No obvious neck abnormalities in headache patients

Do some headaches come from the neck? It seems obvious to half the people who’ve ever had a headache! And some more than others. Consider reader Olivia C, who has occipital neuralgia, a distressed C2 spinal nerve that blasts the back of the head with “lancinating” pain, more of a vicious head stabbing than aching. This awful condition is considered a bit of a special case (see Bogduk 2014), but it is undeniably a head pain with an origin in the neck, and so it’s a useful example of the idea of a cervicogenic headache.

A neck-made headache.

The idea that many more ordinary headaches come from the neck has always been a bit contentious, and an intriguing 2021 study seems to show that the neck is innocent. This revives a decades-old debate about how headaches work. In this post, I explore the significance of this vexing new data.

Is it a surprise? Is this the end for Team Cervicogenic Headache? What will happen? Tune in below! *munches popcorn* 🍿

•

A 2021 study by Masharawi, Mansour, Peled, and Weisman showed no sign of vertebral or intervertebral disc abnormalities in the necks of headache patients, which appears to be at odds with a widely and firmly held belief that some headaches are actually driven by neck pain (cervicogenic headache).

“A comparative shape analysis of the cervical spine between individuals with cervicogenic headaches and asymptomatic controls”
Masharawi et al. Scientific Reports. Volume 11, Number 1, 19413. Sep 2021.

In my headache book, I more or less declare for Team Cervicogenic. I think headaches can “come from the neck,” one way or another. But my opinion is loosely held, and I will gladly follow the science and the experts wherever they lead me over time — which is probably now away from the idea that arthritic cervical spines are a common headache cause.

The whole paper is quite readable, and I enjoyed reading it. Scientific papers aren't always good! They are written reports on something actually done by scientists, and all too often what they did was ill-conceived and poorly executed … and/or the writing is pretentious and tortuous. But this one is a well-written account of a useful experiment. Bravo.

But nothing’s perfect, and even a perfect study would still only bring a single piece of the puzzle of pain into focus. This is “just” a study of spinal structures, after all. The researchers looked “only” for signs of vertebral or intervertebral disc abnormalities. Are there are other possibilities? Sure, and they are also worth studying. But this is an important study of “just” the ways that have been at the centre of the debate for decades! It may not be the whole story, but it’s a big part of it.

The study builds on previous evidence from Knackstedt et al in 2012, who reported that “MRI shows no specific changes of cervical discs or craniovertebral ligaments in cervicogenic headache.” More generally, it also harmonizes with a strong theme in musculoskeletal medicine: pain correlates poorly with obvious body glitches. Jarod Hall, Doctor of Physical Therapy:

“Headaches, and all pain really, are physiological. Sometimes that physiology is driven by gross anatomical structures, but more often it is not.”

The author’s mostly normal neck bones, as revealed by a CT scan. A CT scan is like a deluxe X-ray — like an X-ray, it mainly shows bones, but it has the three-dimensionality of magnetic resonance imaging.

That’s a lot of neck measuring

Masharawi et al took CT scans of 80 people aged 20-40, and measured many specific features of their cervical spines, thousands of measurements in all — nineteen thousand of them. Someone sure stared at a lot of spreadsheets.

Forty of the subjects were headache free.

The other forty not only had headaches, but headaches that were dead-ringers for cervicogenic headaches. Their aching heads all strongly fit a classic clinical description of that kind of headache. And better still: that diagnosis was “confirmed” in every case by the traditional method for confirming the source of pain. They relieved their headache pain with an anesthetic injection in the neck. Such nerve blocks are widely regarded as strong evidence that headaches can indeed be caused by neck issues. (It’s not quite the slam dunk it seems to be, but it certainly is an important clue.)

That method really emphasizes that pain radiating from spinal structures has always been the thing that supposedly makes cervicogenic headaches happen. It’s not just a reasonable-sounding idea; it’s been based on the response to injecting numbing agents into those structures! Which is legitimately rigorous and science-y.

This ideal subject selection is a great strength of this study, making its results far more useful than they would have been otherwise (and no doubt a logistical nightmare). These people had headaches that seemed about as cervicogenic as a headache can possibly seem. If the condition of the spine has anything to do with headaches on average, you would absolutely expect to see some differences in these patients…

The utterly negative results

No differences were found. Diddly and squat. Bup and kis.

Not one statistically significant variation in any measurement. Average values for all measurements were more or less identical, as if the study subjects were all clones. The graphs in the paper tell the story quite nicely (see below).

The obvious interpretation is that the condition of the spine has nothing to do with headaches. If it had anything to do with it, even a little, there would be at least a modest sign of it here. This interpretation strongly confirms a bias against cervicogenic headache, and it was fun to watch some experts pounce on this and declare victory… prematurely.

The obvious interpretation isn’t the only interpretation. This is science! Uncertainty reigns supreme, and this study cannot “settle” anything. It’s valuable data in a legit, ongoing debate, but it's not the coup de grace for cervicogenic headaches.

There are two major limitations of this study, both clearly acknowledged by the authors…

Pedicle shape variation along the cervical spine. What matters here is that all the lines are basically the same for all measurements, in men and women with and without cervicogenic headache. Figure 3 from "A comparative shape analysis of the cervical spine between individuals with cervicogenic headaches and asymptomatic controls" by Masharawi, Mansour, Peled, and Weisman, unchanged, CC BY 4.0.

Limitation one: bone shape versus everything else

The first and worst limitation is that Masharawi et al studied only bone shape and relative position. Hard tissue, not soft. (Although you can infer a lot about intervertebral discs from the vertebrae.) As much as they measured, and as complex as the data crunching was, it falls far short of all the possible spinal abnormalities that could conceivably be linked to headache. There are many other potential structural abnormalities that wouldn't have shown up here, or not ones that are easy to spot. For instance, as reader Dr. Brian James pointed out, who knows what subtle patterns AI might be able to spot in the same data, or even more of it.

And cervicogenic headache might have very little to do with the spinal bones at all, no matter how thoroughly measured or cleverly analyzed. There are many more physiological factors that determine pain. We already know from a lot of other research that the structure of the spine correlates only loosely with chronic pain. This is a notorious truth in this field.

And that’s why, despite being a card-carrying member of Team Cervicogenic Headache, I would have predicted this result. I never believed in the first place that there would be a relatively simple link between headache and spinal bone shape.

Especially in younger people.

Limitation two: Twentysomethings and thirtysomethings

Masharawi et al also studied just forty relatively young people, aged 20–40, averaging almost exactly 30. It’s not a major problem, because they chose their subjects so carefully — these weren’t just any forty younger people with headaches, but the ones with exactly the kind of headache that might be explained by wonky spinal bones.

But it’s still a weakness, because there are more types and causes of headache than there are people in this study. A similar study with more and older subjects might well have had different results. For instance, it is plausible that tissues get stressed years before there is any visible effect on the bones — and even then it might correlate only loosely with it, because there are so many other factors involved.

Given all the variables, your pain-free study subjects will almost certainly be polluted with a few neck abnormalities, and vice versa — and yet, on average, the ones with abnormalities might still be only a little more likely to have pain. You could easily miss such a weak signal in two hundred people in their forties. You might need to look at five hundred people in their fifties.

We actually have low back pain data that tells a story just like this: Brinjikji et al studied a much larger number of older subjects (arguably the better approach) and found that pain correlates poorly with spinal structure … but it does correlate. Correlations don’t have to be strong to matter. Neck pain probably works the same way: visible issues aren’t the whole story, but they are probably part of it!

I am shocked — shocked, I say — that younger people with headaches have normal necks!

I think it’s terrific that we have this data, and it’s a useful and clear piece of the puzzle, and it’s definitely interesting to know that many people can have classic signs and symptoms of neck-powered headaches without having the slightest oddity in their spines.

But, as good as that puzzle piece is, it is also a bit quelle surprise, and not remotely a deal-breaker for the broader idea that some headaches are probably born in the neck, driven by cervical nociception (tissue insult detection) … and that phenomenon should be expected to correlate poorly with spinal bone shape, especially in younger people. There are plenty of ways that necks can cause headaches without showing up on 40 CT scans, no matter how painstakingly measured or cleverly analyzed.

In a way, this was like those studies that confirm, yet again, that Einstein had it right: clever, technical, and important confirmation … of something we already knew.