MUMultiple speciesBmoderateT1hypothesis only
Among 100 jumping horses in active competition, not one had a normal back on radiographs: what a kissing spines finding says, and what it cannot predict.
Executive summary
A kissing spines finding describes close, touching or overlapping spinous processes on a radiograph. It is common in horses without back trouble: of 33 riding horses with no clinical signs, only seven were normal; of 100 jumping horses in competition, none. A higher grade is associated with back pain in horses presented for it, but the measurement depends on posture, and in one surgical series no imaging parameter predicted the outcome.
Key points
Along the back of every horse, the spinous processes of the thoracic and lumbar vertebrae rise like a row of bony sails. Kissing spines is the term used when neighbouring processes come very close, touch or overlap. The usual threshold is a space of less than four millimetres, with contact or overlap. The middle and caudal thoracic spine is most often affected, the lumbar region more rarely. The term therefore describes a shape of bone, not a state of the horse.
A 2025 review sums up the state of knowledge with unusual candour: how often pain really does arise from touching spinous processes remains speculative, no reliable agreement between the imaging finding and clinical signs could be shown, and the way the condition arises is poorly understood despite decades of treatments on offer. This is a narrative rather than a systematic review: it orders what is known, it does not measure it.
No. The finding describes what can be seen on the radiograph. Whether a horse is in pain, and whether that pain comes from the spinous processes, is settled by the clinical examination, as a rule with diagnostic analgesia. Image and complaints are two separate pieces of information that then have to be brought together.
A Swedish study radiographed and scintigraphed 33 riding horses with no clinical signs of back problems. Only seven were entirely normal on both examinations, and eleven had a normal space of more than four millimetres throughout. Twenty-six showed at least one abnormality, usually mild and usually between the thirteenth and eighteenth thoracic vertebrae. The conclusion is blunt: the range of what is seen in horses without complaints is so wide that findings within that range cannot be read as clinically meaningful in horses that are unwell.
Twenty years later the same question was put to an entirely different population: 100 Warmblood jumping horses in active competition, average fence height 1.35 metres, with no observed lameness. Not one of these horses had a radiographically normal back. Narrowing and sclerosis lay mainly between the fifteenth thoracic and the first lumbar vertebra, and changes at the tendon attachments mainly in the region of the withers.
The most unexpected result of that work concerns the direction of the relationship: the highest career level ever reached was weakly but positively associated with narrowing of the spaces and with sclerosis. This must not be read as an advantage. The horses examined are those that lasted in the sport; those that failed do not appear in the sample, and that alone can produce such an association.
Where the changes come from is an open question. One study compared 47 yearlings that had never been backed with 55 trained Thoroughbreds without perceived back pain: the frequency, number and score of the findings did not differ. In the yearlings alone, about a third of the spaces examined were narrowed or in contact, and more than half showed sclerosis, lucencies or remodelling. The authors read this as a pointer towards a developmental origin. That reading is compatible with the data; the mere similarity of two groups does not prove it.
The largest study on this question examined 582 horses presented for perceived back pain or loss of performance, all of which underwent diagnostic analgesia of the limbs, the back and the sacroiliac joints. Both the highest grade and the total grade of the spinous process changes were significantly associated with thoracolumbar pain, as was increased uptake on scintigraphy. Anyone who says the radiograph says nothing at all is contradicting this work.
Two qualifications from the same study weigh just as heavily. First, osteoarthritis of the small facet joints was more strongly associated with pain than the spinous processes alone, and the probability was highest when both were present together. Second, lameness of the fore or hind limbs and sacroiliac pain can mimic primary back pain, which is exactly why diagnostic analgesia remains indispensable. And because every horse had been referred with complaints, no figure for how common the finding is in the wider horse population follows from this work.
A French study compared 102 trotters with signs of back pain and 16 without. Ten of the 16 horses without complaints had changes, against 98 of the 102 symptomatic horses, though theirs were less severe and more localised. For simple contact between the spinous processes, however, the number of affected horses did not differ significantly between the groups: only the number of spaces involved and the severity did. The comparison group numbered 16 animals, which makes any conclusion about small differences uncertain.
A third study, in 223 Warmbloods, shows the same two-part pattern. Total grades on radiography and scintigraphy were higher in horses with thoracolumbar pain. The more finely calculated uptake ratios, by contrast, did not differ between horses with and without pain, and agreement between the scintigraphic methods and the radiographic finding was only slight.
A radiograph is not a measurement but a projection. In six spinal columns from post-mortem examination, 49 spaces were measured 274 times and compared with cross-sectional images. In the backs classified as kissing spines, the spaces appeared clearly narrower on the radiograph than on the sagittal reconstruction. The projection therefore exaggerates the very narrowing on which the finding rests, and it does so under laboratory conditions: in the standing horse the geometry is harder to control, not easier.
The horse's posture shifts the result as well. In seven horses with already diagnosed contact, the thoracic spine was flexed by hand and radiographed in both postures. The space increased at every level from the fifth to the eighteenth thoracic vertebra, at most by a median of 3.1 and 3.0 millimetres, at least by 0.7 millimetres. As a reminder: the threshold below which a space counts as narrow is four millimetres.
A second, independent study points the same way: in seven healthy horses, traction on the abdomen widened the spaces from the eleventh thoracic to the second lumbar vertebra. Both studies cover seven horses; they show a direction, not a correction factor.
| Condition changed | Effect on the measured width | Animals examined |
|---|---|---|
| Flexion of the thoracic spine by hand | Increase at every level from the fifth to the eighteenth thoracic vertebra, by a median of 0.7 to 3.1 millimetres | 7 horses with diagnosed contact |
| Traction on the abdomen with a wide strap | Widening from the eleventh thoracic to the second lumbar vertebra, with a single exception; amount not quantified | 7 healthy, sedated horses |
| Angle of the X-ray beam | Measured width altered, with no consistent pattern across the deviations | 6 spinal columns from post-mortem examination |
| Projection instead of a cross-sectional image | Measured clearly narrower on radiography than on the sagittal reconstruction | 6 columns classified as kissing spines |
Images are read by people, and those people agree to varying degrees. In 100 healthy horses, three experienced assessors graded radiographs of the thoracic and lumbar spine independently of one another. Agreement was good for individual parameters: the width of the space, isolated sclerosis of the upper margin, beak-like new bone at the cranial dorsal aspect and remodelling at the cranial ventral aspect. It was only moderate for the total number of abnormalities per back, which is precisely the summary on which a purchase decision ultimately rests.
The jumping horse study likewise reports only moderate to good agreement between four assessors. Both figures come from study conditions, with experienced readers and standardised images. In everyday practice, between premises with different equipment and different training, rather less agreement is to be expected.
There the radiographic finding supplies a description, not a prognosis. It says how the spinous processes look on that day, on that image. How reliably two specialists arrive at the same overall grade is only moderately documented, and none of the studies analysed here has tested whether this finding predicts the later course of a horse that is bought.
The strongest objection to a large role for the radiograph comes from surgery itself. In a series of 71 horses in which the ligament between the spinous processes had been divided, 56 could be followed for a median of 35 months. Fifty-one of those 56 returned to some level of performance, a little more than nine in ten; 27 of 51 reached their previous level or a higher one.
What matters is what this study did not find. None of the variables tested influenced the return to performance: neither the number of affected spaces, nor the radiographic grade, nor the scintigraphic grade. The examination used to justify the procedure did not predict its outcome. This is the only study analysed here that expressly tested imaging grades as predictors, and 56 horses followed up is few: a weak association could be lost in a group of that size.
A second series, of 102 animals operated on standing, shows what counted instead. Of 66 horses with long-term follow-up, 53 returned to their intended work with no residual clinical signs. The difference lay not in the severity of the back finding but in whether another orthopaedic problem was present before surgery: without one, 28 of 30 horses returned; with one, 25 of 36. Thirty-six of the 102 are missing from the long-term follow-up, which can make the figures look kinder than they are.
The closest thing to a proper test of effectiveness is an analysis of racing performance. 159 operated Thoroughbred racehorses were compared with control horses matched for age, sex, race type, trainer and period, and followed for at least twelve months. The operated horses improved more than their controls. What was measured was racing performance, not pain, and allocation was clinical rather than random. Eight horses developed unilateral wasting of the back muscles with nerve involvement after the procedure.
How much an imaging finding says about pain has been studied in human medicine for decades, and with study designs that equine medicine lacks. A systematic review of 33 studies covering 3,110 people without symptoms found signs of disc degeneration in 37 per cent of those in their twenties and in 96 per cent of those in their eighties. Bulges rose across the same age span from 30 to 84 per cent. Much of this, the authors conclude, belongs to normal ageing.
It does not follow that images mean nothing. A meta-analysis by the same group, covering 3,097 people up to 50 years of age, separated one sign from another: bulge, extrusion, certain signal changes and degeneration were markedly more common in people with low back pain than in controls without symptoms. Spinal stenosis, annular fissures and spondylolisthesis were not. The literature therefore does not say that the image is empty, but that the signs have to be examined one by one.
Two further studies concern the value of imaging itself. A meta-analysis of six randomised trials including 1,804 patients found that immediate imaging for low back pain without warning signs improved neither pain nor function. A systematic review of twelve follow-up studies found no consistent link between imaging findings and later low back pain; the predictive value is therefore unresolved rather than disproved.
One large randomised trial from health services research is particularly sobering. In four American health systems, imaging reports were supplemented with how often the finding in question occurs in people without symptoms; data from more than 238,000 adults were analysed. Use of health care services did not change measurably. Even supplying the ordinariness of a finding, then, barely shifts the decision.
The human counterpart to contact between the spinous processes even has a name of its own: Baastrup's disease. A 2022 systematic review of 35 studies and 1,308 patients describes it as a common and underdiagnosed cause of low back pain, whose recognition calls for imaging that includes functional views in flexion and extension. At the same time it records that the treatments applied rest on no firm basis and that randomised trials are lacking. On this one sign, then, human medicine is no further ahead than equine medicine.
Taken together, the picture that emerges is neither reassuring nor alarming: it is blurred. Touching spinous processes are common in horses without complaints, they are already present in young animals that have never been ridden, and the narrowing that is measured depends on how the image came about. At the same time, in the largest study, a higher grade goes together with back pain.
What remains useful in practice is the order of the questions. An image answers how the bone looks. Whether a horse is in pain and where that pain sits is answered by the clinical examination, by observation in movement and by diagnostic analgesia. Faced with a finding, most is gained by asking which of these steps were actually carried out, and what has changed since the horse last moved soundly.
Original analysis
| Group examined | What was measured | Central result | Study design | What does not follow from it |
|---|---|---|---|---|
| 33 riding horses with no back signs | Radiography and scintigraphy of the spinous processes | 7 entirely normal, 11 with a normal space throughout | Cross-sectional, with no symptomatic comparison group | That abnormalities are harmless in horses that are unwell: the comparison group was missing |
| 47 yearlings never backed against 55 trained Thoroughbreds | Radiography from the seventh thoracic to the third lumbar vertebra | No difference in frequency, number or score | Cross-sectional, a single breed | That training contributes nothing: equal values in two groups establish no cause |
| 100 jumping horses in active competition | Radiography, four independent assessors | No horse without abnormalities; career level weakly associated with narrowing | Cross-sectional, in horses that lasted in the sport | That narrowing benefits performance: horses that failed are missing from the sample |
| 102 trotters with and 16 without signs of back pain | Radiography from the fourteenth thoracic to the third lumbar vertebra | Changes in 10 of 16 horses without complaints and 98 of 102 symptomatic horses | Case-control study with a very small comparison group | That simple contact indicates pain: for that, the number of affected horses did not differ |
| 582 horses at a referral clinic | Radiography, scintigraphy, diagnostic analgesia | Radiographic grade significantly associated with thoracolumbar pain | Cross-sectional, in cases preselected by complaints | A frequency figure for the wider horse population |
| 6 spinal columns from post-mortem examination, 274 measurements | Radiography against cross-sectional imaging, different beam paths | Spaces measured clearly narrower on radiography than on cross-sectional images | Laboratory experiment under controlled conditions | That the discrepancy is just as large in the standing horse: there the geometry is harder to control |
| 7 horses with diagnosed contact, two postures | Width of the space in a neutral and a flexed posture | Increase at every level, by a median of up to 3.1 millimetres | Case series, each horse serving as its own comparison | That flexion changes pain or the course of the condition: a distance was measured |
| 100 healthy horses, three assessors | Agreement when grading | Good for individual parameters, only moderate for the total per back | Cross-sectional, under study conditions | That agreement is just as high in everyday practice |
| 56 operated horses followed up | Return to performance after division of the ligament | A little more than nine in ten to some level, a little more than half to the previous level; no imaging parameter predicted this | Retrospective case series with no unoperated comparison group | That the procedure is the cause of the improvement |
| 3,110 people without back pain | Magnetic resonance and cross-sectional images of the spine | Disc degeneration in 37 per cent of those in their twenties and 96 per cent of those in their eighties | Systematic review of 33 studies | A statement about spinous processes in the horse: another species, another structure, another technique |
The finding on its own says little about that. In a study of 33 riding horses with no back signs, only seven were entirely normal on radiography and scintigraphy, and among 100 jumping horses in active competition not one had a normal back. Abnormalities are therefore the rule in horses without complaints. At the same time, in the largest study, covering 582 horses presented for assessment, a higher grade went together with back pain. Both hold side by side, and the apparent contradiction is not resolved on the image but in the clinical examination of the individual horse.
That is settled neither by a radiograph nor by a study, but by veterinary examination of the individual horse. What the literature contributes is background: 100 jumping horses in active competition all had abnormalities of the spinous processes, and the highest career level reached was weakly but positively associated with narrowing. Those horses were working with the finding. No advantage can be read into that, because horses that failed in the sport are missing from such samples. Neither permission nor prohibition can be derived from the image.
There is no solid body of evidence on that, and this article gives no purchase recommendation. It is known that in 100 healthy horses three experienced assessors agreed well on individual parameters, but only moderately on the total number of abnormalities per back. Yet it is precisely that overall grade which carries a decision. It is also known that none of the studies analysed here has tested whether the finding predicts the later course of a horse that is bought. Anyone weighing it is weighing a description of bone, not a prognosis.
Because reading the image is less clear-cut than the word finding suggests. The width of a space is judged fairly consistently between experienced readers, whereas the overall assessment of a back is only moderately consistent. Added to that is the exposure itself: the angle of the beam alters the measured width, radiography measures narrower than a cross-sectional image, and flexion induced by hand widened the space by more than three millimetres in seven horses, against a threshold of four millimetres. Two images of the same horse are therefore not automatically comparable.
The evidence for that is missing, and one pointer argues against it. A study compared 47 yearlings that had never been backed with 55 trained Thoroughbreds without perceived back pain and found no difference in the frequency, number or score of the findings. In the yearlings alone, about a third of the spaces were narrowed or in contact. The authors read this as a pointer towards a developmental origin. That is not proof: equal values in two groups do not rule out an influence of work with any certainty. On the role of the saddle this study says nothing.
Yes, under the name Baastrup's disease, and the evidence there is no better. A 2022 systematic review of 35 studies and 1,308 patients describes contact between the spinous processes as a common and underdiagnosed cause of low back pain. At the same time it records that the treatments on offer rest on no firm basis and that randomised trials are lacking. The mean age was 59.6 years, which has no counterpart in the sport horse. What transfers is the way of thinking, not the figure.
Terms defined in this text
Equine Health
The percentage ranges attached to radiographic classes come from a veterinary consensus, not a follow-up study: what the data actually show.
HOHorseBmoderate Open access
Equine Health
In 80 dressage horses referred for investigation, reported signs did not differ between horses with and without a lesion. What symptoms are really wor
HOHorseBmoderate Open access
Equine Health
In 392 Quarter Horses, 57 per cent of the healthy horses carried a P variant: what the two assessments of P2, P3 and P4 actually found.
HOHorseBmoderate Open access
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.