Punpanich, Dollapas

AbstractBackground and AimThis study aims to determine the performance of models adding C‐reactive protein (CRP) and procalcitonin (PCT) to the model of end‐stage liver disease (MELD) score for mortality prediction in patients hospitalized with complications of cirrhosis.MethodsA prospective cohort study was carried out in consecutive cirrhotic patients admitted with complications of cirrhosis between September 2012 and December 2013 at Phramongkutklao Hospital, Bangkok, Thailand. All patients had venous CRP, PCT, and laboratory values for MELD score calculation measured at emergency room or admission. Cox regression analysis and the c‐statistic were used to predict mortality. The MELD‐CRP score was externally validated in 818 eligible patients from Mayo Clinic, Rochester, using data from 1288 cirrhotic patients diagnosed between 2010 and 2014.ResultsA cohort of 177 patients with cirrhosis was admitted during the study period. Seventy‐one patients were eligible for analysis. The MELD score was predictive of 90‐day mortality odds ratio (OR) 1.19 (95% confidence interval [CI] 1.09–1.32). Adding CRP and/or PCT to the MELD score improved the predictive of 90‐day mortality: MELD‐CRP OR 2.71 (95% CI 1.66–4.99); MELD‐PCT OR 2.72 (95% CI 1.66–4.99); MELD‐CRP‐PCT OR 2.71 (95% CI 1.67–4.92). The c‐statistics for MELD, MELD‐CRP, MELD‐PCT, and MELD‐CRP‐PCT were 0.81, 0.83, 0.84, and 0.85, respectively. Adding CRP and/or PCT to the MELD score also improved 30‐day mortality prediction. Similar results for the MELD‐CRP score were obtained from the Mayo Clinic external validation cohort.ConclusionThe MELD‐CRP, MELD‐PCT, and MELD‐CRP‐PCT scores may be superior to the MELD score alone in predicting mortality in patients hospitalized with complications of cirrhosis.