CAJ | 학술논문

目的：观察 HBV相关慢加急性肝功能衰竭（ACLF）住院患者基线、第1周及第2周时的终末期肝病模型（MELD）、MELD‐Na和MELDNa评分对短期（3个月）病死率的预测价值。方法收集2014年1月至2015年1月在四川省德阳市人民医院住院的 HBV相关ACLF患者53例，根据3个月的预后情况分为死亡组和生存组，收集两组患者的血液生物化学指标及并发症情况。两组均数比较采用 t检验，计数资料采用χ2检验，各评分模型预测短期病死率的准确性采用受试者工作特征曲线下面积（AUC）表示。结果53例患者中，死亡21例，生存32例。MELD、MELD‐Na及MELDNa评分在基线时预测患者3个月病死率的AUC分别为0．548（95％CI：0．406～0．685，P＝0．555）、0．502（95％CI：0．362～0．643，P＝0．978）和0．523（95％CI：0．381～0．662，P＝0．778）。在第1周和第2周时，其AUC逐渐增加；第2周时的AUC分别为0．881（95％CI：0．762～0．953，P＜0．01）、0．878（95％CI：0．759～0．951，P＜0．01）和0．897（95％CI：0．783～0．963，P＜0．01），均明显高于第1周及基线时相应的值（均P＜0．05）。当阈值分别为26、29和29时，MELD、MELD‐Na及MELDNa评分模型的敏感度和特异度、阳性预测值及准确度均达到最佳，亦高于基线和第1周时相应的值。结论入院第2周时MELD、MELD‐Na和MELDNa评分对HBV相关ACLF患者短期病死率的预测价值优于基线及第1周，而在基线时评分的预测价值可能不佳。
목적：관찰 HBV상관만가급성간공능쇠갈（ACLF）주원환자기선、제1주급제2주시적종말기간병모형（MELD）、MELD‐Na화MELDNa평분대단기（3개월）병사솔적예측개치。방법수집2014년1월지2015년1월재사천성덕양시인민의원주원적 HBV상관ACLF환자53례，근거3개월적예후정황분위사망조화생존조，수집량조환자적혈액생물화학지표급병발증정황。량조균수비교채용 t검험，계수자료채용χ2검험，각평분모형예측단기병사솔적준학성채용수시자공작특정곡선하면적（AUC）표시。결과53례환자중，사망21례，생존32례。MELD、MELD‐Na급MELDNa평분재기선시예측환자3개월병사솔적AUC분별위0．548（95％CI：0．406～0．685，P＝0．555）、0．502（95％CI：0．362～0．643，P＝0．978）화0．523（95％CI：0．381～0．662，P＝0．778）。재제1주화제2주시，기AUC축점증가；제2주시적AUC분별위0．881（95％CI：0．762～0．953，P＜0．01）、0．878（95％CI：0．759～0．951，P＜0．01）화0．897（95％CI：0．783～0．963，P＜0．01），균명현고우제1주급기선시상응적치（균P＜0．05）。당역치분별위26、29화29시，MELD、MELD‐Na급MELDNa평분모형적민감도화특이도、양성예측치급준학도균체도최가，역고우기선화제1주시상응적치。결론입원제2주시MELD、MELD‐Na화MELDNa평분대HBV상관ACLF환자단기병사솔적예측개치우우기선급제1주，이재기선시평분적예측개치가능불가。
Objective To investigate the predictive value of the baseline ,week 1 and week 2 model for end‐stage liver disease (MELD) scores ,MELD‐Na scores and MELDNa scores for 3‐month mortality in hepatitis B virus (HBV) related acute‐on‐chronic liver failure (ACLF) patients .Methods Fifty‐three eligible HBV‐related ACLF patients who received initial treatment in People′s Hospital of Deyang City , Sichuan Province from January 2014 to January 2015 were enrolled in this retrospective clinical follow‐up study .Patients were divided into death group and survival group according to the prognosis after 3 months of treatment .Biochemical data and complications were collected .t test was used to compare the means between two groups and χ2 test was used to compare categorical data .Accuracy of scoring models in predicting mortality within 3 months was performed by area under receiver operating characteristic curve (AUC) .Results Of the fifty‐three patients ,twenty‐one HBV‐related ACLF patients died in 3‐month , while thirty‐two patients survived .The AUC of MELD scores ,MELD‐Na scores and MELDNa scores at the baseline in predicting 3‐month mortality were 0 .548 (95% CI:0 .406 -0 .685 , P= 0 .555) ,0 .502 (95% CI:0 .362-0 .643 ,P=0 .978) and 0 .523 (95% CI:0 .381-0 .662 ,P=0 .778) ,respectively .The AUC increased gradually at the first and the second week .At the second week ,the AUC of MELD scores ,MELD‐Na scores and MELDNa scores were 0 .881 (95% CI:0 .762 -0 .953 ,P<0 .01) ,0 .878 (95% CI:0 .759-0 .951 ,P<0 .01) and 0 .897 (95% CI:0 .783 -0 .963 ,P<0 .01) ,respectively .The AUC of the second week was higher than those of the baseline and the first week (both P<0 .05) .The sensitivity ,specificity ,positive predictive value (PPV) ,and accuracy at the second week were best when the cut off was 26 for MELD ,29 for MELD‐Na and 29 for MELDNa , and were higher than the corresponding values of the baseline and the first week .Conclusions The predictive values of MELD scores ,MELD‐Na scores and MELDNa scores at the second week are superior to the baseline and the first week for 3‐month mortality in HBV‐related ACLF patients ,while the predictive values at the baseline could be poor .