SIRS, Proton pump inhibitors (PPI) are widely used. There are concerns that overuse of PPI might come with adverse effects and may even severely harm patients. 1 In fact, we also found that PPI was commonly used by patients with cirrhosis, and a relevant proportion of patients took PPI for symptomatic reasons, rather than for acute gastrointestinal bleeding episodes. 2 We appreciate the questions raised by Lin and Dong.
3As mentioned by Lin and Dong patients taking PPI were sicker, indicated by higher MELD and Child Pugh scores, and are therefore considered to have a worse prognosis. However, there is little evidence whether PPI use might be an additional risk factor for overall mortality in cirrhotics in addition to well established prognostic parameters such as the MELD score or occurrence of complication of cirrhosis. To consider the disease stage, MELD score and Child Pugh score were included in the multivariate analysis and PPI-usage turned out to be an independent predictive factor for mortality.The lack of data concerning the relation of PPI use to overall survival was the rationale for conducting this study. We acknowledge the suggestions of Lin and Dong, and included additional parameters in the multivariate analysis. We added hepatorenal syndrome (HRS), spontaneous bacterial peritonitis (SBP) and the indication for PPI use (acute bleeding episode, reflux disease or current variceal band ligation) in the multivariate analysis. However, we did not have data concerning the duration of PPI treatment or about antibiotic regimens used in the patients.In the extended analysis SBP [hazard ratio (HR) 2.105, 95% confidence interval (CI), 1.107-4.002, P = 0.023], PPI use (HR 1.926, 95% CI 1.010-3.675, P = 0.047), hepatocellular carcinoma (HCC; HR 5.117, 95% CI 2.948-8.884, P < 0.001), MELD score (HR 1.111, 95% CI 1.069-1.154, P < 0.001), hepatic decompensation (HR 1.865, 95% CI 1.009-3.448, P = 0.047) and indication for PPI use (HR 1.867, 95% CI 1.166-2.990, P = 0.009) were independent prognostic parameters. Lin and Dong proposed to exclude patients with short observation times. We consider it inappropriate to exclude patients who fulfil inclusion criteria which may bias the results of studies. Furthermore, the relation of risk and observation time is valued in Cox survival models. Overall survival was the primary objective of the study. We do not have complete follow-up data concerning bleeding events in patients according to their PPI use at the day of study inclusion.There are two central results of our study which are evident and should be emphasised once more. First, PPI use was common among our cirrhotic patients, with many patients taking PPI without an evidence-based indication. Second, PPI use was associated with a higher risk of death, independent from severity of liver disease and hepatic complications, which is evident and confirmed by multivariate Cox regression analysis. However, although PPI use was a risk factor for death, HCC and MELD were stronger predictors of prognosis.Although Lin and Dong mi...