EVALUATING INITIAL MEDICATION ADHERENCE THRESHOLDS AS PREDICTORS OF CONTINUATION ADHERENCE AMONG PATIENTS WITH MAJOR DEPRESSIVE DISORDER: A CLAIMS-BASED ANALYSIS
Author(s)
Sourab Ganna, PharmD1, Jieni Li, MPH, PhD2, Rajender Aparasu, PhD2;
1University of Houston, Student, Houston, TX, USA, 2University of Houston, Houston, TX, USA
1University of Houston, Student, Houston, TX, USA, 2University of Houston, Houston, TX, USA
OBJECTIVES: Initial medication adherence (IMA) is increasingly recognized as an early indicator of long-term treatment engagement. However, the optimal IMA threshold for predicting continuation adherence (CPA) in major depressive disorder (MDD) is not well established. This study evaluated the predictive performance of multiple IMA thresholds and assessed the association between IMA at optimal CPA.
METHODS: A retrospective longitudinal cohort study was conducted using 2015-2024 Merative Marketscan Commercial data. Adults with MDD initiating antidepressant therapy were identified using a new-user design with continuous enrollment. IMA was calculated as PDC during the first 90 days following treatment initiation and evaluated at thresholds ranging from 70% to 90%. CPA was measured during months 4-12 and was operationalized as optimal if PDC ≥80%. Multivariable logistic regression models were used to examine the association between each IMA threshold and CPA. Predictive performance was compared across thresholds using the area under the receiver operating characteristic curve (AUC), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC), adjusting for demographic, clinical, and healthcare utilization covariates.
RESULTS: Among 1.04 million adults with MDD initiating antidepressant therapy, the proportion achieving IMA at PDC thresholds of 70%, 80%, and 90% was 69.9%, 65.9%, and 58.3%, respectively. Higher IMA was associated with greater odds of achieving optimal CPA (≥80%), with adjusted odds ratios (aOR) ranging from 7.02-16.44 across thresholds. The magnitude of association declined with increasing IMA stringency. IMA ≥70% demonstrated a strong association with CPA (aOR:16.44,95%CI:16.20-16.68) and optimal predictive performance (AUC:0.763). More restrictive thresholds showed no improvement in discrimination and attenuated effect estimates.
CONCLUSIONS: IMA is a strong predictor of CPA in MDD; however, its predictive performance varies depending on the threshold. An IMA cutoff of 70% demonstrated the favorable balance between effect size and model fit, supporting the use of IMA as a valuable measure in MDD research.
METHODS: A retrospective longitudinal cohort study was conducted using 2015-2024 Merative Marketscan Commercial data. Adults with MDD initiating antidepressant therapy were identified using a new-user design with continuous enrollment. IMA was calculated as PDC during the first 90 days following treatment initiation and evaluated at thresholds ranging from 70% to 90%. CPA was measured during months 4-12 and was operationalized as optimal if PDC ≥80%. Multivariable logistic regression models were used to examine the association between each IMA threshold and CPA. Predictive performance was compared across thresholds using the area under the receiver operating characteristic curve (AUC), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC), adjusting for demographic, clinical, and healthcare utilization covariates.
RESULTS: Among 1.04 million adults with MDD initiating antidepressant therapy, the proportion achieving IMA at PDC thresholds of 70%, 80%, and 90% was 69.9%, 65.9%, and 58.3%, respectively. Higher IMA was associated with greater odds of achieving optimal CPA (≥80%), with adjusted odds ratios (aOR) ranging from 7.02-16.44 across thresholds. The magnitude of association declined with increasing IMA stringency. IMA ≥70% demonstrated a strong association with CPA (aOR:16.44,95%CI:16.20-16.68) and optimal predictive performance (AUC:0.763). More restrictive thresholds showed no improvement in discrimination and attenuated effect estimates.
CONCLUSIONS: IMA is a strong predictor of CPA in MDD; however, its predictive performance varies depending on the threshold. An IMA cutoff of 70% demonstrated the favorable balance between effect size and model fit, supporting the use of IMA as a valuable measure in MDD research.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
PCR129
Topic
Patient-Centered Research
Topic Subcategory
Adherence, Persistence, & Compliance