjust for BFS nonresponse. Facility-level differences in BFS-PM scores were calculated before and after weighting to describe the effects of nonresponse adjustment.

Results: The analytic sample consisted of 14,510 Veteran decedents whose family member was eligible to receive a BFS, including 8,698 (59.9%) respondents and 5,812 (40.1%) nonrespondents. Across the 145 VA Medical Centers represented in the sample, the mean facility-level response rate was 40.1%. The mean change in the facility-level BFS-PM score from pre- to postweighting for nonresponse was 1.2% (standard deviation = 3.4%) and ranged from -11.2% to 11.9%.

Conclusion: Adjustment for nonresponse bias in the calculation of facility-level BFS-PM scores continues to be an important and appropriate practice in the evaluation of quality of end-of-life care in VA inpatient facilities.