Development and Validation of a Risk Stratification Model Incorporating Serum Lactate and Albumin for Predicting...

Enteral nutrition feeding intolerance (ENFI) is a common complication in patients with sepsis and may compromise nutritional support and clinical recovery. This study aimed to develop and validate a risk stratification model incorporating serum lactate (LAC) and albumin (ALB) for predicting ENFI in patients with sepsis.

A total of 258 patients were screened for eligibility according to predefined inclusion and exclusion criteria. The final complete-case cohort comprised 230 patients with sepsis who received enteral nutrition between December 2021 and June 2025, including 88 patients in the ENFI group and 142 in the non-ENFI group. Clinical variables and laboratory indicators obtained within 24 hours before enteral nutrition initiation were analyzed. Predictors were screened using least absolute shrinkage and selection operator regression and entered into multivariable logistic regression for model construction. Model performance was evaluated by receiver operating characteristic analysis, calibration analysis, bootstrap internal validation, and decision curve analysis.

Lactate levels were significantly higher in the ENFI group than in the non-ENFI group (median [interquartile range], 2.70 [2.00-3.86] vs 1.82 [1.19-2.48] mmol/L; P < 0.001), whereas ALB levels were significantly lower (27.83 ± 5.02 vs 32.82 ± 4.33 g/L; P < 0.001). Least absolute shrinkage and selection operator regression identified 4 predictors: ALB, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, LAC, and mechanical ventilation. Multivariable analysis showed that decreased ALB (odds ratio [OR] = 0.80; 95% CI, 0.74-0.86), increased APACHE II score (OR = 1.09; 95% CI, 1.01-1.17), increased LAC (OR = 2.03; 95% CI, 1.50-2.85), and mechanical ventilation (OR = 2.52; 95% CI, 1.28-5.10) were independent predictors of ENFI (all P < 0.05). The 4-variable model achieved an area under the curve of 0.854 (95% CI, 0.803-0.905), outperforming single indicators. Bootstrap internal validation yielded a corrected area under the curve of 0.845, and the Hosmer-Lemeshow test indicated good fit (P = 0.299). Decision curve analysis reported a positive net clinical benefit across a threshold probability range of 10% to 70%.

A 4-variable prediction model anchored by LAC and ALB and further incorporating APACHE II score and mechanical ventilation showed good discrimination for predicting ENFI in patients with sepsis. This model may provide a practical tool for early risk stratification and may help support individualized enteral nutrition monitoring in clinical practice.