Disease frequency, study design and causal interpretation
Epidemiology connects a defined population, an observation period and a health event. Incidence counts new events among people at risk, whereas prevalence describes existing cases at a specified time or period. Person-time allows different observation durations to contribute to an incidence rate. A cumulative incidence instead uses an initially at-risk cohort over a defined interval. These denominators answer different questions and should not be interchanged merely because both results are called rates.
A cross-sectional study measures exposure and outcome together and often cannot establish their temporal order. A case-control study starts from outcome status and compares earlier exposure, while a cohort starts from exposure or another defined membership and follows outcomes. A retrospective cohort can use an already completed exposure period without becoming a case-control design. Matching changes the analysis: in a simple paired case-control design, discordant pairs provide the matched exposure odds ratio.
Relative and absolute comparisons provide complementary information. A rate ratio describes proportional contrast, while a rate difference expresses additional events per unit of person-time. An attributable interpretation requires a credible causal argument and appropriate assumptions; removing a measured association does not automatically remove the predicted number of cases. An odds ratio should not be casually relabeled a risk ratio when disease is common.
Selection bias concerns who enters or remains in a study. Recall and measurement bias concern how information differs systematically. Confounding arises when a separate factor distorts the exposure-outcome comparison, and adjustment requires a defensible causal model. Adjusting for a mediator or collider can introduce problems rather than improve every estimate. Stratified results should also be examined for effect modification, which is a substantive difference between effects rather than simply an error to remove.
For a calculation, first draw the table and label every denominator. For interpretation, identify the design, timing and plausible alternative explanations. Temporality is necessary for a cause, but temporal order alone is insufficient. Report uncertainty and limitations rather than use a significant association as proof. The practice questions provide scientific models, not an official national investigation manual or legal reporting schedule.