Validation of body mass index for the diagnosis of malnutrition in patients with liver cirrhosis

Objectives - Simple clinical tools are needed to detect malnutrition in cirrhotic patients. We have validated optimal body mass index (BMI) cut-offs for the diagnosis of malnutrition in accordance with the importance of ascites. Methods - BMI, mid-arm muscle circumference (MAMC) and triceps skinfold thickness (TST) were measured before paracentesis in study (SP) and validation (VP) populations of 875 and 294 cirrhotic patients, respectively with no ascite (NA), mild (MA), tense ascites (TA) (NA/MA/TA: SIP:327, 270, 278 VP: 111, 69, 114). Preserved nutritional status (SP: 259; VP: 93), malnutrition including severe and moderate malnutrition (SP: 251 and 365; VP: 92 and 109) were defined from MAMC and TST measurements. Results - Optimal BMI cut-off values were 22, 23 and 25 kg/m(2) in NA, MA and TA patients, respectively. In the whole SIP and VP, sensitivities of these cut-offs were 86.2% and 89.1%, respectively; the corresponding negative predictive values (NPV) for the diagnosis of severe malnutrition were 92.3% and 93.2%; specificities and positive predictive values (PPV) were 87.7% and 89.9%, 92.7% and 93.6%, respectively for the diagnosis of malnutrition. In the entire VP, peripheral oedema did not change sensitivity and NPV of BMI cut-offs for the diagnosis of severe malnutrition and tended to increase specificity and PPV for the diagnosis of malnutrition. 96.7% of the malnourished TA patients in the VP before paracentesis were correctly identified after removal of ascites. Conclusion - BMI is a reliable parameter to detect malnutrition in cirrhotic patients with the above mentioned cut-offs. Peripheral oedema and removal of ascites do not affect its diagnostic performance.