Objectives: Recently, new prognostic models (Model for End-Stage Liver Disease [MELD model] and Emory score) were proposed for the prediction of survival in transjugular intrahepatic portosystemic shunt (TIPS) patients. Although the MELD model is considered to be superior and has consecutively been applied to priority listing for liver transplantation, these models have never been directly compared in terms of long-term prognosis. We therefore compared the prognostic accuracy of the different models, including the Child-Pugh score, in an unselected cohort of TIPS patients followed long-term.
Methods: Baseline risk scores for 162 unselected consecutive TIPS patients followed until death (n = 81), liver transplantation, or end of observation (n = 81) (mean follow-up 30.7 +/- 26.4 months) were calculated, and respective concordance- (c-)statistics for the predictive accuracy of 3-, 12-, and 36-month survival for the three models were compared statistically.
Results: All three models predicted short-term (3-month) survival with similar accuracy. The MELD model generated the best c-statistics for both 12-month (c-statistic 0.73, 95% CI = 0.64-0.82) and 36-month survival (c-statistic 0.74, 95% CI = 0.64-0.84). The predictive accuracy of the Emory score was significantly lower (c-statistic for 12-month survival: 0.60, 95% CI = 0.52-0.68, p = 0.012 vs MELD). In the statistical comparison of the MELD and the Child-Pugh model, only a trend favoring MELD for the prediction of 1-yr survival in patients with intestinal bleeding could be observed (MELD: c-statistic 0.78, 95% CI = 0.67-0.89; Child-Pugh: c-statistic 0.67, 95% CI = 0.55-0.80, p = 0.059).
Conclusions: The MELD model is superior to the Emory score but only slightly superior to the Child-Pugh classification for the prediction of long-term survival in TIPS patients.