Now we're moving on to our abstract portion of the presentation, and it gives me great pleasure for our first abstract to present Dr. Rohan Ramakrishna from Weill Cornell, who is actually the winner of our Cone PV House Award, and in fact, please come up and accept your award. Congratulations. Do you want a picture? You got to tweet it. Okay, you got to tweet it. You got to tweet it. Okay, here we go. Wait, you got to pick up the award. Okay, everybody's smiling. Awesome. Thank you very much. And you don't need your own computer because you're already loaded. I just have my notes on. Okay. Let's get you loaded up here. There you go. Left is forward, right is back, and you can also scroll if you like. Okay, perfect. Well, thank you all for having me, and again, thanks to the awards committee for this great honor. Again, my name is Dr. Ramakrishna. I'm from Weill Cornell. And the project we worked on in cooperation with our Center for Comparative Effectiveness Research was looking at surgeon volume both from an annual and cumulative perspective. I have no disclosures. So, you know, administrative databases have become an increasingly powerful tool to assess surgeon outcomes and quality. And though they're flawed, they're increasingly influential. It's important as neurosurgeons, as you've heard alluded to in the previous talk, that we take control of this data and we understand it because policymakers are going to look at this data as real-world representations of our outcomes. And there are a number of databases out there. There are things that are state-driven, like the database that I'll go through. Payer databases, commercial products like Bizion, you may have heard of, that hospitals purchase. NISQIP, for example, which is becoming increasingly granular. And then, of course, society-based registries and subspecialty registries to track patient outcomes. Spine, of course, has done a great job at tracking this across multiple institutions and patients. Now, volume has been known across subspecialties to be associated with outcome, with increased volume resulting in improved patient outcomes. But this can be hard to tease out in specialties with the breadth of pathologies and procedures. So neuro-oncology, which is what I looked at in this study, of course, encompasses things from low-grade tumors to high-grade tumors, skull base, et cetera. So defining volume can be a little bit challenging. There are, of course, salient factors that influence outcome in addition to volume, like age, sex, race, insurance status, whether the procedure was elective or not, and, of course, the hospital volume itself. And, you know, these databases ultimately are still flawed because they don't necessarily they're coding. They're based on coding, which, as you know, can be inaccurate or miscoded or undercoded. Just for a bit of history, this kind of started in the late 80s and 90s with the New York State reporting of cardiac outcomes after CABG. And I'll just point you to the two markers there. The risk-adjusted mortality in 1989 for the poor-performing group was 7.1. And then simply by benchmarking and showing that their mortality rates were much worse compared to their peers, in 1992 they were able to get back, you know, within range of the best-performing group. So it shows you the power of reporting, you know, quality metrics like this. So our hypothesis for this study was that both high annual and cumulative volumes would positively affect patient outcomes. You might surmise that for certain procedures there might be a threshold number of cases after which further improvements in patient outcome would not be affected by volume. So as a simple thing, you know, like a simple uncomplicated microdisc, once you've done 1,000, keeping a high volume of 200 or 300 a year may not be that important in terms of patient outcome. Similarly, trauma hemicraniectomy, the procedure itself remains fairly similar no matter the patient. But neuro-oncology, the procedures can be wildly different from patient to patient. And so that's why I wanted to assess the effect of cumulative and annual volume. In terms of our methods, we looked at the New York SPARCS database, which has been around for over 15 years. And our study population were patients undergoing craniotomy for a malignant brain tumor using ICD-9 codes. We excluded patients with previous craniotomies or neurologic conditions within 30 days of the procedure. We stratified surgeons on the basis of their five-year cumulative and annual volume. And you can see the numbers pointed there. And this is on the basis of median surgeon volumes. We looked at 30-day hospital outcomes that are listed of the unique things. Neurologic outcomes after surgery, of course, have been reported previously. But what's unique about our study is that we looked at the range of neurologic outcomes that can be seen after this. So, for example, hemiparesis, hemiplegia, aphasia, need for trach PEG, those sorts of things, which are not really reported in administrative database-type studies. So we have a little over 16,000 patients initially to study from. And after various exclusions related to surgeon outcome with the patients at full outcome data, et cetera, we ended up with a little over 10,000 to study. In terms of demographics, again, we're divided between low-cumulative and high-cumulative, and then between low-annual and high-annual. And what you can see is that the surgeons with the most volume both in terms of high-annual and most experience in terms of high-cumulative had a higher percentage of younger patients, higher percentage of commercial insurance patients, much higher percentage of elective surgery relative to emergent surgery. And also, in terms of race, had a much higher percentage of white patients relative to other demographic groups. Now, there's nothing you can say about that other than that may reflect access issues relative to the communities that these particular surgeons may serve. Over time, I'll just point you to the gray and the black here. The gray represents high-annual, high-cumulative surgeons, and you can see over time they had increasing market share relative to low-annual surgeons. But in neuro-oncology, as you know, a lot of patients show up in the ER and are quickly operated on in that context, and that's why even surgeons with relatively low experience or low-cumulative volumes, not necessarily experience, have a steady share of patient market share for these sorts of procedures. When you get to start looking at outcomes, all of these are significant apart from DVT-PE. And just to go through them quickly, prolonged length of stay, so greater than the 75th percentile in terms of length of stay. High-annual, high-cumulative surgeons had a much reduced length of stay relative to low-annual, low-cumulative surgeons. And I'll show the data, but bottom line, medium length of stay for a high-volume surgeon was five days. Medial length of stay for a low-volume surgeon was almost eight days. So there's huge cost implications, as you can imagine, for a hospital system. High charges, again, defined by greater than 75th percentile. Much lower charges in the high-volume group relative to lower-volume groups. That may reflect post-operative issues. Again, you can't say much more about that. Also, charges may not be uniform across hospitals, so it's a little hard to interpret. Non-routine discharge, either going to a skilled nursing facility or acute rehab. Again, much more common to go home if you're in a high-volume surgeon versus a low-cumulative surgeon. Hydrocephalus, interestingly, was lower rates in the high-volume group relative to the lower-volume groups. Don't have a great explanation from that. Can't really say much from administrative data. And then medical complications were lower in the higher-volume group. Again, when we looked at neurologic complications, this was, again, improved in the higher-volume group. Again, this is unadjusted data, so this is not multivariate. I'll show you the multivariate data in a second. And, you know, the rate of neurologic complications seems enormously high, 44 percent. But that's by diagnosis, so it's unlikely that these represent significant complications. But nonetheless, that's what the numbers are. Surgical complications did not vary. These include things like infection, CSF leak, intracranial abscess, meningitis. But 30-day reoperation and 30-day death were significantly improved in the higher-volume groups. Of note, which I found quite interesting, was that the rate of 30-day death was about 3.3 percent, which is, you know, much higher than I would have thought by looking at, you know, by personal experience. Ultimately, that's what the data is. If we look at just high-volume groups, so just looking at the high-cumulative groups, there are all these things that I just went through. You seem to have better outcomes the higher the volume is. So if you look at surgeons that do over 66 craniotomies a year for a brain tumor, you can see their length of stay, their charges, their discharge, et cetera, are all improved. I'll skip this slide, but in the interest of time. But again, stroke and hemiparesis were significantly improved in high-volume surgeons relative to low-volume surgeons. I will say one thing about the multivariate analysis. In the interest of time, you know, when you adjust for patient preexisting comorbidities, hospital volume, et cetera, the only things that came out in the multivariate analysis were medical complications and length of stay. So not as strong as the unadjusted analysis, but nonetheless leading us tantalizingly confirming our unadjusted analysis. I think the reasons for that may reflect, even though it seems high of 10,000 patients, when you divide it with 430 surgeons, you get, you know, the numbers aren't that high. So in conclusion, you know, I don't think the data necessarily, because of the multivariate analysis, supports regionalization per se. I think that's what people try to do with these studies, but I think it does support internal regionalization within private groups, et cetera. There was a natural trend towards subspecialization over time. And, you know, the univariate analysis showed the things we talked about. So thank you very much, and thank you for the opportunity. Thank you.

surgeon volume

patient outcomes

neuro-oncology

craniotomy

malignant brain tumors