Good and poor adherence: optimal cut-point for adherence measures using administrative claims data.

Area of Science:

• Health Services Research
• Pharmacoepidemiology
• Chronic Disease Management

Background:

• Patient adherence to medication is crucial for managing chronic conditions.
• Administratively derived adherence measures like MPR and PDC are commonly used but require validated cut-off points.
• Identifying optimal adherence thresholds can improve patient outcomes and healthcare resource utilization.

Purpose of the Study:

• To determine the optimal adherence value cut-off for stratifying patients into good and poor compliers.
• To evaluate the predictive value of medication possession ratio (MPR) and proportion of days covered (PDC) for hospitalization outcomes.
• To assess these adherence measures across diverse chronic conditions including schizophrenia, diabetes, hypertension, CHF, and hyperlipidemia.

Main Methods:

• Retrospective analysis of Arkansas Medicaid administrative claims data.
• Inclusion of adult patients with specific chronic conditions and continuous eligibility.
• Assessment of 1-year medication adherence using MPR and PDC, with hospitalization as the primary outcome.
• Logistic regression and ROC curve analysis to identify optimal adherence cut-off values.

Main Results:

• Optimal adherence cut-off values for predicting any-cause hospitalization ranged from 0.63 to 0.89.
• Optimal cut-off values for disease-specific hospitalization ranged from 0.58 to 0.85 across the studied cohorts.
• These findings indicate variability in optimal adherence thresholds depending on the outcome and disease.

Conclusions:

• An adherence cut-off of 0.80 provides a reasonable basis for stratifying patients as adherent or non-adherent.
• This threshold demonstrates validity in predicting subsequent hospitalizations across several prevalent chronic diseases.
• Further research is recommended to explore other outcome metrics for adherence threshold identification.