Skip to main content

Table 3 Reworked example RCT protocol sample size calculation sections

From: Specifying the target difference in the primary outcome for a randomised controlled trial: guidance for researchers

Primary outcome type

Example text

Binary

Men After Prostate Surgery (MAPS) radical prostatectomy trial [28]: The primary outcome is continence. The sample size was based upon a target difference of 15% absolute difference (85 versus 70%). This magnitude of a target difference was determined as both a realistic and important difference from discussion between clinicians and the project management group, and inspection of the proportion of urinary incontinence in the trials included in the Cochrane systematic review [29]. The control group proportion is also based upon the observed proportion in the RCTs in this review. Setting the statistical significance to the two-sided 5% level and seeking 90% power, 174 participants per group are required; 348 in total.

Continuous

Full-thickness macular hole and Internal Limiting Membrane peeling Study (FILMS) [7]: The primary outcome is ETDRS (Early Treatment Diabetic Retinopathy Study) distance visual acuity. A target difference of a mean difference of five letters, with a common standard deviation (SD) of 12, was assumed as five letters is equivalent to one line on a visual acuity chart and is viewed as an important difference by patients and clinicians. The SD value was based upon two previous studies; one RCT and one observational comparative study [30,31]. This target difference is equivalent to a standardised effect size of 0.42. Setting the statistical significance to the two-sided 5% level and seeking 90% power, 123 participants per group are required; 246 in total.

Time to event

Arterial Revascularisation Trial (ART) [32]: The primary outcome is all-cause mortality. The sample size was based upon a target difference of 5% in 10-year mortality with a control group mortality of 25%. Both the difference and control group mortality proportions are realistic based upon a systematic review of observational (cohort) studies [33]. Setting the statistical significance to the two-sided 5% level and seeking 90% power, 1464 participants per group are required; 2,928 participants (651 events) in total. Participants will be followed for 10 years.