[Skip to Content]
[Skip to Content Landing]
Article
October 17, 1990

Studying the Effects of the DRG-Basedn Prospective Payment System on Quality of CareDesign, Sampling, and Fieldwork

Author Affiliations

From the Health Program of the RAND Corp, Santa Monica, Calif (Drs Draper, Kahn, Sherwood, Keeler, Rogers, Allen, Wells, and Brook, Mss Reinisch and Carney, and Mr Reboussin); the RAND Graduate School of Policy Studies, Santa Monica, Calif (Drs Draper, Keeler, Rogers, and Brook); the Departments of Medicine (Drs Kahn, Kosecoff, and Brook), Health Services (Drs Kosecoff and Brook), and Psychiatry and Biobehavioral Sciences (Dr Wells), UCLA; Value Health Sciences Inc, Santa Monica, Calif (Dr Kosecoff); and the Office of Research and Demonstrations of the Health Care Financing Administration, Baltimore, Md (Dr Savitt).

From the Health Program of the RAND Corp, Santa Monica, Calif (Drs Draper, Kahn, Sherwood, Keeler, Rogers, Allen, Wells, and Brook, Mss Reinisch and Carney, and Mr Reboussin); the RAND Graduate School of Policy Studies, Santa Monica, Calif (Drs Draper, Keeler, Rogers, and Brook); the Departments of Medicine (Drs Kahn, Kosecoff, and Brook), Health Services (Drs Kosecoff and Brook), and Psychiatry and Biobehavioral Sciences (Dr Wells), UCLA; Value Health Sciences Inc, Santa Monica, Calif (Dr Kosecoff); and the Office of Research and Demonstrations of the Health Care Financing Administration, Baltimore, Md (Dr Savitt).

JAMA. 1990;264(15):1956-1961. doi:10.1001/jama.1990.03450150056031
Abstract

We have conducted a nationally representative before-after study of the effects of the diagnosis related groups—based prospective payment system (PPS) on quality of in-hospital care for aged Medicare patients. We used a pre-post design with multiple time points in both the pre-PPS (calendar years 1981 and 1982) and post-PPS (July 1985 through June 1986) periods. We gathered clinically detailed data from medical records of patients with one of six diseases and supplemented these data with postdischarge information from Health Care Financing Administration files. We used a stratified multistage cluster sampling design with data gathered on 16 758 patients chosen from 297 hospitals in 30 areas in five states. Our hospital participation rate was 97%; we successfully accessed 96% of the medical records we requested; and our mean item-level reliability score was 0.80. Our sample matches the nation closely on hospital urbanicity, size, teaching status, ownership, and percentages of Medicare and Medicaid patients, and patient demographics and mortality.

(JAMA. 1990;264:1956-1961)

×