Risks of failed Helicobacter pylori eradication and upper gastrointestinal bleeding : a population-based cohort study in Hong Kong

Abstract

Currently, there is a rising trend in the failure rates of clarithromycin-containing triple therapy for Helicobacter pylori (H. pylori). H pylori is a significant contributor to gastric cancer and peptic ulcer disease, and peptic ulcer disease is the leading cause of upper gastrointestinal bleeding (UGIB). Furthermore, the use of anti-coagulants could also lead to the development of UGIB.

This thesis comprised of four studies that utilized the electronic healthcare database in Hong Kong to investigate the risks of failed eradication treatment for H. pylori, as well as the risk of subsequent UGIB after H. pylori eradication. The detailed contents of these studies were as follows: 1. The trend of failure rates of clarithromycin-containing triple therapy for H. pylori during the period of 2003-2017 was evaluated. Treatment failure was defined as the requirement for retreatment within 2 years of initial eradication. Results showed the failure rates of clarithromycin-containing triple therapy increased over the 15-year period, and younger patients were found to have higher retreatment rates as compared to older patients. 2. The role of 11 available machine learning models in predicting the failure of clarithromycin-containing triple therapy for H. pylori were assessed. Hong Kong patients with triple therapy in 2003-2013 were used as training set, whereas Hong Kong patients with eradication therapy in 2014-2017 were used as internal cohort, and United Kingdom (UK) patients with similar eradication therapy in 2012-2017 were used as external cohort. In both internal and external validation set, Extra-Tree (ET) Classifier model showed the best performance of area under receiver operating characteristic curve (AUC) of 0.88 and 0.85, respectively. 3. The long-term UGIB risk in the H. pylori-eradicated patients and matched H. pylori-negative patients were investigated. Findings indicated H. pylori-eradicated patients had 1.6-times higher risk of UGIB as compared to H. pylori-negative patients. This increased risk of UGIB in H. pylori-eradicated patients was particularly evident after 2 years of follow up and in population older than 45 years. 4. Risks of UGIB in patients with H. pylori eradication and newly started on warfarin or direct oral anti-coagulants (DOACs) were explored. Findings revealed new DOACs users had a significantly lower risk of UGIB as compared with new warfarin users among H. pylori-eradicated cohort. The risk of UGIB in new warfarin or DOACs users was similar between H. pylori-eradicated and H. pylori-negative cohorts.

The findings in this thesis discovered an increasing trend in failure rates of clarithromycin-containing triple therapy between 2003 and 2016 in Hong Kong, particularly among younger patients. Machine learning algorithms could be valuable in early identification of high-risk patients for failure of triple therapy. The protective effect of H. pylori eradication for UGIB is limited to younger individuals and the first two years following eradication. DOACs may be a preferable choice to warfarin after H. pylori eradication in patients who required anti-coagulation in terms of UGIB risk. These findings could help to identify patients at high-risk of eradication failure and prevent risk of subsequent UGIB.