Maintaining HIV Viral Suppression for a Decade in a Socially Precarious Community: A Longitudinal Analysis of the Undetectables Intervention

Background and Purpose: The United Nations 90-90-90 goals for global HIV eradication (90% of HIV-positive people tested, 90% of those positive engaged in antiretroviral therapy (ART), 90% of ART recipients becoming virally suppressed) is heavily dependent on high levels of adherence to ART in order to reach viral undetectability. Moreover, HIV cannot be eradicated without addressing populations with the highest prevalence: socially precarious communities negotiating intersecting challenges such as racism, homelessness, substance use and mental illness. Finally, interventions need to sustain effectiveness over time. Implemented in 2014, the Undetectables Intervention (UI) sought to increase adherence to ART among vulnerable people of color with HIV/AIDS (POCWHA) in New York city. UI utilized financial incentives ($100 for every undetectable quarterly assay) along with targeted case management among POCWHA with a high prevalence of homelessness, substance use, and mental illness. Studies indicated significant increases in adherence and viral suppression rates, leading to UI being designated a model intervention by the CDC, and the Fast-Track Cities initiative, a global program launched by the mayors of 330 cities that implements evidence-based interventions to reach the 90-90-90 goals. To examine the sustainability of its effectiveness, our team which designed and tested the implementation of UI, examines its effectiveness ten years after its launch.

Methods: We conducted longitudinal analyses of participants enrolled in UI from 2014 to 2024 (n=2,216), utilizing repeated measures models.

Results: At last follow-up, 89.3% were undetectable (viral load [VL]<200 copies/ml) for those still enrolled in 2024 (n=1,111), while overall, 80% were undetectable for the ten-year sample. The average number of quarterly follow-ups (beyond baseline) was 14. Being undetectable was positively associated with number of follow-ups (p<0.0001). Each additional follow-up increased the odds of being undetectable by 3%. On average, UI increased undetectability odds by 45%.

A cohort effect was detected: those enrolled after 2019 had lower rates of being undetectable at baseline than those enrolled before (73 versus 83%, p<0.0001), but also reduced their VL more rapidly over time (p<0.0001). Moreover, 85% of those undetectable at baseline remained undetectable at last follow-up, while 59% of those detectable at baseline transitioned to undetectability at last follow-up. There was a significant interaction effect between VL over time, and VL status at baseline: VL for undetectables at baseline increased over time, while VL for detectables at baseline decreased longitudinally.

Discussion and Implications: With almost 90% current adherence, UI meets a crucial 90-90-90 goal. Participants were retained for a long time (3.5 years, on average), with time spent in UI significantly increasing the probability of reaching viral suppression. We discuss factors (such as the onset of Covid-19) that explain why a more vulnerable cohort was enrolled in the last five years. The intervention was most effective for the most vulnerable participants: those detectable at baseline, and the cohort enrolled after 2019, during the onset of Covid-19. However, we also detected treatment fatigue over time, especially for those who started when virally suppressed. We discuss the implications of treatment fatigue, and ways to modify UI to address it.