Evaluating the Quality of Hospital Care through Severity-Adjusted Death Rates: Some Pitfalls

Designed to examine the validity of the severity-adjusted death rate proposed recently hy Roemer et al. as an outcome index of hospital quality, this study applies the formulas developed by these investigators to 1971 data for 102 short-term general hospitals in New York City. Analysis of the data suggests that: 1) occupancy-corrected length of stay may not be a valid indicator of average case severity in hospitals; 2) variations in severity-adjusted death rates according to differences in technological adequacy, control status, and teaching status of the hospitals furnish only partial support for the validity of the index; and 3) the index is too dependent of local population and other characteristics to be used as a valid indicator of quality differences between hospitals in different regions of the country. Although attempts to modify the index so as to increase validity may yet be in order, development of an index of hospital quality based directly on case-fatality rates may be more productive in the long run.