Dr. Radecki’s article on steroids in sore throat is too quick to accept the study he cites about the increased risk of short term steroids (BMJ. 2017;357;j1415). In that study, the steroid-treated group was older and had more comorbidities than the group not treated with steroids. It is likely that the increased incidence of adverse events related more to those preexisting differences than to the steroids.

Waljee et al compared non-steroid users and users with matched ICD-9 diagnoses. However, there can be a wide disparity in the severity of symptoms and findings for patients assigned the same ICD-9 code. No severity index or rating was used. It may be that corticosteroids were prescribed to those with more severe symptoms or findings, which would again invalidate the comparison.

Waljee et al also used a self-controlled case series, stating that patients had more adverse events in the six months after receiving steroids than in the six months before. That methodology may be appropriate to track adverse events related to vaccines, since individuals receiving vaccines are presumably at their baseline and the vaccine introduces an isolated new variable. However, the patients in this study received steroids to treat symptoms, conditions, or specific illnesses. Those represent additional variables that completely invalidate the “self-control.”

Waljee et al suggest “alternatives to corticosteroids (eg, non-steroidal anti-inflammatory drugs for acute gout or tricyclic antidepressants for neuropathic pain.” The considerable adverse effects of NSAIDs and tricyclics are widely known. The authors provide no data that their alternatives are safer than corticosteroids.

The study cited found statistically significant increased risk for short-term steroid use, but even if the methodology and results were valid, the relevance of those differences would be questionable. For example, non-steroid user incidence of sepsis was 0.02 percent; for users, it was higher, 0.05 percent. Most individuals who were informed that taking a steroid would likely improve their symptoms, but (according to a flawed study) may increase the incidence of sepsis by three patients among 10,000, would likely still choose steroids.
– Cloyd Gatrell, MD, FACEP
Carlisle, Pennsylvania

Dr. Radecki Responds

Thank you for your comprehensive and insightful comments. I agree with your assessment of the limitations of the Waljee article, many of which impact the validity of the observations and their generalization to the treating population. However, in a Bayesian sense, some of these observations are consistent with expected effects and are reinforcing, even if far from reliable proof of causation.