This is the authors' abstract. We don't add key points for this paper.
Preoperative identification of pneumonia risk after hip fracture surgery may support risk assessment in older adults. We developed a nomogram and evaluated its conditional internal performance while accounting for missing predictors. This single-center retrospective cohort included patients aged ≥ 65 years undergoing hip fracture surgery during 2021–2024. The outcome was first clinically diagnosed pneumonia through postoperative calendar day 7, with surgery designated day 0. A fixed six-variable logistic model used age, COPD/asthma, recorded dementia or cognitive impairment, minimum preoperative SpO₂, albumin, and ln(CRP + 1). Missing SpO₂ and/or CRP were handled using 50 chained-equation imputations with 20 iterations. Internal validation resampled patients before imputation. The pooled equation was scored in the 760 patients with genuinely observed complete inputs, defining a conditional evaluation target (B-observed). Approximate corrected-performance intervals used 1000 outer samples, four inner full-imputation developments per sample, variance-component removal of inner Monte Carlo noise, and a first-order normal approximation. Of 937 eligible patients, 80 (8.5%) developed pneumonia; 177 had missing predictors. In the overlapping complete-input subgroup (65/760 events), apparent AUC was 0.946 and optimism-corrected AUC was 0.939 (approximate 95% CI 0.916–0.962). The corrected calibration slope was 1.583 (1.245–1.921), calibration-in-the-large was 0.047 (-0.116–0.210), and corrected Brier score was 0.0497. The complete-case refit had corrected AUC 0.944. Decision curves were exploratory. A six-variable preoperative nomogram was developed in 937 older adults undergoing hip fracture surgery. It showed high optimism-corrected discrimination in patients with complete observed predictor data, although the calibration slope indicated insufficient dispersion of predicted risks. The findings suggest potential value for preoperative risk stratification. External validation and further evaluation of calibration, missing-input handling, and clinical utility are needed before routine implementation. Not applicable.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: