Discussion

Article Outline

• Copyright

Dr Michael Stoner (Greenville, NC). I enjoyed the presentation and would first like to thank the authors for the ability to review the study prior to the meeting. This continues the trend of the Nationwide Inpatient Sample (NIS) data that we've seen at the meeting in the last couple days, with a similar spin coming out over and over again. These presentations are consistently showing a higher stroke rate for carotid artery stenting (CAS). The purpose of this paper is to show the influence of age on carotid revascularization outcomes, both carotid endarterectomy (CEA) and CAS. I have a few questions for the authors regarding the study.

First of all, I realize that the NIS does not have temporal data with respect to the codes used to select for stroke, so the authors selected procedures within 48 hours of admission. Therefore, they attempted to not select out those patients who are in-house and already have a stroke. I'd like the authors to expand on this. Were most of these procedures done within 24 hours? In other words, can you tighten up the data a little bit to 24 hours or less, and be sure that you're really looking at elective data.

Secondly, your note that the mortality rate was higher in octogenarians. I was wondering if the NIS data will help us answer exactly why that is. Do you note more hemorrhagic strokes in the patients greater than 80 years of age who have a neurological complication from CAS or CEA?

And lastly, how do these data impact your practice? If we used a traditional surgical metric of stroke-death, carotid artery stenting is the loser at all points in your study. What do the authors advocate with respect to the use of CAS in the elderly?

Dr Vogel. Thank you for your questions. First, we decided to limit the study cohort to focus on elective CAS procedures. We chose to evaluate surgeries performed within 48-hours of admission to attempt to include only elective procedures. However, the majority of the procedures were performed within the first 24-hour period.

Regarding the second question, the reality is the stroke codes contained in administrative data are a significant source of confusion based on the limitations of the data. The code most commonly evaluated in administrative studies is the code for “iatrogenic stroke”. This is likely under-coded, based on the ability of the coders at each institution to identify iatrogenic vs all other strokes, and is an inherent limitation of the dataset. Administrative data is based on billing codes and we believe the iatrogenic stroke code is under-coded by hospitals and we have started a validation study to further analyze this. For this reason, we have included other codes for stroke. Stroke is considered an acute diagnosis and by limiting the cohort as we have to elective procedures we feel our data represents a more accurate description of stroke after CAS interventions than other studies using the same NIS dataset. I believe our study demonstrates that octogenarians do not have a higher stroke rate after CAS than other patients, but that CAS has a higher stroke rate than CEA overall. Stroke was the strongest predictor of mortality, CAS was more often utilized in the elderly, and CAS carries a higher stroke rate than CEA. This, we believe, accounts for the higher mortality seen in the elderly.

Finally, regarding your question describing how this data will impact my practice. We have found CAS to be inferior to CEA using the NIS data. I believe the use of CAS should be based on the individual patient with the knowledge that it has a higher stroke rate than CEA in retrospective administrative data. The choice of CEA vs CAS must be tempered with patient comorbidities to make an appropriate decision for each individual patient and this is how we have modified our practice. Prospective studies, such as the CREST trial, are needed to answer this question appropriately.

Dr Martin Back (Tampa, Fla). Hospital's code diagnoses for patients to maximize reimbursement so the more comorbidities they document from a chart, the greater the potential revenue. How accurate do you think these diagnoses are for a given patient? Your analysis for prognostic variables related to intervention technique is dependent on that accuracy to give meaningful results.

Dr Vogel. Yes, the NIS is administrative data based on billing. It is retrospective data. The code for stroke should not be contained in the data unless the patient had a stroke at that hospitalization, as it is considered an acute diagnosis and we limited the cohort to elective CAS procedures. The reason that the fluid and electrolyte were noted is that it is one of the 29 Elixhauser comorbidity used in evaluating administrative data. The Elixhauser comorbidities are a validated tool for adjusting for comorbidities in administrative data.

Prospective trials are needed to obtain the final answer regarding CAS, and results from these studies using administrative data are retrospective and should be considered hypothesis generating as well as reflective of large populations.