The Power of Prehabilitation, the Reporting of Power Calculations in Randomized Clinical Trials Evaluating Prehabilitation in Cancer Surgery: A Systematic Review and Meta-research Study

Abstract

Objective

To assess sample size calculation reporting in randomized controlled trials (RCTs) investigating prehabilitation interventions in oncological surgery patients.

Data Sources

A systematic literature search was performed in multiple medical databases from inception to April 2023, including MEDLINE, Embase, The Cochrane Library, CINHAL, AMED, and PsychINFO.

Study Selection

The inclusion criteria used were RCTs evaluating effectiveness of exercise, nutrition, and/or psychological interventions on postoperative outcomes of adult patients undergoing oncological surgery.

Data Extraction

Two authors (DS and SV) extracted information on the sample size calculation parameters, including type I error (α), power (1-β), mean (or mean difference between randomization arms), and variance (eg, standard deviation) for continuous outcomes, and event rates or event rate difference between randomization arms for dichotomous outcomes. When possible, we recalculated the sample size required using the collected data, given a 10% margin of error.

Data Synthesis

Of the 59 included publications (58 RCTs), 26 (44%) reported sufficient information to complete sample size recalculation. Of those that provided sufficient information allowing us to recalculate the required sample size, 11 (42%) were within a 10% margin of the reported sample size, whereas 9 (35%) were >10% higher than reported sample size and 6 (23%) were >10% lower than reported sample size.

Conclusions

Over half of the published RCTs in this field exhibit poor sample size calculation reporting. Most RCTs that report sufficient sample size information were underpowered. More stringent reporting requirements are necessary.