Adapting the 5-factor modified frailty index for prediction of postprocedural outcome in patients with unruptured aneurysms

Abstract

OBJECTIVE The 5-factor modified frailty index (mFI-5) is a practical tool that can be used to estimate frailty by measuring five accessible factors: functional status, history of diabetes, chronic obstructive pulmonary disease, congestive heart failure, and hypertension. The authors aimed to validate the utility of mFI-5 for predicting endovascular and microsurgical treatment outcomes in patients with unruptured aneurysms. METHODS A prospectively maintained database of consecutive patients with unruptured aneurysm who were treated with clip placement or endovascular therapy was used. Because patient age is an important predictor of treatment outcomes in patients with unruptured aneurysm, mFI-5 was supplemented with age to create the age-supplemented mFI-5 (AmFI-5). Associations of scores on these indices with major complications (symptomatic ischemic or hemorrhagic stroke, pulmonary embolism, pneumonia, or surgical site infection requiring reoperation) were evaluated. Validation was carried out with the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database (2006–2017). RESULTS The institutional database included 275 patients (88 underwent clip placement, and 187 underwent endovascular treatment). Multivariable analysis of the surgical cohort showed that major complication was significantly associated with mFI-5 (OR 2.0, p = 0.046) and AmFI-5 (OR 1.9, p = 0.028) scores. Significant predictive accuracy for major complications was provided by mFI-5 (c-statistic = 0.709, p = 0.011) and AmFI-5 (c-statistic = 0.720, p = 0.008). The American Society of Anesthesiologists Physical Status Classification System (ASA) provided poor discrimination (area under the curve = 0.541, p = 0.618) that was significantly less than that of mFI-5 (p = 0.023) and AmFI-5 (p = 0.014). Optimal relative fit was achieved with AmFI-5, which had the lowest Akaike information criterion value. Similar results were obtained after equivalent analysis of the endovascular cohort, with additional significant associations between index scores and length of stay (β = 0.6 and p = 0.009 for mFI-5; β = 0.5 and p = 0.003 for AmFI-5). In 1047 patients who underwent clip placement and were included in the NSQIP database, mFI-5 (p = 0.001) and AmFI-5 (p < 0.001) scores were significantly associated with severe postoperative adverse events and provided greater discrimination (c-statistic = 0.600 and p < 0.001 for mFI-5; c-statistic = 0.610 and p < 0.001 for AmFI-5) than ASA score (c-statistic = 0.580 and p = 0.003). CONCLUSIONS mFI-5 and AmFI-5 represent potential predictors of procedure-related complications in unruptured aneurysm patients. After further validation, integration of these tools into clinical workflows may optimize patients for intervention.