Clarithromycin-resistant H. pylori: add bismuth to amoxicillin-metronidazole

Design

Multicentre randomised controlled trial, three arms

Population

Adults with clarithromycin-resistant H. pylori, southern Korea

Primary outcome

Eradication rate (ITT and per protocol)

Effect

ITT: PAM 68.2%, PAM-B 84.8%, PBMT 81.8%; nausea 6.8% vs 25.0% (PAM-B vs PBMT)

A multicentre randomised trial in Gut and Liver (September 2026) enrolled patients in southern Korea with clarithromycin-resistant Helicobacter pylori. It compared 14 days of PPI, amoxicillin and metronidazole (PAM); the same with bismuth (PAM-B); and classic bismuth quadruple therapy with PPI, bismuth, metronidazole and tetracycline (PBMT).

By intention to treat, eradication was 68.2% with PAM, 84.8% with PAM-B (P = 0.024 vs PAM) and 81.8% with PBMT. Per protocol it was 75.4%, 96.5% and 94.6%. PAM-B and PBMT did not differ. Nausea and vomiting were far less frequent with PAM-B than PBMT (6.8% vs 25.0%).

The trial was modest in size, conducted in one region of Korea, and registered retrospectively; metronidazole resistance patterns will differ elsewhere. But the result is plausible — bismuth is thought to help overcome metronidazole resistance — and the regimen removes tetracycline, which is often hard to obtain.

Where clarithromycin resistance is known or likely, amoxicillin-based bismuth quadruple therapy is a better-tolerated alternative to classic quadruple therapy.

• Avoid clarithromycin-based triple therapy where clarithromycin resistance is known or common
• Use a 14-day regimen of PPI, bismuth, amoxicillin and metronidazole when tetracycline is unavailable or poorly tolerated
• Do not use PPI-amoxicillin-metronidazole without bismuth — eradication was only 68%
• Warn patients that bismuth turns stools black
• Confirm eradication with a breath or stool test afterwards

Why it matters

It offers a rescue regimen that avoids tetracycline, the component of bismuth quadruple therapy most often unavailable or poorly tolerated.

Don't overread it

One region, a modest sample and retrospective registration; confirm against local metronidazole resistance.

The statistics, in plain English

Intention-to-treat counts everyone randomised, including those who stopped early, so 84.8% is the realistic cure rate; per-protocol (96.5%) counts only those who completed treatment. PAM-B and PBMT were not statistically different (P = 0.64), which with this sample size means similar, not proven equivalent.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.