- Geetha Iyer, graduate student,
- G Caleb Alexander, associate professor
- Correspondence to: G C Alexander
Concerns about the cardiovascular risks of macrolide antibiotics surfaced in the 1980s, with case reports and clinical studies describing arrhythmias and QT prolongation with erythromycin.1 2 These concerns subsequently extended to azithromycin3 and clarithromycin,4 two other members of the macrolide class. For example, a large cohort study using Danish healthcare data showed, on average, a 76% higher risk of cardiovascular mortality with current use of clarithromycin compared with penicillin V among middle aged patients, translating into an adjusted absolute risk difference of 37 cardiac deaths per million courses.4 The authors acknowledged concerns about confounding by indication and residual confounding. However, in a randomised controlled trial without such problems, a two week course of clarithromycin increased the risk of cardiovascular death for three years, compared with placebo, among patients with stable coronary artery disease.5
The study by Wong and colleagues in this issue6 further characterises the association between clarithromycin and cardiovascular events in Hong Kong’s national healthcare database. Unlike most previous studies focused on cardiovascular death, Wong and colleagues chose the diagnosis of incident myocardial infarction as their primary outcome, and arrhythmia, stroke, all cause mortality, and cardiovascular mortality as secondary outcomes. To examine the associations of interest, the authors used three separate study designs—retrospective cohort analysis (with amoxicillin as control), self controlled case series, and case crossover analysis. Each analysis identified an increased short term risk of all cardiovascular events, except stroke, associated with the current use of clarithromycin; no long term risk was identified. The adjusted risk difference for current use of clarithromycin compared with amoxicillin was 1.9 extra myocardial infarctions (95% confidence interval 1.3 to 2.68), 0.95 extra cardiac deaths (0.51 to 1.51), and 0.2 extra arrhythmias (0.04 to 0.49) for every 1000 patients treated.
The cohort analysis, comparing clarithromycin with amoxicillin, included over 300 000 users. Notably, some baseline characteristics of the treatment groups differed, even after adjustment for propensity score, indicating the potential for channeling bias. This was further suggested by the finding that clarithromycin was also associated with an increased risk of non-cardiac mortality (suggesting that adults given clarithromycin were sicker than controls at baseline). Analyses of indications for antimicrobial use, if available, might have been valuable to help evaluate the extent of potential confounding by indication.
The self controlled case series7 design controls for variability between people as only those with both the exposure and the outcome are included. The authors chose clarithromycin treatments for a chronic infection—Helicobacter pylori—as their exposure since acute infections can increase the risk of myocardial infarction. Clarithromycin was associated with myocardial infarction and arrhythmia but not stroke in both self controlled case series and case crossover analyses. These results add credence to the associations found in cohort analysis, despite reflecting use of combined H pylori treatments (clarithromycin, amoxicillin or metronidazole, and a proton pump inhibitor), rather than clarithromycin alone.
The strength and consistency of the associations described across different study designs are persuasive and consistent with most of the literature reporting the cardiovascular risks associated with this class of antibiotic. How should these findings inform clinical practice? Absolute risks and measures such as number needed to harm (NNH) are useful complements to relative risks or hazard ratios when considering the clinical relevance of these statistically significant associations. Although the NNH was not calculated in the current study, Schembri and colleagues analysed two prospective cohorts of older adults admitted to hospital and estimated that for acute coronary syndrome, arrhythmia, or new or worsening heart failure, the NNH for clarithromycin ranged from 8 (among those with acute exacerbations of chronic obstructive pulmonary disease) to 11 (among those with community acquired pneumonia).8 While these values reflect the use of clarithromycin among a relatively narrow and high risk subpopulation, they nevertheless suggest that the potential risks are not negligible.
As patients and clinicians try to make sense of the evidence and incorporate this information into decision making, a few fundamental concepts of treatments are worth keeping in mind. All drugs have risks, most drugs have serious risks, and given that as many as a quarter to half of all antibiotics9 10 prescribed are not clinically indicated, the opportunities to use these products more judiciously without compromising quality of care are manifold. Clinicians and patients should consider these potentially serious adverse events when prescribing macrolides, especially to those at highest baseline risk.
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare: CA is chair of the Food and Drug Administration’s Peripheral and Central Nervous System Advisory Committee, serves as a paid consultant to IMS Health, and serves on an IMS Health scientific advisory board.
Provenance and peer review: Commissioned; not externally peer reviewed.