Marginal structural Cox model for estimating the effect of Chinese medicine on the survival of people living with HIV: a 17-year real-world retrospective cohort study.

Objective: The aim of this study was to evaluate the long-term effect of Chinese medicine (CM) on the survival of people living with HIV (PLHIV) by investigating the potential synergistic effects of combining CM with antiretroviral therapy (ART). Methods: We conducted a 17-year cohort study based on standardized case registry data, using various packages in the R software. PLHIV enrolled in the national CM HIV treatment trial were classified as the CM group, while PLHIV not enrolled were classified as the Non-CM group. CD4+T cell count and ART regimen were collected annually during the period from the cohort start to the endpoint. The cumulative observations of person-years and mortality were computed using life table analysis. The cumulative survival rates and survival curves were compared using Kaplan-Meier and log-rank tests. CD4+T cell count and ART regimen were time-dependent covariates, and a marginal structural Cox model was used to control for these variables in evaluating the effect of CM on the survival of PLHIV. Results: A total of 2,924 PLHIV were included in the analysis, comprising 1,210 in the CM group and 1,714 in the Non-CM group. The mortality was significantly lower in the CM group than that in the Non-CM group (2.92/100 person-years vs. 3.86/100 person-years, P < 0.001). Multivariate analysis showed that the risk of death in the CM group was 0.86 compared with the Non-CM group (95% CI: 0.77-0.97, P < 0.05). The mortality risk of patients who received first-line ART and second-line ART was 0.47 (95% CI: 0.39-0.56, P < 0.05) and 0.49 (95% CI: 0.38-0.62, P < 0.001), respectively, compared with those who did not receive ART. Conclusion: The results demonstrate that after controlling the time-dependent covariates with a marginal structural Cox model, CM could improve the long-term survival rate of PLHIV by investigating the potential synergistic effects of combining CM with ART.