Comparative Analysis of the ASA-PS Score and Clinical Frailty Scale in Predicting Postoperative Intensive Care Unit Requirement in Geriatric Hip Fracture Surgery: A Retrospective Evaluation

Background/Objectives: Identifying geriatric hip fracture patients who will require postoperative intensive care unit (ICU) admission remains a clinical challenge. The American Society of Anesthesiologists Physical Status (ASA-PS) classification and the Clinical Frailty Scale (CFS) are both used in preoperative risk stratification, yet their comparative utility for this purpose has not been well characterized. Methods: This single-center, retrospectively designed study included 243 patients aged 65 years or older who underwent hip fracture surgery between January 2023 and December 2025. The primary outcome was determined as postoperative ICU admission, while the secondary outcomes were in-hospital mortality and postoperative complications. Discriminative performance was assessed by ROC analysis with DeLong pairwise comparison. Multivariable logistic regression analysis was performed to identify independent predictors of ICU admission. Results: Postoperative ICU admission occurred in 72.8% of patients. On multivariable analysis, neither ASA-PS nor CFS independently predicted ICU admission. Age (OR 1.052, 95% CI 1.009–1.096; p = 0.017), coronary artery disease (OR 3.992, 95% CI 1.581–10.083; p = 0.003), and spinal anesthesia (OR 0.363, 95% CI 0.161–0.823; p = 0.015) were found to be independent determinants. The addition of either scoring system to this clinical model did not improve discriminative performance (AUC 0.712 vs. 0.709 for both). For in-hospital mortality, CFS demonstrated a markedly superior discriminative ability compared to ASA-PS (AUC 0.801 vs. 0.623; DeLong p = 0.047). Conclusions: In this study, ASA-PS and CFS demonstrated comparable performance in predicting ICU admission; however, neither scale retained independent predictive value after adjustment for age, coronary artery disease, and type of anesthesia. Moreover, incorporating either score into the existing clinical risk model did not meaningfully improve discriminative performance (AUC: 0.709 vs. 0.712). Conversely, CFS exhibited superior discriminative ability compared to ASA-PS for in-hospital mortality. Given the low number of mortality events (n = 13), this finding should be interpreted cautiously as exploratory and warrants confirmation in larger, prospective, multicenter studies.