Analysis of CD4 Count in People Living with HIV: An Additive Negative Binomial Mixed-effects Modeling of Longitudinal Data

Introduction CD4 cell counts provide insight into the health of a person’s immune system, as well as information about how their disease is progressing. Boosting the immune level of individuals living with HIV through antiretroviral medication is the most effective way to prevent complications and illnesses caused by Opportunistic Infections (OIs). Methods In this study, we conducted a longitudinal cohort analysis of CD4 count in people living with HIV using additive negative binomial mixed-effects models. A flexible Generalized Additive Mixed-effects Model (GAMM) framework was employed to capture complex nonlinear patterns in repeated CD4 measurements. The analysis was based on longitudinal data from the CAPRISA 002 Acute Infection (AI) study at the Centre for the AIDS Programme of Research in South Africa. Key variables, such as age, baseline BMI, and follow-up duration (time), were analyzed nonparametrically, along with other relevant factors analyzed parametrically. Results The study results revealed significant effects of baseline viral load and HAART initiation on CD4 count progression. Patients initiating HAART showed a 1.233-fold increase in expected CD4 count compared to pre-treatment levels. Baseline viral load negatively impacted CD4 count, even with small unit changes (γ =-1.581e-07, p -value=0.00079). Smooth terms of age ( edf = 14.24, p -value < 2e-16), time (edf = 10.343, p -value < 2e-16), and baseline BMI ( edf = 3.044, p -value = 2.21e-06) exhibited significant non-linear relationships with CD4 count. Spline plots indicated gradual CD4 improvement over time, suggesting long-term benefits of HAART, especially in older and higher-BMI patients. Discussion The findings of our analysis offer a deeper understanding of the functional relationship between the outcome variable and key predictors over time. The research found that initiating antiretroviral therapy improves trajectories of CD4 counts, whereas higher baseline viral load significantly impairs immune recovery over time. The modeling further revealed that age, time, and baseline BMI have a significant nonlinear impact on CD4 count dynamics over time. Conclusion The study establishes that BMI has an impact on the progression and immune responses of Highly Active Antiretroviral Therapy (HAART). The significant nonlinear effect of time suggests that the progress of patients’ CD4 count is slow, and higher CD4 count levels are observed after several treatment visits based on the studied data set. HIV patients who do not maintain immunological stability by consistently receiving antiretroviral medication face an increased risk of illness if they contract OIs due to weakened immune response.