As has been highlighted in previous articles of this column, randomized controlled trials (RCTs) command the highest level in the hierarchy of evidence among clinical studies. However, in some situations, RCTs are neither feasible nor ethical. A classic example from the medical field is the assessment of the association between smoking and lung-related diseases. It is easy to understand that it is not ethical to randomize patients into smoking and nonsmoking groups, and follow them to see whether one group is more likely to develop lung-related diseases in the future. An orthodontic example might be to randomize patients needing orthodontic care to treatment and nontreatment groups to assess long-term changes in periodontal status between subjects undergoing orthodontic therapy vs no orthodontic therapy. In this scenario, it would not be ethical to deprive needy patients of therapy. Another example would be to assess the effect of water fluoridation vs nonfluoridation. In this case, it would not be possible to randomize by place of residence (village, town, or city); comparisons would be feasible with participants from areas with and without water fluoridation and, therefore, without random allocation. In those situations, we observe patients and collect data without making decisions about their exposure and the course of events. Those studies are called observational; the Figure shows how they relate to interventional studies (clinical trials).
Observational studies can be prospective, retrospective, longitudinal, cross-sectional, descriptive, or analytic; those characteristics are shown in the Table .
|Prospective||Information on exposure is collected at the beginning of the study and information on the outcome in the future|
|Retrospective||Information on exposure is collected retrospectively|
|Longitudinal||The interest is to see the route of events over time|
|Cross-sectional||Information on the exposure and outcome is collected at the same time (snapshot)|
|Descriptive||Distribution of an outcome at a particular time point is described|
|Analytical||Associations between risk factors and outcome of interest are explored|