H. Hong1, R. R. Kelz2, M. J. Smith1 1University Of Pennsylvania,Neurosurgery,Philadelphia, PA, USA 2University Of Pennsylvania,Surgery,Philadelphia, PA, USA
Introduction: Patients with intracranial aneurysm (ICA) may be suitable for either surgical or endovascular repair. The limited generalizability of existing studies comparing the two treatment modalities underlines the importance of using observational data to delineate optimal standards of practice. The multi-institutional American College of Surgeons’ National Surgical Quality Improvement Program (NSQIP) is considered the best available data source for surgical outcomes research, yet its utility in evaluating care for patients with ICA is unclear. This study assesses whether the current form of the NSQIP is sufficient for analyses on ICA-specific procedures.
Methods: A retrospective cohort study of patients undergoing surgical clipping (CPT codes 61697, 61698, 61700, 61702) or endovascular coiling (CPT 61624, 61635, 75894) of ICA was conducted using the ACS NSQIP participant use file 2006-2013. Prior to the analysis, a focus group of clinical experts was convened to identify significant preoperative and postoperative variables for ICA management. The availability of the data elements was tabulated. Univariate analysis using the chi-squared test was performed to compare patient, disease, and procedure characteristics. A multivariate logistic regression model was then developed to determine the factors associated with the primary outcome measure, 30-day combined death or stroke rate (30dDS).
Results: A total of 974 adult patients undergoing surgical clipping for ICA repair were identified for inclusion during the 8-year period. Zero entry of endovascular repair was found. A considerable number of potential risk factors (e.g. subarachnoid hemorrhage, ventriculostomy need, shunt need, family history of aneurysms, the Hunt and Hess grading, morphology, location, size, and number of aneurysms) and outcome measures (e.g. vasospasm, follow-up radiographic result, modified Rankin scale, Glascow outcome scale) were not available for analysis. For the clipping cohort, the overall 30dDS was 12.7% with 63 deaths and 80 strokes. In the multivariate model, only two variables—hypertension requiring antihypertensive medication and coma state prior to the operation—maintained significant association with higher risk of 30dDS (OR=2.4, 95% CI 1.2-5.1 and OR=6.8, 95% CI 2.2-22.8 respectively).
Conclusion: The study show that the current form of NSQIP is inadequate as data source for evaluating quality of ICA-related procedures. While a rudimentary outcomes study for surgical clipping of ICA is feasible, the database misses endovascularly treated patients and many variables of significant import to patients with ICA. To improve the utility of the database for this disease, the NSQIP might consider an ICA-specific pilot study including patients undergoing procedures outside of the operating room and making use of the expanded custom fields to capture the missing disease-specific risk factors and outcomes.