Text

Introduction: Sample size calculation plays a major role in the planning of a clinical trial. When missing values are expected, a common practice is to inflate the calculated sample size by an estimated dropout rate in order to maintain the desired power. However, it is recommended to include patients with missing values in the analysis via imputation in order to avoid loss of information and potential bias. This results in a discrepancy between the analysis method for which the sample size is calculated and the evaluation method ultimately used.

Multiple imputation (MI) is widely used to replace missing values in the data set as is it allows more variability and uncertainty of the estimator to be represented. Based on the between- and the within-imputation variance, Zha and Harel [1] proposed a power calculation, demonstrating that statistical power can be higher when multiple imputation is used, which has a particular impact on sample size planning. Further research is needed to systematically evaluate how much power can be gained in order to give recommendations beyond the commonly used inflation of the required sample size.

Methods: We extend the simulation study by Zha and Harel [1] with a fixed sample size under the “missing at random” assumption, whereby we compare different imputation methods and vary the number of covariates and their respective relationship to the primary endpoint (weak, moderate, or strong). In each scenario, we compute the power using the provided formula and compare it to the power obtained by a complete case analysis.

Results: We analyse whether the power gain from multiple imputation in the primary endpoint can be robustly quantified in various settings; particularly in relation to the imputation variances. The power gain is then translated to a more efficient calculation of the required sample size. Additionally, we identify scenarios in which MI leads to the most substantial improvements – either in terms of power gain or sample size reduction.

Conclusion: This simulation study aims to contribute to the improvement of sample size calculation for clinical trials that use imputation methods in the primary analysis to replace missing outcome data. It is expected that the required sample size can be reduced. Future work will include a blinded interim analysis and adaptive sample size adjustment.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.

Literatur