Development of an early prediction model for ICU-acquired weakness in sepsis using PNI and SII

ICU-AW is common among sepsis survivors. This study developed and validated an early prediction model for ICU-AW in septic patients and evaluated whether the Prognostic Nutritional Index (PNI) and systemic immune-inflammation index (SII), derived from routine admission blood tests, improve prediction beyond conventional clinical variables.

This single-center retrospective cohort study enrolled adult sepsis patients with an ICU stay ⩾ 48 h and complete laboratory data within 24 h of admission. ICU-AW was diagnosed using a Medical Research Council (MRC) sum score < 48. Candidate predictors, including SOFA, were screened by LASSO regression and entered into multivariable logistic regression to build the model. An ablation model excluding PNI and SII was used to assess their incremental predictive value.

A total of 425 patients were included (training set: 297; validation set: 128), with an overall ICU-AW incidence of 40.8%. After adding SOFA to candidate predictors, the final model included APACHE II score, SOFA score, septic shock, endotracheal intubation, vasoactive agent use, DIC, PCT, and SII/100, whereas PNI/10 showed a non-significant protective trend. The full model achieved AUCs of 0.872 and 0.851 in the training and validation sets, respectively, versus 0.846 and 0.824 for the ablation model, with slightly better calibration and net benefit.

The nomogram may help identify septic patients at high risk for ICU-AW within 24 h of ICU admission. SII appeared to improve prediction, whereas PNI contributed little and should be interpreted cautiously. These findings may support early risk stratification, but the added value of these biomarkers requires further validation.