I haven’t posted here in years.
But what do you think of this study:
Lumbar Epidural Steroid Injections for Radicular Pain With Normal MRI Findings — Do They Work?
It's a master class in how to publish weak science when you have a strong brand and good connections
Small study of 14 per arm
'Propensity' matched 1:1
Two institutions
The selection artifact here is amazing: "
All patients were treated over a 16-year period from January 5, 2009 until November 1, 2024 owing to the “rarity” of the event. Seven of 14 “cases” were recruited from a prior study evaluating the effect of MRI on ESI outcomes (i.e., the group in which the doctor did not look at the MRI before the ESI), with the others referred from surgeons or neurologists for diagnosis and treatment [
14]. Control patients were garnered from the same study and several other prospective studies in which the same data were recorded [
14,
19,
[22],
[23],
[24]]."
I'm not sure what the rarity of the event is but those erudite academic doctors don't ever inject something that doesn't make sense or insurance always said no, but it seems unlikely to me that they did a good capture of all the ESIs they did in that time. They acknowledge that though: "First, although most patients were recruited as part of clinical studies with others accumulated based on referrals, the patients were not necessarily consecutive (i.e., some may have been missed), and the control patients were garnered from a database containing slightly under 500 patients."
I just want to acknowledge that if those two institutions or even just those two faculty did only 500 lumbar epidurals in that time period, I really don't think we should make any generalizations from their findings.
Regardless, 'randos' in one arm with concordant imaging vs mostly clinical trial patients with clean imaging, and no big difference in short term outcomes, but maybe some difference in long term outcomes, and maybe they just selected some randoms that did poorly.
For all of them though, they improved in pain the same amount at all time points. ODI was improved more in people with normal imaging vs abnormal imaging.
The conclusions though in the paper:
"For the main outcomes, there were no significant differences between outcomes for those with near-normal MRIs and the control group with abnormal imaging
suggesting that some of these patients benefit. The lack of a significant difference likely stems from the small number of patients with near-normal MRIs who underwent ESI,
the lower success rates in the propensity-matched controls than in the entire cohort from which it was drawn, and the
high placebo effects associated with interventional procedures which may drown out small signals of efficacy."
And the conclusions in the abstract:
"There were
no significant differences in pain outcomes compared to control patients, though patients with near-normal imaging fared worse than historical controls and the larger cohort from which propensity-matched patients were selected. The
possibility of poorer outcomes should be considered when selecting patients with normal imaging and radiculopathy for ESI."
I guess I could use this to argue that imaging findings don't matter so we should do an epidural in every body with leg pain? Or maybe I should argue that epidurals don't work since imaging findings don't matter? Or maybe I should just say it might be important that imaging is checked, but maybe this just wasn't worth publishing like this?
