Murri and colleagues ask about the distribution of completely antiretroviral-naive patients. This information can be found in our Table 4—9% (2 of 23) of patients in our series who had adherence of 95% or more and 10% (6 of 58) of those with less than 95% adherence had never received any antiretroviral medication.

It is widely agreed that virologic outcome depends both on adherence and other factors, such as viral resistance to antiretroviral medications. All five patients who had virologic failure despite at least 95% adherence had been heavily pretreated with antiretroviral medications. These five patients had been HIV positive for a median of 10 years (compared to a median of 6 years for the entire study sample) and had received six to nine antiretroviral drugs. One patient underwent antiretroviral resistance studies; genotyping showed HIV-1 mutations that were responsible for resistance to indinavir and saquinavir.

We are surprised that Murri and colleagues did not find a correlation between adherence and virologic outcome. The difference in their results and ours may lie in the sensitivity of methods used to assess adherence. We used a microelectronic monitoring system (MEMS TrackCap, Aprex, Union City, California) and found that 72% of patients had less than 95% adherence; Murri and associates used patient self-report and found that only 35% of patients had this degree of nonadherence. In the context of clinical studies, microelectronic monitoring should be considered the gold standard for assessing medication adherence.