As a young scientist, I worked on an early project with an aim to demonstrate the mechanism by which Pseudomonas aeruginosa (P. aeruginosa) caused lethal gut-derived sepsis in mice. This goal was especially important given that P. aeruginosa is one of the most frequent organisms to cause lethal sepsis among burn patients, the critically ill, and those undergoing lifesaving bone marrow transplants and major surgery. We developed a model in which direct introduction of the organism into the gastrointestinal track resulted in lethal sepsis. To accomplish this, we created a novel model of infection (i.e., overnight starvation, surgical injury in the form of a 30% bloodless hepatectomy, and direct cecal injection of live strains of P. aeruginosa).1
However, rather than focus on the host factors that might be causal to this lethal preparation, such as gut barrier function, activation of inflammation, and so on, we focused on the microbe itself. Results indicated that the three-hit model led to the release of host-derived bacterial signaling molecules (i.e., opioids, norepinephrine, and ischemic end-products) into the cecum that activated virulence genes in P. aeruginosa such that its phenotype shifted from that of an indolent colonizer to a virulent pathogen.
We identified the P. aeruginosa quorum sensing-dependent virulence gene that was activated (lecA) in vivo and then demonstrated that it was both “necessary” and alone “sufficient” to cause lethal sepsis in this model.2 To do this required making mutants cured of their ability to express lecA and then demonstrating that the mutants did not express the lethal phenotype in vitro nor kill the mice in vivo.
Finally, to show that lecA was alone “sufficient,” we added extrachromosomal plasmids to the knockout mutants to re-express lecA, making sure that our lecA deletion did not inadvertently knock out any related genes, but rather only the lecA gene. Results indicated that lecA alone was “sufficient” to express the lethal phenotype both in vitro and in vivo. These studies displayed fidelity to the molecular Koch’s postulates of infection pathogenesis as put forth by the preeminent microbiologist Stanley Falkow, PhD.3 Unlike the original Koch’s postulates, he proposed that the mere presence of a strain cannot be claimed to be causative to the disease process unless its disease-producing phenotype (a biphasic response) can be demonstrated to be expressed in vivo.
Thus, to apply this line of reasoning to surgical complications, it is not the mere presence of a soft gland, or small duct, an older patient, “dog ears”4 or ischemia at the anastomotic site that alone is “necessary” (or deterministic) for a complication to occur, but rather how those factors operate in vivo over the course of healing such that the phenotype expressed (e.g., the complication) can be determined to be alone “sufficient” to cause its actual occurrence.
In this quest for truth, there are those who believe that “black box” technology—whereby all visual and audio feed for a given operation can be captured and stored—will solve the much-needed evidence to identify root causes (or necessary and sufficient evidence) of postoperative complications.5 As I see it, there is a fundamental shortcoming with video capture of an operation or act.6 To illustrate this, I played two videos of separate esophagojejunostomies that I personally constructed following a laparoscopic gastrectomy for CDH1 mutation. I explained to an audience of surgeons that the video on the left leaked while the one on the right side of the screen healed without incident. I then asked them to offer their commentaries.
Knowing that the one on the left leaked, many constructive comments, including aspects of my technique, areas of ischemia, the concern over “dog ears,” and so on were made. Once they finished commenting, I informed them that I had reversed the images. The video of the anastomosis on the left did not leak, whereas the one on the right was the anastomosis that actually leaked. This was a good demonstration of hindsight bias7 and the lack of accounting for the process of biologic healing that occurs following surgery, and the reason that makes technique itself “forgiving” to use a common phrase among surgeons.
Pancreaticojejunostomy after pancreaticoduodenectomy leaks presents another such example. In my experience, remarks at M&M conferences have included observations such as “Well, it was a soft gland with a very small duct.” Yet all of the other soft gland/small duct pancreaticojejunostomies that did not leak were dismissed, leaving this comment as a probabilistic assessment of the occurrence rather than a deterministic one.
Such comments and our lack of the ability to continuously monitor biologic healing in vivo over the course of recovery show us that our inability to apply the scientific lexicon of “necessary and sufficient” prohibits us from causally linking our hindsight explanations to the root cause of our complications. It may be for this reason that common complications (e.g., bleeding, leaks, infection) seem to repeatedly surface at M&M conferences. This reality should make us take pause and consider whether we are sufficiently advancing the science of understanding our complications based on our analyses and comments at these meetings such that we are preventing them from occurring in the future. In addition, the examples described in this article should support the assertion that video/audio assessment of intraoperative events likely will suffer from hindsight bias and may fail to provide sufficient evidence that allows us to claim that a specifically viewed image or video is causative to a complication.
Preventing Complications in the Future
Leadership in surgery will surely abjure when confronted with the idea that most expert explanations at an M&M conference are ineffective at preventing the same complications from occurring in the future. Some leaders may argue that expert discussions at M&Ms are useful to trainees so that they can learn from the complications of others. Yet, one might imagine that continuing to hear that anastomotic leaks performed by high-volume expert surgeons are due to an “error in surgical technique”—without actually demonstrating any physical evidence for it—seems incomprehensible when hearing the same argument while the same complication continues to surface over and over.
Similarly, discussions surrounding postoperative bleeding requiring a takeback to the operating room as the operative team declares it was “dry when we closed,” also must seem redundant and ineffective at clearing up why this complication continues to occur. The same could be said for many other complications, including prolonged ileus, delay in diagnosis, and so on. At least to this surgeon-scientist, intraoperative video capture alone, without some type of mechanistic understanding of the biology of events occurring before, during and after surgery, will not provide the evidence needed to change surgical practice such that common complications can be avoided in the future. These include the ever-recurring complications of surgical site infections, anastomotic leaks, bleeding, and delays in diagnoses.
How Can We Do Better?
If a post-M&M questionnaire was required that included the following with a Likert-type scale for answers, perhaps the current format might change.
- Were the appropriate cases among the list of M&M complications chosen for discussion?
- Did the discussion address the root cause of the complication with sufficient evidence to prevent it from occurring in the future?
- Was the discussion balanced, and conclusions made, based on solid, verifiable evidence?
Given that the stated purpose of M&Ms is to openly discuss our complications so that others can learn from and avoid them in the future, a change in format may be needed. To achieve this, a task force should be established to examine how complications are discussed at a typical M&M conference and how the structure and format can be improved.
Disclaimer
The thoughts and opinions expressed in this viewpoint article are solely those of Dr. Alverdy and do not necessarily reflect those of the ACS.
