Skip to main content

You are here

Statistical Considerations

Researchers are recommended to involve a statistician as early as possible in the design stage of a study (before the protocol is developed).  Statisticians can advise on study design, outcomes measures, sample size, randomisation (and stratification factors) and methods of analysis (including interim analysis).

Noclor will ensure that there is an appropriate level of statistical input into a research study when considering a sponsorship request (when a Noclor partner is being requested to act as sponsor). Noclor may be able to offer some statistical support in the design of non-clinical trials so please contact Noclor directly to inquire about this service .

The level of statistical input should be proportionate to the study type. 

Statistical considerations in study design

The HRA guidance Specific questions that need answering when considering the design of clinical trials lays out
questions researchers, sponsors, peer reviewers, and Research Ethics Committees (RECs) should ask when considering their study design. This document is “layered”, providing increasing detail if needed. It is not designed to be read from beginning to end. The reader should first consider the questions in Layer I and then click over the term or phrase if more information is required.

It is an important to get sample size calculation right, if the sample size is too small it will not be able to answer the question posed, and would be a waste of time and money (and unethical).  All study protocols should provide information on sample size with solid reasoning behind the calculations and how they have been determined.  The justification should be a concise summary of how the assumptions used in the calculation were chosen and why they are considered plausible for the planned study. The assumptions should take account data from previous studies and any systematic review of existing relevant evidence.  If there are plans to revisit the sample size during their study (after interim analysis), researchers should include this detail in the protocol.  When a sample size has been calculated, researchers should assess and evidence if there is a realistic chance of recruiting the required number of participants within the time stated. Research has to be workable in real life not just on paper.

The appropriateness of the study design, sample size and justifications will be assessed during the peer review process.

Statistician input - clinical trials

A qualified and experienced statistician should be involved in all clinical trials and their details will be requested as part of the sponsorship request process.

For Clinical Trials of Investigational Medicinal Product (CTIMPs) the trial statistician must also be able to demonstrate relevant statistical qualifications, experience and relevant GCP training (for example awareness of guidance ICH Topic E9 – Statistical principles for clinical trials). 

The roles and responsibilities of the statistician on the study should be clearly defined in an agreement (when employed by organisation external to the sponsor) or through the trial delegation of responsibility log. 

The sponsor may request information (supporting statements) directly from the named statistician throughout the trial, for example confirmation of review and approval of any substantial amendments of the protocol (if their signature is not required on the protocol signature page). 

Statistician input - other research

A small study, using simple statistics and significance tests is unlikely to require a qualified statistician, the Chief Investigator or other member of the study team (as delegated) may have sufficient knowledge and experience