Quality-adjusted efficiency measures for public hospitals

Abstract

Public hospital expenditures are a major cause of rising healthcare costs. Many remedies have been proposed, but a widespread concern is that some of these ideas may lower quality of care. In the literature there are many studies that discuss this very issue, but considerably less attention has been devoted to the way hospital quality should be estimated. This is important because inaccurate estimates of hospital quality may distort results for empirical studies that incorporate quality in the analysis.

This thesis contributes to the literature on estimating hospital quality by focusing on three specific cases. The first part pertains to the use of the standardised incidence ratio (SIR), commonly used in health economics as a risk-adjusted measure of performance. Existing analytical methods tend to capture the statistical dispersion of the SIR in simplistic ways. Bootstrapping or simulations are obvious answers, but they tend to be overly time-consuming. This thesis adopts a number of analytical approximations from medicine and applied statistics, and compares their performance against each other as well as the bootstrapped results. The main findings highlight the importance of recognising the stochastic element in the SIR; methods that explicitly account for this produce more accurate confidence intervals than those that did not do so.

The second part of this study investigates the significance of how hospital quality is defined and incorporated into empirical studies. Two issues relating to hospital quality are examined: the stochastic nature of quality statistics, and the multifaceted nature of quality of care. The first issue is tackled using bootstrap regression, adjusting for the distribution of the quality statistic as measured by the SIR. It is found that with bootstrapping, the statistical significance of the quality parameter increased noticeably. The second issue is addressed by introducing a second quality measure into the production function. Results for the two measures were significantly different, highlighting the importance of carefully defining hospital quality.

Part three of the thesis examines the implications of disease aggregation, a little discussed issue in the literature. This study investigates the importance of disease aggregation via an empirical investigation of the relationship between hospital quality and efficiency. Results indicate an overall inverse relationship, but the magnitude and policy implications change significantly as disease aggregation becomes more refined.