The primary goal of this study is to compare diabetics and non-diabetes regarding oral health status, use of dental services and cost of dental care. The design is a retrospective study utilizing the records of Kaiser Permanente Northwest, a large HMO in the Portland, Oregon area. The principal investigator has conducted a series of preliminary studies to ensure that the databases to be used are 1) available, 2) accurate, and 3) valid. The study will link individuals' data from 5 electronic databases within the KPNW HMO: 1) MIPS, the Membership Information Processing System, which has records on 386,000 individuals dating back to 1982; 2) the KPNW Diabetes Registry which has records on 12,755 active HMO members dating back to 1988; 3) the TEAM Database which has data on all dental office visits and procedures on 155,000 members (no starting date provided); 4) the LIS, the Laboratory Information System, which provides an integrated base for medical laboratory procedures including results interpretation and reporting for all outpatient work including anatomical pathology as well as clinical laboratory test results; and 5) TOPS, The Outpatient Pharmacy System, which records every prescription dispensed by KPNW's outpatient pharmacies dating back to 1986. The study group with diabetes will be drawn from the nearly 5,000 members of this HMO plan who have diabetes and use the HMO's dental plan. The comparison group will be drawn from KPNW members who use the HMO's dental plan. The groups will be stratified on age and gender and matched on these characteristics. Variables of interest to be taken from electronic databases include: 1) use of dental services, 2) mean number of visits, 3) mean number of extractions, 4) mean number of types of dental services (diagnostic, preventive, restorative, periodontal, oral surgery and other), 5) mean costs of dental services (overall and within specific type of dental services), 6) level of diabetic control (as measured by hemoglobin A1c and by the fructosamine (or glycated protein) test, and 7) current diabetic medications. The data for missing teeth and level of edentulism will be obtained from a dental chart of a random sample of individuals selected by age, sex and race from the diabetic and non-diabetic groups. A preliminary study on over 300 patients was conducted of the validity of these radiographic and charting records as compared to counts of missing teeth by a trained clinical examiner. This revealed an accuracy of over 90% with most errors being due to failure to note data that were actually in the dental chart. The data analysis plan is fully described as is the sample size calculation for the determination of sample sizes for the component of the study on both the number of missing teeth and edentulism. The proposal concludes with a section on future studies and hypothesis generation.
