Monday, March 24, 2014

Azithromycin & levofloxacin safety concerns continue

- Jennifer Middleton, MD, MPH

Safety concerns with azithromycin and levofloxacin seem to be mounting of late, and a new study titled "Azithromycin and Levofloxacin Use and Increased Risk of Cardiac Arrhythmia and Death," just published in the Annals of Family Medicine, adds fuel to the fire.

Rao et al conducted a retrospective cohort study, comparing US veterans who had received azithromycin or levofloxacin to those who had received amoxicillin. The researchers looked at outcomes during the 10 days after starting one of these three antibiotics (first dose of antibiotic = day 1). Participants were excluded if they had been on another antibiotic in the month prior to receiving one of the three antibiotics and/or if they had any current "life-threatening" condition. The researchers then looked through participants' charts and identified which ones had been admitted for a serious cardiac arrythmia (long QT syndrome, ventricular tachycardia, ventricular fibrillation, ventricular flutter, or cardiac arrest), been evaluated in the emergency department with one of those cardiac arrythmias, and/or had died of any cause.

They found that, compared to amoxicillin, treatment with azithromycin and levofloxacin was associated with an increased risk of both death and serious cardiac arrythmia.

During days 1-5, the hazard ratio (HR) for risk of death with azithromycin was 1.48 (95% Confidence Interval 1.05-2.09) and the HR for risk of death with levofloxacin was 2.49 (95% CI 1.70-3.64). In days 6-10, the HR for death in the azithromycin group was no longer statistically significant (1.14 95% CI [0.81-1.62]) but the HR for death in the levofloxacin group still was (1.95 95% CI [1.32-2.88]). The data for cardiac arrythmias follows a similar trend and may be viewed here.

The authors do note that a similarly styled study in Denmark last year was reassuring regarding these risks in a young, healthy cohort, but caring for older men with multiple co-morbidities is a significant portion of my patient panel and, I suspect, is for a significant number of other family docs as well. These men are more likely to have pneumonia, for which I frequently use one of these drugs, and azithromycin and levofloxacin are also recommended for patients with severe COPD exacerbations.

This study's findings warrant attention, even if it is a retrospective study. Although many physicians prefer basing practice change on a prospective study, retrospective studies are better suited than prospective studies to detect rare events. With a prospective study, researchers would need to identify large groups of participants and potentially follow them for a long time before having enough cases to meet statistical rigor. That can take a lot of expense and time to accomplish and can be an understandable turn-off for a potential researcher. So, this retrospective cohort study may be as good as it gets for a while.

As for me, I'll be checking out the AFP By Topics on Pneumonia and COPD to investigate alternative antibiotics to use in my older male patients. How about you - will this retrospective cohort study change how you prescribe azithromycin and/or levofloxacin?

1 comment:

  1. Realizing the difficulty in gathering a prospective study group together to evaluate a rare event - the fact is that drawing conclusions from this retrospective analysis is far from definitive. For example - the authors acknowledge potential for bias in antibiotic selection (sicker patients being more likely to have received azithromycin or levofloxacin rather than amoxicillin). How then can this trial "prove" it was the antibiotic that was responsible? Clearly, from what we know - azithromycin and levofloxacin appear more likely to lengthen the QT and predispose certain patients to a potentially malignant ventricular arrhythmias. That said - such events are rare. I don't think cause-and-effect is in any way proven by this retrospective analysis in which potential for bias between patient groups seemed likely.

    ReplyDelete