A hierarchical modeling approach for assessing the safety of exposure to complex antiretroviral drug regimens during pregnancy

Abstract

Combination antiretroviral regimens have achieved tremendous success in reducing perinatal HIV transmission, and have become standard of care in pregnant women with HIV. However, the large variety of combination antiretroviral regimens utilized in practice raises the question of whether some of these highly potent drugs pose other risks to the pregnancy or infant. While HIV-infected pregnant women are almost always exposed to multiple antiretrovirals concurrently, standard safety screening strategies typically consider each individual antiretroviral separately, which fails to account for potential confounding due to simultaneous exposure to other antiretrovirals. In this paper, we evaluate a hierarchical modeling approach which groups antiretrovirals by drug class to screen for the safety of antiretrovirals taken during pregnancy, while still providing individual antiretroviral drug effect estimates. In simulation studies, we observed that the hierarchical approach may be advantageous as compared to considering each antiretroviral drug separately or simultaneously evaluating all antiretrovirals in a fixed effect model, particularly when there is prior evidence suggesting drugs from the same class behave similarly on the outcome. The characteristics of the hierarchical approach are illustrated in an application evaluating risk of preterm birth using a study including over 2000 pregnancies representing over 100 antiretroviral combinations, each involving up to three drug classes.

Surveillance monitoring for safety of in utero antiretroviral therapy exposures: current strategies and challenges

“…In addition, evaluation of each individual ARV drug essentially overlooks the fact that the comparison group is not “unexposed”; they instead reflect different combination regimens not including the specific drug of interest. Correia et al [78] have shown that such an approach can lead to biased estimates for individual ARV drugs, particularly in settings where an ARV drug that does impart increased risk for an adverse outcome is commonly used together with certain other ARV drugs. In this setting, the other ARV drugs commonly included in the same regimen will tend to have biased estimates of risk and lead to inflated false positive findings.…”

“…Based on simulation studies under various “true effect” scenarios, Corriea found that the hierarchical modeling approach consistently outperformed the standard approach of separate models for each ARV drug in terms of reductions in the false discovery rate, and generally had high power to detect true associations when drugs within the same class had similar adverse effects. However, in scenarios in which drugs from the same class had effects in opposite directions, or only one drug in a class had a strong effect, larger sample sizes were required for the hierarchical model to perform well [78]. …”

“…Differences in measurement or definitions of outcomes can lead both to under- and over-diagnosis and have to be considered during primary study design as it not possible to control for differences in meta-analysis or data pooling. It is encouraging that steps have been taken by the research community and by the WHO towards building future sustainable collaborations between key cohorts across the globe [74,78]. …”

Cardiac status of perinatally HIV-infected children

“…A model with so many correlated ARV exposures could lead to multicollinearity and unstable estimates with inflated variances; thus, we also used a hierarchical model including fixed effects for each drug class and specifying a random effect for each drug within its corresponding drug class, as described elsewhere [23]. This hierarchical model essentially estimates individual drug effects as the drug class mean plus a deviation for the individual drug, and generally performs better than separate models and mutually-adjusted models; in particular, there are fewer false positives and estimates are generally less biased and more precise [23].…”

“…Evaluating the association of individual ARVs with cardiac outcomes in PHIV youth is complicated by possible drug interactions, as well as by the large number of regimens currently used. The standard approach of evaluating individual ARVs one at a time can be biased because it does not account for other drugs in the same regimen [23]. In addition, analyzing a large number of echocardiographic measures results in multiple statistical comparisons which must be considered when interpreting results.…”