Lead/Presenter: Kertu Tenso, PEPReC
All Authors: Tenso K (Partnered Evidence-Based Policy Research Center, VA Boston Healthcare System; Boston University School of Public Health), Strombotne KL (Partnered Evidence-Based Policy Research Center, VA Boston Healthcare System; Boston University School of Public Health) Feyman Y (Partnered Evidence-Based Policy Research Center, VA Boston Healthcare System; Boston University School of Public Health) Auty SG (Boston University School of Public Health) Legler A (Partnered Evidence-Based Policy Research Center, VA Boston Healthcare System) Griffith KN (Partnered Evidence-Based Policy Research Center, VA Boston Healthcare System; Vanderbilt University Medical Center)
Objectives:
The COVID-19 pandemic has resulted in significant excess mortality among both the general U.S. population and at the Veterans Health Administration (VHA). However, these national estimates may mask important facility-level variation in excess mortality and its correlation with institutional characteristics. The objective of this research was to identify excess mortality among VHA-enrolled veterans during the pandemic for each medical center, and to correlate these estimates with facility characteristics and rates of COVID-19 cases and deaths within each facility’s catchment area.
Methods:
We obtained data for 11.4 million unique veterans who were enrolled in the VHA and sought care during 2016-2020 using the VHA’s Corporate Data Warehouse (CDW). We estimated a mortality risk prediction model using five-fold cross-validation and Poisson quasi-likelihood regression. Potential predictors of mortality included individual-level demographics, priority group (an eligibility determination which reflects disability related to military service or economic hardship), and major comorbidities. We then estimated excess mortality and observed versus expected (O/E) mortality ratios for veterans assigned to each VHA facility during March-December 2020. We graphed our findings and correlated these estimates with facility characteristics (number of hospital beds, percent rural enrollees, percent highest complexity, number of staff, capacity in clinic days, relative telehealth percent change) and rates of COVID-19 cases and deaths among the general population.
Results:
VHA-enrolled Veterans experienced 52,038 excess deaths from all causes during March-December 2020, equating to a 16.8% excess all-cause mortality rate. However, there was substantial variation in facility-level excess mortality rates, with estimates ranging from -5.5% to +63.7%. Facilities in the lowest quartile for excess mortality experienced fewer COVID-19 deaths (0.7 vs. 1.51, p < 0.001) and cases (52.0 vs. 63.0, p = 0.002) per 1,000 population compared to facilities in the highest quartile. Compared to facilities in the lowest quartile of excess mortality, facilities in the highest quartile were more likely to have more beds (276.7 vs. 187.7, p = 0.02) and a higher percent change in the share of visits that were conducted via telehealth from 2019-2020 (183% vs 133%, p = 0.01).
Implications:
There was wide variation in excess all-cause mortality across VHA facilities during the COVID-19 pandemic, which was only partially explained by the burden of COVID-19 illness within each facility’s catchment area. Facility-level excess mortality was associated with hospital bed size and a likely endogenous expansion of telehealth capacity, but was not associated with rurality, # of staff or clinic days, or patient complexity.
Impacts:
Our work provides valuable insights for performance management and institutional learning at the VHA.