A Guide to Sample Size for Pilot Studies

Sample Size for Pilot Studies 
Maximising the value of pilot data

Pilot Studies are a common strategy to assess the feasibility of a study by previewing the expected outcome in a larger study. However, pilot studies are often too small and may often lead to suboptimal decision-making at the trial design stage.

For example, insufficiently sized pilot studies may significantly mis-specify the pre-trial estimates for the effect size or nuisance parameters used in a study’s sample size determination.

You will learn about:

• Introduction to Pilot Studies
• Common Rules of Thumb for Pilot Study Size
• Sample Size Determination for Pilot Studies
• Internal Pilot for Blinded Sample Size Re-estimation

Sample Size for Pilot Studies

Pilot study sample sizes are often based on simple rules of thumb such as the “rule of 30”. However, these heuristics have been evaluated to be inadequate to achieve the goal(s) of interest for a pilot study. Newer methods allow for the pilot study sample size to be calculated, which integrates the pilot study's objectives and design.

In addition, blinded adaptive design design provides an approach where a pilot study can be directly integrated into the full study. This improves power and ensures the pilot data can be easily utilised in the final analysis. For example, the internal pilot design for sample size re-estimation can adjust for over-optimism in the estimates for nuisance parameters such as the variance or overdispersion at the planning stage.

Watch this tutorial to understand the impact of sample size on pilot study performance, look at the validity of common rules of thumb for pilot size and more formal approaches for sizing and integrating pilot studies.

Data files include:

• T-Test Sample Size Calculation - Example 1 and 2. nqt
• Pilot Sample Size (Whitehead Overall Program) - Example 1 .nqt
• Pilot Sample Size (UCL Manual SD Calculation) - Example 1 .nqt
• Mean Difference Confidence Interval - Example 1 .nqt
• Internal Pilot Two Means - Example 2 .nqt
• Internal Pilot Two Proportions - Example 3 .nqt
• Case-Control Phase IV Study - Example 3 .nqt
• Internal Pilot Two Counts - Example 4 .nqt
  
• Problem Detection Example .nqt
• Internal Pilot Simulation Code .nqt
  
• CountOverdispersionTwo .csv
• CountNoOverdispersion .csv
• BlindedSSRNCT .csv
• BlindedSSR95UCL .csv
• BlindedSSR80UCL .csv
• 250PilotSDEqualOnePointTwoFive .csv
• CONSORT-extension-Pilot-and-Feasibility-Trials-Checklist .doc

• ---

  nQuery helps make your clinical trials faster, less costly and more successful with tools for sample size calculation - refining Frequentist, Bayesian & Adaptive designs. 
  
  Start your unlimited 14-day trial to help design efficient, informative and ethical adaptive trials.