We'd also now invite Dr. Gohl up to the podium to give his talk on predictors of mortality after craniotomy. Thank you for having me. My name's Nick Gohl. I'm a second-year medical student at the University of Pennsylvania. I'll be talking about a study we've conducted looking at surgical complications that predict perioperative mortality following elective craniotomy. So the idea of this study, I think, is fairly straightforward. It's about looking at surgical complications not just as quality improvement endpoints but also as factors that might help predict perioperative mortality much the same way as a medical comorbidity, perioperative mortality being sort of the ultimate quality improvement endpoint. We know that a patient undergoing an elective craniotomy has between a 1% and 2.5% chance of dying within 30 days of their surgery, not an insignificant number by any means. And at this point, there's a fairly robust literature telling us that centers or hospitals with the best rates of surgical mortality do not experience lower rates of surgical complications. Complications occur at just the same frequency, the difference being the rate of which those complications progress to death within 30 days of surgery. It's a concept known as failure to rescue. So what the data tells us is that in terms of preventing perioperative death, the target is not so much preventing surgical complications but it's how we intervene after the first complication occurs, how we break up the pathway from that first complication to death within 30 days of surgery. So sort of toward that aim, one of the goals of this study is to help identify patients with the greatest risk of perioperative death to enable closer monitoring earlier, more aggressive intervention to ultimately prevent some of those deaths. And then sort of a secondary goal is to think about how we can better or differently characterize perioperative complications as measures of neurosurgical quality and neurosurgical safety. The idea being, you know, if you want to ask the question which is worse, a UTI or a surgical site infection after surgery, you could look at the costs incurred to the hospital system or the effect on hospital length of stay, you could also ask the question which of those two, to what degree do they affect the risk for that patient dying perioperatively. It's another metric that we might apply to QI in neurosurgery. So this is quite a large study. We sampled data from almost 33,000 patients undergoing elective craniotomy from 2006 to 2015. We used data from the ACS NSQIP, the American College of Surgeons National Surgical Quality Improvement Program. It's a large publicly available database of surgical outcomes. One that's been well studied and well validated includes data from more than 400 institutions nationwide with large academic centers and smaller community hospitals. And then sort of preoperative factors we considered, things like demographics, medical comorbidities, and there's a long list. And then surgical factors that might affect someone's risk of dying within 30 days of surgery, sort of type of surgery, indication for surgery, things that happen in the OR. We wanted to measure the effect of complications on risk of dying independent of those preoperative and surgical factors. So one way to do that and what we did is to build a model from study data, a predictive model predicting a patient's risk of dying based only on preoperative and surgical factors. And once you have that, you can measure the independent effect that perioperative complications have on mortality risk. And then similarly, we made use of the American College of Surgeons ACS NSQIP surgical risk calculator, a tool also derived from ACS NSQIP data that does just what I've described, predicts a patient's risk of dying based only on preoperative and surgical factors. It's a well-validated tool, one that's been validated specifically within the neurosurgical population. So here we take a look at just the prevalence and the mortality rates associated with the 11 surgical complications. We studied the overall rate of mortality in the study population was 2.1%. And the overall rate of morbidity, so the chance that a patient experienced any complication was 16.6%. The most frequent complication that we studied was bleeding. Bleeding requiring a blood transfusion occurred in 4.4% of patients. And the deadliest complication was cardiac event, which we would define as an MI or cardiac arrest, unsurprisingly associated with a very high rate of mortality, close to 50%. This was one of the rarest complications studied occurring in one out of 200 patients. And then the complication associated with the greatest number of deaths within 30 days of surgery was respiratory failure, which we would define as an unplanned intubation or failure to be weaned from mechanical ventilation within 48 hours of surgery. So out of 671 deaths in the study population, 264, so about 40% were associated with respiratory failure. So here we take a look at the effect of surgical complications independent of preoperative risk. So you can see that the four complications associated with the greatest independent effect on risk of dying, cardiac event, respiratory failure, renal failure and stroke, all the complications we studied were associated with an independent risk of dying aside from surgical site infections. And here we look at the independent effect on mortality independent of the occurrence of other complications. So one potential shortcoming of the studies you could argue how do we know that these complications we're talking about are causally related to a patient's death versus simply being markers of the death process? Are they the reason the patient dies or simply things that happen as a patient is dying? I would argue that, you know, something has to kill the patient whether it's bleeding or infection or respiratory failure. And if we demonstrate a certain complication has a strong relationship with mortality in the absence of all other complications that we look at or independent of the presence of those complications, that's evidence, maybe not proof, but strong evidence that that complication is causally related to that patient's death. And so what we've demonstrated here is that those four complications I mentioned previously as cardiac event, respiratory failure, renal failure and stroke are again the four complications most strongly associated with perioperative death independent of the occurrence of other complications. And I think importantly four out of the 11 complications we looked at showed no statistically significant relationship with death independent of the occurrence of other complications. Those were venous thromboembolism, urinary tract infections, unplanned return to the operating room and surgical site infections. So what the data indicates is that though those complications may occur frequently in sick and dying patients, the data suggests that those complications aren't actually advancing patients along the pathway toward death. They aren't causally affecting those patients' mortality. And here we make use of the ACS and SQIP surgical risk calculator. We've stratified patients based on their preoperative risk into low risk, medium risk and high risk cohorts. So takeaways from this I'd say are those same four complications we mentioned before, cardiac events, respiratory failure, renal failure and stroke show a strong significant increase in mortality across all three risk stratified patient cohorts. Cardiac event in particular shows a strong and roughly equal risk mortality in low risk and high risk patients. So otherwise healthy patient getting neurosurgery or sick patient with many medical comorbidities. If they were to suffer a cardiac event, risk of dying is high for both those patients and almost equal. And if you look at urinary tract infections and surgical site infections in the highest risk patients against sick patients with many medical comorbidities, after suffering those complications the risk of dying is more or less unchanged. Again, suggesting that those complications aren't causally related to these patients' deaths. So to conclude, as we've been saying, there are four complications that show the strongest association with preoperative mortality. So cardiac events, respiratory failure, stroke and renal failure. Show the strongest association on univarity analysis, independent of preoperative risk, independent of the occurrence of other surgical complications and across all risk stratified patient subgroups. And four of the complications we looked at, urinary tract infections, venous thromboembolism, surgical site infections and unplanned return to the operating room weren't associated with mortality, independent of the occurrence of other complications. Thank you.

surgical complications

perioperative mortality

elective craniotomy

neurosurgical quality

perioperative complications