Atrial fibrillation Cycling Wed Exercise inflammation

Think twice before getting an injection for low back pain or sciatica

It’s been a while since I did a Cycling Wednesday topic. As I was skimming thorough the Journal of the American Medical Association last night, I came across this review article on spinal injection therapy for low back pain. It was a shocker.

Two factors brought my attention to the article: First, almost all the cyclists or runners I know have been beset with either back pain or sciatica (referred pain or weakness down the leg). In fact, in some cases, back issues have ended their competitive careers. I too have been flattened with low back pain. It was awful. The other reason to pay attention to the review is the huge numbers of requests I receive from AF patients who are asking to stop their anticoagulant (blood thinner) for a spinal injection.


Low back issues are even more common than atrial fibrillation. In the US, low back pain accounts for 2% of all doctor’s visits and is the fifth most common reason for visits to primary care. That makes perfect sense, doesn’t it? The same risk factors for heart disease, not moving enough and eating too much, also increase the risk of skeletal issues in the low back. The waiting room of the orthopedist looks very similar to the cardiologist.

Since I am no expert in orthopedics, I will write most of this post from a journalistic standpoint. The facts in the review are striking enough; they don’t need much editorial. My comments will be at the end.

Five facts from the review article:

  • There is substantial variation in the use of injections. One study found relatively few providers are responsible for a disproportionately high percentage of injections. The authors of the study called it a “pattern of marked [overuse] by a minority of providers.” There is no data on whether high use rates associate with better outcomes.
  • Another problem with studying injection therapy is that not all injections are the same. There are different targets (epidural, facet, discs, and ligaments for example), and different drugs injected. (local anesthetics, steroids, pain meds etc). Injections are performed in different stages of disease: acute, subacute and chronic. All this variation makes it a tough topic to review.
  • One of the best tools for evaluating the benefits of a treatment is to do a systematic review of all the world’s literature. The Cochrane Collaboration did such a study for spinal injections in 2008. The researchers found that only 6 of 18 trials showed significant results for at least one outcome in favor of the injection. No clear pattern of benefit emerged. The researchers concluded: “There is insufficient evidence to support the use of injection therapy in subacute and chronic low-back pain. However, it cannot be ruled out that specific subgroups of patients may respond to a specific type of injection.”
  • A more recent pooled analysis of 25 studies looked specifically at epidural injections for sciatica. Here, at least there was a very small (6-point on a scale of 0-100) improvement in the short-term. No long-term benefit was noted. The authors concluded: “the small size of the treatment effects raises questions about the utility of this procedure in the target population.”
  • Review of guidelines show only one country (Belgium) recommends spinal injections. The US, UK, and Europe simply do not recommend injection therapy for low back pain. Rather, the guidelines start with education, “back schools,” NSAIDs (editorial comment – be careful), opioid analgesics, back exercises, spinal manipulation, rehabilitation, and behavioral therapy.

The review article concludes:

“Patients with low back pain differ in their clinical presentation and may respond differently to treatments. Injection therapy of any kind may be beneficial in individual cases or subgroups. Nevertheless, given the weak scientific evidence base and the availability of noninvasive and more effective alternatives, physicians and policy makers should not recommend the use of injection therapy for patients with low back pain and sciatica.”

My comments:

This was very surprising. I had no idea that the evidence base for spinal injections was so weak. The concept of putting anti-inflammatory drug right on the source of inflammation makes perfect sense. It hits the problem area without exposing the patient to the risk of systemic exposure to steroids or non-steroidal drugs. You would have thought comparison studies would have strongly favored local injections. But that’s the thing with evidence-based medicine: just because something makes sense, and smart doctors think it so, does not mean it is so.

Spinal injection therapy is an important topic because many of the patients referred for injections are older folks on anticoagulant drugs. Recent studies in the AF literature make it clear that interrupting anticoagulant drugs can be risky. We also know “bridging” patients with (lovenox) shots carries risk. Again we get into net clinical benefit: you don’t mind taking the risk of being off anticoagulation if the benefit is great. This review of the literature on spinal injections suggests otherwise.

The wide-angle overriding view of the matter is always the same in Medicine. It’s best to avoid the need for treatment. Human disease is never 100% avoidable, but a healthy and balanced (emphasis especially on balanced) lifestyle reduces the risk of facing tough decisions. And when disease strikes, if it is safe to wait, taking a conservative approach, giving the body time to heal itself, is often just as good as having a sharp object stuck into you.

A final caveat on the JAMA review article. The authors are PhD researchers from a quality healthcare institute. They are not orthopedists. That might have resulted in a biased selection of the literature. An orthopedist/pain interventionalist perspective might have been different.


10 replies on “Think twice before getting an injection for low back pain or sciatica”

Great post. I think if we really digged up the evidence for a lot of what has become standard of care, we’d scare ourselves. I’d argue letting specialists drive the selection of research would result in bias for the intervention. What bias would a PhD researcher have for or against a procedure (wouldn’t they want to find the “truth”?)?

If your profession rests its laurels on the intervention, of course they will want the intervention to work (we wouldn’t have this problem if we didn’t use a procedure/drug until we had good clinical outcome data, not surrogate markers, AND if we studied procedures against a sham procedure).

You can look at kypho/verterbroplasty to find the same dilemma as you are describing, as well as PSA/Prostate Cancer Screening and baby heart rate monitors during delivery/C-sections (no baby lives are saved). Poor evidence, but routinely done.

Thanks rot the nice words here and on Twitter.

See my comment to RWK.

Outcomes! Outcomes!

I would also add niacin, and all other non-static cholesterol-lowering agents.

A colleague on Twitter, Dr Hluks, commented about the lack of benefit from many rotator cuff tears. There was also the knee arthroscopy study–no better than physical therapy.


A few years ago I had an epidural at L4 and L5. The steroid did pose a systemic problem for me. I ended up with dacryocystitis, two DCRs, maxillary, ethmoid, and frontal sinus infections with subsequent Right Frontal Osteoplastic Flap after a year and a half of life threatening infection.
Spinal injections are not so benign for everyone! Its a good thing I found an alternative that works: acupuncture.

As a physiatrist, interventional spine practitioner, and responsible physician who works side by side with orthopaedic spine surgeons, I feel I need to comment to give my perspective. I have been doing spinal injections for five years.

When I read reviews like this I just cringe. The problem with an editorial review such as this is that it only scratches the surface of spinal pathology, and makes a very broad recommendation against any type of injection therapy based on insufficient evidence. In part, this is the danger with researchers making a clinical recommendation; they overlook many of the more subtle variations in patients, and the reason behind the procedure. With regards to the sciatica review for example, were the patients young, with acute disc herniations causing radicular pain, or older with spinal stenosis and neurogenic claudication in a radicular pattern? Soft tissue lesions tend to improve, and change fairly rapidly. The vast majority of acute disc patients, typically either a contained disc protrusion, focal extrusion, or possiblly free fragment, will actually improve on their own without need for surgery as the disc reabsorbs and stabilizes. When do I have my surgical friends step in? Progressive motor deficit, or lack of improvement in baseline symptoms after six months time, and clinically significant pain. Why not surgery? What is often forgotten is that most patients develop spinal instability years down the road after even a simple discectomy, with further need then for pain therapy, and often surgical intervention. Conservative therapy is definitely preferred regardless of whether that means injections, rehab, or simply tylenol. With regards to the stenosis patient, sometimes these patient responds very well to a simple targeted injection, and yes, for some reason have lasting pain relief, not just a few weeks or couple months. Surgery in these cases is often infinitely more complex with long rehab after decompression or spinal fusion.

This is not my commercial for injection therapy, it really isn’t. Injections are a flawed modality, imperfect at best. They will sometimes provide lasting relief, and minimize the need for analgesics. Sometimes. Cleveland clinic states it’s about 50/50, and this is what I tell patients. It is very difficult to predict who will do well.

We need to narrow our guidelines for treatment, and continue to research the area carefully, and work towards minimizing overuse. Researchers also need to practice responsible evidence based medicine, and realize that overstating a position can be detrimental to patients by limiting access to care. Researchers need to consult carefully with practitioners within that specialty to gain further perspective on various reasons for ordering a certain treatment modality. Yes, at the risk of fearing their bias. Lastly, we need responsible practitioners who appropriately discuss all of the issues with patients seeking care.

I am very open to any comments on the subject. Thank you for your article……

Thank you for this excellent comment. I had hoped someone in the field would weigh in.

I harbor no doubt that spinal injections can and do help selected individuals. As the researchers said at the outset, the diversity of spinal injection makes it tough to study and make overall recommendations. That does not mean, however, that it’s not worth looking at the available evidence. There’s the problem. When patients with spinal disease have been studied in randomized trials there is no compelling evidence of improvement.

Reviews like this should not make anyone in the field cringe. They should serve as stimulus to do as you say “narrow our guidelines for treatment, and continue to research the area carefully.”

In our new world of limited healthcare resources, treatments must be bolstered by evidence. Look at cardiology: AF ablation comes under similar criticism. But we are doing the big studies needed to show long-term mortality benefits. It was the same with acute intervention for heart attacks. More than 50,000 patients with atrial fibrillation have been randomized in clinical trials comparing novel anticoagulants with warfarin. Of course, I know these were industry-sponsored; but I don’t see the problem with comparing spinal injection with physical therapy in a large-scale clinical trials. In Cardiology, much of what we do to patients has been studied and shown effective against the best alternative or placebo. One hope of mine is that in future, we have something of a model proposed by Ben Goldacre. He wants to use EMR and big data to do studies as a matter of normal clinical routine. You got a patient with focal spine pathology; you randomize them to injection or physical therapy and record the outcome months later. Simple. Then you know. No more wily-nily doing stuff because it makes sense. Outcomes will be the driver.

It was a great comment; I appreciate it.

Thanks to all for the comments. I would like to expand the thoughts a little if I may. I do object to the other generalizations by the authors of the JAMA article that “opioids, spinal manipulation, physical therapy, and psychology services” are all viable options in lieu of injection therapy. They of course conveniently failed to produce any data to support this assertion. The reality is there is NO DATA, yes, insufficient evidence to support physical therapy, opioids, or spinal manipulative therapy. How many times have we heard a patient say “physical therapy didn’t help”, or “it made me worse”. There IS data for cognitive behavioral services, and I think this is very important. However, it is very difficult to find carriers who will pay for the behavioral med component, and that’s really unfortunate.
NASS (North American Spine Society), which is multidisciplinary, did an extensive review of the spine treatment literature, and published guidelines on this. They recommend transforaminal injections for a majority of patients with herniated discs contributing to radicular pain. A clinician can consider physical therapy, and spinal manipulation case by case. Opioids are generally only acceptable in the acute phase, not beyond six months. Just another perspective…
Both pain assessment and treatment is very subjective, and I believe this is the primary contributor to such variation amongst studies.
Thank you again for the opportunity to comment.

I sense the answer lies in between comments by Dr. John & RWK. In an ideal world – there would be prospective, randomized, double-blind trials as per the last few sentences in Dr. John’s last comment. That clearly is the GOAL. For “evidence” – such studies ought to be prospective and controlled rather than literature reviews that depend on studies already done (and methodologies that are often hard to asses and lacking). But pain assessment and treatment IS indeed highly subjective, hard to tease out – and patient groups are often far from manifesting homogeneous “focal spine pathology” – so I sense it will be difficult to truly ever attain that nirvana state of a perfect study in this area. We are left with the “art of medicine” to truly individualize and assess what may be best for each patient with active incorporation of informed joint-decision-making along the way. Continuation of the trend where relatively few providers are responsible for a disproportionally high percentage of injections is clearly not optimal (and raises question of the appropriateness of injections by at least some of those providers). On the other hand – selective use of injections for candidates with best chance to benefit (which I believe is RWK’s perspective) may be both indicated and beneficial. Careful record-keeping that documents functional improvement for individuals so treated should help to justify appropriateness of such selective injection treatments until such time that a better objective data set on the pros and cons of spinal injections might be obtained.

As a family physician who sees these patients every day, it is no surprise to see the results. My “gestault” regarding how often the therapy works would be 2/8 patients, rarely giving any long term improvement. More interesting to me though, is just how necessary it is to stop the anticoagulants, especially aspirin and Plavix. I’d love to see a study about that necessity.

I have wondered the same thing. Again to Cardiology: We studied whether it was safe to do PM/ICD implants on fully anticoagulated patients. The dogma was that you must stop the drug and then use heparin. The study showed LESS bleeding when patients underwent surgery (which includes vascular access) without stopping warfarin. I know all surgeries are different, but the point is could we not do a study?

Comments are closed.