Quantitative epidemiology.

We provide guidance for new practitioners in the vocabulary of modern epidemiology and the application of quantitative methods. Most hospital epidemiology involves surveillance (observational) data that were not part of a planned experiment, so the rubric and logic of controlled experimental studies cannot be applied. Forms of incidence and prevalence often are confused. The names "cohort study" and "case-control study" are unfortunate, as cohort studies rarely involve cohorts and case-control studies allow no active control by the investigator. Either type of study can be prospective or retrospective. Results of studies with discrete outcomes (infected or not, lived or died) often are represented best by a form of the risk ratio with 95% confidence intervals. The potential distorting effects of selection bias, misclassification, and confounding need to be considered.