**Background:** Hospital malnutrition is a recognized risk factor for adverse clinical outcomes including increased morbidity, mortality, and healthcare costs, with an estimated 30% of hospitalized patients in the US and Europe presenting with malnutrition at admission. Despite validated screening tools like NRS-2002, MUST, and GLIM criteria, malnutrition remains underdiagnosed in clinical practice. The CONUT (COntrolling NUTritional status) score, calculated from routine serum parameters (albumin, total lymphocyte count, total cholesterol), has shown prognostic value in various diseases but has been less studied for short-term in-hospital outcomes in heterogeneous medical populations, particularly in Western countries.
**Methods:** This single-center, prospective cohort study was conducted at Fondazione Policlinico Agostino Gemelli IRCCS, Rome, Italy, from March 2021 to February 2022. All adults (>18 years) admitted to the Internal Medicine and Gastroenterology ward were eligible. Patients unable or refusing consent were excluded. Upon admission, demographic characteristics, primary diagnoses, comorbidities (Charlson Comorbidity Index, CCI), anthropometric measurements, laboratory values, and nutritional variables (NRS-2002, MUST, nutritional supplementation) were collected. CONUT score was calculated based on serum albumin (g/dL), total lymphocyte count (count/mm³), and total cholesterol (mg/dL), with scores categorized as normal (0-1), mild (2-4), moderate (5-8), or severe (9-12). Patients were grouped into normal-mild (≤4) and moderate-severe (≥5) for inferential analyses. The primary outcome was LOS; secondary outcomes were in-hospital mortality and 30-day re-admission rate. Multivariable logistic regression, Kaplan-Meier curves, and ROC analysis were performed. With an expected incidence of CONUT >4 of 53.1%, a sample of 196 patients was estimated (margin of error 7%, 95% CI).
**Key Results:** 203 patients were enrolled (127 males, 62.6%; mean age 66.05 ± 14.1 years; mean BMI 25.02 ± 4.88 kg/m²; mean CCI 3.02 ± 2.43). 68.5% were admitted from the emergency department. According to CONUT, 44 (21.7%) had normal, 66 (32.5%) mild, 68 (33.5%) moderate, and 25 (12.3%) severe impairment. Mean LOS was 8.24 ± 5.75 days; 38 (18.7%) developed refeeding syndrome; 9 (4.4%) died during hospitalization. Patients with CONUT ≥5 had significantly longer LOS (9.9 ± 6.4 vs. 6.5 ± 4.0 days; p < 0.0001) and higher in-hospital mortality (7.5% vs. 1.8%; p = 0.049) compared to those with CONUT ≤4. No significant difference was found in 30-day re-admission (5.4% vs. 7.3%; p = 0.58). In multivariate analysis, ER admission (HR 2.16, 95% CI 1.48-3.16, p < 0.0001), moderate-severe CONUT (HR 1.52, 95% CI 1.10-2.09, p = 0.01), and RS diagnosis (HR 2.00, 95% CI 1.31-3.05, p = 0.001) were independent risk factors for longer LOS. For mortality, univariate analysis identified CONUT (OR 1.61, 95% CI 1.21-2.15, p = 0.001), RS diagnosis (OR 10.1, 95% CI 2.4-42.6, p = 0.002), and lower BMI (OR 0.82, 95% CI 0.69-0.97, p = 0.02) as risk factors, while nutritional supplementation within 48 hours was protective (OR 0.12, 95% CI 0.02-0.56, p = 0.006). Multivariate analysis for mortality was not feasible due to limited events. ROC analysis showed CONUT had an AUC of 0.831 (95% CI 0.680-0.982) for predicting mortality, with an optimal cut-off of 8.5.
**Clinical Implications:** The CONUT score, derived from routine laboratory tests, is a simple, objective, and reliable tool for stratifying risk of prolonged LOS and in-hospital mortality at admission in an Internal Medicine and Gastroenterology setting. These findings extend previous evidence from Eastern and elderly populations to a younger, Western cohort. The strong protective association of early nutritional supplementation (within 48 hours) with mortality reduction (nearly 90%) underscores the importance of prompt nutritional intervention in at-risk patients. The study's limitations include its monocentric design, small number of mortality events precluding multivariate analysis, lack of body composition data, and uninvestigated confounders (statin therapy, hematological/infective diseases). Further large, multicentric studies are needed to validate CONUT cut-off values across different disease populations.