Your Easy Guide to Clinical Studies

A research question is a scientific question defined by researchers, with the aim to provide an answer. A statistician may support the SP-INV on the statistical aspects needed to answer the research question.

 

Aspects include:

 

• Sample size calculation. The statistician calculates how many study participants are required to answer the study question (i.e. based on the intervention, an expected large effect requires less study participants to detect a potential difference or change, while an expected small effect requires more participants)

 

• Variable selection: The statistician provides guidance on the inclusion of:
  • Predictor variables expected to affect study outcome. Their number may be infinite, however the number of predictors included in a statistical analysis is limited, and depends on the number of outcome variables
  • Confounding variables, which can lead to the misinterpretation of study results. Their numbers may require an adjustment of the statistical analysis and the adoption of appropriate sampling/randomisation strategies

As a SP-INV,

• Discuss with a statistician how your research question can be translated in a statistical question (i.e. the inclusion of relevant study variables). This will ensure that your research question can be answered using statistical methods (e.g. study design)
• Reassess the feasibility of your study based on sample size adaptations. Do not underestimate resources needed to achieve the required sample size defined during the concept phase
• Include in your study budget costs for statistical support

 

Not reaching a targeted sample size may have unwanted consequences. To stop the study prior to reaching the recruitment target may result in a lack in statistical power needed to answer the research question. Extending the recruitment time or adding new sites to boost recruitment, will require additional resources. 

More

References

ICH Topic E9 – see in particular

• 2.1.1 Development plan
• 2.2.6 Surrogate variables
• 3.5 Sample size

 

Publications

PubMed Central – see in particular

• PMC 6691636: Wilson Fandino, Formulating a good research question: Pearls and pitfalls