Adherence without access to antiretroviral therapy in sub-Saharan Africa?

We would like to compliment Gill et al. [1] on their article ‘No room for complacency about adherence to antiretroviral therapy in sub-Saharan Africa’. We agree with their concern that adherence to antiretroviral therapy in resource-limited countries will probably decline as treatment access expands. We suggest, however, that future declines in adherence will be less related to publication bias, as suggested by the authors, than to the end of a ‘honeymoon perioD' defined by two phenomena. First, the initial individuals who gained access to antiretroviral therapy in resource-limited settings may possess greater resources, heightened motivation, and are probably not representative of the larger HIV epidemic. As treatment access expands beyond this select group, population adherence levels will probably fall (although not to the levels of 60–70% reported in north America) [2]. Second, early reports on adherence in resource-limited settings were largely based on individuals who had recently initiated treatment [3–6]. Adherence often declines over time as the benefits of treatment plateau and the long-term side-effects of treatment, such as neuropathy and lipodystrophy, accumulate [7–9]. Gill et al. [1] cited two observations in sub-Saharan Africa suggesting that adherence may decline over time, but these studies may have overestimated the decline. In the intent-to-treat, missing-equals-failure analysis of Laurent el al. [4], individuals who stopped therapy because of toxicity or interruptions in access were counted as non-adherent with detectable virus. In the abstract by Akam [10], also cited by Gill et al. [1], the estimate of 68% adherence among patients receiving self-pay therapy is largely attributable to the 53% who discontinued treatment as a result of financial constraints. Adherence, commonly defined as ‘taking medication as prescribeD', presumes access to therapy. Lack of access to therapy and failure to adhere to therapy are different problems requiring different solutions. The former calls for stable drug supply and distribution, whereas the latter calls for interventions to sustain individual behavior. We agree with Gill et al. [1] that there is no room for complacency about adherence to therapy, but neither is there room for complacency about access to therapy.