Introduction: Staple line leak after sleeve gastrectomy (SG) is a rare but dreaded complication with a reported incidence of 0-8%. Many surgeons routinely test the staple line with an intraoperative leak test, but there is little evidence to validate this practice. In fact, there is a theoretical concern that the leak test may weaken the staple line and increase the risk of a postop leak.
Methods: Retrospective review of all SG performed over a 7-year period. Cases were grouped by whether an intraoperative leak test (IOLT) was performed, and compared for the incidence of postop staple line leaks. The ability of the IOLT for identifying a staple line defect and for predicting a postoperative leak was analyzed.
Results: 542 SG were performed between 2007-2014. 13 patients (2.4%) developed a postop staple line leak. The majority of patients (N=494, 91%) received an IOLT, including all 13 patients (100%) who developed a subsequent clinical leak. There were no (0%) positive IOLTs and no additional interventions were performed based on the IOLT. The IOLT sensitivity and positive predictive value were both 0%. There was a trend, although not significant, to increased leak rates when a routine IOLT was performed versus no routine IOLT (2.6% vs. 0%, p=0.6).
Conclusions: The performance of routine IOLT after sleeve gastrectomy provided no actionable information, and was negative in all patients who developed a postoperative leak. The routine use of an IOLT did not reduce the incidence of postop leak, and in fact was associated with a higher leak rate after SG.
Do you agree with the authors that the routine use of the IOLT was associated with a higher leak rate after sleeve gastrectomy?
I don't, and here's why.
As I tend to do whenever I criticize a paper, I begin with a confession that I have written a lot of marginal papers in my time. It's one of the reasons I maintain my anonymity.
A "trend" has no scientific validity. A comparison is either statistically significant or it is not. Many scientists and statisticians have rightfully criticized our blind faith in p values, but they remain a standard way of comparing research results. That discussion is for another time. Let’s face it—p values will be around for a long time.
The claim that there was a trend toward an increased leak rate with IOLT was based on a difference of 2.6% among 542 subjects. Even if one believed in trends, the p value of 0.6 clearly indicates that there is no difference between the two percentages. Many authors get away with stating that trends exist when p values are 0.051 or 0.06. That's still debatable, but at least close to the magic p of < 0.05.
I was never a big fan of intraoperative leak testing and agree with the authors' finding that postoperative leaks can occur when the IOLT was negative. As they mention in their discussion, leaks often present long after the date of the operation and may be caused by ischemia, cautery injury, or other factors not readily identifiable by an IOLT.
Because the authors didn't find a single leak by doing the IOLT in 494 cases, they suggest that an IOLT is not necessary. But what if they had found one leak and fixed it. Would that have changed their conclusion?
I wonder if everyone at their institution has stopped doing IOLTs.
PS: Don’t just read the abstract; read the whole paper.