To drain or not to drain the infraperitoneal anastomosis after rectal excision for cancer

To the Editor,

We have read with great interest a recent article by Denost et al.¹ published in Annals of Surgery. We would sincerely like to congratulate them, and we admire their effort in evaluating the effect of pelvic drainage after rectal surgery for cancer. However, we believe that their conclusions are not completely consistent with the results of the study, and we are concerned about the extent to which they may lead to negative consequences for patients.

Theirs is a superiority trial comparing drainage to no drainage, with statistically negative (non-significant) results, in which the main criterion for assessment was postoperative pelvic sepsis, including anastomotic leakage, pelvic abscess and peritonitis. According to such statistically negative results, the authors conclude that the use of pelvic drainage after excision of the lower rectum for rectal cancer does not provide any benefits for patients.

Generally, in any superiority trial with negative results, a negative conclusion must be made with great caution: the absence of evidence is not evidence of absence.² Although the effect of drainage seems to be minimal for anastomotic leakage and peritonitis, we do not agree with such a definitive conclusion for the appearance of pelvic abscess. Thirty days after surgery, the authors found a difference in pelvic abscess of 3.7% (95% CI: −2.8 to 10.2%) against no drainage (P = .24). During initial hospitalization, this difference was 4.4% (95% CI: −1.2 to 10.1%, P = .010). Both confidence intervals can be easily calculated from the information in the article, although the authors have not reported them. Thus, in both situations, a 10% difference would be perfectly possible, and this difference is precisely what the authors consider to be clinically relevant. In addition, the probability³ (confidence level) derived from their own data that more pelvic abscesses occur during hospitalization when drains are not used, compared to the use of drainage, is no less than 94%. It is clear that the authors have made a very frequent error of interpretation of a statistically non-significant result.² It is also clear that, in order to facilitate the interpretation of the main results of a study, they should be accompanied by confidence intervals, as stated in the Vancouver Recommendations, and conclusions should not rely on all-or-nothing interpretations of P values.

In conclusion, we believe that the results of the Denost et al.¹ trial do not demonstrate that the use of pelvic drainage does not produce beneficial effects to avoid pelvic abscesses after rectal surgery for cancer. Moreover, given their results, the inverse hypothesis could be perfectly posed. Therefore, this important research problem has yet to be resolved.