Decision Analysis Of Treatment For Patients With Esophageal Varices Bleeding

Objective To investigate the relationship between patient's condition and theresults after treatment in patients with esophageal varices bleeding(EVB) todevelop the decision tree, then combine patient's condition and their willing togain the best treatment method.Methods we reviewed the patients who had experienced esophageal varicesbleeding from 1998 to 2003 ,the clinical data of 93 patients receivingdevascularization with splenectomy ,65 patients receiving endoscopic varicealligation(EVL) . Selected the variable including case hisyory,clinical expressand laboratory examination for the study ,with X~2-test to develop decision treewhich based on the probability .Applying decision tree analysis to calculate theexpected utility of decision alternatives ,and then make sensitivity analysis.Results (1)After devascularization with splenectomy,the parameter of decision treeto judge effectiveness or ineffectiveness were prothrombin time (PT),total oflosing,albumin(ALB),ascites ,degree of spleen swelled,portal vein diameter,billrubin(BIL) .There were eight ends of the decision tree ,two probalities thatineffectiveness could be judged were 85.7%(12/14),83.3%(5/6) respectively,effectiveness were 90.9%(10/11),100%(34/34) respectively,the other four endswere ones that patients couldn't be judged .In confidence interval ,the right rateof the decision tree to judge ineffectiveness was 85.0%(17/20), effectivenesswas 97.8%(44/45), probalities couldn't be judged were 62.5%(5/8),70%(14/20),total right rate was 86.0%(80/93). (2)After endoscopic variceal band ligation, the parameter of decision tree tojudge effectiveness or ineffectiveness were degree of spleen swelled,portal veindiameter, liver function Child-Pugh's classification and total of losing. Therewere five ends of the decision tree,two probalities that ineffectiveness could bejudged were 81.3%(13/16) , 80%(4/5) respectively, effectiveness were96.3%(26/27), the other two ends were ones that patients couldn't be judged .Inconfidence interval ,the right rate of the decision tree to judge ineffectivenesswas 81.0%(17/21), effectiveness was 96.3%(26/27), probalities couldn't bejudged was 64.7%(11/17), total right rate was 83.1%(54/65). (3)Having Applied decision tree analysis for the patients that the resultscouldn't be judged,we knew that if the patients would have receiveddevascularization with splenectomy,the expected utility was 0.824,meanwhile,received endoscopic variceal band ligation was 0.778.Obviously,devascularization with splenectomy was reasonable method,it could gain higherquality of life. (4)After having made sensitivity analysis ,the results was when the deathrate was 12.5%,or rebleeding rate after endoscopic variceal ligation was 23%,we can select any way . (5) Child-Pugh's C patients had a significantly higher rebleeding rate ordeath rate compared with Child-Pugh's B and A respectively(P<0.05);emergency operation had a significantly higher death rate compared withprepared operation(P<0.05).Conclusions (1)Either receiving devascularization with splenectomy or receivingendoscopic variceal ligation ,the results was related with the liver function ,suchas prothrombin time (PT),albumin(ALB),ascites ,billrubin(BIL),degree ofspleen swelled, portal vein diameter, total of losing. (2)The long-term follow-up is required after EVL so that EVL wasengaged in again for the esophageal varices to lower rebleeding rate. (3) Child-Pugh 's C patients or emergency operation should be avoided asfar as possible.