Do Patients from Vulnerable Populations Undermine Overall Hospital Performance?

Background and Objective: Studies from a decade ago concluded that hospitals with a higher proportion of Medicaid patients in their patient population had worse performance scores on processes of care important to Medicare patients. We lack studies based on more recent quality measures from the Medicare Hospital Compare program, which now determine hospital reimbursement. This study analyzes Hospital Compare data from 2013 and 2014 to compare the quality of care and costs at hospitals where a large proportion of patients come from vulnerable populations, compared to hospitals that care for few patients of low socioeconomic status.

Methods: We analyze Hospital Compare information for 2013 and 2014 for all hospitals participating in the Medicare program. Some quality measures apply to only a subset of hospitals, so that the sample size ranges from 884 to 3,669 hospitals. Quality measures include measures of timely and effective care, complication rates, readmissions, and patient satisfaction. We also analyze spending during the hospital stay and 30 days post-discharge. We test whether these outcomes vary according to the share of patients in the hospital who are black, as well as quartiles of the hospital’s disproportionate share index. The analyses control for multiple hospital characteristics.

Results: Preliminary results reveal no, or small differences in most measures of timely and effective care and patient satisfaction for hospitals that have high versus low proportions of black patients under their care; or in hospitals in the highest versus the lowest quartile of the DSH index. Most measures of timely and effective care have little room to vary and average close to 95%. Patient satisfaction measures are on average much lower, but still vary little. However, there are significant differences in waiting time for emergency room care by hospital race composition and hospital DSH index, although we have not yet adjusted for ER volume. Other hospital characteristics are associated with wider variation in outcomes, including hospital bed size and hospital ownership type. We are conducting analyses of complications, readmission rates, and spending. We will update this abstract before the conference to incorporate these results.

Conclusion: CMS launched the value based purchasing program for hospitals in 2012, in an effort to link hospital reimbursement to quality measures. Some policy analysts are concerned that the program unfairly penalizes hospitals that care for disproportionately more minority or low-income patients, because their outcomes may be influenced more by socioeconomic status rather than by hospital quality. We find that hospitals with high shares of black patients or with a DSH index in the upper quartile perform similarly to other hospitals in most measures of timely and effective care and in patient satisfaction. These results suggest that it may not be necessary to adjust the value based purchasing reimbursement rules to account for socioeconomic patient case mix. No such adjustments currently occur. We may modify this conclusion after we complete our full analysis.