Comparison of adherence measures using claims data in the South African private health sector

Abstract

Background. Medication adherence measurement is becoming increasingly important. Biological assays and markers, directly observed therapy, self-reports, pill counts and surveys have been successfully used to assess adherence under various circumstances, but may be limited by cost, ethical concerns and self-reported bias. Administrative claims data, in addition to offering a solution to these limitations, provide access to large study populations under real clinical practice situations, and in a timely and effective manner. With the wide range of adherence measures determined from claims data available – some of which have been found to be mathematically equivalent – researchers are often faced with the decision of choosing which is appropriate. An assessment of the various measures is therefore important for better understanding and to facilitate future adherence studies using administrative data.

Objectives. To compare different adherence measures using data from a medicines claims database in South Africa (SA), employing montelukast for the purpose of illustration.

Methods. This retrospective, cross-sectional research used data from 1 January 2006 to 31 December 2015 from a privately owned pharmaceutical benefits management (PBM) company in SA. Claims for montelukast were identified and adherence was determined using the continuous multiple-interval measure of oversupply (CMOS), compliance ratio (CR), modified medication possession ratio (MPRm), refill compliance rate (RCR), continuous single-interval measure of medication acquisition (CSA) and proportion of days covered (PDC) capped at 1. The measures were compared with the medication possession ratio (MPR) as the reference.

Results. The MPR, CMOS and CR were equivalent, each yielding an adherence value of 86%. The MPRm, RCR and average CSA yielded higher adherence values of 96.9%, 117.2% and 129.0%, respectively, whereas the PDC produced a lower adherence value of 76.0%. The measures that used the entire study period as the denominator produced consistent results compared with the measures that used the difference between claims dates as denominator.

Conclusions. The MPR is considered the most widely used metric to measure adherence using administrative data, but it may not always be applicable owing to the type of data available. Adherence computed using the CR, CMOS and PDC capped was found to be comparable to the MPR, and they may therefore be used as alternatives.