Objective: To develop a diabetes registry from an outpatient pharmacy database to systematically analyze the prevalence of diabetes, patterns of glycemic medication and glucose monitoring, pharmacy costs, and hospital use related to diabetes care in the Veterans Health Administration (VHA) in fiscal year (FY) 1994.
Research design and methods: Veterans with diabetes were identified using a software program that extracted the social security number (SSN) of patients receiving insulin, sulfonylurea agents, or glucose-monitoring supplies. The cumulative FY94 cost for a drug was calculated by multiplying the units dispensed times the unit cost for each fill, using the actual drug cost that was in effect at the time of dispensing. Admission data were obtained by crossmatching the SSN registry with the VHA Austin Mainframe Patient Treatment Files to retrieve associated diagnosis-related groups (DRG), Physicians' Current Procedural Terminology (CPT), and International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) codes.
Results: From among 1,180,260 unique patients, 139,646 veterans with diabetes receiving insulin, oral agents, or glucose-monitoring strips were identified, accounting for a prevalence of 11.83% from 62 Veterans Administration medical centers. There were 63,078 individuals (52%) who received oral agents, of whom 26.3% also received blood glucose-monitoring supplies; 46,664 individuals (39%) received insulin, of whom 53.2% received blood glucose-monitoring supplies; and 9,440 individuals (8%) received both oral agents and insulin during FY94, with 64.4% receiving blood glucose-monitoring supplies. Only 1,482 (1.2%) individuals received monitoring supplies alone, and 129 patients (0.1%) were provided with an insulin pump. Using an adjusted data set, 12% of veterans accounted for 24% of all outpatient pharmacy costs, with an average expenditure of $622 for veterans with diabetes compared with $276 for veterans without diabetes. There was $454 (73%) for non-diabetes-specific prescriptions and $168 (27%) for prescriptions related to glycemic control. Of pharmacy expenditures for glycemic control, $101 (60.1%) was attributed to insulin, oral agents, and supplies, while $67 (39.9%) was attributable to glucose monitoring. Veterans with diabetes were admitted 1.6 times as frequently as veterans without diabetes.
Conclusions: This study demonstrates the feasibility of using a pharmacy-based electronic diabetes database in a payor system that can track both claims and individual classes of medication based on a unique identifier number. While the prevalence of diabetes in the VHA is high relative to other health care systems and the general population, patterns of medication usage, pharmacy costs, and relative admission frequency are comparable to results from the private sector.