Kyle, Kossovsky et al, 2006 in a prospective cohort study (quality rating positive), studied 995 patients admitted to acute care to test the sensitivity and specificity of the following screening tools: Nutritional Risk Index (NRI), Malnutrition Universal Screening Tool (MUST) and Nutritional Risk Screening 2002 tool (NRS-2002). The researchers compared the nutritional assessment tools, using the Subjective Global Assessment (SGA) as the "gold standard." They evaluated the association between nutritional risk determined by these screening or assessment tools and LOS. The mean age of those with LOS one to 10 days was 50±21.9 years; the mean age of those with LOS greater than 11 days was 65.4±18.7 years; the mean age of those with LOS unknown was 44±17.0 years. The age difference between the groups was statistically significant. Proportion of males to females was not statistically significant between the groups.
NOTE: The NRI is described below, but was not used in the conclusion statement, because it did not meet work group criteria for a quick and easy screening tool.
All patients were assessed at hospital admission with the four tools (NRI, MUST, NRS-2002, SGA). Sensitivity, specificity, and predictive values were calculated and compared with the SGA.
The researchers found that the sensitivities of the NRI, MUST, and NRS-2002 were 43%, 61% and 62%, respectively. Specificities of the NRI, MUST and NRS-2002 were 89%, 76% and 93%, respectively. NRS-2002 had greater positive and negative predictive values (85% and 79%) than the MUST (65% and 76%) or the NRI (76% and 66%). All four instruments found a statistically significant association between nutritional status or nutritional risk and LOS. Agreement was moderate between SGA and NRS-2002 (kappa 0.48, P<0.001), fair between SGA and MUST (kappa 0.26, P<0.001) and fair between SGA and NRI (kappa 0.24, P<0.001).
The researchers stated that NRS-2002 had a higher sensitivity and specificity than MUST or NRI, and that there was a significant association between LOS and nutritional risk or nutritional status as determined by SGA, NRS-2002, MUST and NRI. The authors concluded that NRS-2002, MUST or SGA can be used to determine nutrition status at hospital admission.
The study was limited by use of the SGA, since 1) the SGA was not designed to detect mild malnutrition, and 2) sensitivity and specificity of the SGA may be improved by training examiners in nutrition assessment. The study was also limited by the use of LOS as an outcome parameter, since many factors other than nutrition may influence LOS.
Stratton, Hackston et al, 2004 in a cross-sectional study (quality rating positive), studied five consecutive cohorts of patients (Group 1: 50 gastroenterology clinic outpatients; Group 2: 75 medical and surgical patients; Group 3: 171 older adult medical and surgical patients; Group 4: 50 medical patients less than 65 years of age; Group 5: 52 elective and emergency surgical patients) to compare the prevalence of malnutrition risk assessed by the MUST and a variety of other published tools in both hospital outpatients and inpatients. In addition, the study investigated the concurrent validity of MUST with other published tools, assessed whether the same patients are identified as malnourished and compared the ease of use of MUST with other published tools.
MUST was compared with other nutrition screening or assessment tools as follows: Study 1: MEREC Bulletin tool, Hickson and Hill tool (HH tool); Study 2: Nutrition Risk Score (NRS), Malnutrition Screening Tool (MST); Study 3: Short-form Mini Nutritional Assessment-Short Form (MNA-SF tool); Study 4: Subjective Global Assessment (SGA): Study 5: Undernutrition Risk Score (URS).
NOTE: The MEREC Bulletin tool, HH tool, NRS, MST, MNA-SF and URS reference standards are described below, but were not used in the conclusion statement because they did not meet work group criteria for an acceptable reference standard.
The researchers found that the prevalence of malnutrition risk using MUST was 19 to 60%, while prevalence of malnutrition risk using the other tools ranged from 19 to 65%. MUST had 'excellent agreement' with the MEREC Bulletin tool and 'good-fair' agreement with the HH tool (kappa 0.825, kappa 0.647, respectively) in gastroenterology outpatients. MUST had 'good to excellent' agreement with the NRS and the MST (kappa 0.775, kappa 0.707, respectively) in hospital inpatients. 'Fair-good' agreement was found between MUST and MNA-SF in elderly medical and surgical patients (kappa 0.551, kappa 0.605, respectively); however, a significant bias was found when disagreements between MUST and MNA-SF were examined. There was 'excellent' agreement between MUST and SGA in hospital medical inpatients less than 65 years of age (kappa 0.783). Agreement between MUST and URS was 'poor' in general surgical inpatients (kappa 0.255). Examiners reported that the MUST took approximately three to five minutes to administer.
Sensitivity and specificity of MUST with general surgical patients were 73%* and 91%,* respectively. Sensitivity and specificity of MUST with elderly medical patients were 97%* and 60%,* respectively. The Mini Nutritional Assessment-Short Form was used as a reference standard.
*Values were calculated (see Definitions and Criteria) by the Nutrition Screening workgroup using data included in the article.
The researchers concluded that MUST is quick and easy to use and has 'fair-good' to 'excellent' concurrent validity when compared with most other nutrition screening tools tested.
The researchers mentioned that it is difficult to establish the validity of any malnutrition screening tool due to the lack of a universally accepted definition of malnutrition.