Joint modeling of longitudinal cd4 cell count and time to viral rebound of hiv/aids infected patients initiating antiretroviral therapy

Abstract

Back ground: HIV infection leads to severe depletion of CD4 cell with subsequently reduced levels of circulating CD4+lymphocytes in the peripheral blood. CD4 cell counts are the primary laboratory markers used to track the progression of HIV to AIDS. Time-to-viral rebound and CD4 count measures are the outcome variables of HIV patients after starting ART in this study. The time-to-viral rebound from ART is determined by month time interval among dates of ART commencement to rebound, as documented by the health information data administrator. In such follow-up trials, joint models are used because both longitudinal and survival data are generated. Objective: The objective of the study is to compare separate and joint models of longitudinal CD4 cells measurements and to identify factors affecting change in CD4 cell count over time. Methods: A retrospective cohort study design was conducted among 309 HIV/AIDS patients who were 18 years old and who are under ART follow-up from February 1; 2016 to May 30; 2021 at Jimma University Medical Center, West Ethiopia. First, the data were analyzed using longitudinal and survival models separately. Then, based on the separate model's several joint models with different random effects and shared parameters have been explored and compared using AIC score. Results: Among 309 HIV patients considered in this study; 235 (76.1%) of them were rebound while the remaining 74 (23.9%) were censored. The result from the joint model of the estimated association parameter 𝛼 is −0.102, this indicates both outcomes are negatively associated and higher values of the CD4 cell count are associated with better survival. The two outcomes were associated. The joint model was used to handle the associations between them to obtain a valid and efficient estimate. The result of the longitudinal model reviled that age; adherence; functional status; WHO clinical stage; interaction effect of adherence; functional status and WHO clinical stage with linear time had significantly associated with mean change in the square root of CD4 count. Furthermore, from the survival model we found the survival probability of HIV infected patient were determined by age; viral load; adherence; WHO stage and peripheral neuropathy. Conclusion: The joint model reveals an association between time to viral rebound and repeated CD4 cell measurement. When evaluating the overall performance of both the separate and joint models in terms of model parsimony, the goodness of fit, smaller total AIC, and the statistical significance of both the association parameters, the joint model performs better. Thus, we concluded that the joint model is preferred for simultaneous analyses of repeated measurement and survival data. Recommendation: In the future, the study recommends the application of joint model of bivariate longitudinal and time to viral rebound of survival analysis of HIV progression.