Study designs may be observational or interventional/experimental. This video and the chapter below introduce the main types:
Meta-analysis is also a type of study design.
Case series: A case series describes interesting characteristics observed in a group of patients. It’s often used to report a cluster of rare cases or cases with unusual presentations.
For example, a hospital may report a case series of Coronavirus infections in 12 pediatric patients presenting with Kawasaki syndrome.
Typically, case series do not have a control group. They may also be used to describe different treatment approaches used in different patients and compare outcomes. Case series often serve as a starting point for more detailed studies.
Case-control study: A case-control study compares two groups:
The groups are compared for the presence or absence of particular risk factors. This design is retrospective: you select people based on disease status and then look back in time to identify prior exposures.
For example, to study whether cigarette smoking is associated with heart disease, you select one group with heart disease and another group without heart disease. Then you look back and collect smoking histories in both groups and compare them. You may also identify other risk factors associated with heart disease when comparing the two groups.
The two groups should be comparable. For example, you can’t compare risk factors for ischemic heart disease between adults and children. However, you can compare risk factors for ischemic heart disease between two states or countries.
Case-control studies are analyzed using the odds ratio. They can establish an association between a risk factor and an outcome/disease, but they do not establish causality (you can’t conclude that a risk factor caused the outcome). Incidence and prevalence cannot be calculated from a case-control study.
Case-control studies are useful for:
If the exposure is rare, a case-control study is not useful.
Cross sectional or prevalence studies: In a cross sectional study, data is collected from the study group at a single point in time rather than over a period of time. It measures both the outcome and the exposure(s) in a population and may examine their association.
Examples include:
Prevalence of a disease and odds ratio can be calculated from cross sectional studies. Cross sectional studies are relatively fast and inexpensive to conduct, but they cannot establish causal relationships.
Cohort studies: Cohort studies are prospective studies in which groups are selected based on the presence or absence of a risk factor (exposure) and then followed over time to observe outcomes. Cohort studies are of the following types:
i) Typical or classical or prospective cohort studies: In this type of cohort study, all data are collected prospectively. Subjects in both groups are assessed regularly for outcomes and for the natural history of the disease and/or disease process. This design is time consuming and costly.
ii) Historical or retrospective cohort studies: In this type of cohort study, researchers use existing records (for example, hospital or health records) to identify past exposures and then determine outcomes.
This design is sometimes confused with a case-control study. To tell them apart, focus on how participants are selected:
Although a historical cohort study is less time consuming and less expensive than a classical cohort study, it is less accurate.
Cohort studies can be used to establish causation because the temporality between exposure and outcome is well defined. If the exposure is rare, a cohort design is an efficient method to study the relation between exposure and outcomes. Cohort studies are not always useful for rare outcomes. Incidence rate can be calculated from a cohort study.
Experimental studies or clinical trials: Experimental studies that involve humans are called clinical trials.
In a double blind trial, neither the subject nor the researcher knows whether the subject is in the treatment group or the control group. The treatment group receives the experimental drug or intervention, while the control group receives a placebo.
The highest possible degree of blinding should be chosen to minimize bias.
A randomized controlled trial or RCT is a study design that randomly assigns participants to an experimental group or a control group. RCTs are used to study the safety and efficacy of new treatments. An RCT is considered the best study design because it establishes causality with less bias.
Non-randomized trials do not use randomization and are considered weak studies that are prone to bias.
A sequential clinical trial is a trial in which observations are assessed as they are produced. The total number of participants is not predetermined; it depends on the accumulated results. More subjects are added to the study if the statistical analysis requires it.
In a crossover trial, all patients eventually receive the experimental drug/intervention by exchanging groups. For example, group A gets placebo for the first three months while group B gets drug 1. At the end of 3 months, group A gets drug 1 for another three months while group B gets the placebo.