Observational studies
Cross-sectional study
Measures exposure and outcome at one point in time. Useful for estimating prevalence but limited for establishing temporal sequence.
Case-control study
Begins with patients who have an outcome and compares previous exposure with controls.
- efficient for rare diseases
- commonly reports an odds ratio
- vulnerable to recall and selection bias
Cohort study
Begins with exposed and unexposed groups and follows outcomes.
- can be prospective or retrospective
- permits calculation of incidence and relative risk
- useful for uncommon exposures
Experimental studies
Randomised controlled trial
Participants are allocated randomly to intervention and control groups. Randomisation reduces confounding and provides a strong design for assessing treatment effect.
Measures of disease frequency
- prevalence: proportion with disease at a given time
- incidence: new cases occurring over a period of time
Measures of association
- relative risk
- odds ratio
- hazard ratio
Bias and confounding
Study interpretation should consider selection bias, information bias, confounding, loss to follow-up and the adequacy of adjustment.