Background: Conventional epidemiologic data suggest that diabetic patients use more health care resources than nondiabetic patients, yet overall health care use by diabetic individuals has never been fully quantitated. We took a new approach to this issue based on the actual economics of the provision of health care to diabetic insured individuals.
Methods: The claims records in the Mutual of Omaha Current Trends database, which contains information on more than 400,000 individuals, were surveyed to identify patients with diabetes and create the contrast population of nondiabetic patients by exclusion. International Classification of Diseases, Ninth Revision, Clinical Modification, codes and Physicians' Current Procedural Terminology, Fourth Edition, codes were used to determine all diagnoses recorded and all physician services rendered to the contrast populations. Age- and sex-adjusted comparisons were performed using Mantel-Haenszel procedures to determine an adjusted odds ratio (AOR).
Results: A total of 13,304 diabetic individuals and 388,053 nondiabetic individuals who received health care services from January 1, 1988, to January 1, 1989, were identified. Diabetic insured individuals constituted 3.1% of the overall insured population yet accounted for 8.3% of the charges (P < .01). Inpatient charges accounted for 81% of total diabetic charges but only 61.5% of total nondiabetic charges (P < .001). Diabetic insured individuals had twice as many physician office visits (AOR = 1.87; 95% confidence interval [CI], 1.79 to 1.96), with 2.5 times more physician hospital visits [AOR = 2.50; 95% CI, 2.27 to 2.75). However, the increases in physician care were not uniformly distributed across the diagnostic spectrum. The frequencies of well-established complications of diabetes, such as ischemic heart disease (AOR = 3.32; 95% CI, 3.12 to 3.53), peripheral vascular disease (AOR = 3.14; 95% CI, 2.79 to 3.53), and eye disease (AOR = 3.10; 95% CI, 2.94 to 3.27), were threefold higher in the diabetic group, with parallel increases in related medical services, such as cardiac catheterization (AOR = 3.02; 95% CI, 2.27 to 4.0), vascular surgery (AOR = 2.94; 95% CI, 2.64 to 3.27), and ophthalmologic procedures (AOR = 2.94; 95% CI, 2.72 to 3.18). In contrast, most diagnostic categories showed little or no increase. For example, the frequency of neoplasms (AOR = 1.11; 95% CI, 1.03 to 1.19) was minimally increased, and the associated procedural concomitants of therapeutic radiology (AOR = 0.81; 95% CI, 0.47 to 1.39) and chemotherapy (AOR = 0.98; 95% CI, 0.60 to 1.60) were not increased in the diabetic group.
Conclusions: Our most important new finding is that diabetic patients have neither an elevated risk for a wide spectrum of diseases nor an increase in the receipt of physician services for diagnostic categories without increased risk, despite more frequent physician encounters. We provide real-world risk estimates that help in calculating the effect of offering specific insurance to diabetic individuals or including them in group health plans. The techniques we have developed to analyze computerized claims databases in this way may serve to better quantify the true impact of chronic diseases on the health care system.