I was asked to address this subject by one of my more literate physician readers. I would like to remind everyone again that correlation does not equal causation. In particular, if patients with disease B have ingested A more often than those without disease B, we now have a hypothesis. The next scientific step would be to select a random group of matched individuals, have half of them ingest A, and then observe if they developed disease B in statistical excess of those who did not ingest A. It is extremely unlikely that such a clinical study would ever be undertaken. OTOH we have several retrospective studies which suggested that patients who ingested C had disease D less frequently, and then a forward, prospective, single-blind study was done to see if ingesting C prevented or reduced the incidence of disease D.
As a concrete example, several retrospective studies seemed to indicate that women with a low fat diet had a lower incidence of both breast and colon cancer. A five-year prospective study was then done, limiting women in the study group to a diet containing no more than 15% of its calories from fat. No decrease in the incidence of either cancer was noted. The organizers of the study therefore concluded that there was no evidence that a low-fat diet prevented colon or breast cancer. At the same time, the proponents of the benefits of a low-fat diet claimed that the % of fat in the diet should have been 5%, not 15%, while simultaneously admitting that such a diet was almost impossible to achieve in practice.
Now as to the case at hand. It was well known that macrolide antibiotics can lengthen the QT interval (on an electrocardiogram). It is also known that an electric cardiac phenomenon known as "R on T" can trigger ventricular tachycardia and thereby lead to sudden (cardiac) death. If the QT interval is increased, then there is a longer (by milliseconds) time interval in a cardiac cycle during which the "R on T" phenomenon can occur. In theory, therefore, any drug which lengthens the QT interval could pre-dispose the patient to sudden death. An increased incidence of ventricular tachycardia has been demonstrated in patients taking certain cardiac drugs that lengthen the QT interval, and such drugs are required to carry a "black box" warning because of this. Tricyclic antidepressants can also cause an increase in the QT interval, but usually the increase is not large enough to require a discontinuation of the drug.
An article was published recently in the New England Journal of Medicine (2012;366(20):1881-1890). It examined retrospectively the rate of cardiovascular (i.e. sudden) death of patients prescribed a five-day course of azithromycin (Zithromax)(Z-Pak) with those either taking no drug at all or taking amoxicillin, a different antibiotic. They found an excess of sudden death (called "cardiovascular deaths") in those taking azithromycin. The precise numbers were an additional 47 cardiovascular-related deaths per 1,000,000 courses of azithromycin in the patient population, and a further increase to 247 deaths per 1,000,000 in those at elevated risk for cardiovascular "adverse events".
This antibiotic and others of the macrolide class will now have an FDA-mandated warning. Again, this is a retrospective study, not a prospective one. I am just concerned that the study was of Medicaid patients, which means that they received generic Zithromax (i.e. azithromycin) and we have no way of knowing whether or not the brand name formulation is equally dangerous. If the result is a class effect, which it seems to be, then it should be equally dangerous, but no one knows for sure. I may or may not be splitting hairs here, and we will never find out.