Abstract Pay-for-performance contracts in health care are popular, but have been implemented without explicit consideration of the market failures they are addressing, or of how pay-for-performance interacts with existing payment mechanisms. The fundamental insight that I explore in this paper stems from the recognition that some attributes of health care are easier for consumers to observe than others. Traditional, demand-based payment systems result in under-provision of those attributes that are poorly observed by consumers, but not those that are well-observed. I use a model of provider competition on two dimensions of quality to argue that pay-for-performance contracts should target those dimensions of quality that are under-rewarded by the existing system. These will be dimensions that are poorly observed, not necessarily those that are most important for improving patient health. Policymakers often worry about multitasking problems arising from pay-for-performance, but I argue that pay-for-performance actually provides a solution to multitasking problems that exist in traditional payment systems.
Objective. To analyze whether acute care patients with dementia are more or less likely to receive each of five aggressive medical services near the end of life, compared with patients without dementia. Data Sources. Two years of Veterans Affairs (VA) and Medicare utilization data for all 169,036 VA users nationwide age 67 and older who died between October 1, 1999 and September 30, 2001. Study Design. We performed a retrospective analysis of acute care stays discharged in the final 30 days of life. The main outcome measure was the patient's likelihood of receiving each of five aggressive services (intensive care unit [ICU] admission, ventilator, cardiac catheterization, pulmonary artery monitor, and dialysis), controlling for demographic and clinical factors in probit regressions. Principal Findings. There were 122,740 acute-stay discharges during the final 30 days of life, representing 94,100 unique patients (31,654 with dementia). In probit models comparing acute care patients with and without dementia, patients with dementia were 7.5 percentage points less likely to be admitted to the ICU (95 percent confidence interval [CI], 6.9-8.1; percent of stays with ICU admission=36.8 percent), 5.4 percentage points less likely to be placed on a ventilator (95 percent CI, 5.0-5.9; percent of stays with ventilator use=17.1 percent), 0.7 percentage points less likely to receive cardiac catheterization (95 percent CI, 0.6-0.8; percent of stays with cardiac catheterization=2.7 percent), 1.4 percentage points less likely to receive pulmonary artery monitoring (95 percent CI, 1.2-1.5; percent of stays with pulmonary artery monitoring=2.6 percent), and 0.6 percentage points less likely to receive dialysis (95 percent CI, 0.4-0.8; percent of stays with dialysis=4.6 percent). Conclusions. During the final 30 days of life, acute care patients with dementia are treated substantially less aggressively than patients without dementia. Further research is warranted to determine the causes and appropriateness of these patterns of care.
We investigated the determinants of inpatient rehabilitation costs in the Department of Veterans Affairs (VA) and examined the relationship between length of stay (LOS) and discharge costs using data from VA and community rehabilitation hospitals. We estimated regression models to identify patient characteristics associated with specialized inpatient rehabilitation costs. VA data included 3,535 patients discharged from 63 facilities in fiscal year 2001. We compared VA costs to community rehabilitation hospitals using a sample from the Uniform Data System for Medical Rehabilitation of 190,112 patients discharged in 1999 from 697 facilities. LOS was a strong predictor of cost for VA and non-VA hospitals. Functional status, measured by Functional Independence Measure (FIM) scores at admission, was statistically significant but added little explanatory value after controlling for LOS. Although FIM scores were associated with LOS, FIM scores accounted for little variance in cost after controlling for LOS. These results are most applicable to researchers conducting cost-effectiveness analyses.