Azithromycin – Not Guilty of Murder

The FDA has announced it is reviewing the safety of azithromycin in lieu of a recent NEJM article documenting an association between azithromycin and cardiovascular death.  In theory, azithromycin has been implicated in QT-prolongation and pro-arrhythmic effects, leading to torsades de pointes and polymorphic ventricular tachycardia.  The authors of this study therefore hypothesized an association between azithromycin use and cardiovascular death.

This is a retrospective study of computerized data generated from the Tennessee Medicaid program between 1992 and 2006, linking deaths to any concurrent antibiotic prescriptions.  The authors data-mined for a cohort aged 30 to 74 years of age, had no “life threatening non-cardiovascular illness”, did not abuse drugs, and did not reside in a nursing home.  They compared azithromycin prescriptions to non-prescription controls, as well as amoxicillin, ciprofloxacin, and levofloxacin cohorts.  And, after a little statistical maneuvering, they report a death rate of 85.2 per 1,000,000 courses of antibiotics with azithromycin, which compares to a death rate of 29.8 with no antibiotic and 31.5 with amoxicillin.

So, for every ~20,000 prescriptions of azithromycin written, there is one additional death from cardiovascular causes.  This is another one of those cases where the severity of the absolute difference doesn’t quite match the relative difference – it is likely any efficacy difference between a macrolide and a second-line agent results in greater morbidity than the magnitude of effect found in this study.

Then, azithromycin is frequently prescribed for upper and lower respiratory tract infections – conditions that, in the absence of other specific signs, might be non-infectious cardiovascular disease misdiagnosed as having an infectious etiology.  In their non-propensity matched cohorts, 50% more azithromycin prescriptions were written for respiratory symptoms than amoxicillin.  The propensity matching in their statistical analysis attempts to account for this, but 30% of their azithromycin prescriptions had no documented indication – which I think means there’s likely a hidden statistical difference in underlying pathophysiology secondary to unknown indications.

Finally, this runs contrary to a 2005 article “Azithromycin for the Secondary Prevention of Coronary Events” published in NEJM – at one point, it was theorized that azithromycin would be protective for coronary events.  For 4,000 patients who took azithromycin weekly for a year, there was no difference in cardiovascular outcomes as compared to placebo (CI -13% to +13% relative risk reduction).

There are lots of reasons not to prescribe azithromycin, but this study isn’t the one that should change your practice.

“Azithromycin and the Risk of Cardiovascular Death”
http://www.nejm.org/doi/full/10.1056/NEJMoa1003833