Abstract
OBJECTIVE To improve prediction of recurrent bleeding after endoscopic sclerotherapy of esophageal varices. MATERIAL AND METHODS A prospective, observational, case-control study was performed. Immediate and long-term results of endoscopic sclerotherapy of esophageal varices were studied in 91 patients for the period from 2002 to 2016. Multiple regression analysis with binary response model was applied to analyze the prediction models. RESULTS Recurrent bleeding occurred in 80.5 (20; 182) days after sclerotherapy (range 0-2557 days). Spearman's correlation analysis revealed a significant relationship between bleeding recurrence and erythrocyte count (R= -0.32), Child-Pugh class of liver cirrhosis (R=0.49), Child-Pugh score (5-15) (R=0.54), content of amino acids, HPro/Pro ratio (R=0.71). Prognostic indicators were selected by stepwise inclusion of predictors. Thus, the final version of regression equation is as follows: Y=exp (-0.17+0.93×Child-Pugh score-106.42×HPro/Pro)/[1+exp(-0.17+0.93×Child-Pugh score-106.42×HPro/Pro)]. High risk of recurrent bleeding from esophageal varices within 1 year after endoscopic sclerotherapy is determined by Y-value >0.5. An accuracy of this model is 89.6%, Se 94.3%, Sp 79.2%, PPV 90.9%, NPV 86.4%, OR 63.3, LR + 4.53, LR - 0.07. CONCLUSION Thus, the proposed method is highly informative, effective, available and can be widely used in clinical practice.
