Application of NRS2002 to detect malnutrition in a tertiary care centre: agreement between nurse and dietician's screening steps and relation with clinical outcomes.

Objective: To evaluate the agreement between nurse and dietician nutritional risk assessments when using the Nutritional Risk Screening 2002 (NRS2002) protocol, and to explore the relations of falsely labeling patients 'not at risk' for malnutrition and the screening time difference (STD) between nurse and dietician with the length of stay (LoS).

Methods: Included are all patients hospitalized in a tertiary care center between January 2017 and December 2019 and screened for malnutrition by both a nurse and a dietician. The inter-rater reliability is evaluated using Cohen's Kappa. The relation between STD and the patient classification (PCET) is assessed by a linear mixed effect model. The relation between the LoS and PCET is evaluated with the Kaplan-Meier method and multivariable Cox regression including STD with pathology group and severity of illness as random effect.

Results: 9085 patients are assessed by nurse and dietician. 72% of all assessments agree (Kappa = 0.44 [0.43-0.46]). The dietician is involved later for patients falsely labeled 'not at risk' (1.06 [0.92-1.20] days; p < 0.001). Compared to patients where the dietician is involved within 3 days, the LoS is 7.37 days (Hazard Ratio (HR): 0.51 [0.43-0.61]) longer for patients falsely labeled 'not at risk', while only 3.51 days (HR: 0.72 [0.64-0.80]) longer for patients correctly labeled 'at risk'.

Conclusions: Agreement of screening for malnutrition between nurses and dieticians is weak. Avoiding falsely labeling patients 'not at risk' should be a main concern upon patient admission as later involvement of dieticians is correlated with a longer LoS.