Sample size was estimated using the method of Fleiss15 for the difference in proportions. We
established a clinically significant result as an increase in the rate of a completely intact genital
tract (no tissue separation at any site) from approximately 20% at baseline to 30%. In order to
detect this change with 80% power at the 5% level of statistical significance, the required
sample size was 1200 women. This represents 400 women in each of 3 comparison groups
(warm compresses, massage with lubricant, and no touch). After 400 women had been
randomized in labor, the principal investigator, database manager, and statistician assessed
preliminary study findings to determine if any technique was associated with significantly
better or worse trauma outcomes. Since none of the three techniques was associated with
improved perineal outcomes at that time, the study continued with the original sample size
unaltered.
Randomization
Random allocation to one of three perineal management strategies was performed in a ratio