A comparative study of malnutrition screening tools in advanced liver disease: Sensitivity, specificity, and patient acceptability

Background: Patients with advanced liver disease (ALD) are at increased risk of malnutrition, significantly impacting patient outcomes and quality of life. Screening for malnutrition in ALD can be challenging with changes in weight potentially concealed by fluid from ascites and oedema, or muscle wasting. Current validated malnutrition screening tools used in ALD account for physiological symptoms, but there are no known validated tools that also screen for non-physiological symptoms or fluid overload. Aim: To examine the sensitivity and specificity of a novel malnutrition screening tool for adults with ALD, (Child-Pugh score B or C) with or without ascites attending a regional and metropolitan health site, against three established malnutrition screening tools compared with a Subjective Global Assessment (SGA) as the reference standard. Methods: A cross-sectional study was undertaken within the outpatient and inpatient hepatology services of a metropolitan and a regional hospital. Each participant completed four malnutrition screening tools, the Patient Generated Subjective Global Assessment Short Form (PG-SGA SF), Royal Free Hospital Nutrition Prioritizing Tool (RFH-NPT), Malnutrition Screening Tool (MST), and a newly developed novel screening tool (Nutrition-ALD) as well as a diagnostic SGA with the dietitian. Performance of each tool was compared by calculating sensitivity, specificity, and area under the curve (AUC). Wellbeing scores measured by Integrated Palliative care Outcome scale (IPOS) were correlated with the malnutrition screening tool scores. Results: 50 participants with Child-Pugh B and C liver disease [n = 33 (66 %) outpatient setting; n = 17 (34 %) inpatient setting] were recruited for this study. 82 % of participants were diagnosed with mild to moderate or severe malnutrition using SGA. Relative to the SGA, the PG-SGA SF had the highest AUROC, followed by the novel screening tool. PG-SGA SF demonstrated the highest sensitivity (0.88) and specificity (0.89). Participant-reported experience was most favourable for the novel screening tool, followed by PG-SGA SF. Conclusion: A nutrition screening tool should be quick, sensitive, and specific to detect malnutrition risk to allow for early nutrition intervention. In this study population, the PG-SGA SF had the highest sensitivity and specificity, and the novel tool was most acceptable to the participants. Consideration of wellbeing with a holistic and supportive approach to malnutrition screening may improve accuracy and timeliness of referrals for patients requiring nutrition support and improve efficiency in low-resource settings.