Supplementary file 1_Early-admission serum albumin and hospitalization burden in a selected serial-culture cohort of patients with diabetic foot ulcers and subsequent Pseudomonas aeruginosa detection: a retrospective cohort study.docx

Negative binomial and gamma log-link models estimated adjusted within-cohort associations. Exploratory prognostic analyses used bootstrap optimism correction and same-center temporal assessment (development, 2013–2016; assessment, 2017–2018).</p>Results<p>Among 408 patients [261 men (64.0%)], median LOS was 14 days [interquartile range (IQR), 9–19] and median nominal hospital charges were CNY 18,708 (IQR, 12,532–27,979).

Compared with albumin ≥35 g/L, albumin 30 to <35 g/L was associated with longer LOS [adjusted rate ratio (aRR), 1.24; 95% CI, 1.10–1.40] and higher charges [adjusted charge ratio (aCR), 1.28; 95% CI, 1.12–1.46]; corresponding estimates for albumin <30 g/L were 1.52 (95% CI, 1.29–1.78) and 1.67 (95% CI, 1.40–1.99). Per 5-g/L lower albumin, the aRR for LOS was 1.19 (95% CI, 1.14–1.25) and the aCR for charges was 1.22 (95% CI, 1.16–1.28).

Baseline-only estimates changed by less than 3% on the log-ratio scale. Adding albumin increased the apparent area under the curve (AUC) from 0.679 to 0.758 (optimism-corrected, 0.643 to 0.729; ΔAUC, 0.079; bootstrap 95% CI, 0.036–0.120). In the same-center temporal assessment sample (126 patients; 50 events), the AUC was 0.592 for the core model and 0.719 after albumin was added; corresponding Brier scores were 0.234 and 0.204, and calibration slopes were 0.625 and 0.994.

These small-sample performance estimates were treated as descriptive. The median bootstrap-optimal albumin cutoff was 35 g/L (95% bootstrap percentile interval, 33–37 g/L).</p>Conclusions<p>Within this selected serial-culture cohort, lower early-admission albumin was associated with greater hospitalization burden and with numerically better exploratory prognostic performance in same-center analyses. These findings do not establish causality, transportability, a diagnosis of malnutrition, or a clinically actionable albumin threshold.</p>