Determination of the most effective nutritional risk screening tool to predict clinical outcomes in intensive care unit patients

[摘要] ENGLISH ABSTRACT: Introduction: Malnutrition, as defined by the Malnutrition Universal Screening Tool (MUST), Mini-Nutritional Assessment-Short Form (MNA-SF), Subjective Global Assessment (SGA), Nutritional Risk Screening Tool-2002 (NRS-2002), Short Nutritional Assessment Questionnaire (SNAQ), Nutritional Risk Indicator (NRI) and Malnutrition Screening Tool (MST), has been associated with adverse outcomes in hospitalised patients. Therefore nutritional risk screening is recommended for all hospitalised patients to improve the recognition and treatment of malnutrition. However, little is known about the use of screening tools in an intensive care unit (ICU) setting. The aim of this study was to assess which of these screening tools could best predict clinical outcomes in ICU patients and to comment on their feasibility, in order to make suggestions on their applicability for this patient population.Methods: Over an eight month study period all patients (>18years) with a surgical ICU stay of >48 hours were included. Patients were screened within 48 hours of admission using each of the seven screening tools. Clinical outcomes (mortality, APACHE II score, length of stay (LOS), length of ventilation (LOV), complications, serum-albumin, white cell count (WCC) and C-reactive protein) were recorded until discharge or death. Feasibility and applicability of the screening tools were also assessed.Results: A total of 206 patients (62.6% males) were included. The average age was 49.5 ±17.4 years and average LOS was 5.7 ± 5.5 days. Screening was not feasible in 18.3% of patients. The MUST classified 18.9% of patients as at risk of malnutrition and 30.1% as malnourished, but was not predictive of any clinical outcomes. According to the MNA-SF, 52.2% of patients were at risk of malnutrition and 16.5% were malnourished. This was associated with progressively decreasing serum-albumin levels (p<0.01) and WCC (p=0.01). The SGA classified 30.6% of patients as moderately and 18.4% of patients as severely malnourished and was significantly associated with LOS (p=0.03), LOV (p=0.01), mild complications (p=0.04) and serum-albumin (p=0.01). However, except for serum-albumin which progressively declined with a poorer nutritional status, the moderately malnourished patients showed the worst outcomes and the severely malnourished patients the best. According to the NRS-2002, 72.8% of patients were malnourished; and this correlated significantly with LOV (p=0.02) and the development of moderate (p=0.04) and total (p=0.01) complications. A non-significant but consistent trend for worse results in the malnourished group was also seen for the other outcomes studied. The SNAQ classified 35.9% of patients as malnourished or at risk thereof. This was associated with lower serum-albumin levels (p=0.04), but also with decreased LOV (p<0.01). The NRI classified 2.3% of patients as mildly malnourished, 21.0% as moderately malnourished and 75.0% as severely malnourished and only effectively predicted serum-albumin (p<0.01). The MST classified 78.2% of patients as malnourished and this was predictive of developing more complications (p<0.01). Almost all of the other variables also showed worse outcomes for the malnourished group, but this was not significant.Conclusion: Screening in an ICU seems to have only moderate feasibility and applicability and limited value. Only the NRS-2002 and MST showed potential for predicting clinical outcomes in ICU patients.

[发布机构] Stellenbosch University