Prospective validation of five malnutrition screening and assessment instruments among medical inpatients: Secondary analysis of a randomized clinical trial

Background & aims
Screening for malnutrition upon hospital admission is the first crucial step for proper nutritional assessment and treatment. While several nutritional screening and assessment instruments exist, there is a lack of head-to-head validation of these instruments. We studied the ability of five different nutrition screening and assessment instruments to predict 1-year mortality and response to nutritional treatment in participants of the EFFORT randomized trial.
Methods
In this secondary analysis of a Swiss-wide multicenter, randomized clinical trial comparing individualized nutritional support with usual care nutrition in medical inpatients, we prospectively classified patients as low, intermediate, and high nutritional risk based on five nutritional screening and assessment instruments (NRS 2002, SGA, SNAQ, MNA and MUST).
Results
Overall mortality at 1-year in the 1866 included patients was 30.4%. There were significant correlations and a significant concordance between all instruments with r-values ranging from 0.23 to 0.55 and kappa values ranging from 0.10 to 0.36. While high nutritional risk was associated with higher mortality in all instruments, SGA and MNA showed the strongest association with adjusted odds ratios of 3.17 (95%CI, 2.18 to 4.61, p < 0.001) and 3.45 (95%CI, 2.28 to 5.22, p < 0.001). When comparing mortality in intervention group patients to control group patients stratified by severity of malnutrition, there was overall no clear trend towards more benefit in patients with more severe malnutrition, with NRS 2002 and SGA showing the most pronounced relationship between the severity of malnutrition and reduction in mortality as a response to nutritional support.
Conclusion
Among all five screening and assessment instruments, higher nutritional risk was associated with higher risk for mortality and adverse clinical outcome, but not with more or less treatment response from nutritional support with differences among scores. Adding more specific parameters to these instruments is important when using them to decide for or against nutritional support interventions in an individual patient.