Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.701745
Title: Sample size justifications for pilot trials of publicly funded randomised controlled trials
Author: Whitehead, Amy
ISNI:       0000 0004 5993 2232
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
A sample size estimate for a clinical trial is an important issue as incorrectly estimating it could have both ethical and financial implications for the trial. Calculating the required sample size for a trial with a continuous outcome requires an estimate of the population variance. A pilot trial can be used to get an estimate of the population variance. However, pilot trials are often small and may give imprecise estimates; adjustment methods are discussed which allow for this imprecision. Theoretical minimum values for the overall trial sample size when using an adjustment method to design the main trial after an external pilot trial are provided. Using the results recommendations for external pilot trial sample size are presented which aim to minimise the overall trial sample size. It was found that the optimal pilot trial sample size increases with the size of the main trial, therefore stepped rules of thumb are proposed. For a 90% powered main trial this method indicates that the sample size for a two-armed pilot trial to minimise the overall sample size should be 150, 50, 30 and 20 for standardised effect sizes (δ) of δ < 0.1, 0.1 ≤ δ < 0.3, 0.3 ≤ δ < 0.7 and δ ≥ 0.7 respectively. The work is extended to allow for unequal cost per patient between the two trials. The results show that when the pilot trial is less expensive per patient than the main trial the optimal pilot trial sample size increases, giving more precision for the variance estimate and a relatively small main trial. The opposite is true when the main trial is less expensive than the pilot trial. For a 90% powered main trial this method indicates that the sample size for a two-armed pilot trial to minimise the overall sample size should be between 40- 260, 20-80, 20-40 and 20-30 dependent on the relative cost of the pilot and main trial per participant for standardised effect sizes (δ) of δ < 0.1, 0.1 ≤ δ < 0.3, 0.3 ≤ δ < 0.7 and δ ≥ 0.7 respectively. For internal pilot trials it is shown that the restricted sample size recalculation procedure raises the average sample size and power of the main trial. Aiming to minimise the overall trial sample size, it was found that the optimal pilot trial sample size rises as the main trial size increases. The work presented aims to help researchers choose sample sizes for pilot trials and to assess the impact selected methods have on the power and required sample size of the subsequent main trial.
Supervisor: Julious, Steven ; Cooper, Cindy ; Campbell, Michael Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.701745  DOI: Not available
Share: