Discussion

Article Outline

• Copyright

Dr John A. Schwartz (Medford, Ore). Pesident Ballard, members, and guests. The authors have presented their review of the trends in admissions for and the management of ruptured abdominal aortic aneurysm (rAAA) over a 4-year period. They have performed a detailed analysis of the results of surgical intervention for rAAA in nearly 10,000 patients over the 2-year period of 2003 and 2004. These data were obtained from the Nationwide Inpatient Sample. This provides nationwide estimates from a 20% sample of discharge data from hospitals in 37 states.

The detailed analysis of the data set revealed that 90% of repairs for rAAA were open and 10% endovascular. Endovascular aneurysm repair (EVAR) of rAAA resulted in an 11% decrease in overall mortality compared with open repair and was associated with similar numbers of secondary procedures, cost, length of stay and a higher probability of discharge to home rather than a skilled care facility. Seventy-two percent of EVARs were performed at teaching hospitals compared with only 28% in community nonteaching centers. Ninety percent of the centers were categorized as urban. In the manuscript, the region of the Western Vascular Society apparently contributed a very low percentage of EVARs to this particular sample.

The authors concluded that EVAR is being increasingly utilized in the emergency management of ruptured AAA with a steadily declining mortality. EVAR is associated with improved mortality and outcomes compared with open repair, but results in nonteaching hospitals were substantially worse than those obtained in teaching centers.

This paper is important. It adds to the growing literature confirming the feasibility and efficacy of EVAR for ruptured AAA. The concept of endovascular damage control to arrest hemorrhage and serve as either an adequate treatment or a bridge to definitive surgical or secondary endovascular intervention is very appealing. However, the results of the present study, like most of the literature regarding comparisons of open and endovascular repair for rAAA, are difficult to interpret because often patients with symptomatic aneurysms—but not ruptured—were included and the preprocedural hemodynamics of these patients are not considered in relationship to the outcome.

Moreover, the interpretation of results for like patients is limited by selection bias and the anatomic restrictions of current endovascular devices. One cannot account for self-selection of more robust patients who tolerate transport to tertiary hospitals or teaching institutions. The authors state that the regression model confirmed the benefit of EVAR as well as the worst outcome at nonteaching facilities after adjusting for both patient comorbidity and disease severity. However, they do not describe which variables from this administrative database provided this information. The manuscript did not elaborate on the proprietary mortality score, nor did it explain what that number really means in any relative clinical perspective. My guess is that if I were a patient, I would not want to have a score of 2000.

The main limitation of the study, as the authors point out in the manuscript, resides in its foundation—an administrative database. The accuracy of the data depends on the precision of the coding that clerical personnel enter into the database and is subject to errors due to nonmedical or lay interpretation. There are numerous pertinent coding issues that are not accounted for, such as pararenal aneurysms, complex iliac anatomy, aneurysm neck length, and angulation, all of which enter into the selection of the type of intervention and its eventual outcome.

The use of EVAR for rAAA is clearly expanding; however, due to the nature of the data available for this analysis, there is no way to account for the apparent discrepancies in outcome. The authors enumerate their speculations in their discussion of the results in the manuscript. For many reasons, a retrospective analysis of results between the presented groups is much less of a direct comparison and more a simple reporting of results. Like many studies, this paper raises more questions in my mind than it answers. I have several questions for the authors; one more than violates the Zierler rule.

One, you noted geographic differences in utilization of EVAR for ruptured AAA. Why do you think that percentage is so low in the West? Are there any regions underrepresented in the Nationwide Inpatient Sample, or more specifically which 13 states are not included?

Two, since you have reported the outcome of intervention for ruptured AAA by teaching and nonteaching hospitals, does that allow you to assess the effect of individual institutional volume on the outcome?

Three, in your analysis you assume that patients coded for both open and endovascular procedures were conversions and grouped as EVARs. In light of the significant increase of morbidity and mortality with conversions of elective EVAR to open repair, how many patients did this represent and what was the mortality rate in this subgroup?

Four, do you have any information from the database regarding the nature of the secondary procedures in the detailed analysis? A mean of approximately five procedures per patient seems quite high. Did any of these involve the treatment of secondary rupture from endoleak?

Finally, I would like to congratulate the authors on this study and fine presentation, and thank the Western Vascular Society for the opportunity to discuss this paper.

Dr Kelly Lesperance. Thank you, Dr Schwartz, for reviewing our manuscript and providing those insightful comments. The first question posed by Dr Schwartz addresses the discrepancy in use of endovascular repair seen in different regions of the country. These differences were not found to be predictive of type of repair performed on multivariate analysis. During our study period, between 986 and 1004 hospitals from 33 to 37 states were sampled by the NIS. States excluded from each year group were not identical from year to year.

The second question addresses whether individual surgeon performance in endovascular AAA repair may contribute to the difference in outcomes between teaching and nonteaching hospitals. Although this question cannot be answered from our database, a separate analysis of our data examining outcomes associated with relative hospital volume in AAA repair may shed some light. Increasing numbers of elective AAA repairs were associated with better outcomes with use of EVAR for ruptured AAA compared to those which used this technique less frequently in the elective setting.

Third, our analysis did not include an in-depth review of the patients that were coded with both open and endovascular procedures. Instead, they were grouped into the EVAR group, for intention-to-treat purposes of our analysis. We agree that these patients likely had worse outcomes and a review of these patients may be beneficial. However, our inclusion of these patients in the EVAR group would then be expected to worsen the outcomes in this group, and strengthen our overall conclusions.

The fourth question addresses the nature of secondary procedures necessary during hospitalization. Procedures most commonly coded for in these patients include imaging such as computed tomography scan, need for mechanical ventilation, tracheostomy, and exploratory laparotomy. While we may infer that patients who received EVAR whose charts also included an exploratory laparotomy may have suffered a complication of their repair, it would be difficult to tell this definitively from the NIS database.

In discussion of some concerns about this study presented by Dr Schwartz, I will address the topic of selection bias, an inevitable factor in choosing a repair type for patients with this emergency condition. Selection bias, in this case, reflects the inherent limitations to the use of EVAR for rAAA. The limitations include the ability to obtain preoperative imaging, which is somewhat dependent on the patient's presenting hemodynamic status. Availability of resources such as a collection of stent grafts, operating room resources such as an imaging table, knowledgeable support staff and a vascular surgeon comfortable with endovascular techniques all play a role.

Mortality score is a proprietary scoring system used by the Nationwide Inpatient Sample to help adjust for disease severity between groups. This score takes into account comorbid conditions, diagnoses, and procedures performed. Although we do not have all the information for every patient that is computed into the score, it does provide a slight buffer to this selection bias and shows that patients who did undergo open repair tended to be the more unstable or sicker patients. Our analysis did validate that this score strongly correlated with patient outcome regardless of the type of repair.

Certainly we recognize that any database that relies on ICD-9 coding is prone to errors, such as those mentioned by Dr Schwartz. Assuming that these coding errors are randomly distributed, the large size of our data set should compensate for this coding error.

Thank you for the opportunity to present our research.