97how Robust Are Frailty Assessments? A Systematic Review of the Application of Frailty in the Surgical Population

Introduction: Frailty is increasingly used as a proxy for surgical risks and predictor of postoperative outcomes. Yet, uncertainties still remain in its clinical implementation in the perioperative setting, such as which frailty concept or score to use in the surgical population. Method: We undertook a systematic review to determine the most prevalent frailty instruments used and in which clinical setting it was applied. Search strategy was performed by a trained bibliographic specialist. Since we expected high heterogeneity in frailty and outcome definitions, we did not perform a meta-analysis. Results: Two reviewers independently performed title screening, abstract review and data collection. Out of 6979 study titles screened, we included 118 studies in our final analysis. Reported data account for a total of 2,500,040 surgical patients. Frailty was most prevalent in lung transplantation (40.6%, n = 2 determinations) and least prevalent in gynaecological surgery (1.5 %, n = 2 determinations). The most frequently-used instrument was the modified Frailty Index (mFI) from the National surgical Quality Improvement Program (NSQIP) dataset (24%) followed by the Fried frailty phenotype (17%). The Rockwood model had the least variation (30% to 50%, average of 40.5%) in the prevalence of frailty compared to Fried phenotype and Complete Geriatric Assessment. The majority of studies did not specify the timing of frailty assessments (75%). The overall calculated OR of death for frail patients compared to non-frail patients were 3.3. Conclusion: Frailty is very prevalent in patients undergoing elective surgical procedures and is independently associated with adverse outcomes. The Rockwood model however appeared to have the most stable estimates of prevalence between studies and within different surgical procedures. No consensus exists as to the timing of frailty assessment in relation to surgery. Further studies should also focus on patient-centred long term outcomes such as functional decline in order to better inform the decision making process.