Investigation of the method of "mixed" pill counts as a tool to detect deliberate masking of non-adherence to antiretroviral therapy at Ntshembo Clinic, Mamelodi Hospital

Abstract

High levels of adherence to ART are essential for maximal suppression of viral replication and avoidance of drug resistance. Pill counts are an indirect, objective method of assessing adherence. Patients can invalidate pill counts by manipulating the number of tablets returned. This paper describes a pilot study which investigated the ability of ‘mixed’ pill counts to detect deliberate masking of non-adherence to ART at a public sector ARV Clinic in Pretoria, South Africa. Seventy-eight adult patients on a first line regimen of ART were recruited. At the first return visit, a standard pill count was performed and adherence (% of tablets taken) was calculated. For the repeat prescription, three days’ extra supply was dispensed without the patients’ knowledge. At the second return visit, a ‘mixed’ pill count was performed and adherence was calculated. Patients were grouped into three categories based on calculated adherence: truthfully non-adherent (<100% adherence), adherent (100% adherence) and ‘over-compliant’ (>100% adherence, i.e. returning to the clinic with fewer tablets than required). Exploratory interviews were conducted with truthfully nonadherent and over-compliant patients to obtain explanations for discrepancies in pill counts. Twenty-nine (37%) patients completed the study. Reasons for drop-out or discontinuation from the study included the issue of prescriptions for 2-3 months’ ARV supply, missed appointments, regimen changes and failure to return remaining tablets to the clinic. Eleven patients (38%) were identified as over-compliant in one or more of the ARVs in their regimen. Nine of these patients agreed to be interviewed, of which three admitted to manipulating their tablet numbers. Reasons for manipulation included: being ’fine now’ and not in need of ARVs; changes in body shape; possibility of the social grant being terminated if non-adherent; getting a new supply and no need for remaining ARVs; knowing that the tablets would be counted for the study. This pilot study indicated that the ‘mixed’ pill count method is capable of detecting deliberate masking of non-adherence. Applying this method to a larger sample may better estimate the frequency of pill count manipulation by patients and help gain insight to reasons for this behaviour and the extent of actual non-adherence. Key words: dumping, manipulation, masking, over-compliance, HAART