Journal Logo


Of Fistula and Football

Casciani, Fabio MD; Trudeau, Maxwell T. BS; Vollmer, Charles M. Jr MD

Author Information
doi: 10.1097/SLA.0000000000004228
  • Free

The act of transecting and reconnecting hollow viscera is central for most alimentary tract procedures. Despite an ever-burgeoning portfolio of anastomotic techniques, devices, and management practices to achieve optimal tissue healing, anastomotic leaks are still frequent, dreadful, and unsolved sequelae. Clinically-relevant pancreatic fistula (CR-POPF) is the primary procedure-specific complication after pancreatectomy, responsible for substantial morbidity, longer hospital stays, and increased costs. At its greatest severity, Grade-C POPF is the chief contributor overall to mortality (36%) after pancreatic resections.1

Although currently poorly understood, potential biologic drivers of pancreatic fistula include tissue ischemia, inflammatory changes, ineffective operative techniques, and the influence of microbiome. Nevertheless, for pancreatoduodenectomy, anastomotic construction poses time zero in fistula physiology. In this moment, the patient's actual physiologic state is manifest. With this understanding, surgeons will face multiple management decisions: type of anastomosis, bowel alignment, use of stents, application of biologic sealant and/or autologous tissue, drain placement and type, and Somatostatin-analogues administration. Employment of these purported “fistula mitigation strategies” may have a pivotal impact on the success of the anastomosis. Given this abundant toolkit, it is not surprising that significant variability of practices exists amongst surgeons, institutions, and countries.2 Therefore, the most practical and influential time to appraise the patient's specific fistula hazard profile (risk assessment) is intraoperatively.

Although certain risk scores are based on preoperative information, this process (risk presumption) is realistically limited, and unlikely to modify surgical strategy or disqualify patients for resections. Similarly, so-called prognostic factors acquired in the postoperative period have almost no utility for fistula prevention. At this point, in fact, the prevention window has likely already passed, and development, if inevitable, will soon be clinically apparent. Instead, they realistically represent the odometer to the biologic process and the surgeon converts to a reactive management mode.


The Fistula Risk Score (FRS) was conceived to assess CR-POPF risk at the time of anastomosis construction3 – offering surgeons a simple, quantitative risk appraisal tool for real-time decision-making (proactive approach to prevention and mitigation).

The FRS considers 4 elements, 3 of which are attendant to the patient (gland texture, main duct diameter, pathology), and a modifiable variable (estimated blood loss). Altogether, they equate to a weighted sum of 0 to 10 points. Therefore, the FRS calculation does not return a simplistic raw probability of fistula occurrence: instead, the 0 to 10 scale correlates with stepwise escalations of CR-POPF rates, overall morbidity burden, mortality, and resource utilization.4 Compared to other predictive scores, which require specific calculators, the FRS has the advantage of being easily computed by fingertips or, alternatively, using a “cheat sheet” located in the operating room.

Furthermore, the FRS can be segregated into 4 discrete zones consisting of: negligible (FRS 0), low (FRS 1-2), intermediate (FRS 3-6), and high (FRS 7-10) CR-POPF risk. Interestingly, assessing data from the Pancreas Fistula Study Group – a large, multi-institutional, international series – a sizable proportion of cases (30%) resides in the lowest risk zones (FRS 0-2), with CR-POPF incidence varying from 0.7% (negligible) to 5.2% (low risk). Conversely, high risk scenarios only occur 10% of the time, but with daunting fistula rates exceeding 30%. In between, the majority (60%) of cases fall in the intermediate zone, with a 15% leak rate.5

Most importantly, this classification has been successfully used for risk stratification when evaluating fistula mitigation strategies and surgeon/institution outcomes (risk-adjusted analyses). Using this framework, best practice strategies that significantly reduce CR-POPF occurrence have been discerned. Some of the salient insights, mostly tailored to discrete risk zones, include: the validation of a selective drain placement protocol with early removal in the case of low/negligible risk scenarios,6 the ineffectiveness of internal stents,7 and the definition of an optimal treatment approach for high-risk patients (concomitant adoption of pancreatico-jejunostomy with external stenting, prophylactic drain use, and Octreotide omission).8

These and similar studies demonstrate that the search for a singular, ubiquitous technique for fistula prevention may not be sensible. To this point, even well-designed RCTs have provided inconclusive results, as they have thus far omitted appropriate risk-stratification and focused on single strategies in isolation. Similarly, as demonstrated for the low/negligible risk cohort, a precautionary “more is better” approach might even be deleterious. Thus, when evaluating the efficacy of surgical techniques to reduce CR-POPF occurrence, it is imperative for strategies to be analyzed according to a holistic approach to the individual patient's risk profile. In accordance with this concept, the use of the FRS has offered the ability to advance from a first-generation, unsophisticated, one-size-fits-all thought process to a contemporary, situational decision-making approach.


In American football, situational game planning defines a strategy whereby teams constantly adapt their play-calling to specific circumstances they face in a given moment during the game. Although a pregame preparatory strategy can certainly be tailored to the opponent's perceived strengths and weaknesses (risk presumption), many unplanned factors might emerge thereafter.

Quoting 8-time Super Bowl winning coach Bill Belichick, champion of situational football: “Even in some situations that are similar, there is usually something - the conditions of the field, or the game, or the wind, or something else - that adds another variable in there (…) There is really a lot of moving parts there, a lot of factors. When you put it all together, again, we have our guidelines and there is certainly a feel, if you will, for certain things. Then again, there are definitely some strategical situations that trump all of that.”9

Considering the sum of situation-specific factors (risk assessment), a wise, flexible coach will adopt proper mid-game adjustments. For example, pregame strategies will be abandoned dependent on field position, remaining time in the game, turnovers, current score differentials, weather conditions, and injuries. Unfortunately, football coaches have not yet developed a basic summary statistic which can be calculated during the game that predicts optimal strategies to increase the odds of winning. Instead, they must rely on nonobjective, gut instincts. Fortunately, pancreatic surgeons have just such a tool on their side in the FRS.

In fact, the FRS introduces the concept of precision medicine to pancreatic surgery. Moving further from risk categories, a FRS Catalogue of 80 different possibilities (scenarios), and their associated outcomes, has been derived combining all the possible levels of the 4 elements composing the score.5 Each scenario represents a unique biological situation (phenotype) a surgeon can encounter on any given day. Certain common, or most vulnerable, clinical scenarios have been independently scrutinized to discover best-practice interventions for fistula mitigation.5


Football also offers another interesting illustration to understand the relevance of proper risk stratification – the kicking game. 2013, the season in which Matt Prater set the record for the longest field goal ever (64 yards), was also the best ever statistically for National Football League (NFL) kickers. That year, the average Field Goal Completion (FGC) rate reached 86.5%, meaning that only 13.5% of attempted kicks were missed.10 Curiously, such a rate equals to the overall CR-POPF incidence after pancreatoduodenectomy.5

Missed field goal rates demonstrate great variability when considering different field distances.10 In fact, the FGC rate was 97.6% for field goals attempted within the 30-yard line (25.5% of all kicks), and 100% within the red zone (20-yard line). 60.2% of all attempts were within an intermediate distance (30–49 yards) with 13.6% being missed. On the other hand, long distance (50+ yards) attempts were infrequent (14.3%), with a full third of them failing. Figure 1 shows the correlation between NFL FGC and CR-POPF rates, segregated by discrete risk zones. Such data indicates distance from the goal as the major contributor to success for placekickers, although it has been shown that, for pancreatic fistula, the most influential variable is the FRS (OR 1.46).2

Comparison between Missed Field Goal rates across incremental distances from the goal (from the National Football League official statistics, season 2013)10 and CR-POPF rates across FRS-derived risk zones (from the Pancreas Fistula Study Group international collaborative, 5533 Whipple from 17 Institutions).5 The histogram (bottom) displays the CR-POPF rate escalation across the FRS. FRS indicates Fistula Risk Score.

In fact, there are situations in which the odds are overwhelmingly favorable, so that is very unlikely to miss the goal. This happens with the extra-point scenario, when scoring with an “easy” kick is almost automatic (1262/1267 in 2013 =0.4% missed). The analogy for surgeons is the negligible risk zone of FRS, when they face a firm gland, with a dilated duct, in a smooth operation with minimal blood loss for pancreatic cancer or pancreatitis. This single negligible risk scenario (FRS 0) is the most frequent situation pancreatic surgeons can encounter on any given day (8.1% chance) and proves just a 0.7% CR-POPF rate.5 Only technical errors, such as gross mistakes in anastomotic technique or incredible botched snaps for kickers, might interfere with a predicted 100% success rate.

Accuracy becomes essential in moderate risk scenarios (30–50 yards). In such situations, once distance is measured kickers must consider additional factors (wind, turf conditions, hashmarks, and the possibility of a blocked kick) and consequently embrace the most accurate strategy for success (blocking scheme, kicking style, run-up length, air trajectory…). Similarly, once surgeons recognize both glandular and operative factors as being challenging (FRS 3–6), they should apply - on a scenario-by-scenario basis - the best strategies and their particular combination to mitigate fistula development.

Furthermore, there are infrequent (≈10%–15%), high-stakes scenarios where distance becomes nearly prohibitive (50+ yards) or, analogously, the pancreatico-enteric anastomosis is daunting (FRS 7–10). Only selected placekickers have the “leg” to make such field goals, just as perhaps only the most skilled, experienced surgeons can have mastery over a 1-mm duct in a mushy pancreatic remnant after a tough, bloody operation. In fact, in such intimidating circumstances, the surgeon may decide instead to “punt” by not actually performing an anastomosis (eg, duct occlusion, external drainage, conversion to total pancreatectomy), rather than incur the consequences of a disintegrated anastomosis.

Not surprisingly there is variance in quality amongst NFL kickers - whose FGC rates ranged between 70% and 96% in 2013.10 Furthermore, risk assumption differs between kickers. In 2012, Greg Zuerlein attempted 13 kicks >50 yards (making 54%), whereas his peers averaged 3.8 attempts and 62%. Similarly, both surgeons and institutions display unique FRS risk assumption profiles, and their fistula performance rankings differ significantly between both raw and risk-adjusted rates.2

Finally, there has been an inexorable improvement in kicker's FGC rates over time. For instance, the >50 yards FGC rate has improved over 40% over the last decade (48%–68%). This, no doubt is the product of specialization. It is not a coincidence that the first of just 2 Pro Football Hall of Fame placekickers, Jan Stenerud, was originally a soccer player. His dedication to refining kicking's technical aspects (creating the so-called “soccer style” technique) enabled him to exceed his peers’ performance, such that he concluded his 19-season career with an outstanding 66.8% FGC rate (compared to his peer's 52.6%), yet was just 27% from long-distance. Today's kickers have constantly refined their craft to the point they now average 87% made overall and an astounding two thirds of those attempted >50 yards. Although not easily measurable over time, it is undeniable that specialization in pancreatic surgery has clearly diminished the frequency of its Achilles heel – pancreatic fistula.


Two common, although not necessarily productive, mindsets persist in the surgical consciousness, namely the aforementioned “more is better,” and “one-size-fits-all” perspectives. Now, more than ever, coaches and general managers in all sports are following the lead of the New England Patriots, who construct a roster with great flexibility to address the strengths and/or weaknesses of their opponents on any given Sunday, not to mention other variables like weather, stadium conditions, or rest intervals. This is an extension of the concept of “in-game” situational football to being able to maximize competitiveness over the full course of a schedule and season – including preparations for optimal performance during the playoffs. For the pancreatic surgeon, this translates to familiarity, facility with, and appropriate application of a full array (tool-box) of options for fistula mitigation. Contemporary data analytics is helping to overcome habitual surgical behaviors, thus allowing more tailored patient-level, targeted management.


Professional football represents a domain in which analytics are constantly employed to identify opportunities for improvement. By combining huge amounts of observational data, outcomes are predicted so that personalized training schemes and strategic decisions can be affected. Compared to other CR-POPF prediction models, the FRS is a simple tool offering an in-game, situational approach to pancreatoduodenectomy – allowing for selection of optimal mitigation strategies according to the patient's specific risk profile at the time of anastomotic construction. It is reasonable to project that CR-POPF may never be eliminated entirely, because of the complex technical nature of pancreatoduodenectomy, menacing pancreatic constitution and physiology, and enigmatic anastomotic healing processes. However, one can ask themselves if they are just satisfied with a relatively low, but not so insignificant, 13% to 20% leak rate? Quoting another iconic NFL coach, Vince Lombardi: “Perfection is not attainable, but if we chase perfection, we can catch excellence.” Thus, by applying analytics to their own practice, every surgeon should get gradually closer to the lowest CR-POPF rate attainable through an exercise of continuous improvement. Notably, the FRS constitutes an example of how real data analysis obtained from a long-lasting, international collaboration can be utilized to investigate performance, improve outcomes, and refine practice. We suggest such a paradigm could be adopted to generate similar prediction scores in other high-acuity specialty domains, such as esophageal or colorectal surgery, where anastomotic leaks also constitute dreadful complications.


1. Vollmer CM Jr, Sanchez N, Gondek S, et al. A root-cause analysis of mortality following major pancreatectomy. J Gastrointest Surg 2012; 16:89–102.
2. McMillan MT, Soi S, Asbun HJ, et al. Risk-adjusted outcomes of clinically relevant pancreatic fistula following pancreatoduodenectomy: a model for performance evaluation. Ann Surg 2016; 264:344–352.
3. Callery MP, Pratt WB, Kent TS, et al. A prospectively validated clinical risk score accurately predicts pancreatic fistula after pancreatoduodenectomy. J Am Coll Surg 2013; 216:1–14.
4. Miller BC, Christein JD, Behrman SW, et al. A multi-institutional external validation of the fistula risk score for pancreatoduodenectomy. J Gastrointest Surg 2014; 18:172–179.
5. Trudeau M, Casciani F, Ecker BL, et al. The Fistula Risk Score Catalogue: toward precision medicine for Pancreatic fistula after pancreatoduodenectomy. Ann Surg. 2020 Jun 11. [Epub ahead of print.] doi: 10.1097/SLA.0000000000004068.
6. McMillan MT, Malleo G, Bassi C, et al. Multicenter, prospective trial of selective drain management for pancreatoduodenectomy using risk stratification. Ann Surg 2017; 265:1209–1218.
7. Sachs TE, Pratt WB, Kent TS, et al. The pancreaticojejunal anastomotic stent: friend or foe? Surgery 2013; 153:651–662.
8. Ecker BL, McMillan MT, Asbun HJ, et al. Characterization and optimal management of high-risk pancreatic anastomoses during pancreatoduodenectomy. Ann Surg 2018; 267:608–616.
9. NBC Sports. Bill Belichick goes deep (really deep) into situational football. September 13, 2016. Available from: Accessed February 17, 2020.
10. Pro Football Reference. Available from: Accessed February 10, 2020.

anastomotic leak.; fistula risk score; pancreatoduodenectomy; postoperative pancreatic fistula

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.