Choosing a research study design and selecting a population to study

Abstract

Epidemiological studies have been standardised into a group of 'designs'. The descriptive study describes disease by time, place and person and can develop hypotheses about associations between disease and possible determinants. The analytic study tests these hypotheses. The cross-sectional study measures the disease and determinants at a single point in time. The cohort study identifies those within a group with or without a determinant, and observes the occurrence of disease in the two groups. The case-control study identifies a group of patients with a disease and selects a group of persons from the same population who do not have the disease, comparing the presence of a determinant in the two groups. The experimental study, a type of cohort study, is one in which the investigator 'assigns' the determinant (a treatment) to one subgroup in a population and compares the occurrence of a disease between those with and those without the determinant. All such studies must ensure that the comparisons made have relevance to a defined population. This is done by selecting a 'representative' sample from that population. Carefully selecting a study design and population facilitates the creation of new knowledge while avoiding, as far as possible, important errors.