Table 1

Description of epidemiological study designs (adapted from Detels8)

NameAims and key aspects of designAdvantagesDisadvantages
*Incidence: number of instances of illness starting, or of persons falling ill, during a given period in a specified population.1
Odds ratio1: exposure odds ratio—the ratio of the odds in favour of exposure among cases to the odds in favour of exposure among non-cases; disease odds ratio—the ratio of odds in favour of disease among the exposed to the odds in favour of disease among the unexposed.
Prevalence: number of events—for example, instances of a given disease or other condition, in a given population at a designated time.1
Risk: the probability that an event will occur—for example, that an individual with become ill or die within stated period of time or by a certain age or with a certain exposure to a risk factor.1
Ecological studiesDocument the co-occurrence of disease and other factors in a population using existing statisticsCheap. Relatively easy to conduct. Provide rationale for undertaking more expensive analytical studiesRisk factors and disease may not be occurring in the same people
Cross sectional surveysEstablish the magnitude of disease and factors in a community by collecting dataCan document the co-occurrence of disease and suspected risk factors in specific individuals. Useful for studying chronic diseases which have a high prevalence* but an incidence* that is too low to make a cohort study feasibleExpensive. Subject to problems of information and measurement bias and uncontrolled confounders. Not useful for studying diseases with low prevalence/short courses. Unless historical information is obtained from all the individuals surveyed, the time relationship between the factor and the disease is not known
Case-control studiesCompare the prevalence of suspected causal factors in cases and controls and identify associationsCan estimate odds ratio*. Cheaper and easier to perform than cohort and experimental studies. Method of choice for studying rare diseases. Indicated when a specific health question needs to be answered quicklyCannot measure risk*. Very prone to selection bias and recall bias. Not useful for determining a spectrum of health outcomes resulting from specific exposures (since a definition of a case is required in order to perform the study)
Cohort studiesMeasure risk* of disease association with exposure to a factor in a prospective designCan establish the temporal relationship between an exposure and a health outcome. Can determine spectrum of disease resulting from exposure to a given factorExpensive and time consuming. Subject to confounding. Prone to loss to follow up. Can have complications in the analysis when exposure varies over time. Not feasible for diseases with low incidence*. Do not prove causality
Experimental studies/intervention studies/clinical trialsProvide strong evidence, if not proof, of a causal relationship between an exposure factor and diseaseOften considered as providing the most reliable evidence from epidemiological research. Confounding factors that may have led to the subjects being exposed in cohort studies are not a problem here as investigators make the decision about who will be exposed to the factor based on the specific design factors to be employed (for example random allocation, matching)Expensive. Can cause ethical problems in human studies