Once you have set a research question there are many types of study design that can be used to allow you to answer this question. The process of thinking through the question, then deciding what you need to measure in order to answer that question will often determine the most obvious study design. However, you might need a statistician if it is a more complicated situation that requires randomisation, for example.

Broadly studies are often described as interventional or non-interventional, intervention being clinical trials and non-intervention, observational. There are other broad descriptions of studies involving human subjects such as sampling only, audit and social science studies. The followings are examples of clinical study designs.

Case Study
This type of study relies on literature review or uses a physician’s clinical cases to evaluate the possibility of an association between an observed effect and a specific environmental exposure. This type of study is useful when the disease is uncommon and when it is caused exclusively or almost exclusively by a single kind of exposure.

Case-Control Study
A Case-Control study is an observational epidemiological investigation. This type of study compares a group of subjects with a particular disease or disorder (“cases”) with a group of subjects without (“controls”). The proportion of each group having a history of a particular exposure or characteristic of interest is then compared. This type of study is retrospective and demonstrates association, but not cause and effect. It has been used, for example, to demonstrate the association between the use of oral contraceptives and thromboembolism.

Cohort Study
A Cohort study is a longitudinal investigation in which subjects with differing exposures to a suspected factor are identified and then observed for the occurrence of certain health effects over some period, commonly years rather than weeks or months. The occurrence rates of the disease of interest are measured and related to estimated exposure levels. Cohort studies can either be performed prospectively or retrospectively from historical records. This type of study also demonstrates association, but not cause and effect.

Randomised Controlled Trial
In this type of study, subjects are randomly assigned into the control group or the investigational group. This study type is generally considered the most rigorous study design.

• Active controlled: The control group receives the typically used or approved treatment while the investigational group receives the treatment or intervention being studied.
• Placebo controlled: The control group receives an inactive product (placebo) while the investigational group receives the treatment or intervention being studied.
• Blinded Studies: When a study is “blinded”, subjects do not know which group they are randomly assigned to. Blinded study designs are meant to eliminate bias toward or against the therapy or placebo. In a double-blind study, neither the physician nor subject knows to what treatment they are assigned. In a single-blind study, only the physician knows what treatment the subject is receiving.

Randomisation
Whatever the method of randomisation used, blinding is also key. Firstly, it is important that the person recording the allocation to different groups does not know which allocation comes next as they may consciously or unconsciously choose which patients to randomise based on which the next group would be. An easy way to avoid this is to make sure that the random allocation is set in sealed, sequentially numbered envelopes and the envelope is only opened after each patient has been recruited, in the order of the number sequence. This reduces biases in allocation, but will not completely eliminate this as the person doing the allocation may try to read the allocation through the envelope, by holding it up to the light, for example. If this is a concern, then a telephonic randomisation service can be used, whereby the person enrolling the patient calls a number to receive the allocation at the point at which randomisation should be done.

Where simple randomisation in small trials is likely to lead to unequal distributions in small studies, the participants might be randomised in smaller blocks of, for example, four participants where there is an equal number of control and intervention allocations (in this case two of each) randomly assigned in the blocks. This means that you will not end up with a significantly unequal allocation in the study overall.

For more information on randomisation, visit: http://www.bmj.com/content/316/7126/201