Tuberculosis in prisons. Importance of considering the clustering in the analysis in cross-sectional studies

This is the supplementary material that includes: 1) Systematic search in PubMed to identify those publications whose main objective was to estimate the prevalence of TB and the risk factors associated with TB. 2) The two datasets used to show the impact of adjustment versus failure to adjust by the cluster structure and make researchers aware of diseases in prisons, especially infections such as tuberculosis. We compare log-binomial and Poisson regression models, Generalized Estimating Equations, and mixed-effects regression models. We illustrate our methodology with two studies of tuberculosis in prisons