“John, This cough and congestion is killing me. It’s turning thick and green. Can you write me a Z-pak? It always works for me.”
If you write a blog on medical decision-making and heart rhythm matters, it seems an incredible omission not to opine on the FDA warning concerning the commonly used antibiotic azithromycin (the drug in a Z-Pak). Quoting directly from the FDA warning:
[Azithromycin] can cause abnormal changes in the electrical activity of the heart that may lead to a potentially fatal irregular heart rhythm.
Should we say this more clearly: that simple antibiotic you are taking for a minor infection could be lethal!
This is something, isn’t it?
In the following paragraphs, I will present the facts, describe the study that led to warning and then offer my bottom-line take home.
It all started in May 2012 when a group of researchers from Vanderbilt published this widely publicized and controversial study in the New England Journal of Medicine.
As background, the research team knew that azithromycin, which belongs to a class of antibiotics called macrolides, has the propensity to prolong the QT interval. The QT interval is the time from the onset of heart muscle activation to completion of relaxation. It’s an important interval because excessive lengthening of the QT can lead to lethal cardiac arrhythmia, which we give the French name torsades de pointes (twisting of the points.)
An individual’s susceptibility to QT prolongation is highly heritable. There are two forms of long-QT syndrome: congenital long QT syndrome (due to many different types of mutations of heart ion channels) and the more common variety of acquired long QT syndrome.
The most common cause of acquired QT prolongation is drugs. A close second is electrolyte abnormalities, low potassium levels being the main culprit. Whether one develops drug-induced QT prolongation depends not just on the drug’s ability to prolong the QT interval; it also depends on an individual’s genetic susceptibility to the drug. Sadly, there is no reliable way to know one’s susceptibility beforehand.
This sets the stage for the research interest in a commonly prescribed drug that might have potential cardiac dangers. Few drugs are more commonly prescribed than the Z-pak. Think about the science from a population perspective. Even a tiny risk from a drug could be important if it’s given to millions.
Now to the 2012 study: It was a statistical behemoth. The research team studied a “cohort that included patients who took azithromycin (347,795 prescriptions), propensity-score–matched persons who took no antibiotics (1,391,180 control periods), and patients who took amoxicillin (1,348,672 prescriptions), ciprofloxacin (264,626 prescriptions), or levofloxacin (193,906 prescriptions).”
Their findings were highly provocative: “During 5 days of therapy, patients taking azithromycin, as compared with those who took no antibiotics, had an increased risk of cardiovascular death (hazard ratio, 2.88; 95% confidence interval [CI], 1.79 to 4.63; P<0.001) and death from any cause (hazard ratio, 1.85; 95% CI, 1.25 to 2.75; P=0.002).” Another way of communicating the results is to say that for every 1 million prescriptions for azithromycin an additional 47 deaths occurred.
This is a very important study. Let’s look at it from a couple different perspectives:
Is this good data? Should we even give the study consideration?
There were significant limitations. My colleague, Dr Wes Fisher unabashedly writes about many of them in this candid post. In short, here are my top 4 reasons to be careful with interpreting too much form the study:
- It looked only at a Medicaid population in Tennessee;
- It was a look back rather than look forward study;
- The groups compared were not randomized;
- Data was taken from computer records only.
Association v Causation: These problems lead to the issue of confounding variables. In other words, yes, there may be an association between a drug and an increased death rate, but this does not mean the drug caused the higher death rate. This is a huge distinction. Just remember that association does not equal causation; and the more confounding variables there are, the less likely that the relationship is causative.
But on the other side of the coin is plausibility. Here, there is a case to be made. Consider the facts: Though rare, azithromycin can prolong the QT interval. It is bad to have a prolonged QT interval. Individuals vary in their susceptibility to drug-induced QT prolongation. We have no means of detecting QT susceptibility before giving the drug (at least not until personalized genomic medicine matures.) So it’s possible that a tiny risk of a drug could be amplified over a population. The findings are plausible.
Now let’s look at the data again in light of biologic plausibility. The researchers made a valiant effort to combat the problem of confounding. By matching the comparison groups on 153 variables (things like age, gender, co-existing diseases, race etc) they ended up with two very similar cohorts. Further, they added a third study group composed of patients who took other antibiotics. The fact that Ciprofloxacin (a known QT prolonger) increased risk while amoxicillin (no QT effects) did not, enhances the study’s findings on azithromycin.
Clinical impact: Is azithromycin a good drug? Are Z-paks really that dangerous?
Azithromycin is a useful antibiotic. It’s well tolerated, easy to take, inexpensive and offers a broad spectrum of coverage against many of the most common bacterial infections. But it is a drug, a chemical with real biologic actions. Treating anything in Medicine means considering the trade-offs. On the upside, the benefits of azithromycin are obvious: feeling better faster and the avoidance of complications of unchecked bacterial infection. With treatment, however, comes risk. In the case of azithromycin, this study suggests the possibility, albeit slight, of lethal heart rhythm disturbances.
This presents a really tough situation for both patients and doctors. First of all, many of the illness for which azithromycin is used are caused by viruses—not bacteria—and thus antibiotics are useless. Second, many of these illnesses, even if bacterial, will resolve with no therapy. Third, Z-paks have reached legendary status and ’empowered’ patients seek out the drug. Fourth, profit-driven walk-in clinics, where many of these infections are treated, are incented to make patients happy—and Z-paks please. Finally, in the US, it is much easier (and perhaps less risky) to treat people with a drug than it is to explain the disease and that it may resolve without a drug.
I don’t want to sound too loud an alarm on azithromycin. It’s not a bad drug. Drugs are not bad or good. They all have benefits and risks and alternatives. The FDA has decided that the risks of azithromycin warrant mention to the public.
I like the warning. Though the NEJM study is not statistically strong enough to quantify the risk of azithromycin, it does focus attention on the risk/benefit analysis of medical decision-making. Namely, that there is no free lunch for treatment—not even Z-paks.
More and more, with the medicalization of everything, patients and doctors need to better understand that taking medicine or having surgery means accepting trade-offs.
For the record, as a heart rhythm doctor, I am respectful of the QT interval. You should be too.