CONFIDENCE INTERVALS FOR MEASURES OF EFFECT

Studies that measure the effect of an exposure involve the comparison of two or more groups. Cohort studies may be conducted using a fixed follow-up period for each person. These studies allow direct calculation of risks, which may then be compared. Alternatively, cohort studies may allow for different follow-up times for each person, giving rise to data from which incidence rates may be estimated and compared. Case-control studies also come in more than one variety, depending on how the controls are sampled. Usually, the analysis of case-control studies is based on a single underlying statistical model that describes the statistical behavior of the odds ratio. Prevalence data, obtained from surveys or cross-sectional studies, usually may be treated as risk data for statistical analysis because, like risk data, they are expressed as proportions. Similarly, case-fatality rates, which are more aptly described as data on risk of death among those with a given disease, may usually be treated as risk data.