Development of a health data derived frailty index as a predictor of adverse outcomes in older patients with pancreatic cancer

Abstract

Pancreatic cancer is a prevalent disease among older adults. Well-selected patients, based on a geriatric assessment for risk stratification, could be good candidates for chemotherapy and/or curative resection. Deficits accumulation frailty indices (FI) utilize readily available clinical data and easily obtained patient-reported information to predict hospitalization and mortality of older individuals. Retrospective data from 440 older adults (median age 76 years) with pancreatic cancer, obtained from electronic health records, was used to develop a FI and its ability to predict mortality and other geriatric and cancer related outcomes was tested. Fatigue (n = 45), infection (n = 40) and neutropenia (n = 36) were the most common registered adverse events of treatment; 153 subjects had no adverse events. The mean FI score was 0.26, 112 subjects were fit (0.0 < 0.2), 255 pre-frail (0.2 < 0.35), and 73 frail (≥ 0.35). Median survival was twelve months for the whole sample; at one year 62.5% of fit patients, 46.3% of pre-frail, and 26% of frail patients were alive. The FI categories correlated with institutionalization (p < 0.001) and non-planned hospitalization (p < 0.001). The FI categories did not correlate with the presence of Common Terminology Criteria for Adverse Events (CTCAE) grade 3–4 adverse events (p = 0.377). We conclude that patients with pancreatic cancer classified as frail with our FI had worse survival than those fit and pre-frail. Non-fit patients were also more prone to be institutionalized and have non-planned hospitalizations. The items used for this FI can be usually acquired from electronic health records and could be automated in the future, which could simplify its use as a helping tool for decisions in older patients with pancreatic cancer.