Mixed-effects Modeling of Association of Syndemics With Hiv Viral Loads in Nairobi Sex Workers Cohort on Antiretroviral Therapy

Abstract

Background: Sex workers are amongst the key populations substantially burdened by HIV/AIDS. However, there is definitive proof that antiretroviral therapy (ART), if consistently used, has individual and public health benefits due to viral suppression and onward HIV sexual transmission reduction. The standard HIV/AIDS clinical care approaches mostly emphasize on biomedical and health interventions and may ignore synergistic epidemics (syndemics) which, when present, may be associated with the likelihood of elevated HIV viral loads. Objectives: To examine and model the association of syndemics with HIV viral load outcomes as well as antiretroviral therapy (ART) adherence among sex workers living with HIV in the Sex Worker Outreach Project (SWOP)–City cohort in Kenya using linear mixed-effects. Method: Data collected between 2013—15th October 2018 from SWOP–City were modeled using linear mixed-effects for progression of viral load marker. Results: None of HIV-syndemics considered was statistically significant on univariable random intercept model. On the multivariable model, only HIV-STI syndemic and condom use were statistically significant effect (p<0.01). A unit increase in the months since baseline was associated with 0.055110 Box-Cox transformed viral load in the random intercept model, however, this effect wasn’t statistically significant (p>0.005). Poor ARV adherence variable was associated with 3.408471 increase in Box-Cox transformed viral load (p= 0.0225) and 1.237 positive change in transformed viral load (p=0.0119) baseline effects. An interaction of poor last ARV adherence with month yielded 0.1528 negative change suggestive of programme intervention along the follow-up but not significant (p=0.0990). The effect of longitudinal ART adherence in lessening syndemics was associated with a reduction in Box-Cox and log10 transformed viral loads, nonetheless, these effects were not statistically significant (p>0.05). An increase in linkage to psychosocial support was associated with 0.10077 reductions in log10 viral load but not significant (p=0.454). Linkage to psychosocial support played a key role in modifying the relationship between log HIV viral and poor adherence (p<0.01). Conclusions: The effect of syndemics was not directly associated with poor viral loads in the presence of adherence and psychosocial support. Linkage to psychosocial support modified the effect of syndemics on viral load evolutions. The study underscores the need of enhancement of linkage to psychosocial support.