Statistical Collaboration and Statisticians for UNC Investigator Initiated Studies
Any UNC Investigator Initiated Trial (IIT) which qualifies for PRC review (Expedited or Full) must have some form of statistical justification. This includes smaller studies, such as those that are sometimes referred to as ‘feasibility’ or ‘pilot’ studies. A smaller study does not abrogate the need for statistical justification. (More about small studies later)
To underline the need for statistical justification and statistical design input, the PRC enacted a rule whereby all protocols submitted to the PRC for review require a statistical sign-off. It had been shown time and again that many studies submitted without statistical input were simply not ready for review. And so, to encourage investigators to seek out statistical input and collaboration, the statistical sign-off was enacted.
While all projects submitted to the PRC for review require a statistical sign-off, that sign-off need not come from a LCCC statistician. However, it should be noted that LCCC statisticians are uniquely qualified to assist UNC investigators, since they have extensive experience in designing studies that meet the requirements required to pass the rigor of PRC Review. Excellent statisticians from the SPH Biostatistics Department and other university statisticians may be available to you. In any event, it is likely in your project’s best interest to have a statistical collaborator involved.
If you are unsure about whether or not you need to use a LCCC statistician, a good first step would be to visit the BIOS Walk-In Clinic. This would give you the opportunity to discuss your project and converse with a LCCC statistician about what steps you might want to take next. Or you could go to the Lineberger BIOS Core web page and email our BIOS Core contact person to help you find an appropriate statistician for your project. Your initial contact statistician may not be the one you ultimately work with on your project, but this person would be a good first step to help you find a good fit for you and your project.
Most researchers are familiar with the use of power and sample size calculations that are used for the statistical justification for a comparative study that tests a primary research hypothesis. However, not all studies are about hypothesis testing. Sometimes the researcher may be interested in estimating a parameter of interest. In this case, the ‘precision’ of this estimate would provide the statistical justification for the study. In all study descriptions, there has to be a measureable ‘something’ that informs whether the study was ‘successful’ or not. It is critical that the PRC be able to easily see what defines and makes your study a ‘success’, and how this success was measured.
Small Study Design
The following describe ‘feasibility’ and ‘pilot’ studies. It should be noted that many small studies contain aspects of both.
A ‘feasibility’ study is a small study that is often conducted just to see if some aspect of the proposed research is ‘doable’. For example, an investigator may want to see if a questionnaire can be completed by patients at a particular point in time during their treatment. The primary measure of success for this feasibility project would be the proportion of completed questionnaires. The measure of precision for this objective would be the width of the 95% confidence interval of this proportion.
A ‘pilot’ study is a small study with a sample size of often no more than 5 to 15 subjects. A pilot study is often done as a precursor to a larger study. While data from a small study of this kind rarely produces results that are publishable, these studies may gather crucial information that will be used in power and sample size calculations for a larger study, and may also provide important insight into the study design. Regardless, a pilot study must still contain a clear primary objective and a description of what constitutes a successful completion of the study. A successful pilot study may gather important estimates about measures of interest and their statistical variability. Again, these numbers would be useful in future power and sample size calculations.
Often, a small study will contain components of both these studies. Sometimes these studies may be referred to as ‘exploratory’ studies. The main point to remember is that ALL studies, regardless of size, must have a definition of success that is clear and measureable, and must have some justification for the sample size and/or a measure of precision for the primary objective.