Reconsidering Dexamethasone in Headaches
Longtime followers of this blog may recall a kerfuffle over a year ago, related to a graduating senior’s incendiary journal club discussion, in which he examined a small but well-conducted trial of using dexamethasone in benign headaches. At the Journal Club and in my online review, concluded that a one-time dose of this pretty safe medication would be of value in reducing headache recurrence, and were dismayed when an esteemed reviewer for Journal Watch brushed aside the study as too small and unstructured (her review is reprinted in the blog post comments section).
Several residents complained in the comments section, and the term “nihilism bias” was coined. Months went by, but on the night of St. Patrick’s Day, the editor-in-chief of Journal Watch and a leading figure in EM jumped into the fray and advised us that changing practice based on a single, small study is not the kind of care he teaches or wants to be subjected to.
Steroids in headache, which had been a hot topic at SAEM last year and subject of some additional trials in the interim, was addressed again in Journal Watch this week. The very same editor reviewed a new meta-analysis of dexamethasone for acute migraines (Colman, BMJ 2008 Jun 14; 336:1359) and wrote:
“…The authors conclude that when added to standard acute migraine treatment, a single parenteral dose of dexamethasone is associated with a 26%reduction in headache recurrence within 72 hours. An accompanying editorial notes that patients with diabetes should be treated with caution because of the potential for prolonged elevation of blood sugar…
Despite the limitations of a meta-analysis, these results suggest that a single dose of dexamethasone is a reasonable addition to the treatment of acute migraine episodes that has little downside for patients without contraindications to steroids. Although dexamethasone had no benefit for initial pain relief, the number needed to treat to prevent one recurrence was only 9, suggesting that treatment has significant potential to help patients remain functional and avoid repeat emergency department visits. This study tested only parenteral administration, but an oral dose might be just as effective in nonvomiting patients.”
Perhaps a lot has changed in 18 months, perhaps not so much (the Baden 2006 study that was the basis of our original journal club was the smallest in the meta-analysis, and showed the most favorability of dexamethasone), but one things for sure — you can now order a dose of dex with a clear conscience.
Posted
on Monday, July 14th, 2008 at 4:55 am by Nick. Filed under
Headache, Blog.
You may post a comment.
Nick et al.
Though Dr. Denny’s saturnine disposition may lead to identifying him with the term more than most, the nihilism bias has been spoken about by the ebm folks for years. I think it was first mentioned in BMJ.
As to the steroid controversy, the noted EM figure was right in his assertions, though if I remember correctly, they were not expressed in the nicest way possible. One study should rarely change practice unless its conclusions are very robust and generizable. I do not think it in any way contradicts his argument to publish what he did.
This is the case with adopting new treatments. What I find more worrisome is when the same concept is applied to potential harm. In the most recent annals, there is a debate re: etomidate in sepsis. Many studies point to, but do not definitively show that this drug may be harmful, not in the ED but later in the hospitilization. The end of his argument essentially was that until harm is shown beyond a shadow of a doubt, we should not abandon the drug in this setting. This is not in keeping with EBM: you need to prove a new drug is helpful resoundingly; a much smaller burden of evidence is placed on harm.
Comment by EMCrit on July 14th, 2008 at 10:20 pm
Your points on etomidate and the burden of evidence for harm are well taken. Some of the residents are discussing this controversy over email and a similar conclusion was drawn (we have evidence that etomidate alters biochemical pathways are in a potentially harmful way, why not be prudent and use other agents?)
For the record, a google search for “nihilism bias” returns only these pages. The phrase is also not found in pubmed, though there are entries that describe a bias towards nihilism.
I do look forward to the M&M where, rather than conformation bias or commission bias, the error is ascribed to nihilism bias.
In other news, I found this wonderful essay by Jeff Mann recently (via Dr. Andrus’ Links on critical-care.info) about whether small RCTs can be clinically significant and scientifically conclusive.
Comment by Nick on July 20th, 2008 at 10:15 am