Before running your first experiment or enrolling your first research participant, it’s good to review your study design and solicit input from all members of the research team to ensure that you have included all the measures needed for data collection. Here is a review of some of the most common types of study designs and the critical variables to consider from a data collection standpoint. (Although the terms here are used most frequently in the context of clinical trials, they incorporate general principles that can be applied to a wide range of scientific query.)
- Cohort study—These studies follow groups of subjects over time, examine cause and effect, and can be either retrospective or prospective. Key data measures in a cohort study include predictor variables and outcome variables. Predictor variables are risk factors, health, or lifestyle characteristics that are being studied for their effect on outcome variables, which are the incidence of a disease or condition.
Examples:
—Prospective cohort study: Cigarette smoking as a risk factor for lung cancer over a 20-year period
—Retrospective cohort study: History of fluoridated water supply in a sample of patients with periodontal disease
- Observational study—Unlike cohort studies, observational studies examine a group of subjects at only one point in time to evaluate the prevalence of a disease and associations of risk factors with having the disease. Although this type of study also measures predictor variables and outcome variables, because it does not follow subjects over time, it can demonstrate association but not causality.
Examples:
—Association of particular smoking, eating, and exercise habits in a group of lung cancer patients
—The incidence of birth defects in frogs as a function of their habitat distance from a chemical plant
- Randomized controlled trial—Widely considered the gold standard of clinical trials, randomized controlled trials differ from cohort and observational studies in that they actively impose an intervention on a study group in the hope of changing an outcome rather than passively studying predictor and outcome variables before or after an outcome. These trials, in which the results of the intervention on the study group are compared with the data from a control group, are much more likely to prove causality than a cohort or observational study.
The term “randomized” in a randomized controlled trial refers to the way in which subjects are divided into treatment groups. These trials may be open, in which case both investigator and subjects are aware of which treatment or intervention is being used, or they may be blinded, in which case either the subjects or the subjects and the investigator are unaware of the treatment or intervention being used. (Some researchers, particularly those in the field of ophthalmology, prefer to use the term masked to avoid confusion with the literal state of blindness.)
Blinded studies are useful to control for the placebo effect, in which factors that may influence the subjects’ response to a treatment have nothing to do with the treatment itself. A placebo is a physiologically inert treatment or a substance with no pharmacological effect used as a control to test the efficacy of an investigational intervention or treatment.
Examples:
—An investigational drug tested against standard chemotherapy in two groups of patients with lung cancer
—The efficacy of an experimental antiretroviral drug versus a standard cocktail treatment in two groups of HIV-infected mice
Although the study designs presented here are probably the most common, especially for studies involving human subjects, there are innumerable variations. The following reference texts offer more information about these and other types of study design to consider depending on the specific parameters of your research:
- Day, Robert A. How to Write & Publish a Scientific Paper. Phoenix: Oryx, 1998.
- Good, Phillip. The Design and Conduct of Clinical Trials. Hoboken: John Wiley & Sons, 2006.
- Hulley, Stephen et al. Designing Clinical Research. Philadelphia: Lippincott, Williams & Wilkins, 2007.
- McFadden, Eleanor. Management of Data in Clinical Trials. Hoboken: Wiley-Interscience, 2007.
