Didscussion

Article Outline

• Copyright

Dr Jon S. Matsumura (Chicago, Ill). I enjoyed the study. I think it's very important, because you've shown a reduction in mortality with the endovascular compared with open repair in a pretty large dataset.

In studies that I am familiar with, more than two thirds of the early deaths after endovascular repair occur after the patient goes home, so they are no longer inpatients. Because the deaths happen within 30 days, they are counted as surgical.

Can you adjust the mortality rate in your endovascular group to account for the shorter length of stay? Can you compare it with a similar time period for the open? Do you have a 30-day mortality rate for the endo and open groups?

Dr Patrice L. Anderson. We are unable to estimate a 30-day mortality rate; we can only estimate in-hospital mortality. Our access to the New York State dataset (SPARCS) did not include patient identifiers; consequently we were unable to follow re-hospitalizations or evidence of deaths out of the hospital. It is important to reemphasize that we have only analyzed in-hospital mortality rates for this presentation.

Dr Satish C. Muluk (Pittsburgh, Pa). I have two questions for you. Do you think the differences between endovascular and open mortality results could be related to the difference in the type of aneurysms being done among the open patients in this time period? And second, the increase in numbers that you saw over the time period, could that have any connection with patients being sent in from out of state to the big centers in New York that would be doing endovascular repair during this period?

Dr. Anderson. I fully agree with your first statement. It may be that the patients who are getting open repair have technically difficult anatomy and that they were not candidates for endovascular repair because of the complexity of their case. The type of aneurysm and the technical difficulty are clinical details that are not available in large administrative datasets such as SPARCS.

I also agree with the second statement. Patients may have migrated into New York State to the centers capable of doing endovascular repair. What we found is that centers performing endovascular repairs did approximately 60% of the open cases in 2000, but by 2002 they were doing about 78% of the open repairs. This may indicate that there is some concentration of open repairs in the institutions that are capable of endovascular intervention.

Dr Richard M. Green (Rochester, NY). That was nicely presented, and I appreciate your sending me the manuscript.

I think these results are of critical importance, because they show that with the introduction of the new technology, and no basis of past experience, community-based surgeons can already get better results than with conventional therapy. And I think you're to be congratulated for bringing this to our attention.

Many of us have criticized the 1-day, 2-day training programs, but clearly they work. Can you tell us more about the learning curve though? I'd be very interested to know, on day 1 are the results as good as at year 5 of experience?

Dr Anderson. From our current data I am unable to conduct a longitudinal analysis of how many cases an institution has done in total, but that would be very interesting for us to look at in the future.

I can say that the smaller volume centers, performing one to five cases per year in 2002, have a comparable mortality rate to the teaching institutions and the centers doing a large number of cases.

The relationship of volume to outcome is always a very interesting and intriguing concept. A challenge with this dataset is that we do not have patient identifiers, which would be needed to adequately adjust for preexisting medical conditions. Consequently, we do not have the basis to say that the mortality at the community level is the same, better, or worse than the rate in a teaching institution, because there may be sicker patients going to the teaching institutions, or vice versa. Though we did look at the mortality rates between low-volume and high-volume centers, and found them to be comparable, we're limited in how much we can accurately say about the volume outcomes.

Dr Alan Dardik (New Haven, Conn). I enjoyed your talk greatly. I have three, I hope simple, questions.

First, were you able to verify the accuracy of the endovascular code; ie, were all the hospitals across the entire state of New York instantaneously, in October 2000, coding accurately their endovascular repairs as endovascular, or were some of them coded as open repairs?

Second, the New York database has instructed us on the importance of both surgeon and hospital volume. Although you said that you couldn't tell us any information about surgeon volume, how about hospital volume in this state?

And last, do you have any information for us about the hospital charges? How did the cost of endovascular repair compare with open repair during this time?

Dr Anderson. Starting at the very beginning, certainly in 2000 fewer hospitals were clear on how to code for endovascular repair. Endovascular repair was approved in 1999 by the FDA; however, the 3971 ICD-9 code was not provided until October of 2000. In our 2000 data we identified only 131 cases coded as elective AAA repair under ICD-9 code 39.71. We suspect that there were more endovascular cases that year and that most of these ended up in the 39.52 category, which is a kind of a catch-all category. Some may have also ended up in the open 38.44 category. Though we did include some of the 2000 data in the manuscript, we were very reluctant to make any kind of comparisons between the two repairs, because the coded data were limited for endovascular repair, and the open data may have included endovascular cases.

Coding accuracy improved in 2001 undoubtedly, and was fairly well established in 2002. Regarding the relationship between hospital case volume and outcome, we would need to analyze risk-adjusted outcomes. This would require access to patient iden-tifiers, which we do not have. In the institutions that performed the top 10% of the endovascular repairs, the mortality rate was slightly higher, though not significantly greater, than the crude mortality that we presented for the state. Thus we have no indication of a volume-outcome relationship.

Last, the economic comparisons between these two repairs will be very interesting for future research. One way to look at this would be to see how patients are coded under the DRG system. There is a large difference in hospital reimbursement, depending on whether patients are coded with or without comorbidities and complications. What we are seeing with our data is that endovascular patients have a significant number of comorbidities. Thus, if patients are properly coded under DRG 110 (with comorbidities and complications), as opposed to 111 (without comorbidities and complications), there is approximately a $10,000 increase in hospital reimbursement. Therefore accurate coding has the potential to make endovascular repair more cost-effective.

Dr John J. Ricotta (Stony Brook, NY). Two questions. This database allows you to track where the patient came from. Do you have any data on migration of patients from one zip code or one area to another area to get this new treatment, particularly the endo treatment?

And then the second question is, do you have any data on the specialty of the providers providing endovascular repair, whether they were all surgeons or whether there were nonsurgical specialists doing it?

Dr Anderson. The answer to the first question is no. When I looked at the dataset, I looked at simply what counties were performing the repair. I did not run what counties the patients were coming from. It would be interesting to look at the migration.

As far as who is performing the endovascular interventions, we do not have any access to physician identifiers, so I cannot tell you whether they were performed by interventionalists or surgeons.