More than just a cutoff: Medication refill adherence is linearly related to biomarkers of treatment response in hypothyroidism

Abstract

Medication adherence is a complex set of time-varying behaviors—filling, ingesting, and refilling medication at the right time and in the right manner—necessary to effective pharmacotherapy. Proper methods of quantifying adherence and linking it to patient outcomes are foundational for any study seeking to understand its effects on human disease. Common adherence measurements use pharmacy refill records to estimate how much medication patients have on-hand in a given time period. Many studies use fixed adherence cut points to reduce this continuous measurement to a dichotomous variable: Adherent versus non-adherent. Using Veterans Health Administration data from a large national cohort of Veterans, we demonstrate a linear relationship between levothyroxine refill adherence and later probability of elevated serum TSH values. We found no breaks or elbows in the relationship that suggested an appropriate cutoff distinguishing adherent patients from non-adherent based on outcome. Each 0.1 decrease in adherence (as proportion of days covered) leads to a 0.045 decrease in probability of a serum TSH value below laboratory reference highs. Levothyroxine treatment indication, and weight-based dose of levothyroxine received, modify the relationship between adherence and TSH values. Patients with more severe disease (thyroid cancer) and lower levothyroxine doses show better TSH control.