BackgroundThe aim of this study was to analyze four pre-operative physical frailty indicators from a geriatric assessment (GA) independently and combined in a physical frailty index, in their ability to predict postoperative 30 d-complications. Materials and methodsIn this secondary analysis of data from the published PERATECS study (ClinicalTrials.gov: NCT01278537), the predictive value of four physical frailty indicators from a defined GA battery was examined with univariable and multivariate logistic regression models in a sample of 493 onco-geriatric surgical patients. The primary endpoint was incidence of major (Clavien–Dindo ≥ grade 2 [CD ≥ 2]) complications within 30 postoperative days.Predictors of the first model included self-reported exhaustion (SRE), body mass index (BMI), Timed Up-and-Go (TUG) and handgrip strength (HGS) independently, and a second model combined these four items as a Physical Frailty Index (4i-PFI). Both regression models were adjusted for age, gender, American Society of Anesthesiologists (ASA) status, tumor sites, duration of surgery time and Mini Nutritional Assessment (MNA) score. ResultsA total of 233 patients (47 %) developed CD ≥ 2 complications. In addition to ASA score, length of surgery, and gynecological and upper gastrointestinal tumor sites, the first model showed that SRE (OR 1.866) predicted CD ≥ 2 complications, but not TUG, BMI and HGS. In the second model, the 4i-PFI predicted CD ≥ 2 complications (OR pre-frail = 1.808, frail = 3.787). ConclusionsPhysical frailty indicators as SRE revealed a better ability to predict CD ≥ 2 complications than BMI, TUG and HGS. However, prediction of CD ≥ 2 complications was enhanced when these parameters were combined in a novel 4i-PFI.