How effective are 3-month and 12-month predictive models of mortality after first liver transplantation?

    loading  Checking for direct PDF access through Ovid

Abstract

BACKGROUND

Postoperative mortality after liver transplantation (OLT) is influenced by a number of factors. Accurate predictive models of mortality that are based on multicenter data, and that predict mortality at different times postsurgery are lacking.

OBJECTIVES

To assess mortality and develop models to predict 3-month and 12-month mortality after OLT.

DESIGN AND INTERVENTION

This study analyzed data from the European Liver Transplant Registry. Exclusion criteria included recipients <15 years old, combined organ transplants, and a lack of data. Eligible data were randomly divided into 2 datasets: a training set (∼70% of data), which was used to identify factors influencing 3-month and 12-month postoperative mortality and to generate mortality models, and a validation set (∼30% of data), which was used to assess the adequacy of fit of the models. The following factors were considered: recipient and donor age and sex, disease cause, year of transplantation, United Network for Organ Sharing score, recipient HCV antibody status, hepatitis B surface antigen status, bypass type, graft type, total ischemia time, and size of transplant center. Donor age was categorized into ≤40 years, 41-60 years, and >60 years. Transplant center size was classified according to number of transplants performed per year: small (<37), medium (37-69), and large (≥70). A mortality score was calculated for each individual.

OUTCOME MEASURES

The main outcome measures were mortality and the identification of factors influencing mortality.

RESULTS

In total, the data of 34,664 individuals were analyzed. The 3-month and 12-month mortality groups contained 21,605 individuals (2,450 [12%] of whom had died by 3 months) and 18,852 individuals (3,391 [18%] of whom had died by 12 months), respectively. Increased risk of postoperative mortality at 3 months was associated with donor age >60 years (odds ratio [OR] 1.16; 95% CI 0.99-1.16), older recipient age (OR 1.12 per 5 years; 95% CI 1.10-1.14), total ischemia time >13 h (OR 1.38; 95% CI 1.21-1.57), acute liver failure (OR 1.61; 95% CI 1.34-1.93), compatible (OR 1.22; 95% CI 1.05-1.42) or incompatible donor-recipient blood group (OR 2.07; 95% CI 1.47-2.91), split or reduced graft (OR 1.96; 95% CI 1.61-2.35), and low United Network for Organ Sharing score (score 1: OR 2.43; 95% CI 2.07-2.85; score 2: OR 1.67; 95% CI 1.43-1.94). A better outcome at 3 months was associated with a large center size, donor age ≤40 years, and the presence of hepatitis B, hepatitis C or primary biliary cirrhosis, alcoholic cirrhosis, and cirrhosis with hepatocellular carcinoma. Similar factors were associated with 12-month mortality as were associated with 3-month mortality. When assessed for adequacy of fit with the validation population, the 3-month and 12-month mortality model showed good discrimination between those who died and those who survived (c statistic 0.688 and 0.677, respectively). The 12-month mortality model did, however, underestimate the risk of mortality in this group of patients.

CONCLUSION

The 3-month and 12-month mortality models can effectively predict post-transplant survival in patients undergoing first OLT.

Related Topics

    loading  Loading Related Articles