Utilization trends and risk factors for hospitalization in diabetes mellitus.

A population-based prevalence cohort of 1,111 residents of Rochester, Minnesota, who had diabetes mellitus on Jan. 1, 1975, was subjected to follow-up assessment for hospitalizations through Dec. 31, 1980. On the basis of these data, hospitalization rates were calculated for various clinical types of diabetes, and a risk factor analysis was done for non-insulin-dependent diabetes mellitus (NIDDM) to identify high-risk persons for subsequent intervention studies. The adjusted incidence density of hospitalization was 141.6 per 1,000 person-years for NIDDM and 331.3 per 1,000 person-years for insulin-dependent diabetes. Although the modeled clinical characteristics accounted for little variability in NIDDM-related hospitalization, age modified by the effect of gender was the strongest risk factor found (multivariate hazard ratios: 1.0 and 1.43, respectively, for male and female patients younger than 65 years old; 1.88 and 1.83, respectively, for male and female patients 65 years old or older); coronary heart disease, diabetic retinopathy, and persistent proteinuria were associated with a 50% increased risk. Although older patients with NIDDM (especially men) are at greatest risk for a first hospitalization, clinical factors alone seem inadequate to account for these hospitalizations. The effect of Medicare's prospective payment systems (PPS) was studied by using a data base for Olmsted County, Minnesota, to determine whether PPS decreased the rate of hospitalizations among patients with diabetes. Among Olmsted County residents 65 years of age or older, the adjusted rate of diabetes-associated hospitalizations decreased from 26.5 per 1,000 person-years in 1980 to 16.7 in 1985, whereas the adjusted rate of all other hospitalizations increased from 259.5 per 1,000 person-years to 261.9. Thus, PPS may have reduced hospitalization rates in elderly patients with diabetes.