Group-based trajectory modeling to determine long-term HIV viral load trends among children with HIV in Kenya

Abstract

Background: Identifying determinants of longitudinal HIV viral load (VL) trajectories using group-based trajectory modeling (GBTM) can inform clinical strategies and mechanisms of non-adherence among children. Methods: Children under 12 months of age who were newly diagnosed with HIV were enrolled in the Optimizing Pediatric HIV Therapy (OPH; NCT00428116) from 2007-2010. Children initiated antiretroviral therapy (ART) at enrollment, and VL was assessed every 3 months for 24 months post-ART and 6-monthly thereafter up to 8 years of age. VL trajectory groups were defined using GBTM. Fisher’s exact and Kruskal-Wallis tests were used to determine correlates of each trajectory group compared to the sustained-low VL group. Results: Five VL trajectory groups were identified among 89 children with 522 VL visits from 6-24 months: sustained-low VL (63% of children), sustained-very-high (16%), sustained-high (9%), low-to-high (7%), and high-with-periods-of-low (6%). Children in the sustained-high group were more frequently on a first-line protease inhibitor (PI)-based regimen (63% vs 38%; p=0.03) and had younger caregivers (median: 22 vs 28 years; p=0.02). Among 54 children with 560 VL visits followed from 48-96 months, 5 trajectory groups were identified: sustained-low (74%), mid-range (4%), periods-of-low (7%), high-to-low (7%), and sustained-high (7%). Those in the high-to-low group had younger caregivers (21 vs 29 years; p=0.01). Conclusions: GBTM identified unique VL patterns among children with unsuppressed VL. Caregiver and regimen-related characteristics were associated with patterns of non-suppression. Younger caregivers may benefit from tailored counseling to help them support child ART adherence. Palatable regimens are necessary for viral suppression among children with HIV.