Abstract 116: Temporal Trends and Hospital-level Variation in the Incidence and 30-day Mortality of In-Hospital Cardiac Arrest within the Veterans Health Administration

Background: Improving the quality of in-hospital cardiac arrest (IHCA) care within the Veterans Health Administration (VHA) has received significant attention. Yet there are no national VHA data on the incidence and mortality outcomes of IHCA to guide or evaluate these efforts. We sought to determine overall trends and hospital-level variation in the incidence and 30-day mortality of IHCA within the VHA. Methods: Among 2,731,295 patients hospitalized at 115 VHA hospitals between 2008 and 2012, we defined IHCA using specific ICD-9 procedure codes for cardiac arrest and cardiopulmonary resuscitation. Among patients suffering IHCA, we used the VA Vital Status file to identify 30-day mortality from hospital admission. A severity of illness score was used to account for case-mix and determined from a logistic multivariate adaptive regression spline (MARS) model fit to our mortality outcome with covariates for age, race, gender, admission diagnosis category, 29 comorbid conditions, and 11 lab values drawn within 24 hours of admission. Hospital-level IHCA incidence and 30-day mortality rates were compared using empirical Bayes random effects estimates from multi-level regression models after risk- and reliability-adjustment. Results: 8,565 (0.3%) patients suffered IHCA between 2008 and 2012 and there was no significant trend in the rate of IHCA over this time period. The hospital-level incidence of IHCA varied and was statistically significantly higher than the median rate at 38 (34%) hospitals and significantly lower at 24 (21%) hospitals (Figure A, p<0.05 without adjustment for multiple comparisons). Among patients suffering IHCA, the overall 30-day mortality rate was 68.6% and the risk-adjusted 30-day mortality rate decreased from 71.2% in 2008 to 66.1% in 2012 (p for trend <0.01). Hospital-level 30-day mortality was significantly higher than the median rate at 5 (4%) hospitals and significantly lower at 7 (6%) hospitals (Figure B). Conclusions: Within the VHA, the incidence of IHCA has remained stable while 30-day mortality has improved. However, hospital-level variation in IHCA incidence and mortality rates suggest variation in care processes related to IHCA and a target for future investigation to improve patient outcomes.