Confounding by indication invalidates most observational studies on side effects of glucocorticoids

Maarten Boers, Professor of Clinical Epidemiology Epidemiology,

Other Contributors:

January 31, 2012

Dear Editor,

We read with interest the paper by Dixon et al, and the associated paper published in 2011 on the risk of infection in glucocorticoid (GC) use in RA.1,2 Whilst their analyses add somewhat to the evidence of the risk of infection caused by GC in RA, because they are retrospective observational studies of uncontrolled treatment decisions we have strong reservations on the extent of their interpretations and conclusions. We posit that:

(1.) The presence of strong confounding by indication severely limits the usefulness of observational analyses on the risk of GC use in RA; and this is especially so for analyses based on administrative datasets that typically lack important detail on clinical variables;

(2.) The correction for such confounding (by number of rheumatologist visits in the past 6 months and NSAID use) in this dataset is weak and perhaps wrong, as many physicians will now adjust or even stop NSAID in patients on GC;

(3.) The suggested explanations for their finding of an increased risk of infection, say, 2 years after cessation of GC lacks biological plausibility. Wouldn't the authors agree that it is much more likely is that GC use is a strong (perhaps even the strongest) marker for disease severity in RA, especially in countries such as Canada where GC use in RA is traditionally low due to strong opinion leaders with large concern over the safety of GC in RA?

Any observational analysis on the benefit-risk ratio of GC in RA should start and end with a pointer to confounding by indication, an issue which will always weaken any such analysis. Rather, it would be better to collect data from all those randomized controlled trials that can be found by an appropriate literature search and synthesize the findings, using meta-analysis if appropriate. This is currently being undertaken (Jensen et al, in preparation). If the question remains unanswered, then the best way to find a definitive answer would be through a a large and and sufficiently long pragmatic trial of GC in RA.Studies such as those by Dixon et al should be interpreted with great caution.


1. Dixon WG, Kezouh A, Bernatsky S, et al. The influence of systemic glucocorticoid therapy upon the risk of non-serious infection in older patients with rheumatoid arthritis: a nested case-control study. Ann Rheum Dis 2011;70:956-60.

2. Dixon WG, Abrahamowicz M, Beauchamp ME, et al. Immediate and delayed impact of oral glucocorticoid therapy on risk of serious infection in older patients with rheumatoid arthritis: a nested case-control analysis. Ann Rheum Dis 2012.

Conflict of Interest:

both authors, consulting fees from Horizon and Mundipharma, Ltd

Conflict of Interest

None declared