Malnutrition Universal Screening Tool is Not Sufficiently Valid in Screening for Risk of Malnutrition in Patients with Newly Diagnosed Esophago-Gastric Cancer: a Validation Study Compared with Patient-Generated Subjective Global Assessment

Background Malnutrition is prevalent in patients with esophago-gastric cancer and is associated with poorer outcomes. Malnutrition screening tools are used to screen for risk of malnutrition and to determine whether full nutritional assessment is required. The optimal malnutrition screening tool to use in esophago-gastric cancer is unknown. Objective To assess the validity of the Malnutrition Universal Screening Tool (MUST) in esophago-gastric cancer by measurement of its sensitivity, specificity, positive and negative predictive values for detecting risk of malnutrition. Design A validation study comparing MUST with full malnutrition assessment using Patient-Generated Subjective Global Assessment (PG-SGA) as the reference standard. Participants /setting. 80 patients with newly diagnosed esophago-gastric cancer awaiting radical treatment at an oncology specialist hospital (Royal Marsden NHS Foundation Trust, London, United Kingdom) were recruited between December 2011 and June 2013. Main outcome measures Validity (accuracy) of MUST compared with PG-SGA as a reference method. Association of individual MUST features (BMI, percent weight loss) with PG-SGA score. Statistical analyses Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of MUST compared with PG-SGA. Spearman’s rank correlation (ρ) between BMI and percentage weight loss with PG-SGA score. Results MUST was found to have agreement with PG-SGA in 53/80 (66.3%) patients, misclassifying 19 (23.8%) patients as being at low risk of malnutrition when they were not and 8 (10%) at risk when they were not. MUST demonstrated a sensitivity of 61.2% (95% CI: 46.2-74.8%) and a specificity of 74.2% (95% CI: 55.4-88.1%). PPV for MUST was 78.9% (95% CI: 66.5-87.6%) and NPV was 54.8% (95% CI: 44.6-64.6%). Lower BMI was weakly associated with higher PG-SGA score (correlation -0.259, p=0.021) while greater percentage weight loss was moderately associated with higher PG-SGA score (correlation +0.653, p< 0.001). Conclusions MUST did not meet a priori criteria (sensitivity and specificity ≥70%) for validation against PG-SGA. MUST is unlikely to be an appropriate malnutrition screening tool in patients with newly diagnosed esophago-gastric cancer.