Choice of statistical model in observational studies of ART.

Analyzing data on ART presents unique and sometimes complicated challenges related to choosing the unit(s) of analysis and the statistical model. In this commentary, we provide examples of how these challenges arise and guidance for overcoming them. We discuss the implications of different ways to count treatment cycles, considering the perspectives of research questions, data management and analysis and patient counseling. We present the advantages and disadvantages of different statistical models, and finally, we discuss the definition and calculation of the cumulative incidence of live birth, which is a key outcome of research on ART.