Return to 2001 Abstacts List
89. Predictor Variables of Pharmacological Treatment of LDL-C in Veterans with Diabetes
L Eaton, University of Medicine and Dentistry of New Jersey, VA New Jersey Health Care System; M Safford, University of Medicine and Dentistry of New Jersey, VA New Jersey Health Care System; G Hawley, VA Healthcare Analysis and Information Group; L Pogach, University of Medicine and Dentistry of New Jersey, VA New Jersey Health Care System
Objectives: Detection and management of hyperlipidemia is a key strategy in reducing cardiovascular risk in persons with diabetes. Our objective was to evaluate, using electronic data bases, disparities in achieving a LDL-C target value <130 mg/dl based upon gender, race, age, marital status and diagnosis of coronary artery disease (CAD).
Methods: Using linked FY98 laboratory, pharmacy and administrative databases obtained by the VA Healthcare Analysis Information Group, veterans with diabetes were identified if they had >0 inpatient or >1 outpatient 250.xx ICD9-CM codes, or were prescribed an oral agent or insulin. These files were merged with data on each patientís LDL-C; use of lipid-lowering medications (LLM); age, sex, race and marital status; and ICD9-CM codes (404-404.03) for CAD. LDL-C data from 111 out of 145 medical centers was available, and 110,844 patients (>98% male) were identified as having diabetes and at least one LDL-C value associated with a triglyceride <400 mg/dl, permitting use of the Friedewald equation. High Risk was defined as an LDL-C >130 mg/dl and non-LLM use. Crude odds ratios were calculated for each patient characteristic using Chi Square, and a multiple logistic regression model was used to simultaneously control for patient characteristics.
Results: For this model 70,479 (64%) of the original 110,844 patients were used. Excluded from the analysis were all patients who had an LDL<130 with no LLM usage, and those with missing demographic information. Patients under 65 years of age made up 41% of selected patients, 65-74 years constituted 40%, and 75 years and over comprised 19%. An association was found to exist between age and high risk status (p<.001). Crude odds ratios show that patients under 65 years were 1.4 (95% CI, 1.32-1.42) more likely than those between 65-74 years of age to be high risk while patients 75 years and older where 1.3 (95% CI, 1..25-1.38) times more likely to be high risk than patients between 65-74 years. Patient characteristic adjusted odds ratio suggest that patients under 65 years were 1.2 (95% CI, 1.12-1.23) more likely than those between 65-74 years of age to be high risk while patients 75 years and older where 1.4 (95% CI, 1.35-1.51) times more likely to be high risk than patients between 65-74 years.
Using multivariate analysis, insulin only (OR, 0.69; 95% CI, 0.0.607-0.796) and oral agent only (OR, 0.70; 95% CI, 0.617-0.802) were associated with a decreased likelihood of being high risk. Age under 65 years (OR, 1.2; 95% CI, 1.12-1.23), 75 years and older (OR, 1.4; 95% CI, 1.35-1.52), female (OR, 1.15; 95% CI, 1.008-1.302), absence of CAD (OR, 3.7; 95% CI, 3.51-3.95), unmarried (OR, 1.2, 95% CI, 1.18-1.29), and non-white (OR, 1.5; 95% CI, 1.40-1.53) were associated with an increased likelihood of being high risk.
Conclusions: Although increased odds ratios of having an untreated LDL-C level >130 mg/dl were seen for women, non-white, and veterans aged <65 or >75 years, the major predictor was a non- diagnosis of CAD.
Impact: The finding suggests the need to increase clinician awareness that diabetes should be treated as a CAD equivalent.