Marginal Structural Models for Estimating the Effect of Highly Active Antiretroviral Therapy Initiation on CD4 Cell Count

Stephen R. Cole ) 2 Miguel A. Herna n 1 Joseph B. Margolick 0 Mardge H. Cohen 4 James M. Robins 1 3 0 Department of Molecular Microbiology and Immunology, Bloomberg School of Public Health, Johns Hopkins University , Baltimore, MD 1 Department of Epidemiology, School of Public Health, Harvard University , Boston, MA 2 Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University , Baltimore, MD 3 Department of Biostatistics, School of Public Health, Harvard University , Boston, MA 4 Cook County Hospital , Chicago, IL The effect of highly active antiretroviral therapy (HAART) on the evolution of CD4-positive T-lymphocyte (CD4 cell) count among human immunodeficiency virus (HIV)-positive participants was estimated using inverse probability-of-treatment-and-censoring (IPTC)-weighted estimation of a marginal structural model. Of 1,763 eligible participants from two US cohort studies followed between 1996 and 2002, 60 percent initiated HAART. The IPTCweighted estimate of the difference in mean CD4 cell count at 1 year among participants continuously treated versus those never treated was 71 cells/mm3 (95% confidence interval: 47.5, 94.6), which agrees with the reported results of randomized experiments. The corresponding estimate from a standard generalized estimating equations regression model that included baseline and most recent CD4 cell count and HIV type 1 RNA viral load as regressors was 26 cells/mm3 (95% confidence interval: 17.7, 34.3). These results indicate that nonrandomized studies of HIV treatment need to be analyzed with methods (e.g., IPTC-weighted estimation) that, in contrast to standard methods, appropriately adjust for time-varying covariates that are simultaneously confounders and intermediate variables. The 1-year estimate of 71 cells/mm3 was followed by an estimated continued increase of 29 cells/mm3 per year (estimated effect at 6 years: 216 cells/mm3), providing evidence that the large short-term effect found in randomized experiments persists and continues to improve over 6 years. acquired immunodeficiency syndrome; antiretroviral therapy, highly active; bias (epidemiology); causality; CD4 lymphocyte count; confounding factors (epidemiology); HIV Abbreviations: AIDS, acquired immunodeficiency syndrome; CI, confidence interval; GEE, generalized estimating equations; HAART, highly active antiretroviral therapy; HIV, human immunodeficiency virus; IPTC, inverse probability-of-treatment-andcensoring; MSM, marginal structural model. - A declining number of T-lymphocytes expressing the CD4 molecule (CD4 cells) is an important marker of the progression of human immunodeficiency virus (HIV) disease among infected persons (1). Randomized trials conducted in the 1990s indicated a dramatic effect of highly active antiretroviral therapy (HAART) on CD4 cell count (2, 3). The clear evidence of a survival benefit meant that the HAART trials had to be stopped for ethical reasons. Thus, randomized trial data bearing on the long-term effectiveness of HAART on the evolution of CD4 cell count are neither available nor likely to become available. However, HAART was approved by the US Food and Drug Administration in 1996, and data from the Multicenter AIDS Cohort Study and the Womens Interagency HIV Study are available through 2002. In this paper, we use these observational data to estimate the effect of HAART on mean CD4 evolution over a period of 6 years. Estimation of the effect of HAART on CD4 evolution is challenging for the following reason. Current treatment guidelines (4, 5) suggest that physicians use plasma HIV type 1 (HIV-1) RNA level (i.e., viral load) and CD4 cell count to determine the timing of HAART initiation. However, current viral load and CD4 cell count are known predictors of subsequent CD4 cell counts (6). Therefore, to obtain an unconfounded estimate of the total (i.e., direct and indirect) effect of HAART on CD4 cell count, it is necessary to adjust for viral load and CD4 cell count before HAART initiation. A standard approach is to include past viral load and CD4 cell count as time-varying covariates in a regression model for the mean of the current CD4 cell count, conditional on past treatment and confounder history. Unfortunately, this standard approach fails because evolving viral load and CD4 cell count are strong intermediate variables on the causal pathway from past HAART treatment to current CD4 cell count. For example, the biologic effect of HAART is known to be largely mediated through its effect on viral load (7): HAART dramatically reduces the load of circulating virus by blocking replication at multiple points in the viral life cycle. Thus, the standard approach can, at best, only estimate the relatively small direct effects of past HAART treatment on current CD4 cell count at time t that are not mediated through the reduction in viral load and the increase in CD4 cell count prior to time t. Moreover, the standard approach may additionally induce selection bias, because CD4 cell count is affected by previous HAART use (810). However, the above difficulties can be surmounted. Robins (1113) has developed methods based on marginal and nested structural models to adjust for variables, such as viral load, that are time-varying confounders affected by prior treatment. In a previous report (14), we estimated the total effect of HAART initiation on time to acquired immunodeficiency syndrome (AIDS) or death using a marginal structural Cox model based on observational data from the Multicenter AIDS Cohort Study and the Womens Interagency HIV Study. Here, we use a marginal structural mean model to estimate the effect of HAART on the evolution of CD4 cell counts from 1996 to 2002 in these two ongoing cohort studies. MATERIALS AND METHODS Study population and measurements In this analysis, we used information from the Multicenter AIDS Cohort Study (15), which, beginning in 1984, enrolled 5,622 homosexual men in four US cities (Baltimore, Maryland; Chicago, Illinois; Pittsburgh, Pennsylvania; and Los Angeles, California), and the Womens Interagency HIV Study (16), which, beginning in 1994, enrolled 2,628 women in five US cities (New York, New York; Chicago, Illinois; Los Angeles, California; San Francisco, California; and Washington, DC). Every 6 months, participants in both studies completed an extensive interviewer-administered questionnaire with information on antiretroviral therapy use and provided a blood sample for the determination of CD4 cell count and viral load. Institutional review boards approved all protocols and informed consent forms, which were completed by study participants in both cohorts. Results presented here are limited to the 1,763 men and women who were alive, HIV-positive, and under follow-up in April 1996 when HAART became available. Each participant contributed a maximum of 12 personvisits beginning with the first semiannual study visit after April 1996 (the baseline visit) and ending (...truncated)