Introduction: This population study aimed to test the sensitivity and specificity of nutritional risk index (NRI), malnutrition universal screening tool (MUST) and nutritional risk screening tool 2002 (NRS-2002) compared to subjective global assessment (SGA) and to evaluate the association between nutritional risk determined by these screening tools and length of hospital stay (LOS).

Methods: Patients (n=995) were assessed at hospital admission by four screening tools (SGA, NRI, MUST and NRS-2002). Sensitivity, specificity and predictive values were calculated to evaluate NRI, MUST and NRS-2002 compared to SGA. Multiple logistic regressions, adjusted for age, were used to estimate odds ratios (OR) and confidence interval (CI) for medium and high, compared to low risk in patients hospitalized >11, compared to 1-10 days LOS.

Results: The sensitivity was 62%, 61% and 43% and specificity was 93%, 76% and 89% with the NRS-2002, MUST and NRI, respectively. NRS-2002 had higher positive (85%) and negative predictive values (79%) than the MUST (65% and 76%) or NRI (76% and 66%, respectively). Patients who were severely malnourished or at high nutritional risk by SGA (OR 2.4, CI 1.5-3.9), MUST (OR 3.1, CI 2.1-4.7) and NRS-2002 (OR 2.9, CI 1.7-4.9) were significantly more likely to be hospitalized >11 days, compared to 1-10 days, than patients assessed as low risk.

Conclusion: NRS-2002 had higher sensitivity and specificity than the MUST and NRI, compared to SGA. There was a significant association between LOS and nutritional status and risk by SGA, NRS-2002, MUST and NRI. Nutritional status and risk can be assessed by SGA, NRS-2002 and MUST in patients at hospital admission.