Construction and validation of a frailty screening model for older adults with cataract: a cross-sectional study

Abstract

Objective To develop and internally validate a frailty screening model for older adults with cataract that relies only on information already collected during routine ophthalmology outpatient assessment. Methods In this cross-sectional study, 516 patients aged ≥ 60 years were enrolled and randomly divided into training (n = 361) and internal validation (n = 155) sets. Frailty was defined by the FRAIL scale, with non-frail and pre-frail categories combined for the primary binary analysis. Univariate screening was followed by multivariable binary logistic regression, with LASSO and a 5,000-resample bootstrap of the full development procedure as stability checks. Results The frailty detection rate was 29.07%. Five factors were independently associated with frailty: advanced age, IADL impairment, poor self-rated health, depression, and history of falls. Discrimination was good (training AUC 0.841, 95% CI 0.796–0.887; validation AUC 0.816, 95% CI 0.745–0.888), the optimism-corrected C-statistic was 0.810 (95% CI 0.780–0.841), and calibration was acceptable (validation Brier score 0.1553). Decision curve analysis showed positive net benefit across a broad threshold range. Lowering the operating point from the Youden cutoff (0.463; sensitivity 57.8%, specificity 86.4%) to 0.181 raised validation sensitivity to 86.7% with a negative predictive value of 92.1%. Conclusions Because all five predictors are already recorded during routine cataract assessment, the model offers an opportunistic, zero-burden triage step for identifying concurrent frailty and prompting comprehensive geriatric assessment. External validation and comparison with the FRAIL scale are required.