Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials

• Published in: Clinical nutrition (Edinburgh, Scotland) Scotland), 2003-06, Vol.22 (3), p.321-336
• Main Authors: KONDRUP, JENS, RASMUSSEN, HENRIK HØJGAARD, HAMBERG, OLE, STANGA, ZENO
• Format: Article
• Language: English
• Subjects: clinical outcome; enteral; Enteral Nutrition; food; hospital; Humans; Likelihood Functions; Logistic Models; Malnutrition - diagnosis; Malnutrition - therapy; Mass Screening - standards; Meta-Analysis as Topic; nutrition; Nutrition Assessment; Nutritional Support; Outcome Assessment (Health Care); parenteral; Parenteral Nutrition; patients; Randomized Controlled Trials as Topic; Risk Factors; ROC Curve; screening; Sensitivity and Specificity; Severity of Illness Index
• ISSN: 0261-5614; 1532-1983
• DOI: 10.1016/S0261-5614(02)00214-5

Background & Aims: A system for screening of nutritional risk is described. It is based on the concept that nutritional support is indicated in patients who are severely ill with increased nutritional requirements, or who are severely undernourished, or who have certain degrees of severity of disease in combination with certain degrees of undernutrition. Degrees of severity of disease and undernutrition were defined as absent, mild, moderate or severe from data sets in a selected number of randomized controlled trials (RCTs) and converted to a numeric score. After completion, the screening system was validated against all published RCTs known to us of nutritional support vs spontaneous intake to investigate whether the screening system could distinguish between trials with a positive outcome and trials with no effect on outcome. Methods: The total number of randomized trials identified was 128. In each trial, the group of patients was classified with respect to nutritional status and severity of disease, and it was determined whether the effect of nutritional intervention on clinical outcome was positive or absent. Results: Among 75 studies of patients classified as being nutritionally at-risk, 43 showed a positive effect of nutritional support on clinical outcome. Among 53 studies of patients not considered to be nutritionally at-risk, 14 showed a positive effect ( P=0.0006). This corresponded to a likelihood ratio (true positive/false positive) of 1.7 (95% CI: 2.3–1.2). For 71 studies of parenteral nutrition, the likelihood ratio was 1.4 (1.9–1.0), and for 56 studies of enteral or oral nutrition the likelihood ratio was 2.9 (5.9–1.4). Conclusion: The screening system appears to be able to distinguish between trials with a positive effect vs no effect, and it can therefore probably also identify patients who are likely to benefit from nutritional support.