Proton pump inhibitors in cirrhosis: a retrospective five-year analysis of increased risks of hepatic decompensation and infections

Proton pump inhibitors (PPIs) are among the most commonly prescribed medications worldwide, valued for their efficacy in managing symptoms from acid-related diseases. They are particularly effective in controlling conditions such as gastro-oesophageal reflux disease, peptic ulcer disease and for the treatment of Helicobacter pylori infections when used in combination with antibiotics.¹

In patients with chronic liver disease, PPIs are often prescribed, albeit frequently without a definitive indication. A review conducted in 2017 suggests that up to 60% of chronic liver disease patients receiving PPIs do so without clear clinical justification.¹ The commonly approved indications for PPI use in this demographic include peptic ulcer disease, variceal bleeding and reflux disease.^1,2

Emerging evidence has highlighted potential adverse outcomes associated with PPI use in patients with cirrhosis, a severe form of chronic liver disease. Notably, studies have indicated an elevated risk of hepatic encephalopathy and spontaneous bacterial peritonitis in this patient group.^3,4 These findings suggest that while PPIs are effective for their intended uses, their application in cirrhotic patients may necessitate cautious consideration due to these potential risks.

Additionally, there is a broader concern among clinicians regarding the association of PPI use with an increased risk of all-cause infections.² This is particularly relevant in the context of chronic liver disease, where patients are already at heightened risk of infections due to compromised liver function and other comorbidities.

Despite the global insights into the risks of PPI use in chronic liver disease, there is a notable paucity of data specific to New Zealand. This lack of localised data is significant given the high prevalence of chronic liver disease within the New Zealand population. Understanding the implications of PPI use in this setting is critical for optimising patient outcomes and guiding clinical practice.

This study aims to fill this gap by analysing the outcomes associated with PPI use in patients with cirrhosis over a 5-year period at a major tertiary centre in New Zealand. By examining mortality rates, the incidence of liver-related events and infection rates, this research seeks to provide valuable insights into the safe and effective use of PPIs in the management of chronic liver disease within the New Zealand healthcare context.

Methods

This retrospective observational study analysed data from patients presenting with chronic liver disease at Middlemore Hospital in Auckland, either as inpatients or outpatients, in the year 2014. Ethical approval was obtained from the Health and Disability Ethics Committee.

Patients diagnosed with liver cirrhosis were identified using the International Classification of Diseases 10th Revision (ICD-10) coding for a diagnosis of “cirrhosis” or “chronic liver disease” from electronic patient records for the year 2014. The cohort was divided into two groups: those who received PPI therapy and those who did not. Both groups were followed over a 5-year period to monitor primary and secondary outcomes. The primary outcome was mortality, while secondary outcomes included incidences of hospitalisations due to hepatic decompensation (such as hepatic encephalopathy, ascites, variceal bleeds) and all-cause infections (including spontaneous bacterial peritonitis).

Inclusion criteria required patients to have a clear ICD-defined diagnosis of cirrhosis. Exclusion criteria were not specifically outlined; however, only patients meeting the diagnostic criteria were included. Data already collected as part of patient management at Middlemore Hospital were utilised.

Baseline data extracted for both groups included demographics (age, gender, ethnicity), causes of cirrhosis, comorbidities (including ischemic heart disease, heart failure, renal disease, chronic obstructive pulmonary disease, asthma, hepatocellular carcinoma [HCC] and non-HCC malignancy), relevant blood test results (liver function tests, full blood count, renal function) and concurrent medications. The severity of liver disease in both groups was assessed using the Child–Pugh and MELD scores. This was used in analysis to correct for liver disease severity. The Charlson Comorbidity Index was employed to evaluate the systemic health of patients with cirrhosis. Indications for PPIs prescribed were also collected.

Patients in the PPI group had their PPI therapy classified by dose, duration and type of PPI. In New Zealand, only three types of PPIs are available (omeprazole, pantoprazole and lansoprazole). Medication exposure was quantified using the defined daily dose (DDD) recommended by the World Health Organization, which represents the assumed average maintenance dose per day for a drug used for its main indication in adults. For the purpose of this study, the maintenance dose was defined as omeprazole 20mg. Consequently, 20mg of omeprazole taken once daily for 3 months equated to a cumulative defined daily dose (cDDD) of 3 months. Potency equivalence among the three PPIs was standardised as omeprazole 20mg = pantoprazole 40mg = lansoprazole 30mg.

Primary outcomes recorded for this study were mortality in the PPI group vs non-PPI group. This was split into all-cause mortality and liver-related mortality (as identified by the primary cause of death recorded on patient file pertaining to their chronic liver disease).

Secondary outcomes recorded included incidence of liver-related events such as spontaneous bacterial peritonitis, hepatic encephalopathy, variceal bleeding and ascites that required hospital admission in the PPI group vs non-PPI group.

Furthermore, primary and secondary outcomes were analysed in the PPI group with respect to cDDD and whether increased cumulative exposure to PPIs increased risk of mortality of liver-related events.

Statistical analysis

Demographic and clinical characteristics were presented as counts and proportions for categorical variables and as means with standard deviations or medians with interquartile ranges for continuous variables. For comparisons between groups, analysis of variance (ANOVA) or Kruskal–Wallis tests were utilised for continuous variables, and Chi-squared or Fisher’s exact tests were employed for categorical variables, as deemed appropriate. Logistic regression analyses were conducted to calculate the odds ratio (OR) with a 95% confidence interval (CI) of the association between PPI use and the outcomes. The model was adjusted for demographic characteristics, causes of liver disease, comorbidities, liver function tests and medications to determine if the difference in PPI use and any risk factors remained significantly associated with the outcomes. Model selection was performed using forward and backward selection methods, retaining those risk factors with a p-value of less than 0.15 in the model. A two-tailed p-value of less than 0.05 was considered statistically significant.

Association between PPI dose and mortality

A multivariable logistic regression model was employed to assess the relationship between cumulative PPI dose (cDDD) and liver-related mortality. The dependent variable was mortality (binary: 1 = death, 0 = survival), while the independent variables included cDDD and covariates such as age, Charlson Comorbidity Index, aetiology of cirrhosis, cirrhosis severity, antiviral use and beta-blocker use. A constant term was added to the model, and ORs with 95% CIs were calculated for each covariate. Statistical significance was determined using a p-value threshold of <0.05.

Association between PPI dose and liver-related events

To evaluate the association between cDDD and liver-related events (ascites, hepatic encephalopathy, spontaneous bacterial peritonitis, variceal bleeding and other infections) Poisson regression models were used. Each event was treated as a count outcome (e.g., number of occurrences per patient). The predictor variables included cDDD, age, Charlson Comorbidity Index, cirrhosis severity, aetiology of cirrhosis, antiviral use and beta-blocker use. A log link function was used to model the relationships, and the coefficients were exponentiated to produce incidence rate ratios. The statistical significance of the association between cDDD and each outcome was evaluated using a p-value threshold of <0.05.

All statistical analyses were carried out using the Statistical Analysis Software version 9.4 (SAS Institute, Cary, North Carolina).

Results

Overall results

A total of 392 patients diagnosed with cirrhosis were included in this study. Of these, 304 patients (78%) were administered PPIs, while 88 patients (22%) were not. The mean duration of PPI usage was approximately 71 months (standard deviation 88), encompassing use for up to a year prior to the commencement of the study.

Within the PPI group, only 32% (n=60; peptic ulcer disease n=23; variceal bleeding n=8; gastro-oesophageal reflux disease n=29) of patients were prescribed PPIs for an evidence-based indication. Seventy percent of patients (n=216) were prescribed PPI therapy for unspecified indications. The remaining patients were prescribed PPI therapy for non-evidence-based indications, including non-ulcer-related gastrointestinal bleeding, dyspepsia and oesophagitis unrelated to reflux disease (Figure 1).

View Figure 1–5, Table 1–2.

Baseline characteristics

The average age of patients across both cohorts was 59 years and was comparable between the groups. At baseline, patients in the PPI group had more severe liver disease as seen by the liver function markers, MELD score (p=0.000) and Child–Pugh score (p=0.010) for the patients in this group.

Hepatitis B was the most common cause of cirrhosis in both groups, followed by alcoholic liver disease. There was a marginally higher prevalence of hepatitis B in patients not exposed to PPIs (p=0.007), while alcoholic liver disease was more prevalent in those exposed to PPIs (p=0.011).

More patients in the PPI group were also on a non-selective beta blocker agent (p=0.034) compared to the non-PPI group. This is likely to reflect the more severe nature of liver disease in these patients with complications such as portal hypertension and variceal bleeding.

Regarding comorbidities, the overall Charlson Comorbidity Index score indicated a higher comorbidity burden in the PPI group (p=0.002) compared to the non-PPI group (Table 1).

Outcomes

The primary outcome assessed was mortality. Ninety-one patients (30%) in the PPI group died during the 5-year follow-up period, compared with 17 patients (19%) in the non-PPI group. After adjusting for confounding factors significantly different at baseline and factors significant on univariate analysis (age, aetiology of cirrhosis, Charlson Comorbidity Index, medications), the final multivariate adjusted ORs indicated no significant difference in mortality between the two groups (OR 0.85, p=0.668) (Figure 2).

The primary outcome was further analysed looking specifically at liver-related deaths. Of the patients who had died during observation period, 32 deaths (35.2%) were attributed to liver-related causes in the PPI group and three deaths (17.6%) were attributed to liver-related causes in the non-PPI group. After adjusting again for confounding variables, the final OR and p-value indicated no statistical difference in liver-related mortality between both groups (OR 1.79; p=0.365) (Figure 3).

The secondary outcome evaluated was the number of hospitalisations due to hepatic decompensation and all-cause infections over the 5-year follow-up period. There were 94 (31%) hospital admissions in the PPI group and nine (10%) admissions in the non-PPI group. The multivariate adjusted analysis revealed a significant difference in hospital admissions between the groups (OR 2.95, p=0.046). Furthermore, the incidence of infections was statistically significant between the groups, with 94 (30%) hospital admissions for infections in the PPI group and 10 (11%) admissions in the non-PPI group (OR 2.42, p=0.025) (Figure 4, Figure 5).

Outcomes with cDDD

Primary and secondary outcomes were further analysed in the PPI group with respect to cumulative dose exposure to PPIs using the cDDD.

When mortality in the PPI group was analysed using logistic regression, with other factors corrected for, there was a small statistically significant decrease in mortality in the PPI group (OR 0.0044, p=0.037).

When liver-related events were analysed using Poisson regression in the PPI group with other factors corrected for, it was found that cumulative PPI use is significantly associated with increased risk of all liver-related events and other infections (Table 2).

Discussion

In this study conducted in a tertiary centre, we observed a significant association between the use of PPIs and the incidence of infections in patients with liver cirrhosis. Specifically, the literature supports a link between PPI use and spontaneous bacterial peritonitis, which may be attributed to bacterial dysbiosis caused by alterations in the gut microbiome.^5–8 PPIs disrupt the normal gut flora by altering gastric pH, leading to bacterial overgrowth and changes in the microbiome composition.^5–8 These changes can compromise the intestinal barrier, facilitating bacterial translocation into the bloodstream.^5–8 This mechanism might explain the higher incidence of infections, such as spontaneous bacterial peritonitis, and an increased risk of hepatic encephalopathy observed in PPI users with cirrhosis.^8,9 It is worth noting that our study demonstrates an increased risk of infections and liver-related events with cumulative increase in PPI dose, further supporting this hypothesis. One study carried out in Singapore also looked at cDDD of PPI in cirrhosis and found increased events of hepatic decompensation.¹⁰ Our study is in accordance with these findings.

Our study does not reflect a significant increase in mortality associated with PPI use, and in fact our study suggests a slight decrease in mortality associated with increasing cumulative dose of PPIs. Previous studies have suggested that PPI use may be associated with increased mortality in patients with cirrhosis. Dultz et al.¹¹ reported PPI use to be an independent predictor of mortality in cirrhosis. The study carried out in Singapore also showed an increase in mortality with increased cumulative dose exposure to PPI.¹⁰ However, this was within an observation period of 1 year. Another retrospective study performed in Scotland showed no difference in mortality between PPI users and non-PPI users.¹² A further retrospective study carried out in Taiwan looked specifically at mortality in patients admitted with hepatic encephalopathy on PPIs and reported an increased short-term and long-term mortality risk.¹³ It is difficult to make confident conclusions on the association between PPI use and mortality in chronic liver disease patients. Some studies have demonstrated an increased risk while others have not. A vast majority of these studies are retrospective in nature and as such correcting for confounding factors will be challenging to establish clear association. This was explored by China et al.¹⁴ in a post hoc analysis of a large multi-centre randomised trial (the ATTIRE trial) of patients with cirrhosis. The outcomes of this analysis were that there was no significant increase in infections, renal disease or mortality in patients taking PPI therapy, but the study acknowledged the large real-world difference in baseline characteristics between cirrhosis patients that contributed significantly to confounding factors in analysis, and this should be considered in any further recommendations made with regards to PPI use in these patients.¹⁴ Given these conflicting findings in the literature, this topic remains debated and it is difficult to say with certainty whether PPIs increase mortality in patients with cirrhosis. This will ideally need to be further explored with prospective studies in future.

Our study stands out due to its reasonable sample size of 394 patients with liver cirrhosis and an extensive follow-up period of 5 years. This allows for a robust assessment of the relationship between PPI use and clinical outcomes in liver disease. Both groups were comparable at baseline in terms of demographics such as age, ethnicity and aetiology of cirrhosis, thereby minimising selection bias. However, it is noteworthy that the severity of liver disease was significantly greater in the PPI group compared with the non-PPI group, which could have contributed to the poorer morbidity outcomes observed in that cohort.

Additionally, our study revealed a high prevalence of hepatitis B among the participants, which contrasts with other studies carried out internationally.^{1–4,10–15} This discrepancy highlights the unique demographic and clinical characteristics of our study population, emphasising the need for region-specific research to understand the diverse impacts of PPI use in different populations.

Despite these strengths, our study is not without limitations. As a retrospective cohort study, it inherently possesses certain constraints that may impact the validity and generalisability of its findings. Retrospective designs are particularly susceptible to confounding variables, and although efforts were made to adjust for known confounders, it is acknowledged that not all confounding factors can be accounted for in such a design. In this study, while we were able to establish exposure to PPIs using cDDD, we acknowledge that we were unable to accurately explore the effect of the different PPIs available in New Zealand on outcomes individually. Additionally, retrospective studies rely on the accuracy and completeness of existing records. In our study, we did not have access to community/general practitioner (GP) records for patients and as such were unable to accurately ascertain the indication for initiating PPIs and indication for dose changes that occurred in the community. We also were unable to capture any decompensation events that may have been managed in the community not requiring hospital admission. Furthermore, the retrospective nature of the study limits our ability to establish causality. While associations between PPI use and adverse outcomes can be identified, definitive clinical conclusions regarding the causative effects of PPIs cannot be drawn from this type of study. Prospective, randomised controlled trials are necessary to further elucidate these relationships and provide more conclusive evidence on the impact of PPI therapy in patients with liver cirrhosis.

PPIs are commonly prescribed for cirrhotic patients to manage complications such as gastro-oesophageal reflux disease and peptic ulcers, and to prevent gastrointestinal bleeding. However, in our study, a significant number of patients were prescribed PPIs for unspecified indications. This raises questions about the appropriateness of PPI use in this population and suggests a potential over-prescription issue. It is important to note that the majority of PPIs were initiated in the community by the patient’s GP rather than in a hospital setting. This finding underscores the need for education and awareness among all practitioners and careful evaluation of PPI indications in cirrhotic patients to avoid unnecessary exposure to potential adverse effects.

Moreover, while we stated that patients who were prescribed PPIs had more severe liver disease, this aspect was not explored in depth. The greater severity of liver disease in the PPI group could be a confounding factor contributing to the higher incidence of adverse outcomes observed. Patients with advanced liver disease are more susceptible to infections and other complications, which might explain the increased risks associated with PPI use in this cohort. Future studies should aim to stratify patients based on the severity of liver disease to better understand the impact of PPIs in different sub-groups of cirrhotic patients.

Conclusions

In conclusion, our study showed that over 5 years in a large tertiary centre, PPI use in cirrhosis was associated with a significantly higher risk of hepatic decompensation and an increased risk of infection without a significant increase in mortality.

Given that only 32% of the PPI prescriptions met the recommended indications, there is a strong case for more judicious use of PPIs in patients with cirrhosis. Emphasising the indication, duration of therapy and close monitoring for adverse effects are crucial steps. Further studies should focus on establishing clearer guidelines and investigating the pathological processes by which PPIs influence disease progression and infection risks in liver cirrhosis. These studies should also aim to delineate the safe use of PPIs in this vulnerable patient population.