From PDC to integral quantitative metrics: an innovative approach to assessing adherence to combination pharmacotherapy in outpatient practice

Abstract

Introduction: Current methods for assessing adherence to combination therapy based on aggregated measures (e.g., PDC, proportion of days covered) poorly capture treatment intensity and sustainability over time. We attempted to describe adherence trajectories for combination therapy using integral metrics. Aim: To develop a set of integral metrics analogous to key pharmacokinetic (PK) parameters – area under the concentration-time curve (AUC, area under the curve), maximum concentration (C_max), and time to maximum concentration (T_max) – to assess adherence to combination therapy as time-based exposure on the PDC methodology among outpatients.

Materials and Methods: As a specific example and a substrate for method development, we used results from a retrospective pharmacoepidemiologic cohort study of patients with heart failure (HF) who had experienced an acute myocardial infarction. Data were extracted from the Unified Medical Information and Analytical System of Moscow (Russia). Three patients with a follow-up duration of ≥24 months were randomly selected. We analyzed benefit-covered electronic prescription fills for key HF therapy classes and calculated adherence for each class using PDC across four half-year intervals. We constructed a stepwise trajectory of combination therapy as the function N(t), where N is the number of therapy classes for which the patient was covered at a given time point based on dispensed supply (accounting for days’ supply and overlaps), and t is follow-up time. By analogy with the “concentration-time” curve, we considered N(t) as a dynamic “therapy class coverage curve” and used it to calculate PK-analog adherence metrics: exposure to actually dispensed combination therapy, AUC_N(t) (class-months); normalized AUC_norm (0-1); C_maxN(t) (number of classes); and T_maxN(t) (months). In addition, we set a target threshold of N(t)≥3, determined the time to first attainment, T_optN(t), and calculated the time (or proportion of time) with N(t)≥k, denoted as T_N(t)≥k. We also quantified therapy “losses” between intervals based on a decrease in the modal value of N(t) and failure to reach the threshold over the follow-up period.

Results: AUC_N(t) values were 26.0, 87.37, and 36.67 class-months, and AUC_norm values were 0.27, 0.74, and 0.39 in patients with internal IDs 13, 39, and 110, respectively. The proportion of time with N(t)≥3 was 0.079, 0.96, and 0.23 (1.9, 22.9, and 5.5 months), and the modal N(t) values by half-year corresponded to patterns of 2→0, 3→3→4→4, and 3→2→1→0. The numeric profiles were consistent with the stepwise N(t) trajectories and reflected three distinct adherence patterns.

Conclusion: The proposed PK-associated approach enabled quantification of complex adherence trajectories for combination therapy in HF by representing them as an interpretable set of integral numeric metrics. These metrics characterize time-based exposure to combination therapy while accounting for adherence. The results support the conceptual validity of the method and may serve as a basis for further studies of its prognostic value.

Graphical Abstract