Average adherence (Av-Adh) and log-transformed treatment interruption (TI) to ARV were prospectively measured over 6 months using electronic drug monitoring (EDM) in several cohorts of PWH. Multivariate linear regression models including baseline confounders explored the influence of EDM-defined adherence (R2) on 6-month log10 HIV-RNA. Multivariate logistic regression models were used to compare the risk of HIV-RNA detection (VR) within subgroups stratified by lower (≤95%) and higher (>95%) Av-Adh.

Three hundred ninety-nine PWH were analyzed with different ARVs: dolutegravir (n = 102), raltegravir (n = 90), boosted PI (bPI; n = 107), and NNRTI (n = 100). In the dolutegravir group, the influence of adherence pattern measures on R2 for HIV-RNA levels was marginal (+2%). Av-Adh, TI, and Av-Adh × TI increased the R2 for HIV-RNA levels by 54% and 40% in the raltegravir and bPI treatment groups, respectively. TI increased the R2 for HIV-RNA levels by 36% in the NNRTI treatment group. Compared with the dolutegravir-based regimen, the risk of VR was significantly increased for raltegravir (adjusted odds ratio [aOR], 45.6; 95% CI, 4.5–462.1; P = .001), NNRTIs (aOR, 24.8; 95% CI, 2.7–228.4; P = .005), and bPIs (aOR, 28.3; 95% CI, 3.4–239.4; P = .002) in PWH with Av-Adh ≤95%. Among PWH with >95% Av-Adh, there were no significant differences in the risk of VR among the different ARVs.

These findings support the concept that dolutegravir in combination with 2 other active ARVs achieves greater virological suppression than older ARVs, including raltegravir, NNRTI, and bPI, among PWH with lower adherence.