Observational studies are studies where the effect of a risk factor or intervention is observed without changing who is or isn't exposed to the risk factor.
A case-control study compares exposure histories (e.g., medication use, vaccination, or surgical procedure) between participants who have the disease or condition under study (cases) and participants who do not have the disease or condition (controls) and is particularly useful in investigating relatively rare diseases (e.g., cancer).
A cohort study compares the incidence of outcomes (e.g., disease and mortality) between patients with different exposure status (with/without exposure or exposed at different levels). Depending on the frequency of the outcome, it may require a large number of participants and a long-term follow-up (e.g., for cancer). However, if information on exposure and outcome is already known (e.g., from existing records or databases), the analysis can be conducted retrospectively with significant saving in time and money.
Several variants of cohort and case-control designs (nested case-control, self-controlled case series and case-crossover designs) exist and might be more appropriate in studying a specific hypothesis.
A cross-sectional study examines the relationship between the exposure (or intervention/treatment) of interest and the outcome (e.g., diabetes) in a defined population at a particular time. Researchers may not know whether the exposure occurred before or after the outcome. These studies measure prevalence of the outcome.
In ecological studies, both outcome and exposure are measured for groups of people rather than individuals. Therefore the conclusions, while applicable at the group level, may not not be applicable to the individuals (the "ecologic fallacy").
Learn more about the advantages and disadvantages of observational studies.