Thank you for allowing us the opportunity to respond to the question raised by Dr. Mercier regarding our randomized trial evaluating subcuticular suture type for skin closure after cesarean delivery.1
Randomized clinical trials are the gold standard for assessing superiority of interventions, and randomization ensures two (or more) treatment groups differ only in regard to the interventions being compared. Intention-to-treat (ITT) analysis is important, because it includes every patient who was randomized according to her assignment. The benefits of this primary analysis include maintaining the prognostic balance that was generated from the original random allocation, preserving sample size and statistical power, and ignoring protocol deviations or noncompliance; also, the estimation of treatment effect is, in general, conservative and unbiased. Current CONSORT guidelines recommend analysis by the ITT principle.2 Often, the statistical analysis will also include a per-protocol analysis. This analysis was performed in our population, and results are seen under “by suture received.” Combining these two approaches with similarly reached conclusions increases confidence in the study results.
As suggested in your letter, there are also arguments against ITT in certain scenarios, such as inclusion of patients who did not receive any treatment, treatment effect dilution and increased susceptibility to type II error, heterogeneity, and marked differences in compliant compared with noncompliant patients with lack of generalizability. Although crossover was seen in our study, we had almost no missing outcome data points. In this situation, our a priori ITT analysis accompanied with “by suture received” was chosen. Finally, the addition of 20 patients to the polyglactin 910 group to meet the proposed sample size requirements could potentially have tipped the results to significance; however, studies are designed with ending points, and misallocation to groups occurs in the majority of prospective randomized trials.