| Background. Adherence is difficult to measure in routine practice, despite its importance in diabetes control. Persistence, the continuous refill of medications, is one measurement that could be integrated into clinical care.;Objective. To characterize oral hypoglycemic persistence patterns, identify predictors of non-persistence, and determine the association between persistence and glycemic control among incident diabetics in the Veterans Administration.;Methods. Eligible veterans were ≥ 18 years; filled a prescription (rx) for oral hypoglycemic agents (OHAs) for the first time between January 1, 2000 and December 31, 2002; and filled at least 1 rx for any drug (but no diabetic medications) during the 6 months preceding the index date. The index date (date of entry into the cohort) was defined as the date the OHA was first dispensed. Persistence over 12 months, the primary outcome, was defined as the number of days supply/study period, as well as using Steiner's medication possession ratio. Patient, regimen, and clinical level variables were examined for their association with non-persistence. Multivariable models were used to assess the association of persistence and glycemic control.;Results. 109,449 veterans had a median age of 63 yrs, were mostly male (97%) and had substantial comorbidity, with 56% having hypertension and 36% hyperlipidemia. Baseline hemoglobin (Hgb) A1c was 7.6%. Participants received a median of 1 OHA over 1 year, with metformin being the most common index medication (41%). The proportion of patients with poor persistence (< 80%) increased throughout the year, ranging from 18% of patients in the first quarter to 43% in the last with the biggest decline occurring in the first 3-6 months. White race, endocrinology care, number of OHAs, and comorbidity were associated with good persistence (≥80%). Diagnoses for alcohol and drug abuse were negatively associated with persistence, while depression had no significant effect, although the model explained only 7% of the persistence variation. Subjects with good persistence were more likely to achieve the goal of Hgb A1c ≤ 7.0% at one year (OR 1.33, 95% CI 1.28, 1.38).;Conclusions. By 1 year, almost half of incident diabetics had poor persistence, despite the strong association of good persistence with improved glycemic control. Rather than targeting patients based on risk of non-persistence, direct patient-specific persistence measures should be used to identify those needing adherence counseling and to tailor adherence interventions in clinical practice. |
