Hierarchical Model to Predict Length of Stay and ICU Outcome According to Post-Operative Respiratory Failure after Liver Transplantation

    loading  Checking for direct PDF access through Ovid

Abstract

Length of stay in ICU and factors predictive of ICU outcome after liver transplantation have not been fully investigated.

Post-operative respiratory failure (PRF), defined as failure of weaning within 48 hours after surgery or respiratory failure within the further 48 hours after planned extubation, is a serious pulmonary complication after Liver transplantation (LTx).

PRF risk was assessed by combining two classification approaches to risk factors: systematic classification (recipient-related preoperative factors; intraoperative factors; logistic factors; donor factors; post-operative ICU factors; post-operative surgical factors) and an innovative patient/organ classification (general factors related to the patient; native liver factors; new liver factors; kidney factors; heart factors; brain factors; lung factors). The end-points were: to identify PRF predictors; to assess length of stay (LoS) in ICU and 90-day survival according to PRF.

Two-hundred adult chronic patients were included. Missing analysis and multicollinearity tests were performed. For each category, the best predictive variable was identified by ROC curves.The competitive role of each factor was hierarchically defined (4 levels: at the bottom, at the univariate analysis, then at the ROC curve analysis, then at the collinearity exclusion, at the top, at the multivariate analysis. Among the initial 20 significant parameters (univariate analysis), 13 resulted relevant at ROC curves, 10 at collinearity exclusion and 7 at logistic regression analysis.

Recipient age (OR=1.046; p=0.012), female gender (OR=2.761; p=0.019), MELD (OR=1.089; p=0.000), restrictive lung disease (OR=2.609; p=0.020), intra-operative veno-venous bypass (OR=3.199; p=0.005), pre-extubation PaCO2 (OR=1.106; p=0.003) and MEAF (OR=1.375; p=0.000) resulted independent risk factors for PRF (p=0.792 at Hosmer-Lemeshow test). Patients with PRF had higher LoS (p<0.001) lower ICU survival and inferior day-90 survival (99.2% versus 86.5%, p<0.001) than others.

At the top of the hierarchich multifactorial model predictive of PRF there were the factors related to native liver, those related to new liver and those related to lung performance. Pre-transplant donor-dependent factors, kidney, heart and brain have no impact on early respiratory recovery and length of ICU staty of LTx-patients.

The methodology adopted provides to hepatologists and surgeons an adequate tool to estimate the prognosis of ESLD listed patients. Moreover, patients with high MELD, and/or restrictive respiratory pattern and/or portal thrombosis should be informed on their poor prognosis, eventually worsened by possible early allograft dysfunction. Likewise, our model may help intensivists to identify the optimal extubation time according to the patient risk. Understanding, PRF determinants can allow identify risk mitigation strategies. Finally, hospitals’ accountants and health managers could exploit this model to predict variable costs and better resource allocation according to the different risk categories.

Related Topics

    loading  Loading Related Articles