The Colorado Studies

The claim

The Colorado studies put a nerve root on a bench and measured what pressure does to it

Between 1973 and 1986, a laboratory at the University of Colorado in Boulder did something nobody in the manual therapies had managed before. It took the spinal nerve root, loaded it, and recorded what happened. The engineer Chung Ha Suh built the geometry. The neurophysiologist Seth Sharpless ran the electrophysiology. The neuroscientist Marvin Luttges carried the work forward for another decade after both of them had made their point. The headline result, published in 1975 in the federal monograph The Research Status of Spinal Manipulative Therapy, was that spinal roots lose conduction under pressures so small that no clinician would ever call them an injury.

Ask what that commits you to. If a nerve root can be quieted by a pressure of five to ten millimeters of mercury, then the interesting variable is not damage. It is load. Damage is a threshold you cross once. Load is a condition a tissue lives inside, hour after hour, and it can be raised or lowered without anything breaking and without anything showing on a film. That distinction carries the whole argument of this page. Colorado did not find a new injury. It found that function changes long before structure does.

The money

The first money was chiropractic money, not federal money

The Colorado program is routinely described as the first federally funded chiropractic research. That is not what happened, and the real sequence is more interesting. In 1973 the Foundation for the Advancement of Chiropractic Tenets and Science, the research arm of the International Chiropractors Association, paid to put Suh to work inside a state university. Scott Haldeman, writing in the Journal of Chiropractic Humanities in 2015, described it as the first chiropractic-sponsored research program into the theories on which chiropractic was based, carried out by scientists who were not members of the profession (Haldeman 2015).

Federal money arrived on a separate track. Congress set aside roughly two million dollars and asked the National Institute of Neurological and Communicative Disorders and Stroke to examine spinal manipulation directly. The result was a workshop at the National Institutes of Health in Bethesda, Maryland, in February 1975, and a 310 page volume edited by Murray Goldstein and issued through the Department of Health, Education, and Welfare (Goldstein 1975). Sharpless contributed the chapter Susceptibility of Spinal Roots to Compression Block. So the profession did not win a federal grant. It won a federal audience, which is harder to get and worth considerably more.

The engineer

Chung Ha Suh turned the vertebra into coordinates before anyone could test it as tissue

Suh was a mechanical engineer, and he treated the spine the way an engineer treats any linkage. His 1974 paper in the Journal of Biomechanics, volume 7, pages 161 to 169, set out the fundamentals of computer aided X-ray analysis of the spine (Suh 1974). It addressed the geometric distortion built into every radiograph, the problem of locating a vertebra in three dimensions from a two dimensional shadow, and the description of one segment moving against its neighbor. Before Suh, spinal position was a clinical opinion delivered with confidence. After Suh, it was a coordinate set with an error bar attached.

This matters more than it sounds. A field cannot measure what it cannot describe, and description is what turns an argument into an experiment. Suh also gave the field a habit it badly lacked, which was to publish where your critics read. The 1974 paper ran in an engineering journal with no stake in the outcome rather than in a house organ. The people gathered around him at Boulder were engineers and neuroscientists, not clinicians defending a position, and Haldeman is explicit that they were not members of the profession whose theories they were testing. Independence was the whole point. A result you cannot be accused of buying is worth more than a result you can.

The neurophysiologist

Sharpless arrived already an authority on what happens when input is reduced

Seth Sharpless was not recruited as a friendly voice. He was a University of Colorado neuroscientist with a serious record on one specific question: what does a nervous system do when its normal input is withdrawn? His 1964 review in Annual Review of Physiology, volume 26, pages 357 to 388, titled Reorganization of Function in the Nervous System, Use and Disuse, is a founding survey of denervation supersensitivity (Sharpless 1964). He returned to the theme in 1975 in Federation Proceedings, volume 34, pages 1990 to 1997, writing on supersensitivity-like phenomena in the central nervous system. He had also worked on differential nerve block by cooling, published in the Journal of Physiology in 1972 (Byck 1972).

Ask why that background matters here. Sharpless already knew that reducing input does not simply subtract function. It changes the gain of everything downstream. Tissue deprived of its normal traffic becomes more responsive, not less, and the response it gives back is scaled differently than before. So when he put pressure on a spinal root, he was not asking whether a nerve could be broken. He was asking a subtler question. How little pressure does it take to change what the system downstream expects? That is a question about state, and state is the vocabulary this entire history runs on.

The measurement

Spinal roots block at pressures a peripheral nerve would shrug off

The 1975 chapter reported a clean comparison. Peripheral nerves tolerate compression reasonably well. Spinal roots do not. Sharpless applied light, sustained pressure to roots in anesthetized animals and watched the compound action potential fall away. The pressures were tiny. In the recorded discussion at the Research Workshop on Neurobiologic Mechanisms in Manipulative Therapy, held at Michigan State University in October 1977, the figure quoted back to him was five to ten millimeters of mercury. For scale, that sits below the pressure inside a resting vein, and well below anything a hand could apply and still feel it as contact.

Now be precise about what is solid and what is not. The pressure range is well attested and repeats consistently across the textbook literature that followed. The percentage decrement does not. Later authorities restate Sharpless as producing a fifty percent, a sixty percent, or a seventy five percent reduction in the recorded potential. Scott Haldeman gives seventy five percent in Principles and Practice of Chiropractic. Others give half. This page will not choose one figure and present it as the number, because the original results varied with the preparation, with which root was loaded, and with how long the load stayed on. A range honestly stated is more useful than a decimal invented for confidence.

The anatomy

Roots are fragile because they are barely armored

Why would a root fail where a nerve holds? Anatomy answers it. A peripheral nerve is wrapped in perineurium, a dense sheath that carries load and defends an internal chemical environment. Spinal nerve roots have no perineurium. They travel through the subarachnoid space and the intervertebral foramen with far less connective tissue around them, which is why Sunderland and Bradley had already reported in 1961 that roots yield to tensile stress more readily than nerves do (Sunderland and Bradley 1961). Colorado put numbers on that intuition, and the numbers are stark enough to end the argument.

In 1986 Beel, Stodieck and Luttges published two companion papers in Experimental Neurology, volume 91: one on the biomechanics of spinal nerve roots at pages 30 to 40 (Beel 1986), one on their protein composition at pages 41 to 51 (Stodieck 1986). They elongated mouse roots to mechanical failure and calculated proportional limit stress, strain, and apparent modulus from force records and photographs taken through the test. Their own summary called the resulting data indicative of an extremely weak material. The comparison with peripheral nerve was not close, and the reason they gave was the sheer difference in how much connective tissue each structure carries.

The model

MacGregor, Sharpless and Luttges explained the block with a pressure vessel, not a pipe

A measurement without a mechanism is a curiosity. In March 1975, the same year as the monograph, the Colorado group published a mechanism in the Journal of the Neurological Sciences, volume 24, pages 299 to 304 (MacGregor 1975). Ronald MacGregor, Seth Sharpless and Marvin Luttges argued that a nerve fiber under compression behaves less like a rigid walled tube and more like a pressure vessel. The fiber resists deformation because displaced fluid inside it pushes outward against an elastic membrane, and the membrane, rather than any rigid wall, sets the limit on how far it can go.

Follow the consequence. In a pressure vessel, wall stress scales with vessel size, so MacGregor's analysis predicts that percentage deformation rises with fiber diameter. Large fibers deform proportionally more than small ones under the same external load. That is a testable claim, and it explains something clinicians meet constantly. Compression does not silence a nerve evenly. It reorders it, taking the large fast fibers first and leaving the small slow ones running. What arrives centrally is not less signal. It is differently weighted signal, and a system tuned to the old weighting will read the new one as something wrong.

After the root

Luttges found that injury changes tissue that was never injured

The most under-quoted Colorado result is not about pressure at all. In 1979 Gerren and Luttges published in Experimental Neurology, volume 65, pages 587 to 607, under a title that gives away the whole finding: functional changes in undamaged sciatic nerves and spinal cord of mice following nerve damage (Gerren and Luttges 1979). Injure one nerve and untouched nerve elsewhere changes. So does the cord. A 1976 paper in the same journal, volume 50, pages 706 to 733, had already tracked degenerative changes in mouse sciatic nerves by electrophoretic and electrophysiologic methods together, which is how they could see chemistry and signal move in step.

The line ran on for a decade. In 1982 Triano and Luttges published a model of sciatic neuritis in Spine, volume 7, pages 129 to 136 (Triano and Luttges 1982), built on chronic irritation rather than crushing. In 1984 Beel, Groswald and Luttges measured how crush injury alters the mechanical properties of peripheral nerve, in the Journal of Biomechanics, volume 17, pages 185 to 193. In 1986 Luttges, Stodieck and Beel reported postinjury changes in the biomechanics of nerves and roots in mice. Read together, these papers say something Colorado never reduced to a slogan. A local mechanical event produces a distributed functional change.

The limits

Later work with tighter controls did not reproduce the lowest thresholds

Honesty about a favorable result is the only thing that makes it worth citing. In 1992 Pedowitz, Garfin, Massie, Hargens, Swenson, Myers and Rydevik published a careful replication in Spine, volume 17, pages 194 to 199 (Pedowitz 1992). They compressed the pig cauda equina at 0, 50, 100 and 200 millimeters of mercury for two or four hours, monitored efferent compound motor action potentials and afferent compound nerve action potentials together, and followed recovery for a further ninety minutes afterward. What they found does not match the popular retelling of Sharpless, and the popular retelling is wrong.

So the threshold is not a constant of nature. It moves with species, with preparation, with which root is loaded, with how long the load lasts, and with whether you are recording motor or sensory traffic. Anyone quoting ten millimeters of mercury as a fixed biological law is quoting a slogan rather than a result. What survives replication is the direction and the ordering. Roots are more vulnerable than nerves. Sensory and motor fibers do not fail together. Duration counts as heavily as magnitude, which is the single most useful sentence in the whole literature.

Circulation first

Pressure takes the blood supply before it takes the signal

There is a mechanism underneath the mechanism, and Rydevik found it. In 1981, in the Journal of Hand Surgery, volume 6, pages 3 to 12 (Rydevik 1981), Rydevik, Lundborg and Bagge watched intraneural microcirculation directly under vital microscopy while a purpose built minicompression device squeezed a rabbit tibial nerve. They graded the pressure and recorded which vessels stopped first. The ordering they found is the reason the low numbers elsewhere in this literature are not absurd, because a vessel gives out long before an axon does.

Venules fail before arterioles. Arterioles and intrafascicular capillaries fail before the nerve stops conducting. At 60 to 80 millimeters of mercury no flow could be observed at all, and nerves compressed at 400 millimeters of mercury for two hours still showed stagnant flow three and seven days later. Congestion comes first, then ischemia, then conduction loss. A pressure far too small to injure a nerve mechanically can still change what that nerve does over hours, because the nerve is not being crushed. It is being starved, slowly, in a way no imaging study will show you.

Colorado and the model

The right tissue, measured with the trial grammar of its era

Colorado picked the correct tissue and could not yet run the correct trial. That is the verdict, and it is not a criticism of Suh or Sharpless. The design available to them, and to nearly everyone since, delivers one predetermined input to every subject and reports the average. The Unified Model of Tone holds that outcome depends on the correspondence between an input and each individual constraint structure. An unstratified trial therefore averages a well matched intervention and a mismatched one. Responders and non responders collapse into a mean that describes neither, and a genuinely large effect reads as weak.

The model states this as a prediction rather than a complaint, and the prediction is specific. Stratify a sample by a tone measure recorded before any input is given. Specify each person's leverage point from that measure. Then randomize between an input delivered there and the identical input delivered to a site the measure did not select. The model predicts a substantially larger effect in the matched arm and a modest one in the mismatched arm, and it predicts that pooling the two reproduces the small average the literature keeps reporting.

State the confirming finding with the same clarity. The separation between the two arms confirms the prediction, and confirms with it the claim that correspondence rather than force is the active ingredient. The discipline that keeps this honest is fixing the site in advance. A leverage point identified after the result is known explains everything and forecasts nothing, and the model does not claim that privilege. A site written down first is a forecast. A site chosen afterward is only a story.

The measurement belongs to Colorado. The pressures of 5 to 10 mm Hg are theirs, as are the pressure vessel model of 1975 and the root strength at 10 percent of peripheral nerve in 1986. So is the null result at 50 mm Hg that Pedowitz reported in 1992. The extension is the model. What Colorado established is that the tissue changes state under loads far beneath injury, so the load is a variable rather than an accident. What the model adds is where to put the load and how to know in advance. The field does not lack evidence so much as it lacks evidence organized around the correct variable.

The dime

The most repeated version of this research is the one nobody measured

Somewhere between the monograph and the seminar room, ten millimeters of mercury became the weight of a dime. The claim circulates in a dozen forms, usually as a coin resting on a nerve and cutting its function by sixty percent. That is not a research finding. It is a unit error. Pressure is force divided by area, and nobody can convert the weight of a coin into millimeters of mercury without stating a contact area, which the story never does. Sharpless worked on exposed roots in anesthetized animals under a calibrated device. No dime was ever involved.

Correcting it costs nothing, because the real result is stronger than the slogan. The durable contribution of the Colorado studies to the story of tone is a category rather than a number: the nervous system changes state under loads far below the threshold of injury. That is what a regulated system looks like, as opposed to a machine with a breaking point. Suh gave the field coordinates. Sharpless gave it a threshold worth arguing about. Luttges showed the argument does not stay local. The next chapter of this history, on Prigogine and self organization, explains why systems held far from equilibrium behave exactly this way.

What the record shows

Twenty years of Colorado data, and the numbers worth carrying forward

• 1973. The Foundation for the Advancement of Chiropractic Tenets and Science paid to put Chung Ha Suh to work inside the University of Colorado at Boulder. Haldeman in 2015 called it the first chiropractic sponsored research program into the theories on which chiropractic was based, carried out by scientists from outside the profession (Haldeman 2015).
• 1974. Suh set out the fundamentals of computer aided X ray analysis of the spine in the Journal of Biomechanics, volume 7, pages 161 to 169 (Suh 1974). The paper replaced clinical opinion about segmental position with a coordinate set carrying an error bar.
• 1975. Sharpless contributed Susceptibility of Spinal Roots to Compression Block to the 310 page monograph edited by Murray Goldstein, reporting conduction loss in spinal roots at roughly 5 to 10 mm Hg (Goldstein 1975). Later authors quote the decrement at fifty, sixty and seventy five percent, so the range is the honest citation.
• 1975. MacGregor, Sharpless and Luttges described the compressed fiber as a pressure vessel in the Journal of the Neurological Sciences, volume 24, pages 299 to 304 (MacGregor 1975). MacGregor's analysis predicts that percentage deformation rises with fiber diameter, so large fast fibers fail first.
• 1981. Rydevik, Lundborg and Bagge recorded venular flow impaired at 20 to 30 mm Hg and arteriolar flow impaired at about 40 to 50 mm Hg. No flow at all could be observed at 60 to 80 mm Hg (Rydevik 1981).
• 1986. Beel, Stodieck and Luttges measured nerve root strength at 10 percent of peripheral nerve and root stiffness at 20 percent. Their companion papers ran in Experimental Neurology, volume 91, pages 30 to 40 (Beel 1986).
• 1992. Pedowitz and colleagues compressed the pig cauda equina at 0, 50, 100 and 200 mm Hg for two or four hours. They found no significant conduction deficit at 50 mm Hg, and clear deficits at 100 and 200 mm Hg (Pedowitz 1992).