Introduction The incidence of malnutrition in patients with cirrhosis is high. However, it often goes undetected as many screening tools are based on measurement of body mass index (BMI), which is a poor nutritional marker in this population as patients tend to be centrally obese yet muscularly depleted. The gold standard for the assessment of malnutrition in this population is the Royal Free Hospital Global Assessment (RFH-GA). The Malnutrition Universal Screening Tool (MUST), which is based on BMI, is still used in some UK Liver transplant Units, although it may not be valid in this setting. Hence the aim of this study is to validate the MUST tool against the gold standard RFH-GA for use in patients with cirrhosis.

Methods Multicentre validation was undertaken in a cohort of 133 patients, (98 men: 35 women; age 56 [23–73] yr) with cirrhosis across five UK liver transplant units. Nutritional status was screened using the MUST tool and then categorisation of nutritional status was determined by using the RFH-GA. The analysis of descriptive data, cross-tabulation, performance variables, 95% CIs and κ values were calculated using standard methods. κ Values were interpreted according to Altman, 1999.

Results Eight-four (63%) of the 133 were categorised, using the RFH-GA, as being moderately or severely malnourished. In contrast the MUST tool identified only 45 (34%) patients as being at nutritional risk. Thus the sensitivity and specificity of MUST for determining nutritional risk were 34% (95% CI 20 to 51) and 94% (95% CI 86 to 97); respectively; the κ value was 0.19 demonstrating a poor level of agreement. The sensitivity and specificity of MUST improved when the patients with fluid retention were excluded from the analysis, 100% (95% CI 46 to 100) and 91% (74–98); respectively. The performance of the MUST also improved as the accurate dry body weight was better calculated but still did not reach 100% sensitivity indicating body weight alone is not a good marker of nutritional status in this patient population.

The performance of the MUST utilising alternative weight adjustments in patients with fluid retention

Conclusion The performance characteristics of the MUST tool in this setting are poor. This tool, can not be recommended for screening patients with chronic liver disease for nutritional risk.

Competing interests None declared.