Background
Older adults are at increased risk of HIV acquisition, yet few studies examine correlates of their HIV prevention. It is important that we understand correlates of being diagnosed at an older age, as well as factors associated with viral non-suppression (non-VS) among those diagnosed at an older age for U=U prevention efforts. However, there is no universal cut-off for classifying a person as “older” at the time of their HIV diagnosis. Our objective was to determine a cut-off for older age at HIV diagnosis and use that cut-off to evaluate correlates of non-VS among those diagnosed above that cut-off.
Methods
The DC Cohort is a longitudinal cohort of people with HIV (PWH) in Washington, DC. We evaluated cohort participants enrolled from 2011-2024 who were diagnosed with HIV in the year prior to enrollment. We used logistic regression to determine the direction of the relationship between age and each demographic variable. The direction was used to predict the demographics of participants diagnosed with HIV after age cut-offs from 20-70 years old. Receiver operating characteristic (ROC) curves were used to evaluate the accuracy of predictions. We used generalized estimating equations to evaluate correlates of non-VS, defined as a viral load > 200 copies/mL, among those diagnosed with HIV above the age cut-off with the best predictive accuracy.
Results
We evaluated 1,240 DC Cohort participants who were newly diagnosed with HIV in the year prior to enrollment. Participants were mostly non-Hispanic Black (78%), male (71%), heterosexual (37%), and the mean age was 37 (SD: 13.5). We found that an age 40 cut-off had the best predictions across demographics. Therefore, we evaluated the 487 participants diagnosed with HIV at age ≥ 40. We found that females (OR (95% CI): 1.8 (1.1, 2.8)), those with public insurance (OR: 1.6 (1.0, 2.4)), and heterosexual mode of HIV acquisition (OR: 1.7 (1.1, 2.7)) were more likely to be non-VS. These relationships were not significant in the adjusted model. Transgender participants were less likely to be non-VS in both the unadjusted and adjusted models (aOR: 0.2 (0.1, 0.7)).
Conclusions
An age cut-off for HIV diagnosis of 40 years or older resulted in the most accurate predictions of participant demographics. Among those diagnosed age 40 or older, gender, mode of HIV transmission, and insurance were associated with non-VS in the unadjusted analysis. These results can inform both primary and secondary prevention measures targeted towards those age 40 and older.