J M Kim1, Y Y Jo2, S W Na1, S I Kim3, Y S Choi1, N O Kim1, J E Park1, S O Koh4. 1. Department of Anesthesiology and Pain Medicine, and Anesthesia and Pain Research Institute, Yonsei University College of Medicine, Seoul, Korea. 2. Department of Anesthesia and Pain Medicine, Gachon University Gil Hospital, Incheon, Korea. 3. Department of Surgery, Yonsei University College of Medicine, Seoul, Korea. 4. Department of Anesthesiology and Pain Medicine, and Anesthesia and Pain Research Institute, Yonsei University College of Medicine, Seoul, Korea. Electronic address: sokoh@yuhs.ac.
Abstract
BACKGROUND: Acute renal failure (ARF) after liver transplantation requiring continuous renal replacement therapy (CRRT) adversely affects patient survival. We suggested that postoperative renal failure can be predicted if a clinically simple nomogram can be developed, thus selecting potential risk factors for preventive strategy. METHODS: We retrospectively reviewed the medical records of 153 liver transplant recipients from January 2008 to December 2011 at Severance Hospital, Yonsei University Health System, in Seoul, Korea. There were 42 patients treated with CRRT (20 and 22 patients received transplants from living and deceased donors, respectively) and 115 were not. Univariate and stepwise logistic multivariate analyses were performed. A clinical nomogram to predict postoperative CRRT application was constructed and validated internally. RESULTS: Hepatic encephalopathy (HEP; odds ratio OR, 5.47), deceased donor liver donations (OR, 3.47), Model for End-Stage Liver Disease (MELD) score (OR, 1.09), intraoperative blood loss (L; OR, 1.16), and tumor (hepatocellular carcinoma) as the indication for liver transplantation (OR, 0.11) were identified as independent predictive factors for postoperative CRRT on multivariate analysis. A clinical prediction model constructed for calculating the probability of CRRT post-transplantation was 1.7000 × HEP + [-4.5427 + 1.2440 × (deceased donor) + 0.0830 × (MELD score) + 0.000149 × the amount of intraoperative bleeding (L) - 2.1785 × tumor]. The validation set discriminated well with an area under the curve (AUC) of 0.90 (95% confidence interval, 0.85-0.95). The predicted and the actual probabilities were calibrated with the clinical nomogram. CONCLUSIONS: We developed a predictive model of postoperative CRRT in liver transplantation patients. Perioperative strategies to modify these factors are needed.
BACKGROUND:Acute renal failure (ARF) after liver transplantation requiring continuous renal replacement therapy (CRRT) adversely affects patient survival. We suggested that postoperative renal failure can be predicted if a clinically simple nomogram can be developed, thus selecting potential risk factors for preventive strategy. METHODS: We retrospectively reviewed the medical records of 153 liver transplant recipients from January 2008 to December 2011 at Severance Hospital, Yonsei University Health System, in Seoul, Korea. There were 42 patients treated with CRRT (20 and 22 patients received transplants from living and deceased donors, respectively) and 115 were not. Univariate and stepwise logistic multivariate analyses were performed. A clinical nomogram to predict postoperative CRRT application was constructed and validated internally. RESULTS: Hepatic encephalopathy (HEP; odds ratio OR, 5.47), deceased donor liver donations (OR, 3.47), Model for End-Stage Liver Disease (MELD) score (OR, 1.09), intraoperative blood loss (L; OR, 1.16), and tumor (hepatocellular carcinoma) as the indication for liver transplantation (OR, 0.11) were identified as independent predictive factors for postoperative CRRT on multivariate analysis. A clinical prediction model constructed for calculating the probability of CRRT post-transplantation was 1.7000 × HEP + [-4.5427 + 1.2440 × (deceased donor) + 0.0830 × (MELD score) + 0.000149 × the amount of intraoperative bleeding (L) - 2.1785 × tumor]. The validation set discriminated well with an area under the curve (AUC) of 0.90 (95% confidence interval, 0.85-0.95). The predicted and the actual probabilities were calibrated with the clinical nomogram. CONCLUSIONS: We developed a predictive model of postoperative CRRT in liver transplantation patients. Perioperative strategies to modify these factors are needed.
Authors: Scott M Sutherland; Lakhmir S Chawla; Sandra L Kane-Gill; Raymond K Hsu; Andrew A Kramer; Stuart L Goldstein; John A Kellum; Claudio Ronco; Sean M Bagshaw Journal: Can J Kidney Health Dis Date: 2016-02-26