Abstract
Background
Proton pump inhibitors (PPIs) are commonly prescribed medications which are indicated in various different gastrointestinal (GI) diseases, including peptic ulcer disease, gastroesophageal reflux disorder and upper GI bleeding. There is some evidence to suggest that PPI use in cirrhosis may predispose to the development of hepatic encephalopathy (HE) and spontaneous bacterial peritonitis (SBP), albeit with some controversy.
Aims
We aim to conduct a retrospective epidemiological analysis of the association between PPI use in hospitalized patients with cirrhosis and prevalence of HE and SBP.
Methods
This was a retrospective cohort study of 953 adult patients (mean age 62.3 (SD=12.2)) with cirrhosis admitted to the Ottawa Hospital between January 1, 2011 and December 31, 2015. A chart review was conducted and relevant information extracted.
Results
Average MELD-Na on admission was 17.4 (SD = 7.5) with no significant differences when stratified by in-hospital PPI use (p=0.53). 14.7% of patients had a prior history of HE, 5.4% SBP, 31.6% ascites, 9.3% hepatocellular carcinoma and 1.2% hepatorenal syndrome. 26.4% of patients had a history of varices, of which 34.4% had previous variceal bleeding. 45.4% of patients were on a PPI prior to admission and 69.8% during their hospitalization. Patients with a previous history of HE (16.8% vs. 9.6% p=0.006), varices (31.4% vs. 14.2% p<0.001) and variceal bleed (11.6% vs. 3.1% p<0.001) were more likely to be exposed to a PPI in-hospital. Mortality rate during index admission was 19.4%.
There was no significant association noted between the incidence of HE (31.2% vs. 25.0% p=0.06) or SBP (10.4% vs. 8.0% p=0.25) and in-hospital PPI use. There was also no significant association between PPI use and infectious complications, including bacteremia (8.6% vs. 7.6% p=0.63), pneumonia (12.7% vs. 12.5% p=0.95), urinary tract infections (9.2% vs. 6.6% p=0.19) and clostridium difficile (3.8% vs. 2.1% p=0.18). Patients on a PPI had greater in-hospital mortality (22.0% vs. 13. 5% p=0.002), although there was no significance between cause of death (p=0.31) nor death from infectious complications (18.5% vs. 23.1% p=0.52) between groups. Length of stay (LOS) was longer in patients exposed to PPI (median (IQR) 7 (4–17) vs 6 (3–13) p=0.03). Similar findings were noted on subgroup analysis of decompensated patients.
Conclusions
We did not observe a significant difference in HE, SBP or infectious complications among this cohort of cirrhotic patients by in-hospital PPI use. However, there was a significantly higher mortality rate noted in hospital and longer LOS, despite similar baseline MELD-Na and causes of death. Further study and judicious PPI prescribing practices in this vulnerable population of patients is warranted.
Funding Agencies
None
Fatal spontaneous bacterial peritonitis and necrotizing fasciitis with bacteraemia caused by Bacillus cereus in a patient with cirrhosis
