A nutritional risk screening model for patients with liver cirrhosis established using discriminant analysis

Objective To establish a nutritional risk screening model for patients with liver cirrhosis using discriminant analysis. Methods The clinical data of 273 patients with liver cirrhosis who were admitted to Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine from August 2015 to March 2016 were collected. Body height, body weight, upper arm circumference, triceps skinfold thickness, subscapular skinfold thickness, and hand grip strength were measured and recorded, and then body mass index (BMI) and upper arm muscle circumference were calculated. Laboratory markers including liver function parameters, renal function parameters, and vitamins were measured. The patients were asked to complete Nutritional Risk Screening 2002 and Malnutrition Universal Screening Tool (MUST) , and a self-developed nutritional risk screening pathway was used for nutritional risk classification. Observation scales of the four diagnostic methods in traditional Chinese medicine were used to collect patients' symptoms and signs. Continuous data were expressed as mean ± SD (x ± s) ; an analysis of variance was used for comparison between multiple groups, and the least significant difference t-test was used for further comparison between two groups. Discriminant analysis was used for model establishment, and cross validation was used for model verification. Results The nutritional risk screening pathway for patients with liver cirrhosis was used for the screening of respondents, and there were 49 patients (17. 95%) in non-risk group, 49 (17. 95%) in possible-risk group, and 175 (64. 10%) in risk group. The distance criterion function was used to establish the nutritional risk screening model for patients with liver cirrhosis: D1 =-11. 885 + 0. 310 × BMI + 0. 150 ×MAC + 0. 005 × P-Alb-0. 001 × Vit B12 + 0. 103 × Vit D-0. 89 × ascites-0. 404 × weakness-0. 560 × hypochondriac pain + 0. 035 ×dysphoria with feverish sensation (note: if a patient has ascites, weakness, hypochondriac pain, or dysphoria with feverish sensation, this index equals 1, and otherwise it equals 0) . Cross validation showed that the probability of misclassification of the this model was 10%.Conclusion The established screening model may provide a reference for clinical nutritional risk screening in patients with liver cirrhosis.