The five-point modified frailty index (mfi-5) predicts surgical outcomes in pancreatic ductal adenocarcinoma (pdac) patients undergoing pancreaticoduodenectomy

Abstract

Background: Predicting surgical outcomes has become increasingly important. Multiple predictive models have emerged, however, their adoption into practice has been slow. Nowhere is the need more pressing for identifying appropriate surgical candidates than in complex surgical procedures such as pancreaticoduodenectomies (“Whipple procedures”). Velanovich et. al. reported an eleven-point modified frailty index (mFI) using variables collected by the National Surgical Quality Improvement Project (NSQIP) to predict outcomes for emergency general surgery procedures in elderly patients. Modification of the mFI has resulted in a simplified 5-point version mFI (mFI-5). This study evaluates the predictability of the mFI-5 in pancreatic cancer patients undergoing pancreaticoduodenectomies. Methods: We conducted a retrospective study of patients undergoing pancreticoduodenectomy (CPT: 48150, 48152, 48153, and 48154) for PDAC using the ACS-NSQIP Participant User File from 2014-2016 to evaluated frailty and its relationship to postoperative outcomes. Using the mFI-5, each patient was assigned a numeric score (0 or 1) dependent on the presence of the following preoperative variables: functional status, diabetes, history of COPD, history of CHF, or hypertension requiring medication. Patients were then categorized, on a scale of 0 to 5, as no frailty (0), low frailty (1-2), intermediate frailty (3-4), or frail (5). Primary outcomes were evaluated according the following variables: length of stay (LOS), unplanned return to OR (RTOR), 30-day readmission, and 30-day mortality. Results: 6040 patients (age 20–89) were identified with no (37.5%), low (61.2%), or intermediate frailty (1.3%). There were no patients identified within the cohort identified within the frailty 5 category. A significant age difference was noted between the no (63.1), low (68.2), and intermediate (69.2) frailty groups (p < 0.0001). There were fewer women identified in the low and intermediate frailty groups compared to the no frailty group (p = 0.0001). Table 1 depicts the surgical outcomes according to the mFI-5 classifications. LOS was greater in the low and intermediate frailty group compared to the no frailty groups (p < 0.0001). Readmission rates were greater in the intermediate frailty group than in the no frailty group (24.7% vs 14.0%, p = 0.014). Although the 30-day mortality was greater in the low (2.6%) and intermediate (5.2%) frailty groups than in the no frailty group (1.4%), it was only statistically significantly different in the low frailty group. Conclusion: The small subset of preoperative medical co-morbidities included in the mFI-5 significantly predict differences in postoperative outcomes such as longer median LOS, readmission rates, and mortality between frailty groups. The mFI-5 is a simple and excellent predictor of outcome in PDAC patients undergoing pancreaticoduodenectomy. A larger study, including additional patients with higher frailty, is required to fully elucidate the validity of the mFI-5 for predicting surgical outcomes within the intermediate frailty or frail groups. A concise clinical prediction tool such as the mFI-5 has the ability to be easily widely applied in surgical practice to aid surgeons in making appropriate risk stratifications for patients with pancreatic cancer undergoing pancreticoduodenectomy.