1022 — Longitudinal Effects of Medication Adherence on Glycemic Control in Veterans with Type 2 Diabetes

Mueller M (Charleston VA Targeted Research Enhancement Program (TREP)), Egede LE (Charleston VA TREP), Mauldin PD (Charleston VA TREP), Durkalski VL (Charleston VA TREP), Chen GJ (Charleston VA TREP), Moran PW (Charleston VA TREP)

Objectives:
To examine the longitudinal effects of medication adherence on glycemic control in veterans with type 2 diabetes.

Methods:
Data on a cohort of 7,597 veterans with type 2 diabetes were analyzed. The cohort included subjects with type 2 diabetes from June 3, 1986 to April 11, 2006 using established algorithms for identifying type 2 diabetes in VA administrative data. For each subject, the first outpatient visit with a hemoglobin A1C (HBA1C) value was used as the baseline visit, and then subjects were followed forward in time until the last available HBA1C. Medication non-adherence was calculated as the summation (in days) of excessive refill gaps (15 days or more) over multiple intervals within the year. A mixed linear regression model was used to examine the change in HBA1C over time in the adherent and non- adherent groups. HBA1C was entered as the dependent variable and medication adherence was entered as the primary independent variable. The model was adjusted for baseline A1C value, change of time from baseline to account for unequal time intervals between consecutive A1C values, demographic variables (age, gender and race: Non-Hispanic White or Non-Hispanic Black / African-American) and comorbidities (CHD, hypertension, stroke, cancer and mental disorders coded as 1 for present and 0 for not present). SAS was used for statistical analysis.

Results:
53% of the sample were considered medication-adherent. Mean age was 66 years. 97% were men. 72% were non-Hispanic whites, 28% were non-Hispanic Blacks. 64% were married and 50% were unemployed. Mean follow-up period was 4.6 years. Mean HBA1C at baseline were not significantly different (adherent: 7.80 [sd 2.06]; non-adherent: 7.60 [sd 2.22], p= < 0.0001). Over the follow-up period, HBA1C among adherent veterans decreased over time by approximately 0.5 (mean change=-0.57), while those for non-adherent veterans declined at a slower rate (mean change=-0.32). After adjusting for covariates, HBA1C in adherent veterans were significantly lower than those in non-adherent veterans with a mean difference in HBA1C over time of 0.136 (95% CI: [0.11; 0.17], p < 0.0001).

Implications:
Type 2 diabetic veterans who are non-adherent with their medications have significant and persistently worse glycemic control over ~5 years of follow-up.

Impacts:
This is one of the first studies to show that medication non-adherence has persistent effects on glycemic control. To reduce the morbidity and mortality associated with type 2 diabetes among veterans, VHA needs to direct more resources toward effective interventions to improve medication adherence.