Background
Sustained retention in HIV care is critical for viral suppression and epidemic control, yet sources of heterogeneity in longitudinal care engagement, like clinical complexity and chronic comorbidity management, remain underexplored. Using a standard retention metric, we employed latent class analysis to describe retention trajectories and associated clinical profiles among people with HIV (PWH) in North America.
Methods
We analyzed data from adult (≥18 years old) PWH comprising the clinical cohorts of the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) from 2010 to 2023. We defined the date of loss to follow-up as the last encounter, CD4+ cell count, or HIV viral load reported before a ≥18-month gap. We defined clinical retention using a standard metric of ≥2 encounters (HIV primary care or telemedicine), ≥90 days apart, classified yearly from study entry until December 31, 2023, loss to follow-up, or death. We determined the optimal number of latent classes, k, using intercept-only models and selected the k-class solution minimizing a model fit criterion while yielding favorable class size proportions, relative entropy, and profile distinctness. We used posterior class membership probabilities to generate posterior-weighted distributions of demographics (age at study entry, sex, and race/ethnicity), HIV acquisition risk factors, clinical measures at study entry (CD4+ cell count, HIV viral load, and history of any AIDS-defining illness, treated hypertension, or diabetes), and clinical cohort membership.
Results
Among 117,174 PWH from 18 cohorts, the median age was 46.5 years (interquartile range, 35.7–54.5), 84.8% were male, and 37.9% were non-Hispanic Black. Based on our multicriteria approach, we selected a 7-class solution as optimal (Figure 1A). The most pronounced imbalances in class membership were cohort-driven. Compared with the class most consistently retained (class A; 46.2% of participants), the class least consistently retained (G; 3.0%) skewed toward younger PWH (median, 42.5 vs. 48.1 years old) and those with no history of any AIDS-defining illness (11.3% vs. 16.7%), treated hypertension (13.4% vs. 24.6%), or diabetes (5.4% vs. 10.9%; Figure 1B).
Conclusions
Distinct retention trajectories in HIV care demonstrate the importance of differentiated strategies for sustaining long-term engagement based not only on demographics and HIV management, but also on chronic comorbidity management and cohort-specific practice differences.
Table or Figure
