Discussion
Our results demonstrate that the probability of emergent drug resistance decreased steadily during 1996–2004. Incomplete adherence and nonboosted PI–based or NNRTI‐based antiretroviral regimens were associated with a greater probability of the development of drug resistance. In contrast, boosted PI–based regimens were significantly associated with a lower emergence of resistance, even after adjustment for pVL and CD4 cell count. Of note, the latter remained the case at all levels of adherence.
Suboptimal adherence levels (80% to <95%) were associated with the highest risk of resistance in any drug category (hazard ratio [HR], 4.15; ) or in multiple resistance categories (HR, 6.99; ) (survival analysis data not shown). Recent studies have also shown that different ART classes have unique adherence‐resistance relationships. Bangsberg et al., gathering data from several studies, demonstrated that, with regimens containing nonboosted PIs, most drug resistance occurred in patients with adherence levels between 70% and 80%. The data obtained here show that boosted PI–based regimens are associated with relatively low levels of resistance development across all adherence strata, consistent with previous observations that boosted PI–based regimens may have a more “forgiving” profile in terms of virological suppression. Also, it is interesting to note that we observed a 2‐fold increase in the risk of resistance in patients with a history of injection drug use; this contrasts with an only slightly increased risk noted previously. The reasons for this difference are not clear.
There are several features of the present study that should be highlighted. First, this study was based on a large sample of patients within a provincewide treatment program, in which all individuals had free access to medical attention, combination ART, and laboratory monitoring. We are confident, therefore, that our results are not influenced by access to therapy, a factor that has often compromised the interpretation of similar population‐ and cohort‐based studies. Second, the very simple methodology of logistic regression was sufficient to show the simultaneous effects of adherence, pVL, and antiretroviral regimen on the development of drug resistance. Third, this study was based on individuals who were initially naive to ART, ensuring that our results were not confounded by previous therapy use.
There are some important potential limitations in our study. First, study participants who had samples with a pVL <1000 copies/mL were assumed to have no drug‐resistance mutations. To assess whether this introduced a possible source of bias, we conducted a sensitivity analysis excluding these individuals. This analysis showed that this assumption did not bias our results. Second, pretherapy genotypes were not available for the assessment of transmitted resistance in all study participants. We conducted a sensitivity analysis in the subset of individuals who had pretherapy genotypes without transmitted resistance and found that our original findings still held. Third, we used pharmacy‐refill compliance as a surrogate for adherence; however, this measure of adherence has been found to be independently associated with HIV suppression and survival among HIV‐infected individuals enrolled in the BC‐CfE DTP.