Dissertations and Theses

Date of Degree

1-1-2022

Document Type

Dissertation

Degree Name

Doctor of Public Health (DPH)

Department

Health Policy and Management

Advisor(s)

Elizabeth Eastwood

Committee Members

Alexis Pozen

Naomi Zewde

Subject Categories

Health Services Research | Public Health | Quality Improvement

Keywords

hospital readmissions reduction program, hospital ownership, mortality, pay-for-performance

Abstract

Background: The Hospital Readmissions Reduction Program (HRRP) is an aspect of Patient Protection and Affordable Care Act (ACA) of 2010. HRRP requires Medicare to reduce payments to hospitals with relatively high readmission rates. While the implementation of the program has been accompanied by reductions in readmission rates, some have expressed concerns about unintended consequences, including harm to patient care and even death. Under HRRP, hospitals serving a high proportion of socioeconomically disadvantaged patients, like public hospitals, have been more likely to receive financial penalties, potentially impeding their ability to invest resources in improving patient outcomes. There is previous research on the relationship between hospital ownership and quality of care as well as between hospital ownership and patient outcomes. There is also research on the relationship between HRRP implementation and mortality. This project is distinguished by its focus on the relationship between penalties assessed under HRRP and 30-day risk standardized mortality rate (RSMR) modified by hospital ownership.

Methods: To estimate the odds that a hospital incurring a penalty performs worse than the national 30-day RSMR for each condition (Aim 1), I conducted multiple logistic regression analyses for each condition separately for each year. To estimate the association between penalty and RSMR compared by hospital ownership (Aim 2), I conducted multiple linear regression analysis stratified by ownership for each condition separately for each year as well as multivariate linear regression analysis with fixed effects to examine change in RSMR for a given hospital. A multiple linear regression with a term representing the interaction of hospital ownership type and payment adjustment as well as multivariate linear regression with fixed effects was conducted for Aim 3.

Results: Multiple logistic regression showed that, with some exceptions, penalization was generally associated with a decrease in the odds of exceeding the RSMR. However, the relationship was significant only for COPD in 2016, 2017, and 2018 and heart failure in 2016 and 2017. Hospital ownership, on the other hand, was, with one exception, associated with increases in the odds of exceeding national 30-day RSMR for publicly owned hospitals across all conditions and all years with significance for AMI in 2016 and 2017, heart failure in 2017 and 2018, and pneumonia in 2018. Under the stratified multiple linear regression analyses, average penalties were consistently associated with significant percentage point reductions in the average 30-day RSMR for COPD over all measurement periods for all three types of hospital ownership. Although not all of the associations were significant, average penalties were also consistently associated with percentage point reductions in the average RSMR for heart failure across hospital ownership types. However, under the fixed effects model, each percentage point increase in average penalty was associated with a significant 0.49 percentage point increase in the heart failure RSMR for any given public hospital. Examining the interaction of hospital ownership with average penalty and its association with 30-day RSMR through a multiple linear regression of 30-day RSMR by condition for each year and a multivariate linear regression of 30-day RSMR with fixed effects found some consistency between the two models. For public hospitals, the directions of the associations were uniformly positive for CABG (which was significant in 2018 as well as under the fixed effects model). In the case of heart failure RSMR, however, the direction of the association reversed. Under the linear regression conducted for each year, there was an inverse relationship between average penalty and heart failure RSMR among public hospitals. Under the fixed effects model, each additional percentage point increase in average penalty was associated with a significant 0.51 point increase in heart failure RSMR.

Conclusions: Results from the stratified linear regression, which found that the average penalty was significantly associated with decreases in RSMR for COPD across all hospital types for all years, initially reinforced results from the logistic regression model which found penalization significantly associated with a decrease in the odds of exceeding the RSMR. While not as consistently significant, results under the logistic regression and the stratified linear models for some other conditions, like heart failure, suggested that perhaps some conditions were more responsive than others to quality initiatives. Tests of the interaction of hospital ownership with average penalty rarely demonstrated that the association between penalization and RSMR varied significantly by ownership type. Analyses employing fixed effects found that the regression of RSMR on penalization stratified by ownership as well as that on the interaction of hospital ownership with penalization was significant for heart failure among public hospitals. But, the reversal of the direction of the associations suggested that failure to control for time-invariant characteristics may not sufficiently characterize the effect of penalties on 30-day RSMR.

Available for download on Saturday, December 10, 2022

Share

COinS