Enough with the gratuitous spinal steroid injections already (Member Post)

Epidural steroid injection overview: what exactly is it?

ESI is a steroid injection right into the spinal canal, the most common of the interventional pain medicine procedures, which are mostly injections that aim to stop pain at the source.2 A long needle is guided into the fairly tight space just outside the last major barrier around the spinal cord itself, the spinal cord’s “sheath.” That layer is the “epidural” space, the space “outside the dura mater,” but still inside the vertebra. It’s a small target, especially in the neck — anatomy is packed tight. The further up the spine, the smaller the epidural space gets, and the more difficult and risky it is to stick a needle into it.3

There are also injections that go deeper, through the dura and into the inner sanctum of the spinal cord, the “intrathecal” space — more potential, more danger. But they both still involve a needle deep in the spine.4

The goal is to hose down inflamed nerve roots with the soothing balm of the most potent and indiscriminate anti-inflammatory medication we know of: corticosteroids. (And sometimes other things, but steroids are usually the active ingredient.5) This has always been a delicious-sounding idea. If nerve root inflammation is causing the pain, why not go straight to the source and soak it with steroids? Park the medication right where the trouble is! Just thinking about it makes me want some.For decades, this seemed like elegant, targeted medicine. Epidural injections are about a century old, and injecting steroids started in the 1950s.6 It just seemed to make so much sense that no one thought they had to do any trials. And then they finally did some trials…

Ballengee et al. don’t think ESI works very well — and for good reason

Although they do review the evidence, Ballengee et al. also clearly believe that the work has already been done and ESIs are already on medicine’s naughty list: multiple RCTs, meta-analyses, and every major guideline body (ACP, NICE, ASA) say they don’t provide meaningful long-term benefit for non-specific chronic low back pain.789

Who disagrees and why? Mostly the doctors who make a good living injecting spines. We’ve seen them react similarly to the disappointing evidence on the “block and burn” treatments.10 The obligatory quote for this:

It is difficult to get a man to understand something when his job depends on not understanding it.

Upton Sinclair

Maybe you can make the case that ESI is sometimes worth trying, and you can call it “nuance” to disarm skepticism. As with block and burn, it’s not outrageous to judiciously use ESIs with more serious cases or specific indications, with informed consent. Even the strongest negative evidence is usually not the whole story.

But you can say that about practically anything short of gargling with bleach.

Robust treatment benefits rarely lurk in the statistics, and bad clinical trial news is mostly as bad as it seems. There is no question that ESI is a low-value medical practice for most people, most of the time. The happy endings are exceptions that are greatly outnumbered by the optimistic prescriptions, and the outliers mostly aren’t very important.

Spinal steroid injections as a speculative treatment for unexplained back pain are mostly just bad medicine.

Yes, actually, there is a “chance” … just not a very good one. I think of this meme every time I see healthcare pros objecting to evidence that their treatments don’t work, looking for ANY way to avoid the bad news.

Slightly less bad news: short-term relief is real, and the odds are better with sciatica

First, ESI may provide genuine short-term relief — and a few weeks of reduced pain isn’t nothing. If you’re in the kind of acute agony that makes sleep impossible, enough relief to function while time does its work has real value. A short-term ESI experiment is more defensible than the overall evidence against it might imply, especially for patients who urgently need to be functional in the short term.

Second, the evidence is worst for chronic non-specific low back pain — the most common kind, with no clear cause. That’s the main target of Ballengee et al.’s criticism, and deservedly so. But what if a distressed nerve root is almost certainly the culprit? What if you have pain spreading down the leg, sciatica? For radicular pain, which isn’t rare, the ESI evidence is … less terrible. The long-term picture remains unimpressive, but short-term pain relief shows up reliably enough in the better trials to take seriously.

The caveat is that the benefit is not large even when it does exist (which is why I didn’t lead with this). But the signal is there.11

The responsible clinical position, then: maybe ESI, for this patient, with this clear structural cause, as a short-term measure. The mistake — repeated approximately 9 million times a year in the US alone — is treating ESI as a routine option regardless of mechanism. Ballengee et al.:

“For some patients, particularly those with acute radicular symptoms, ESIs may offer temporary relief that allows participation in physical therapy or return to activity. However, they have also become a mainstay in the treatment of chronic low back pain despite limited evidence of benefit in this population.”

Why don’t spinal steroid injections work better than they do? Five possible reasons ESI fails

It’s not like we actually understood the problem or its solution well to begin with, so perhaps it’s not surprising that it doesn’t work well. In 2009, Benyamin et al wrote: “The underlying mechanism of action of epidurally administered steroid and local anesthetic injection is still not well understood.”12 So what might be going wrong between “it just makes sense” and “well that was underwhelming”?

• Inflammation may not be the primary driver in many cases. The ESI premise assumes nerve root inflammation is the main driver of pain, but chronic spinal pain is typically more complex — many possible lesions (and sometimes a combo), psychosocial amplifiers, central sensitization — and corticosteroids can’t touch any of that. The injection is hosing down inflammation that may be a minor player, a bystander, or missing entirely.
• The anatomy is less cooperative than it looks on paper. The epidural space isn’t a tidy, uniform chamber — it’s irregular, fat-filled, and laced with connective tissue. Getting medication to a specific nerve root is harder than you think. Distribution is unpredictable, and the target may simply not get the dose the theory requires.
• Inflammation in this context may be self-limiting anyway. Acute nerve root inflammation from a disc herniation resolves on its own — the herniated material gets resorbed, the immune response calms down as reliably as a cut scabs over. ESI is often late to a party that was already ending, producing apparent short-term benefit that’s mostly just the natural history of the condition doing its thing. Good trials control for that, and tell a more discouraging story.
• It might just be the dilution from injecting any fluid, flushing the local chemical environment — which could help explain why sham injections (saline, contrast dye) often do surprisingly well in trials.
• Repeat injections may also irritate, undermining any anti-inflammatory benefit. There’s some evidence that corticosteroids are actually toxic to neural tissue at sufficient doses — and this treatment is routinely offered in a series of three injections.

Five harms of spinal steroid injections (beyond just being a waste)

1. Mostly the usual for corticosteroids: repeated injections cause bone loss, fracture risk, and adrenal suppression. One or two isn’t going to do much — the dose makes the poison — but why would you want to even start to menace your spine like that without a very clear benefit?
2. Complications are rare but real, as you’d expect with a delicate and precise injection right next to the “skin” of the spinal cord. What goes wrong most often is probably puncturing the dura mater, a “wet tap,” which causes a cerebrospinal fluid leak and a nasty headache. It passes, but yuck!
3. Sometimes you get a pain flare. The opposite of a “soothing balm,” it’s more like onion in your eye!
4. Catastrophic outcomes, like a spinal cord infarction (oxygen starvation), are truly rare and mostly occur in the neck where precision is — but they do happen even in the lumbar spine, and that’s a hell of a thing, given how (over) confidently these are prescribed. Even rare complications crop up quite a bit when you do them many millions of procedures annually.
5. Reliance on ESI also delays and distracts from anything else that might help — the opportunity cost.

But what actually helps instead? We are talking about a notoriously untreatable condition here — so I was braced to be disappointed by the so-called alternatives. But not braced enough, because Ballengee et al. abandoned all skepticism for this part of the paper. They recommend replacing ESIs with physical medicine (physio, chiro, massage); mindfulness/yoga/t’ai chi; cognitive behavioural therapy; and, sigh, acupuncture.

They do not acknowledge that most of that stuff is no more evidence-based or promising than epidural injections themselves, and in some cases much worse.13 That double standard undermines the paper’s serious and worthwhile goal of de-implementation.14

Why do steroid injections persist despite the very discouraging evidence? It’s not just the money

“It’s the money” is only the most obvious reason. Of course fee-for-service reimbursement rewards ESI prescription, but it’s also the more insidious investment and identity problem. Interventional pain specialists were trained to do ESI — some of them many years before the discouraging evidence — and it’s a very human thing to not only be shaped by your training, but also defined and limited by it. A challenge to what professionals do isn’t just a threat to their bottom line, but to who they are.

ESI is also baked into the infrastructure of medicine, a trap made of red tape. Electronic Health Record “order sets” are like templates or macros for ordering a batch of related tests, medications, or referrals with a click. If “epidural injection” comes standard, it becomes the path of least resistance — it’s just there, easy and normalized.

Patient pressure is part of the equation too: just as doctors have been trained to do ESI, we patients have also learned to expect to be “fixed” by procedures like this, like taking a car to the shop, and what we expect we often demand. We want someone savvy to say “well there’s your problem” and tell us, like a confident mechanic, what they’re going to have to do to make it go away. We assume it’ll be worth every penny — if it works, and of course it will, or why would it even be an option?

And the other major treatment options — like physical therapy — are often less available, more work to refer to and work with, and less generously reimbursed.

And so ESI is also often the path of least resistance for both patients and professionals, with all the seductive power of the “easy fix.”

How to solve the problem, maybe, someday, in just 378 easy steps

The profitability of doing ESI, the ease of being reimbursed for it, is the biggest piece of the puzzle, and so “value-based care” is the biggest theme in Ballengee et al.’s recommendations: helping healthcare professionals get paid for what helps people the most rather than services reflexively rendered. I first brought the need for VBC up a few months ago in my series about “Why is it so hard to get good help for pain? Part 1 (Member Post)” — and that’s partly why I wanted to follow-up with this paper about a great concrete example.

There’s already a significant shift in this direction in healthcare. Ballengee et al. note that about 28% of payments are already in value-based models, and they frame that as momentum to build on.

But changing how people get paid is a very heavy lift. While the medical-industrial complex slogs through that bureaucratic quagmire, here is what most patients and clinicians need to know:

ESI is not a good value — and although happy exceptions surely do exist, your case probably isn’t one of them. Patients and professionals alike need to curb their enthusiasm for this procedure. Patients: mostly don’t ask your doctor if spinal steroid injections might be right for you! And doctors, please, mostly just don’t suggest it.

Notes

1. Ballengee LA, Lentz TA, Goertz C, et al. De-implementing Spinal Injections in Pursuit of Value-Based Care: Rethinking Pain Relief. J Pain. 2026 Feb:106245. PubMed 41759692 ❐ PainSci Bibliography 49188 ❐
2. Interventional pain medicine (or m-for-management) is a loose group of minimally invasive procedures — less than surgery, more than conservative — used to diagnose and treat pain in an allegedly “targeted” way, mainly nerve blocks and radiofrequency ablation for spinal joints (the “block & burn” therapies) and epidural steroid injections for radiculopathy and stenosis, but also some implants like spinal cord stimulators and intrathecal pumps. The conceptual glue here is neuraxial and neural targets with the goal of modulating pain transmission.
3. Because of this, epidural injection has long been more common and more suitable in locations from the upper back down, and their reliability and safety for neck pain has been considered dubious even by its proponents. However, it is becoming more accurate, practical, and safe in the neck with new techniques. Fluoroscopic guidance! Ask your surgeon if she’s using it — and, if not, why not.
4. Word nerd note: Both “dura” and “theca” refer to the same membrane, but one of them in Latin and the other in Greek — a very medicine thing to do. So epidural is Latin for “outside the membrane” and intrathecal is Greek for “inside the membrane.”

5. Local anesthetics are almost always co-injected, usually just a little (low concentration lidocaine or bupivacaine). This serves as a diagnostic signal (rapid pain relief confirms correct placement) and may potentiate the steroid. They often make up the bulk of the injectate.

   Saline is often used as diluent or as a volume expander; some practitioners do “epidural saline” injections as a control or as a standalone hydraulic treatment (the “epidural washout” concept).

   Opioids are occasionally added (morphine, fentanyl), but that’s more common in post-surgical or obstetric epidurals than in pain clinic ESIs. Not standard in outpatient interventional pain medicine.

6. Epidural anesthesia as a technique dates to around 1900. The Spanish military surgeon Fidel Pagés is usually credited with the first epidural block, around 1921, though he died shortly after and his work was forgotten for a decade before being independently replicated by the Italian Achille Dogliotti in the 1930s, who got more of the credit historically.

   The epidural space had actually been accessed earlier — caudal epidural injections (via the sacral hiatus, a somewhat easier route) were described as early as 1901 by Sicard and Cathelin in France, working independently and nearly simultaneously. That was the cocaine-as-wonder-drug era of medicine, so they injected cocaine, as you do. Someday we may look back on the injection of corticosteroids as foolish too.

   Cortisone was isolated in the late 1940s (Hench and Kendall, Nobel Prize 1950), and the first reported epidural steroid injection for back pain appears to have been done by Lievre et al. in 1953 or 1957 depending on which paper you count. So the corticosteroid application is a postwar innovation grafted onto a technique that was already half a century old.

   The procedure then proliferated more or less ahead of the evidence — still a bad habit even in the modern era of interventional pain medicine and surgery, decades after evidence-based medicine had generally become the de facto standard (at least in theory).

7. Wang X, Martin G, Sadeghirad B, et al. Common interventional procedures for chronic non-cancer spine pain: a systematic review and network meta-analysis of randomised trials. BMJ. 2025 Feb;388:e079971. PubMed 39971346 ❐

   This review of all the common “block and burn” treatments for neck and back pain was resoundingly bad news. All of the treatments studied were either radiofrequency ablation of nerves (“kill it with fire”) or some combination of injections of steroids or anaesthetics into muscles, joints, or epidural space. It’s generally a mixture of decent evidence-of-absence and murky but suspicious absence-of-evidence. Zooming in on steroid injection, there is moderate quality evidence that they “probably provide little to no difference in pain relief.”

8. Staal JB, de Bie RA, de Vet HCW, Hildebrandt J, Nelemans P. Injection therapy for subacute and chronic low back pain: an updated Cochrane review. Spine. 2009 Jan;34(1):49–59. PubMed 19127161 ❐

   A review of all kinds of injection therapies for the low back and finding “no strong evidence for or against the use of any type of injection therapy.” This is more of an absence-of-evidence than truly “disproved,” but some of those gaps have been filled since 2009, and it was a damning lack of signal even then.

9. Friedly JL, Comstock BA, Turner JA, et al. Long-Term Effects of Repeated Injections of Local Anesthetic With or Without Corticosteroid for Lumbar Spinal Stenosis: A Randomized Trial. Arch Phys Med Rehabil. 2017 Aug;98(8):1499–1507.e2. PubMed 28396242 ❐ This is the highest-profile ESI trial, and it has a slam dunk bad-news result for the active ingredient in most spinal injections for both the short term and the long haul: it has no clinically significant effect at any point. This falls a little short of being clearly negative evidence specifically about ESI for non-specific low back pain — but spinal stenosis is the clearest structural indication for which ESI should work. If it doesn’t work for stenosis, it’s extremely unlikely that it’s going to work for back pain with no known specific cause.
10. Busse JW, Genevay S, Agarwal A, et al. Commonly used interventional procedures for non-cancer chronic spine pain: a clinical practice guideline. BMJ. 2025 Feb;388:e079970. PubMed 39971339 ❐

    This clinical guideline paper is based on Wang et al (see earlier footnote), a major new review of interventional pain medicine (IPM). It’s summarized by a good infographic. It concludes that “well-informed people would likely not want” these treatments, provoking thirty-four medical societies to call for its retraction. Their outrage is loud, but not evidence-based, mostly the usual “in my experience” stuff. There might be reasonable arguments for using these procedures in some cases — but there is no excuse for trying to silence inconvenient science!

11. Oliveira CB, Maher CG, Ferreira ML, et al. Epidural Corticosteroid Injections for Sciatica: An Abridged Cochrane Systematic Review and Meta-Analysis. Spine (Phila Pa 1976). 2020 Nov;45(21):E1405–E1415. PubMed 32890301 ❐

    The conclusion: “A review of 25 placebo-controlled trials provides moderate-quality evidence that epidural corticosteroid injections are effective, although the effects are small and short-term.”

12. Benyamin RM, Singh V, Parr AT, et al. Systematic review of the effectiveness of cervical epidurals in the management of chronic neck pain. Pain Physician. 2009;12(1):137–57. PubMed 19165300 ❐
13. Cashin AG, Furlong BM, Kamper SJ, et al. Analgesic effects of non-surgical and non-interventional treatments for low back pain: a systematic review and meta-analysis of placebo-controlled randomised trials. BMJ Evid Based Med. 2025 Mar:bmjebm–2024–112974. PubMed 40101974 ❐

    This big new review of common non-medical back pain treatments is bad news. It’s a sequel to a 2009 version, and, like any good sequel, it delivers more of what made the first one good: the most rigorous science, a meticulous and sensible review of placebo-controlled trials only, a high bar to clear. And it seems to be just about as discouraging as it could possibly be. But tilt your head like a curious doggo for a different view, and maybe it’s not as bad as it seems? Here’s some good news about the bad news:

    • Given the amount of garbage out there, getting some relief from 1 in 10 treatments doesn’t actually strike me as being all that bad. And the winners are: NSAIDs, exercise, spinal manipulation, taping, antidepressants, and TPRV1 antagonists (like capsaicin in spicy rubs).
    • There’s a lot of “garbage in, garbage out” here, and it mostly confirms that we have inadequate rather than negative evidence. The review had such high standards that there weren't enough high quality studies for many conclusions. (And yet consumers are spending billions of bucks on those not-clinically-proven treatments. 😬) But the silver lining here is that there could be good treatments that simply haven’t been tested enough yet.

    And now the bad news about the bad news. Unfortunately, if tilt your head the other way, the bad news looks even worse than it seemed at first. Quite a bit worse.

    • The lucky handful of treatments that are being touted as effective here are offering really minor relief, all very damned-with-faint-praise.
    • And some of those small benefits are likely illusory — evidence that registered as a thumbs up in this review, but likely won’t stand up to scrutiny/replication.
    • And the reems of “inconclusive” evidence here? Most of that is extremely unlikely to ever get conclusive.

    The bottom line is clear: back pain is largely immune to treatment. See also Wang et al, another 2025 review focusing on interventional medicine — with similarly high standards, and disappointing results.

14. Ballengee et al. apply rigorous evidentiary standards to ESIs (“not an area of scientific controversy”), but then they recommend alternatives with softer language: “modest improvements,” “moderate benefits for some patients,” “shown efficacy in reducing pain intensity.” They cite “systematic reviews and clinical guidelines” for CBT and mindfulness without engaging with how thin that evidence base actually is for chronic non-specific LBP.

    Their flattery of the alternatives is somewhat defensible: several of them are merely damned with faint praise by mixed and mediocre evidence or even an absence of (good) evidence … whereas ESIs are much more fully damned by stronger evidence of absence. But it’s still a bizarre flip-flop from a scorching medical disavowal of ESI to the casual endorsement of milquetoast options like CBT, tai chi, and acupuncture, without so much as a raised eyebrow. “A full review of possible options is beyond the scope of this Focus Article,” they write.

    Well, sure, but that doesn’t make it a good idea to just recommend all of them!