Featured Article: “A Randomized Prospective Multicenter Trial of Pancreaticoduodenectomy With and Without Routine Intraperitoneal Drainage"
Annals of Surgery volume 259, pages 605-612 , April 2014.
This is a randomized trial comparing outcomes of patients undergoing pancreatoduodenectomy(PD) with and without intraperitoneal drain placement. Surgeons from 9 academic hospitals (all considered “high-volume pancreatic centers” because they perform around 50 PD/year) agreed to offer participation into the trial to all their patients undergoing PD or distal pancreatectomy. 282 patients were randomized over a 15-month period, of which 226 were resected. 137 of these underwent PD and are the cohort described in the study. 68 patients were randomized to the “drain arm”, and had one or two closed-suction drains left intraperitoneally at the conclusion of surgery (the number, size and brand were at the discretion of the surgeon), and 69 patients had no drains left. In the group with drains, they were removed once the output was 20 ml/day for 2 consecutive days and/or the amylase concentration was < 3 x the upper limit of normal serum amylase, which turned out to be sometime between postop day 5 and 13 (median of 7). All complications within 60 days were recorded, and patients were followed up to postop day 90 to assess mortality. The study originally intended to enroll 752 patients in order to detect a 10% difference (that would be statistically significant) in the rate of grade II or more complications between groups. However, the study was stopped early by the Data Safety Monitoring Board because of excess mortality in the patients undergoing PD without drains.
As expected in a study with good randomization, the two groups were comparable regarding gender, age (mean 63 y), comorbidities, BMI, pathology (50% had ductal adenocarcinoma), laparoscopic approach (8%), need for vascular resection (18%), and type of pancreatic anastomosis (the most frequent was an end-to-side, duct-to-mucosa). Operative time (416 min) and need for blood transfusion (7%) were also similar in both groups.
The overall number of complications and the complication severity were significantly higher in patients undergoing PD without drainage. Specifically, delayed gastric emptying (42% vs 24%), intra-abdominal abscess (25% vs 12%), other intra-abdominal fluid collections (12% vs 2%) and severe diarrhea (17% vs 2%) were significantly higher in the no drain group.
Although the 90-day mortality was 4 times higher in the no-drain group (12% vs 3%), this was not statistically significant (P = 0.097). The lack of significance was also true with the reoperation rate, which was 9% in the no-drain group and 3% in the drain arm (P = 0.274).
The authors concluded that elimination of intraperitoneal drains after PD increases severity and frequency of complications, and may directly increase mortality, and therefore cautioned against their abandonment.
1) The “magical” P number for statistical significance is <0.05, yet, the Data Safety Monitoring Board made the decision to stop the trial even though the difference in mortality had a P value of only 0.097. Why do you think they did do? It is a lot of work to organize a multi-institutional surgical trial, and one could argue it would have been better to continue the trial until a significant difference in mortality and/or reoperation were found. Can you define what a P value of <0.05 means? Are all statistically significant differences “clinically” significant and vice versa? What is a Data Safety Monitoring Board? Do all clinical trials have to have one? The mortality in the drain-group was 3%, and in the no-drain 12%. Assume that the mortality in the no-drain group (which is an acceptable mortality for PD) remained the same, how many consecutive patients without mortality would have to be enrolled in the no-drain arm to level off the mortality?
2) Discuss what is unique in pancreatic surgery, and in particular after PD, that most surgeons around the world continue to leave intraperitoneal drains, whereas routine drainage has been abandoned in practically all other intra-abdominal operations. To what were the majority of deaths attributed? One patient in the no-drain group had a laparoscopic PD, was discharged on POD 10, and returned on POD 24 in shock. He was found to have hemorrhage from the hepatic artery and died. Can you give potential explanations for this event?
3) A previous single-institution randomized trial comparing drains vs no drains in PD had in fact shown worse outcomes when drains were used. Some surgeons in that institution (about half, according to a recent paper published by them) and others in different parts of the world don’t use drains after a PD. Do you think it is likely they will change their practice? Those surgeons seem to have good outcomes without using drains; can you come up with potential reasons to explain why the no-drain group in the current study did not do as well? Is there anything you would have done differently if you were designing this trial? Name strengths and pitfalls of a multi-institutional vs a single institution trial.
Please feel free to comment on any or all of the questions above. We look forward to hearing from you, the Annals readers. This article can be accessed for free.