Long-term Risk of Upper Gastrointestinal Bleeding After Helicobacter pylori Eradication

A Population-based Cohort Study

Fang Jiang; Chuan-Guo Guo; Ka Shing Cheung; Wai K. Leung


Aliment Pharmacol Ther. 2021;54(9):1162-1169. 

In This Article


In this large real-world cohort study, we found that the long-term risk of UGIB in patients after H. pylori eradication was still 1.6-times higher than H. pylori-negative patients. Notably, we showed that the risk of UGIB after H. pylori eradication remained comparable to H. pylori-negative patients during the initial 2-year but then significantly increased in subsequent years. Moreover, younger (18–45 years) patients appeared to benefit more from H. pylori eradication in the prevention of subsequent UGIB than older (>45 years) patients. The results remained consistent in various analyses including using peptic ulcer bleeding as outcome; in a subgroup of patients with or without prior history of peptic ulcer or UGIB; and treating concurrent medications use as time-varying covariates, which further testify the robustness of our study findings.

Thus far, few studies look into the long-term effects of H. pylori eradication on subsequent risk of UGIB. The current population-based study included more than 125 000 PS matched H. pylori eradicated and H. pylori-negative patients with median follow up of 7–8 years, which could better reflect the real-world situation after H. pylori eradication. Although meta-analysis[13] of controlled clinical trials showed that eradication of H. pylori could prevent recurrent bleeding from peptic ulcers as compared with anti-secretory non-eradicating therapy, most studies[20–27] focused on patients with peptic ulcer bleeding with a relatively small sample size and short follow-up period (usually <2 years). There were also a wide reported range of rebleeding rates that ranged from 0% to 10%.[20–27]

With the longer follow up, our study showed that there was a significant increase in risk of UGIB after the first 2-year in the H. pylori eradicated group. Within the first 2-year after eradication, the risk of UGIB, including peptic ulcer bleeding, was comparable between H. pylori eradicated and H. pylori-negative patients. However, the risk of UGIB significantly increased after the initial 2-year. The results were still consistent even after adjusting for concomitant medication use during the follow-up period by time-varying analysis, suggesting that these findings could not be accounted by immortal time bias with higher consumption of medications that increase the bleeding risk in the H. pylori eradicated group (Table 2). Arguably, some patients could have reinfection of H. pylori after 2-year and resulting in UGIB. However, as it is standard clinical practice to test for and eradicate H. pylori if present in Hong Kong, particularly in patients with UGIB, these patients with any additional eradication therapies after the index clarithromycin-containing regime should have been identified and excluded.

In our previous study,[19] we have shown that patients who failed eradication therapy, as inferred by the need of retreatment for H. pylori, had a 1.5-fold increase in the risk of UGIB than those with successful eradication. In this study, we further found that when compared to patients who were not infected with H. pylori, post-eradicated patients were still at higher risk of UGIB. Together, these findings support that H. pylori eradication could reduce, but not completely eliminate, the risk of UGIB, particularly during longer term follow-up.

Another important observation of this study is the age-stratified analyses of UGIB risk after H. pylori eradication. When compared to patients older than 45 years, there was no significant increase in UGIB risk in younger (18–45 years) patients after H. pylori eradication. Although it is generally recommended to eradicate H. pylori once identified, there is so far no consensus on the optimal age of H. pylori eradication. For gastric cancer prevention, our previous study showed that even older (>60 years) patients could benefit from H. pylori eradication.[14] In the present study, it is however shown that H. pylori eradication in younger (<45 years) patients appeared to benefit most in preventing subsequent UGIB. The possible explanation for this may be due to the fact that younger patients usually have less severe gastric mucosal inflammation from chronic H. pylori infection, which are potentially reversible after H. pylori eradication. Moreover, older age and the presence of comorbidities would increase the needs of medications such as NSAIDs, aspirin, anti-coagulants, SSRIs and bisphosphonate that elevate the risk of UGIB. These findings may again support the benefits of early H. pylori eradication in young patients to prevent future complications including gastric cancer and UGIB. Continuous gastroprotection by PPIs or H2RAs may still be needed in older patients even after H. pylori eradication to prevent subsequent UGIB.

The strength of our study was the inclusion of a large population cohort of H. pylori-eradicated and a matched H. pylori-negative cohort with long-term follow-up data, which can best reflect the real-world outcome of H. pylori eradication in diverse age groups. The comprehensive public healthcare database allows the retrieval of all hospitalisation and prescription records for detailed analysis. The use of PS matching analysis and Cox model, including univariable and multivariable Cox regression model, were performed to adjust for potential confounders. Additionally, medication uses were treated as time-varying covariates in the multivariable model to minimise immortal time bias. Use of PPI and H2RA in the last 6 weeks prior to UGIB were also excluded to reduce indication bias in this study.

This study has some limitations. First, the outcome of H pylori eradication could not be directly identified from the electronic database, and was only inferred from patients who needed retreatment for H. pylori. All patients who required any retreatment after primary eradication were excluded from this study and the overall retreatment rate was 11.3%, which was consistent with prior prospective studies.[28] The exclusion of these patients who required retreatment of H. pylori could substantially reduce the possibility of overestimating the UGIB risk by including patients who failed previous eradication or had recurrent H. pylori infection and developed UGIB, as these patients, if tested positive for H. pylori again would be given another course of eradication therapy. Due to the anonymous nature of the CDARS database, we had validated the post-eradication H. pylori status of patients from our hospital. In this validation analysis, we found that only 3.9% of these patients were found to have H. pylori infection after H. pylori eradication.[16] Second, the overall UGIB rate was 1.62% among our H. pylori eradicated patients, which was also consistent with previous meta-analysis results.[13] Third, although the CDARS captured all prescription records, over the counter use of medications could not be retrieved. However, nearly all medications that would modify the risk of GIB including PPIs, aspirin, NSAIDs and SSRI required doctor's prescription in Hong Kong and could not be obtained over the counter, which limit their impacts on the analysis of GIB risk in this cohort. Fourth, all non-variceal UGIB was used as the primary endpoint in our study, but sensitivity analysis using peptic ulcer bleeding as the endpoint yielded consistent results. Fifth, patients enrolled in the last 1 or 2 year of the study period might have lost the chance to receive the further eradication therapies and would be included in the successful eradication group. Since our patients were included from those who received eradication therapies from 2003 to 2012 and followed up to 2016, the median follow up period after eradication was 7.8 years and most patients would have a minimum of 4 years follow up. Sixth, factors such as smoking or excessive alcohol intake were not available in the electronic database and could not be included in the PS matching. Lastly, all H. pylori-negative patients included in this study had undergone endoscopy and found to have no H. pylori. As most of these patients had indications for endoscopy including UGIB, dyspepsia and anaemia, the bleeding risk of these H. pylori-negative patients could be higher than the average population. This was confirmed in the sensitivity analysis after excluding patients with prior history of UGIB or peptic ulcer, which showed even higher HRs than the primary analysis (Table S2).