There were a number of things that surprised me in this study. the first was how many adults get short courses of oral steroids in the USA (and for all I know in other countries too). Waljee AK, et al. Short term use of oral corticosteroids and related harms among adults in the United States: population based cohort study. BMJ. 2017;357. In their cohort of a huge number of US adults (1.5 million) aged 18 to 64 enrolled in a commercial insurance plan, over 1 in 5 (21%) had an outpatient prescription for a short course of oral steroids (less than 30 days) over a 3 year period. Most of the prescriptions were for a 6 day methylprednisolone dosepak, and the median prednisone equivalent dose was 20 mg.

I can’t fathom why 1 in 5 adults would need potent oral steroids: most of the prescriptions were for URIs and back pain, with allergies and bronchitis following.

Which might be OK if they were safe.

None of the adverse events they were monitoring in this study, sepsis, venous thrombo-embolism, and fractures, were very frequent, but all were more frequent in the 90 days after a patient received short course oral steroids (compared to the 90 days before). Sepsis was 5 times more frequent during the first 30 days, and remained 3 times more frequent between 30 and 90 days after the prescription, when no-one was still taking steroids! Similarly for venous thrombo-embolism, and fractures. The risks were increased by steroids and remained higher for at least 90 days.

This study shows that you sometimes need an awful lot of data to find adverse effects. Even if the baseline risks are much higher, such as for sepsis in preterm babies, huge data sets may be needed to show increased adverse outcomes.

Why would you give steroids for a sore throat? There are some very new data from an RCT of a single dose of dexamethasone, showing that a secondary outcome (proportion with complete symptom resolution by 48 hours) was improved by the steroid from 37% to 27%, but there was no difference in the average time it took to achieve complete symptom resolution.  The primary outcome variable, symptom resolution at 24 hours, was not affected. In other words a few more people get better between 24 and 48 hours, but on average it takes just as long to recover.

That doesn’t sound to me like a good risk/benefit balance, a chance of maybe having a shorter duration of discomfort, with an increased (if tiny) risk of septicemia. So I think I’ll just take some non-toxic symptom relief from the pharmacy next time I have a sore throat, and stay well away from short course steroids.

As for our patients? We need to be sure that there are clinically important benefits, important to families, whenever we consider prescribing steroids. Shortening the duration of oxygen therapy (and thus the diagnosis of “BPD”) is not necessarily worth the risk, and quantifying the risks, unless they are frequent, is very difficult.

It is true that not all the trials of steroids have shown an increase in sepsis, for example. But to be sure there is no increased risk of sepsis (especially after the first few weeks of life of a very preterm baby, when the absolute risks of sepsis are falling and when we are more likely to consider post-natal steroids) requires large numbers, very large numbers, of patients.