At the end of the chapter, the reader will be able to:
Identify the general purposes of research
Discuss the important principles of study design
Distinguish between experimental, quasi-experimental, and observational approaches to pharmacoepidemiologic research
Describe various quasi-experimental study designs used in pharmacoepidemiology
Describe various observational study designs used in pharmacoepidemiology
Discuss the relative advantages and disadvantages of the various study designs
Describe the role of meta-analysis in pharmacoepidemiology
Before selecting a particular study design, it is important to consider the general goal of performing a given research study. It can be useful to categorize a study as serving one of three general purposes: description, identification/exploration of associations, or determination of causal relationships (Table 3-1). Because certain study designs may be better suited than others for a given research purpose, identifying what is expected from the standpoint of the research objective can help in guiding the selection of an appropriate study design. For example, if the goal is to identify potential risk factors for a given outcome, selecting a study design that merely describes the occurrence of the outcome will be of little value.
Table 3-1. General Purposes of Research and Examples. |Favorite Table|Download (.pdf)
Table 3-1. General Purposes of Research and Examples.
Potential study designs
Example research questions
Case studies, case series, prevalence studies, cross-sectional studies
- How many patients are adherent to the drug regimen?
- How many physicians switch drug therapy because of adverse effects?
Identification/exploration of associations
Quasi-experiments, cohort studies, case-control studies, cross-sectional studies
- What factors are associated with patient adherence to drug therapy?
- Are physicians who switch drug therapy more likely to have patients reporting adverse effects than those who do not switch therapy?
Determination of causal relationships
Randomized controlled trials, quasi-experiments
- Does a new patient adherence program result in increased adherence?
- Does the introduction of a new electronic alert reduce the rate of the prescribing of inappropriate medications?
Another way of conceptualizing the purpose of research is to determine whether one wants to develop potential hypotheses (i.e., hypothesis generation) or to formally test hypotheses that were previously developed (i.e., hypothesis testing). The traditional epidemiologic designs (e.g., cohort and case-control) are useful in allowing a researcher to develop hypotheses. The process of formally testing a hypothesis is aimed at making some causal statement (e.g., taking drug X will cause a reduction in blood pressure) and is frequently the underlying purpose of conducting research.1 Some sort of interventional study, such as a randomized controlled trial or a quasi-experiment, is usually considered necessary to test a hypothesis formally and arrive at a causal conclusion; however, advances in statistical techniques have increased the strength of causal statements from some observational study designs.2 A more formal discussion about principles of causality is provided in Chapter 6.
Principles of Study Design