A recent study, from the Karolinska Institutet in Stockholm, a retrospective, case-matched control analysis of 2933 Swedish patients with celiac disease (CD), is the first to establish a correlation between antibiotic use and CD. The methodology, recognizing the limits of retrospective analysis without randomization, is well designed. The investigators had the advantage of the Swedish national health system, which warehouses health data centrally and is available for study. Plus, 99% of all dispensed prescription drugs are available through the Swedish Prescribed Drug Register.
All CD study subjects were biopsy-proven, demonstrating small-intestine villous atrophy (Marsh grade 3). Each subject was assigned 5 controls who did not have CD, matched for age, sex, calendar period of birth, and county of residence. The primary analysis evaluated the strength of the association between the use of any systemic antibiotics and subsequent CD. Subanalyses were performed for antibiotic types (penicillin V, extended-spectrum penicillins, quinolones, macrolines, and all other systemic antibiotics), and dose-dependency was measured in 2 other analyses—the association when individuals had 1 to 2 courses of antibiotics versus 3 or more courses.
Median age at CD diagnosis was 28 years; 27% of CD patients had received at least 1 course of antibiotics during the study period compared with 21% of controls. Antibiotic users had an odds ratio of 1.40 (95% CI, 1.27 to 1.53) for subsequent CD. Dose-dependent effect was observed; high antibiotic users had an odds ratio of 1.58, while low-level antibiotic users had an odds ratio of 1.36 (confidence intervals were similar among all antibiotic users). The type of antibiotic used had no statistical effect.
It is important to remember that despite the investigators attempts to match case-controls, this study only establishes correlation, not causation. Causality is biologically plausible: we know that the intestinal microbiota influences the development of the intestinal immune system, the establishment of oral tolerance, and mucosal barrier function, and that CD subjects have a less bio-diverse microbiota than healthy controls. A normal microbiota may have a beneficial protective effect on the gut. No one doubts that antibiotics, especially when used repetitively, affect intestinal milieu.
The study’s most important limitation is whether there might be what biostatisticians call “reverse causation.” In this case, it would manifest as a propensity for antibiotic use in persons destined for a positive CD diagnosis, but as of yet without a diagnosis. Several studies have shown a mean delay of 5 to 11 years from onset of CD symptoms until diagnosis. Does presentation to primary care with digestive complaints predispose to antibiotic use? Quite possibly; hence the limitation.
And so, it’s time for a prospective study on the effect of antibiotic use on subsequent CD development. But even before causation is more clearly established, let’s continue to be on our guard against unnecessary antibiotic use. It’s always a bad idea, but given the growth of the celiac population over the past several decades, now might be a great time to step up efforts to combat antibiotic use for illnesses unlikely to be caused by bacteria.
Mrild K, Ye W, Lebwohl B, et al. Antibiotic exposure and the development of coeliac disease: a nationwide case-control study. BMC Gastroenterology. 2013;13:109. (Full-text)