Large-scale Helicobacter pylori eradication aspirin trial (HEAT): results of a real-world outcomes trial in primary care

Problem

Helicobacter pylori may be pivotal in the pathogenesis of upper gastrointestinal (UGI) ulcer bleeding from aspirin therapy. The Helicobacter Eradication Aspirin Trial (HEAT) investigated whether H.pylori eradication reduces ulcer bleeding

Approach

HEAT was conducted using novel real-world methodology developed by the Simple Trials for Academic Research (STAR) group. Participants aged over 60 and taking aspirin (≤325mg daily) for at least 4 months were recruited from primary care between 2012 and 2017. H. pylori positive participants were randomised to one week of eradication treatment (twice daily lansoprazole 30mg, clarithromycin 500mg and metronidazole 400mg) or matching placebos. Recruitment was managed using a bespoke web-based database that communicated directly with a programmed search tool downloaded at participating practices that identified all suitable patients and automatically sent trial information and an invitation to participate. The primary endpoint was hospitalisation due to definite or probable peptic ulcer bleeding. A random 10% sample had end of trial breath testing. Electronic follow up including Hospital Episodes Statistics (HES) and GP databases identified potential endpoints. All episodes mentioning GI bleeding or peptic ulcer were evaluated by an independent blinded adjudication committee. The primary endpoint was analysed using a Cox proportional hazards model. Data are still blinded and presented as Group A and B.

Findings

In total, 1,208 GP practices across the UK sent 188,875 invitation letters; 30,166 patients were consented to the trial, of whom 5,353 H. pylori positive participants (17.8%) were randomised. Mean age at randomisation was 73.6 (SD 7.0) and 73.8% of participants were male. In the retest sample 90.7% in Group A tested H. pylori negative vs 24.3% in Group B. During nearly 30,000 person-years of follow-up 607 episodes of clinically significant GI bleeding were identified and 45 were adjudicated as definite or probable UGI ulcer bleeds after randomisation. The rate of presentation varied significantly by time. Events occurred more frequently in the first 2- years of follow up in group B. There was a significant difference between the groups in the early period which was not evident later.

Consequences

Full analysis and unblinded results will be available for the conference. STAR methodology shows that large-scale real world trials of issues of clinical interest can be conducted in primary care.