Association between degree of exposure to the Hospital Value Based Purchasing Program and 30-day mortality: experience from the first four years of Medicare’s pay-for-performance program

Table 2 Difference in 30-day risk adjusted mortality trends between high and low Medicare share hospitals (2009–2016)

	Average 30-day risk-adjusted mortality
	High Medicare share hospitals	Low Medicare share hospitals
A. Acute myocardial infarction
Pre-HVBP: Mean (SD)	15.9% (1.7%)	16.0% (1.8%)
Post-HVBP: Mean SD)	14.8% (1.6%)	14.7% (1.5%)
Pre- to Post change	−1.1%	−1.3%
Difference-in-differences (adjusted)	0.03% (95% CI: −0.13, 0.19%)
B. Heart failure
Pre-HVBP: Mean (SD)	11.0% (1.6%)	10.9% (1.6%)
Post-HVBP: Mean (SD)	11.6% (1.5%)	11.4% (1.6%)
Pre- to Post change	0.6%	0.5%
Difference-in-differences (adjusted)	0.04% (95% CI: −0.08, 0.16)
C. Pneumonia
Pre-HVBP: Mean (SD)	11.4% (1.9%)	11.4% (1.9%)
Post-HVBP: Mean (SD)	11.7% (1.8%)	11.6% (1.8%)
Pre- to Post change	0.3%	0.02%
Difference-in-differences (adjusted)	0.11% (95% CI: −0.04, 0.27)

Sources: Authors’ analysis of Hospital Compare (2009–2016), American Hospital Association Annual Survey (2009), and Final Impact Rules (2009–2016) data
Observed average mortality rates reported for pre-HVBP and post-HVBP periods. Difference in average mortality rate between post- vs. pre-HVBP periods reported for high and low Medicare share hospitals based on linear random effects model regressing mortality rate on post-HVBP indicator; heteroscedasticity-robust standard errors clustered at the hospital level
Difference-in-differences estimates from random effects model reported; heteroscedasticity-robust standard errors clustered at the hospital level; covariates in the model include teaching hospital status, ownership, bed size, year and region
SD Standard deviation
CI Confidence interval
p < 0.05

ISSN: 1472-6963