LAUSR

case, they have suggested that the research is methodologically flawed because it finds that octreotide prophylaxis possibly contributes to elevated rates of clinically relevant postoperative pancreatic fistula (CR-POPF). They now have taken us on a metaphorical tour of the great outdoors, from mountains to molehills, through forests and trees, now reaching the sands where we fear their heads may be buried in dogma. The correspondents continue to harken to the results of a Cochrane metaanalysis of randomized controlled trials (RCTs) that they believe irrefutably support the use of octreotide prophylaxis. We reinforce again the limitations of that study, namely that of the 21 RCTs evaluated, only 3 were performed following the advent of the International Study Group on Pancreatic Fistula (ISGPF), which established a universally accepted definition for CR-POPF. Although none of these 3 studies showed significant differences in terms of fistula development, the CR-POPF rates were actually higher in the octreotide cohort for 2 of the 3 studies, congruent with the findings of our investigations. We agree that it is unusual that octreotide prophylaxis was associated with higher fistula rates in our high fistula risk score (FRS) cohort, but can only present the data as it occurred in real practice. We utilized the most updated statistical methods— namely multivariable regression modeling and propensity score matching—to adjust for confounders. They lament, and we surely acknowledge, that these methodologies are inferior to RCT design, as they cannot account for unmeasured sources of bias. For the sake of making progress in the field on this clinical challenge, we would certainly appreciate the opportunity to confirm these findings in a randomized fashion. However, there is unlikely such an option, as rare occurrence limits prospective validation. High-FRS cases constitute just one-tenth of all pancreatoduodenectomies (PDs) performed. An RCT powered to demonstrate the observed reduction in CR-POPF from 33.5% to 13.2% would require 134 high-risk patients (given a 1⁄4 0.05; 1-b 1⁄4 0.80). Therefore, a prohibitive 1340 patients would be required to be approached preoperatively, if all could be enrolled, given that the FRS can only be calculated intraoperatively. As a point of reference, the extremely well conducted RCT by VanBuren et al regarding drain placement during PD took 15 months to accrue 137 patients operated upon across 9 highvolume centers. Overall enrollment rates were on the order of one half of all patients screened. Doing this math would indicate it would take 25 years to accrue enough highrisk scenarios, simply an unfeasible endeavor. We do not wish to quibble anymore about study design and degree of evidence. Instead, we have already moved on and indeed used our data for hypothesis generation, which the correspondents have rightly acknowledged is a baseline value of the work. We have progressively realized tangible change by learning from the principles our research has revealed. Through the development and subsequent implementation of the FRS, and an adoption of the optimal management practices subsequently identified, the senior author has decreased his CR-POPF rate 4-fold down to a rate of 3.2% over the last 185 PDs, along with improvements in complication burden and resource utilization. Importantly, and germane to the current dialogue, this includes no fistulas in the last 13 high-risk scenarios. Two principle, deliberate changes in practice undertaken to achieve this improvement have been the complete omission of prophylactic octreotide and the abandonment of internalized anastomotic stents. We believe that these results substantiate the value of our original retrospective but well-analyzed and robust collection of actual, real-world experience. We suggest to the correspondents, readers, and other surgeons performing PD that you too might be able to improve on currently expected, average outcomes by becoming more flexible and adapting your current approach to this substantial clinical dilemma.