4 Idiopathic Scoliosis
4.1 Epidemiology of Late-Onset Idiopathic Scoliosis
4.1.1 Scoliosis in the Community
Those of us who were young orthopaedic surgeons in the 1960s and 1970s will remember their first steps into the minefield of learning about spinal deformities and of course in particular idiopathic scoliosis. Our surgical forefathers had laid down fairly strict rules based on their very narrow and hitherto unchallenged views about management and very importantly what they were based upon. 1, 2 Fortunately we all helped change the situation over the years. Even then we knew that only very few small curves progressed 3 and therefore there was nothing wrong with the observation for these. If however the curve got much beyond, say, 20 degrees, then the Milwaukee brace was empirically prescribed. 1 Moe led the way by recommending that curves of 25 degrees or more should be braced and said they indeed “responded well to bracing.” This then produced the epidemic of school screening programs. Meanwhile if curves did progress they weren’t allowed to get more than 60 degrees because of literature mainly from Scandinavia to the effect that if the curve went much beyond that figure patients risked cardiopulmonary dysfunction which, if severely progressive, could jeopardize life as well as quality thereof 4– 7 (▶ Table 4.1).
Under 25 degrees
60 degrees plus
Thus, the notion about bracing was to try and prevent curves that had progressed beyond 25 degrees from getting up to or beyond 60 degrees. So these were the rules and they were applied quite rigidly. If surgery was recommended, then in the pre-Harrington era localizer casts were applied under traction to obtain the best correction and then a window was cut out in the back of the cast through which spinal fusion was performed. The cast was then worn for at least 3 if not 6 months! With the advent of Harrington instrumentation 8 this mitigated all the previous fuss (traction, casts, windows, etc.) before carrying out spinal fusion. This was the “Harrington revolution.” Although it was known that not all curves of 25 degrees or more progressed 3 (four-fifths do not), pioneers of brace treatment dictated that bracing should be worn for up to 23 hours a day because: (1) faith in brace treatment was unquestionable and (2) allowing progression to 60 degrees or more would possibly endanger their patients’ lives. That was the perceived wisdom of the day and was effectively unchallengable. 9, 10 If death or potential heart failure might be the outcome then, not surprisingly, both providers and recipients of health care happily endorsed this treatment program.
If there really was such a serious outcome and patients presented de novo clinically with curves of 30 or 40 degrees then it seemed quite reasonable to identify less severe curves in the community. This was the basis for the introduction of school screening programs, primarily for health reasons but fortunately, as an aside, we began to learn something about prevalence rates and progression potential of the condition of idiopathic scoliosis. At the beginning of school screening these epidemiological matters were relegated for what seemed perfectly reasonable health concerns. 11– 14 Belief in bracing was so strong that it would have been quite unethical to conduct a controlled trial. Quite frankly the dictators of the time simply would not have condoned such a thing. However, over the past few decades, these rules have been challenged to the point where it is difficult to believe that we carried on with these beliefs for as long as we did. Yet there was clear evidence in the literature to go against the notion that healthy young adolescents with a bendy back could die of cardiopulmonary dysfunction.
Back in the 1960s there were several publications about the long-term follow-up of patients with untreated scoliosis. In Sweden, Nachemson 4 and Nilssonne and Lundgren 5 and in America, Collis and Ponsetti 6 reported on long-term prognosis. Nilssonne and Lundgren traced 102 patients with curve magnitudes greatly in excess of that which is normally associated with late-onset idiopathic scoliosis (LOIS) and reported that almost two-thirds of deaths were attributed to cardiopulmonary disease and that they had been disadvantaged in other ways throughout life: 76% were unmarried, 90% had back symptoms, 30% were on a disability pension, and 17% were disabled. The same can be said for Nachemson’s study reporting severe cardiac and pulmonary disease and deaths. 4 Then Collis and Ponsetti reported that two-thirds of patients with thoracic curves greater than 60 degrees had diminished vital capacity. 6 Bengtsson then looked in detail at psychological and psychiatric considerations in untreated scoliosis and found much higher rates of divorce, lower marriage rates, and higher suicide rates. 7 If at the age of, say, 10 or 11, you have a curve magnitude of 130 degrees or so, and these sort of curve magnitudes in these ages typified these investigations, then the only conclusion that can be drawn is that these cases were of early-onset idiopathic scoliosis (EOIS) (or indeed congenital) and not those of late onset. Meanwhile personal experience of patients with LOIS clearly demonstrated that pulmonary function is not abnormal even if curve magnitude is well in excess of 60 degrees; the need to routinely assess cardiopulmonary function before surgery finding invariable normality in our late-onset cases. In Oxford in 1970s we wanted to give up routine cardiopulmonary testing preoperatively for teenagers about to undergo scoliosis surgery. This was regarded as being too drastic a move. Where were these so-called cardiopulmonary compromised 30 and 40 year olds? They were supposed to be filling intensive care units but of course they weren’t. They were out enjoying life! Then, more recently, Pehrsson et al reported another long-term study of mortality recognizing that organic health problems of idiopathic scoliosis belonged to a much younger age of onset whereas no patient with adolescent onset scoliosis died from respiratory failure. 15
Why then this marked discrepancy? The answer was really evident and had already been described in detail in the 1965 Philip Zorab Meeting in London by the distinguished pulmonary pathologist Lynn Reid. 16 Her observations were then published in more detail in 1971. 17 She carried out postmortem examinations on children who had perished from cardiopulmonary compromise at the Brompton Hospital in London. She found very hypoplastic lungs resembling those that occur in infants with congenital diaphragmatic hernia in which the abdominal contents severely compress lung space. It was these infantile “malignant” idiopathic progressive curves (as they were called at that time) that were the main problem in producing severe cardiopulmonary compromise and fatality. Davies and Reid showed that pulmonary alveolar reduplication occurred in the first 2 or 3 years of life and certainly ended by the age of 8 17 (see ▶ Fig. 2.26). Very significant thoracic deformities can therefore cramp the developing lungs preventing further reduplication. Then Margaret Branthwaite who was a consultant chest physician at the Brompton Hospital at that time, later a distinguished barrister, published her findings about cardiorespiratory consequences of untreated idiopathic scoliosis. 18 She found that curves that started before the age of 5 years, and generally in infancy, had the bad cardiopulmonary reputation whereas curves that started beyond the age of 5 had no such cardiorespiratory concerns in the future.
As a result of being aware of these important findings it was perfectly obvious to us in Leeds why there were such bad reports about idiopathic scoliosis in the 1960s—simply because age of onset was not being considered. It was early-onset very progressive scoliosis (idiopathic or congenital) that had the bad reputation for organic health while LOIS had no such health consequences. So we thought it very important to regard idiopathic scoliosis as being either EOIS (before the age of 5) or LOIS (after the age of 5) 19 (▶ Fig. 2.25) and therefore we should reclassify accordingly. Notwithstanding, screening for scoliosis in adolescents was still championed well into 1970s, 1980s, and even 1990s. The process seemed unstoppable despite the strong retrospective evidence base against it.
What about bracing? We shall look at the efficacy (or rather lack of efficacy) later, but there is no evidence in favor of this practice now and there never was. However, screening for scoliosis did produce a valuable by-product of trying to learn more about the natural history although we do not know much more today than we knew 20 years ago. That might be because of the diminishing school screening programs but although the loss is to epidemiology the gain is that hordes of quite normal children no longer go through this harmful and unethical process.
4.2 Screening for Scoliosis
4.2.1 Definitions and Criteria
If you are going to go screening there are certain niceties to be observed. You cannot just go swanning into Marks and Spencer on a Saturday morning and start measuring blood pressure! For decades there have been strict criteria for screening. “Screening is the presumptive identification of an unrecognised disease or defect through the application of tests, examinations or other procedures that can be applied rapidly.” 20 The World Health Organization (WHO) has defined several criteria that should be met before informing an unwitting individual that he or she has a problem and there are 10 of them 21 (▶ Table 4.2).
As regards scoliosis there are a number of prerequisites and so let’s pick out a few from the WHO’s list. First, the condition should be an important health problem for the individual and community. Certainly that was thought to be the case in the late 1960s and early 1970s when it spread through the scoliosis community that children might die from untreated idiopathic scoliosis. We have known all about hypoplastic infantile idiopathic lungs for decades but we knew that far too long before screening faded out. Second, there should be an accepted or useful intervention for patients with the disease. School screening detected droves of children with minor curves, less than the Scoliosis Research Society’s definition of scoliosis which should be a curve of 11 degrees plus with concordant rotation (vertebral bodies turning into the curve convexity). 2 Neither is there any accepted treatment for these very small curves nor do such individuals need treatment. They are not cases of scoliosis and are better termed “schooliosis” (irrelevant curves detected in schools).
There should be a latent to early symptomatic stage of the disease and of course for LOIS there are no symptoms other than deformity. Then there should be a suitable and sensitive screening test for examination, both forward bending tests (▶ Fig. 4.1) and scoliometer readings 22 leading to far too many curves detected that are too small to be called scoliosis. In Leeds we have used the Quantec Surface Shape Measurement system (▶ Fig. 4.2) for a lot of our epidemiological and natural history work, and while this records more than 250,000 data points on surface shape in a fraction of a second it does not tell you how deformed an individual is on, say, a 0 to 100 scale basis. 23 Moving on to age criteria, the rules say that treatment at an early stage should be of more benefit than treatment started later but unfortunately we have no adequate early stage or conservative treatment.
Fig. 4.1 The forward-bend test. (a) A true lateral radiograph showing the essential lordosis. (b) On forward bending this lordosis is compressed and the spine therefore buckles, enhancing the rib hump (see ▶ Fig. 3.9). (Reproduced with permission from Newton P, O’Brien M, Shufflebarger H, et al. Idiopathic Scoliosis: The Harms Study Group Treatment Guide. Stuttgart/New York: Thieme; 2010: 53.)
Fig. 4.2 A patient with a right thoracic idiopathic scoliosis whose surface shape has been registered by the Quantec system.
Then while the question of costs may not be important for fully privately funded health services, it is very important, for example, in the UK where health service costs are most definitely finite. Screening for scoliosis would then compete economically with, for instance, cervical cancer screening and would fail dismally.
Children harvested then had the deformity corroborated on clinical examination and many then underwent X-ray examination of the spine. That you had scoliosis (whatever that was to the child and family) was then permanently imprinted on your medical records, rather like a criminal record. Even chiropractors were getting in on the act. R.A.D. well remembers John Hall in Boston telling him in the late 1970s in a rather facetious manner that he spent more time telling young children and parents that they didn’t have scoliosis than actually treating those that did (as a result of these nonsensical screening programs). Mercifully the U.S. Preventive Services Taskforce eventually recommended eliminating school screening for scoliosis. 24 (Hopefully it has gone for good and we shall never see its like again.)
4.2.2 Natural History
Although nearly all epidemiology surveys have involved female schoolchildren between the ages of 10 and 14, the recent MRC study in Leeds of 16,000 schoolchildren of both genders chose an age group from 6 to 14 so as to encompass all the years of LOIS, 25 the 10-year-olds reaching maturity by the end of follow-up. All of 2.2% of girls aged 12 to 14 years had idiopathic scoliosis, a curve of 11 degrees or more with concordant rotation (▶ Table 4.3). With increasing curve magnitude, thoracic curves and females became greatly over-represented with 70% of curves in excess of 15 degrees (▶ Table 4.4 and ▶ Table 4.5).
Age group (years)
No. of patients
Prevalence in girls (%)
Prevalence in boys (%)
Size of curve
Source: Reproduced with permission from Weinstein SL. Adolescent idiopathic scoliosis: Prevalence and natural history. In Weinstein SL (ed) The Pediatric Spine. Principles and Practice. 1994. Copyright © by Lippincott-Raven.
Comparing the results of the Oxford study 26 with those of the Leeds study almost 20 years later, it would appear that LOIS is perhaps pursuing a more benign course with the passage of time (2.5% of teenage girls in Oxford versus 2.2% in Leeds).
Then we know that bigger curves do tend to progress more than smaller curves and this would accord with Euler’s laws of flexible columns as well as common sense (e.g., the Leaning Tower of Pisa—the more it leans the more it will carry on doing so). In addition, some useful information about other curve patterns is helpful. Right-sided thoracic curves and left-sided lumbar curves are more likely to progress, as is well known, but why? The answer lies in the transverse plane geometry of the thoracic and lumbar vertebrae. Thoracic vertebrae are asymmetric to the right in the transverse plane thus favoring a right thoracic scoliosis (due to the constant pressure of the descending thoracic aorta 27, 28) (▶ Fig. 3.2 and ▶ Fig. 3.3) but there is also an inbuilt coronal plane deformity in “normal children” with curvatures in the right and left directions being equally represented for small curves (<10 degrees). If therefore the preexisting normal coronal plane deformity is to the left it will counter the adverse effect of right-sided transverse plane asymmetry with the two effectively canceling each other out. If, however, there is a preexisting right-sided coronal plane deformity, then right-sided transverse plane asymmetry may give momentum to a preexisting thoracic lordosis. Just in case you’re wondering, the inbuilt transverse plane deformity is in the opposite direction in situs inversus 29 (see ▶ Fig. 3.3), which more than adequately confirms the Leeds theory about curve directions. 30, 31
A lot of natural history information comes from screening programs although nearly all harvested are very small curves. Of 134 patients detected by Brooks only 5% increased and 22% decreased. 11 Rogala detected 603 and only 7% progressed, with 2% progression in those less than 10 degrees and 10% progression in those that were greater. A total of 52 immature patients whose curves were between 20 and 30 degrees progressed. 14 In 1984 Lonstein, a key figure in the epidemiology of idiopathic scoliosis, published a very important natural history paper. 32 A total of 727 patients with LOIS curves measuring from 5 to 29 degrees were reviewed having been followed either to the end of skeletal growth or curve progression: 23% progressed while 11% showed improvement in the curve of 5 degrees or more. Risser 0 and curves of less than 15 degrees favored improvement while curve progression was related to bigger curve magnitudes, a more advanced Risser sign, and greater chronological age. Double curves progressed more than single (27% versus 18%) and single thoracic curves were the most progressive.
Much emphasis was placed on the Risser sign (see ▶ Fig. 2.23) and indeed the Risser sign is very widely used in the clinical management of patients with LOIS. 33 He noted on posteroanterior (PA) X-rays of the pelvis that the iliac apophysis appeared laterally and anteriorly (capping) and with continued growth it developed posteriorly in its excursion of ossification across the iliac crest to dip down to contact the ilium medially at its junction near the sacrum (completion). Meanwhile closure of the line between the apophysis and the ilium was of no growth significance but might take a further 2 or 3 years. The average time of completion of excursion of the iliac apophysis to its medial and posterior attachment was about a year although there was considerable variation. The average chronological age when the iliac apophysis excursion was completed was 14 years in girls and 16 years in boys although in girls it could vary from 10 to 18. Spinal growth is slow in the period from 5 to 10 years of age and curve progression was about 4 to 5 degrees a year while in the pre-adolescent phase of 10 to 15 years the average annual curve increase was 10 degrees.
Weinstein, a key figure in so much that we know about the behavior of idiopathic scoliosis, and Ponseti published about curve progression after maturity in patients with LOIS. 3 They followed 102 patients with 133 curves for an average of more than 40 years. These were evaluated as regards progression by X-rays at maturity and then at follow-up although some had intervening radiographs. More than two-thirds of curves progressed after skeletal maturity while curves of less than 30 degrees tended not to progress regardless of curve pattern. Important prognostic factors for thoracic curves were Cobb angle, apical vertebral rotation, and the RVAD of Mehta 34; while in the lumbar spine additional factors, such as the relation of the fifth lumbar vertebra to the intercristal line and translatory shifts, were important for progression (translatory shifts are where with, say, lumbar curves, L3 appears to be displaced laterally to L4). It should be noted however that skeletal maturity was regarded as either Risser 4 (iliac apophysis completely ossified) or 5 (iliac apophysis fused to the ilium). Of course the vertebral end plates do not fuse until up to the middle of the third decade 35 and this prompted the Leeds Scoliosis Study Group to measure in an epidemiological survey sitting height for several years after Risser maturity and found that teenagers grew beyond Risser maturity by an average of 2 cm. 36 It was therefore felt that starting at Risser maturity, possibly 10 years before spinal maturity, may not be the best starting point for analysis of curve progression in adult life.
This unrecognized gap between pelvic maturity and spinal maturity led to publications to the effect that pregnancy was bad for curve progression. 37– 39 Nachemson even went to the length of suggesting the avoidance of pregnancy in such women in their 20s, particularly in those who have been brace treated. 40
Notwithstanding there were several of Weinstein’s cases where there was an interim film beyond which there was still significant curve progression. 3 Curves that measured between 50 and 75 degrees at Risser maturity, particularly thoracic curves, progressed the most. Some of the examples in his paper were quite striking although it was not possible to tell whether this was due to coincidental disk degeneration.
More recently Weinstein summarized the state of knowledge of natural history in a review article with more than 250 references and this should be obligatory reading for all scoliosis surgeons. 41 There are easy-to-remember tables about prevalence, probability of progression, and progression factors (e.g., ▶ Table 4.6). For curves of more than 10 degrees the prevalence rate is 2 to 3% with a quite even female to male ratio which the Leeds Scoliosis Study Group confirmed. 25 With increasing curve magnitude to curves of more than 30 degrees the female to male ratio is now 10:1 and the prevalence rate 1–3:1,000. ▶ Table 4.6 is interesting as it shows the probability of progression changes with curve magnitude and age.
Curve magnitude at detection (*)
Age at detection
*Values are percentages.
Source: Reproduced with permission from Nachemson A, Lonstein J, Weinstein S. Report of the Scoliosis Research Society Prevalence and Natural History Committee 1982. Presented at the Scoliosis Research Society Meeting, Denver.
The matter of back pain in relationship to scoliosis has challenged scoliosis surgeons ever since surgery was introduced. We all assumed without any statistical analysis that long Harrington fusions to L4/L5 produced back pain making us take the metalwork out and carry out extension osteotomies in several cases although whether this altered the natural history was certainly unclear. It made sense to pay attention to sagittal curves and to restore lumbar lordosis but was flattening of the normal sagittal profile the reason for the pain or was it over-stressing the available joints below? Nowadays by paying attention to sagittal curvatures with modern instrumentation we hope the problem of long-term low back pain will be very much less but we certainly aren’t in a position to say so at the moment. In any event we have an annual incidence of back pain in the general population varying from about 60 to 80%. We all know this and if more specifics are required then you only have to consult Weinstein’s 1999 Natural History paper for the exact references. 41 Incidentally, incidence and prevalence are not interchangeable but are quite different. They are both rates—incidence rate being the rate of new cases and prevalence rate being the number of existing cases. It is generally accepted that the problem of back pain in scoliosis patients is no greater than in the general population. 42 However, a quarter of scoliosis patients had reported to their doctor with back pain mainly those with lower curves and lumbar curves with translatory shifts. In this Iowa long-term series whereas at Risser maturity only 2% of patients had evidence of osteoarthritis, at age 40 at follow-up all of 38% of patients had radiographic evidence of degenerative joint disease of the spine and at 50 at follow-up in all of 91%. Again this is very difficult to separate from the natural history in the general population bearing in mind that the great majority of scoliosis patients are female who do have “the gene” for primary generalized osteoarthritis of which significant facet joint arthropathy in the lumbar spine is an important part. Furthermore, we have to look at comparable aged nonscoliotic females to assess the prevalence of degenerative changes in the lumbar spine to form a proper comparison group. One would also have to look at those who have been treated surgically and those who have not. The evidence would suggest that with the use of long straight Harrington rods decades ago it was a problem but whether it is now or will be in the future with modern instrumentation and recognition of the importance of sagittal shape one simply does not know (see later under Specific Curve Patterns).
4.3 Late-Onset Idiopathic Scoliosis
4.3.1 Clinical Presentation and Evaluation
It is important to look at LOIS in some detail because this condition is the “bread and butter” of scoliosis surgeons worldwide in terms of occupying their clinical time. More than 90% of cases of scoliosis we see are of the LOIS variety. Furthermore, the treatment methods for LOIS are the cornerstone of treatment for scoliosis of many other diagnostic categories. Whereas 40 years ago, and indeed for the next 10 or 15 years thereafter many patients presented because they had been noted to have some asymmetry of the torso on school screening, the only mode of presentation now and for the last at least 20 years has been because of concern by patient or family about body shape. Presentation is quite different from, say, an underdeveloped 11-year-old girl to a relatively mature 14-year-old. For the 10 or 11 year old, it is more often the parents who see a shape abnormality and they are very concerned about their daughter whereas it is the 14-year-old more mature girl that herself is very worried about the fact that she is not of normal shape like her friends. The common denominator however is that the spine is buckling during adolescent growth and they want to know why this has happened, what does the future have in store, and if there is any satisfactory treatment. Compared to 30 to 40 years ago, we now have good sound answers to these very important questions.
The history is almost universal, namely that the last time we saw our daughter or she saw her shape herself in the mirror her back was normal but now it has become unsightly, with usually either a rib or loin prominence (▶ Fig. 4.3). With a thoracic curve there may be concern that one developing breast is bigger or smaller than the other. With a thoracic curve the whole torso is twisted (plagiothorax) so that with a right thoracic curve then the left breast may seem larger than the right. It has even been suggested ludicrously that this may be indicative of the cause of idiopathic thoracic scoliosis with the underlying ribs having hypervascularity and so asymmetric rib length is secondarily produced. 43 This spurious hypothesis even predicated rib length surgery for LOIS. One might ask how this correlates with idiopathic scoliosis in boys not to mention infants! With a thoracolumbar or lumbar curve there is often marked waist asymmetry and this may be a presenting feature. Indeed, girls may find the ugly waist just as concerning as the rib hump. Then asymmetric shoulder height is also common with the right shoulder being higher for a right thoracic curve while a higher left shoulder may indicate the characteristic signe d’epaule 44 of a double thoracic curve (▶ Fig. 2.17). Decompensated patients also present with a significant list of the torso (▶ Fig. 2.19). Whatever feature of body disfigurement is the presenting symptom it is a matter of deformity only and the family can be reassured that there are no organic health consequences.
Fig. 4.3 The deformity of LOIS. (a) The unsightly rib hump of a thoracic curve. (b) The ugly waist asymmetry of a lumbar curve.
On further questioning the child is absolutely normal with no associated problems. That is certainly the case in the vast majority of LOIS patients. There are however a number of red flags to watch out for. Back pain should not be a problem in children to begin with, let alone those with a scoliosis, and is a red flag for primary health care and accident and emergency doctors. For the scoliosis patient this is particularly important. Pain, particularly if present at night, would indicate a pathological basis for the scoliosis and mandates further examination and imaging. The main suspects are a spinal cord tumor or syrinx (▶ Fig. 2.38) or a bony tumor such as an osteoblastoma or osteoid osteoma. These must be excluded by MRI or skeletal scintigraphy and CT. More often than not plain X-rays will localize the lesion. For example, with an osteoid osteoma/osteoblastoma there is enlargement of the pars/transverse process junction (▶ Fig. 4.4) while with an intradural tumour or syrinx there may well be widening of the interpedicular distance at the site of the lesion. MRI is definitive in the diagnosis of spinal cord tumors and syrinxes while CT scanning and isotope scanning are strongly positive for bone lesions. For individuals with these lesions, on examination the curve is more often very stiff whereas with LOIS they remain very flexible until curve size is considerable. The most important neurological abnormality for intradural lesions is absence or asymmetry of the abdominal reflexes whereas no neurological abnormalities would be expected with bone lesions. With osteoid osteomas the often late diagnosed cause of night pain can lead to psychological or behavioral problems during adolescence and it is remarkable how curative for these nonorganic problems, as well as pain, surgical removal is. The scoliosis in association with these bone tumors is never great, generally not exceeding 20 or 30 degrees. Curve regression after removal is the rule and these curves never become autonomous in their own right unless the process of posterior excision tethers the back of the spine.
Fig. 4.4 Plain film appearances of an osteoblastoma with enlargement of the right L2 pedicle/transverse process junction and no clear pedicle (“winking owl” sign).
Because the typical idiopathic-type scoliosis deformity is also present in other diagnostic categories it is important to observe the musculoskeletal system in its totality. A very tall slim child should be checked for arachnodactyly, a high arched palate, and joint laxity for Marfan syndrome, whereas some café-au-lait spots and, in particular, axillary freckling would demand considerable further investigation as regards neurofibromatosis type 1 (NF1). Collagen diseases and neurofibromatosis very often present with an idiopathic-type scoliosis as the first clinical feature. It is only those with the severe dystrophic bone change that have the short sharp angular curve with lots of rotation diagnostic of NF1. Meanwhile a high stepping gait would indicate a peripheral neuropathy whereas a wide-based unsteady gait would suggest Friedreich’s ataxia. It is very important for the scoliosis surgeon not to divorce themselves from the rest of the musculoskeletal system and so in this book we cover the general orthopaedic principles of these various conditions in which a scoliosis might also be an important clinical feature because it is often the spine surgeon to whom these patients first present.
Other than for the above nonidiopathic conditions that might present or masquerade as idiopathic scoliosis, far too many investigations are performed and, importantly, far too many X-rays are carried out unnecessarily, dangerous in particular to the developing thyroid and breasts. 45, 46 The only reason for radiographing the spine of the typical idiopathic case at first visit is to make sure that it is idiopathic and does not belong to another category, such as, to exclude a congenital anomaly of the spine. One full length PA radiograph of the patient erect and one lateral with a reference grid is quite sufficient for this initial consultation (▶ Fig. 4.5). These views are principally to exclude a congenital spinal anomaly and not to register the deformity which can be clearly seen on examination of the patient. The lateral X-ray is commonly misinterpreted as showing a kyphosis when of course all these deformities are lordoscolioses (see ▶ 3). A PA X-ray of the left hand and wrist to measure bone age is useful to titrate these appearances with chronological age so as to more accurately measure maturation status and growth velocity. 47
Fig. 4.5 Erect PA X-ray spine (whole) with grid.
Registering Cobb angles on the initial film is important to guide the understanding of natural history but measuring Cobb angles on an annual basis is quite unnecessary and should indeed be an inadmissible radiation hazard. What matters is not what the X-ray tells you but what the patient and family tell you about the deformity itself. First of all, reassurance is the most important thing so that the family can understand that there is no disease present, that the spine is perfectly normal except that it has decided to buckle with growth, that this condition is very common affecting more than 2% of normal growing girls, and that it produces no organic ill-health consequences whatever. Then the family should be taken through what the future may hold. There is then only one question to be asked: is the deformity acceptable or not? Acceptability must mean that the patient and family are quite happy with the deformity as it is now, provided it doesn’t worsen significantly. Unfortunately, we have no way of preventing progression nonoperatively. Unacceptability means that the patient and the family are not satisfied with the deformity to the point where they will undergo surgical correction to improve/correct the situation.
Nowadays with modern instrumentation systems not only is correction a real possibility it is very much what surgeons and their patients expect. Of course full correction of the deformity does not necessarily mean a Cobb angle of zero degrees. The reason the initial X-ray looks so bad at 30 degrees or so is that the spine, fairly deep inside the torso, has to buckle appreciably from the midline before it alters surface shape and so the essential principle of surgical correction is to put it back to where it was before it impinged on the surface of the body.
The surgeon has no role at all to decide whether the deformity is acceptable or not, that is for the patient and family to decide. What the surgeon must do is to provide all the information necessary to the family so that they can make this often difficult decision in a proper risks and rewards equation. It is quite inappropriate for the surgeon to look at the X-rays, and, tell the patient the deformity measures 45 degrees and that surgery is necessary. 48 It is for the family and patient to decide upon surgery regardless of Cobb angle and if they have gone through this very important risks and rewards equation with the surgeon then the only decision that they can make is the right one. Fortunately, nowadays rewards greatly outweigh risks—the latter being minimal. Families do often come nowadays with some knowledge about scoliosis from the Internet and the sensible way forward is to provide them with knowledge that is really essential to their case and then afterward deal with any questions that might arise from their Internet experience.
If the deformity is considered acceptable it is very unfortunate that we do not have a method of treatment that will maintain acceptability. It was thought in the days of Milwaukee and Boston bracing that these contraptions might alter natural history favorably but unfortunately that has not proved to be so. In that situation observation is the only way forward and that should not be radiographically measuring Cobb angle, say, every 6 months or yearly, but should be by inspection of the patient and, most importantly, listening to how the patient and family regard the deformity and how intrusive it is in their way of life. At the end of the visit standing height can be measured using a calibrated stadiometer and the reading recorded on standard centile charts, 49 kept in the records.
It is quite unnecessary to take any more spine X-rays on the first visit and indeed for evermore unless the patient’s deformity becomes unacceptable when of course, as we shall see, there are important images to take before surgery that help operative planning. For idiopathic patients, flexibility is a function of curve size and the bigger the curve gets the less flexible it will be simply because the component parts have become more and more three-dimensionally deformed with growth. There is no point in looking at it radiologically before the decision for surgery when it is perfectly self-evident on clinical inspection.
The first visit is extremely important and the patient and family should not be overdosed with information about scoliosis that is not particularly germane at this point in time. An early introduction to scoliosis can well be supplemented by further important information at the second and subsequent visits. Normality, with the exception of the spinal deformity, is the name of the game and should be stressed repeatedly as consultations continue.
It is commonly asked what the cause of this condition is and this is something that can be discussed more and more as the consultations progress. At the first consultation it should be impressed that this is not a simple right–left growing problem or wearing school bags on one shoulder rather than the other, but a front–back problem with the spine growing more quickly at the front than the back. This is abundantly clear when they look at their child’s back and see that overall it is very flat sideways on, more so than others of their age. Then to produce a deformity the spine simply has to buckle with growth and that is perhaps all they need to know about the etiology of this problem. With a 14-year-old more mature girl there may already be considerable psychosocial distress and it is often very helpful to have a few words with the girl in the presence of the nurse but without the rest of the family, indeed many such girls request this.
One of the research fellows in the Leeds Scoliosis Study Group, Dr. Fiona Smith, an honors psychology graduate doing her PhD, looked at degree of deformity in girls with idiopathic scoliosis in relation to psychosocial distress using validated psychological questionnaires to tease out and measure their problems. 50 One of the parameters she measured was body mass index (BMI) and for those with severe deformities eating disorders and a low BMI were common. The normal prevalence rate of eating disorders in adolescent schoolchildren is of the order of 2 or 3% but this was 10 times higher, 20 to 30%, in those with severe idiopathic scoliosis. In the worst cases there was frank osteoporosis just when peak bone mass in females should be achieved. Occasionally one finds a patient so distressed as to be grossly underweight and almost emaciated (▶ Fig. 4.6). Such individuals may need hyperalimentation before they can be deemed fit enough to undergo surgical intervention. We are therefore not relegating this condition of idiopathic scoliosis as simply a matter of cosmetic deformity but rather there is very much more to having a deformity than perhaps meets the eye. It is of course the amount of psychosocial distress that drives the patient with the deformity toward surgery and as this condition is a matter of appearance and deformity only then that is all that matters. Patients nowadays tend to present with deformities of between 20 and 30 degrees and so at initial presentation it is unusual for the deformity to be unacceptable although acceptability might be a 50-degree curve in one more stoical patient and family and a 35-degree curve may be unacceptable in another. It is important not to suddenly stick up the X-ray on the viewing box as many patients and families express shock and guilt that they could have allowed the deformity to become so apparently large. They should therefore be forewarned that the appearance on X-ray grossly exaggerates the surface shape deformity lest you frighten the life out of them. It is obsession with X-rays and Cobb angles that drives scoliosis surgeons to unnecessarily take radiographs on every visit which only upsets them. Reassurance is the name of the game.
Fig. 4.6 (a) This 15-year-old girl had severe anorexia and was severely undernourished and had amenorrhea. (b) DEXA scanning revealed a T score (–2) on the border between osteopenia and osteoporosis.
New classifications seem to be irresistible; however, of course, there is nothing wrong with the original SRS classification 2 (▶ Table 4.7). We don’t really see any need to produce new classifications if they are limited in their applicability or indeed validity. The King classification 51 (▶ Table 4.8) is to do with posterior surgery for thoracic curves only while the Lenke classification (▶ Table 4.9) is to do with posterior surgery for all curve patterns 52 although there are problems with the perception of kyphosis.
Double structural right thoracic and left thoracolumbar or lumbar curves with the lower curve bigger than the thoracic curve.
Double right thoracic and left thoracolumbar and lumbar curves of equal magnitude or the thoracic curve being bigger.
A single thoracic curve.
Unbalanced thoracic curves with significant decompensation to the side of the convexity of the curve.
Double thoracic major curves with a main right thoracic curve and above that a left upper thoracic curve.
Double thoracic and thoracolumbar/lumbar where both curves are of
much the same size
Triple major—double thoracic curve with a structural thoracolumbar or lumbar curve below
Single thoracolumbar or lumbar curve
Double structural curve—main thoracic and thoracolumbar or lumbar curve below where the main thoracic curve is less and may be nonstructural
We shall see later how the introduction of anterior spinal surgery for idiopathic scoliosis has affected selection of fusion levels.
Before anterior surgery came in the generally perceived wisdom with posterior spinal surgery was to fuse all vertebrae in the structural curve. A look at any PA radiograph of a thoracic curve would indicate that the first neutral (unrotated) vertebrae may be one or even two above and below the upper and lower end vertebrae, respectively (▶ Fig. 2.7). Therefore, quite obviously, a fusion from end vertebra to end vertebra will be too short leaving out two or three vertebrae in the structural curve that are not incorporated in the fusion. Not surprisingly this was a fundamental reason why curves could progress after operation until the end of growth, quite simply because the whole curve hadn’t been fused. In addition, this could lead to imbalance and decompensation. In the old days we said that too short a fusion could lead to “adding on”—that is, the curve size as measured by Cobb angle could progress above or below the central fused area. With early-onset curves too short a posterior fusion was said to lead to the crankshaft phenomenon 53 of further deformation which indeed was perfectly obvious all along without introducing the phrase crankshaft phenomenon if you don’t curtail anterior spinal growth at the same time. In the past if you wanted to do posterior surgery for a double curve with long instrumentation down to the mid or lower lumbar spine then these fusion levels, neutral to neutral, more often than not led to a long stiff spine, flat in the sagittal plane obliterating the lumbar lordosis, and adding low back pain to the problem for which subsequent surgery in the nature of lumbar extension osteotomy was often required to try and relieve symptoms.
The whole idea of the King classification 51 came from the great John Moe himself, the founder of the world famous Twin-Cities Scoliosis Center, and he figured that these long posterior flattening fusions were not good practice and therefore introduced the concept of selective fusion 54, 55 whereby the thoracic curve could be instrumented and fused sparing the lower spine to allow continued lumbar spine mobility and hopefully to reduce postoperative back pain. The original paper is quite wordy but in essence led to an alteration in the recommended posterior fusion levels based on full radiographic analysis (standing, supine, bending, etc.). 51 That’s why you don’t need those sets of X-rays until you’ve decided on surgery. Of course we were all doing selective thoracic fusions anyway based upon curve size and side-bending correction clinically as well as radiologically.
In days gone by Harrington himself always insisted that the lower end of his posterior fusion should be “in the stable zone of Harrington.” 56 If vertical parallel lines are drawn through the lumbosacral facets then the vertebral bodies that remained within these lines are in the stable zone. The authors of the King paper 51 felt as the study progressed that a more accurate determination could be gained by a single vertical line drawn through the center of the sacrum perpendicular to the iliac crests which they called the center sacral line (CSL). This important line has persisted to the present day although now is referred to as the center sacral vertical line (CSVL). The vertebra that is bisected or almost bisected by this line is determined and is recorded as being the so-called stable vertebra (▶ Fig. 4.7).
Fig. 4.7 The center sacral vertical line (CSVL). The figure has been modified to include the stable vertebra (SV) and Harrington’s stable zone (shaded area).
(From the Spinal Deformity Study Group Radiographic Measurement Manual, page 47, published by Medronic, Inc., 2004)
Then one of the most important criteria was the degree of flexibility on side bending and they introduced the concept of the flexibility index. If the percentage correction of the thoracic curve was then subtracted from the percentage correction of the thoracolumbar or lumbar curve, this was designated as the flexibility index. They then produced the King classification of different curve types based upon this, as shown in ▶ Table 4.8, for posterior thoracic scoliosis surgery only.
Type 1 curves were double structural right thoracic left thoracolumbar or lumbar curves with the lower curve bigger than the thoracic curve and the thoracic curve more flexible. Not surprisingly these did not do well with a selective thoracic fusion so had to be fused all the way down to the bottom of the lumbar curve. However, they do very well with an anterior selective instrumentation of the lower curve.
Type 2 curves were double structural curves where the thoracic and lumbar curves were of equal magnitude, or the thoracic curve was bigger, and most of these can be treated by selective thoracic fusion. This is based upon the premise that the greater flexibility of the lumbar curve would allow the spine below to restack, or indeed improve in response to upper curve correction.
Type 3 were thoracic curves that were subdivided according to the basic pattern of the thoracic and compensatory curves. In essence, a type 3 curve was a straightforward single thoracic curve which of course could be dealt with by a selective fusion.
Type 4 curves were unbalanced thoracic curves listing way to the side of the convexity of the curve and for this a longer posterior fusion was required.
Type 5 curves were what we used to term Moe double thoracic majors, 55 whereby the thoracic curve to the right has to be treated with metalwork going up to the cervicothoracic region on the right side, or beyond, to maintain shoulder symmetry.
That really was the prescribed treatment for idiopathic thoracic curves going back to the early 1980s. However, without being unnecessarily critical it really didn’t seem to add anything to our knowledge if one persisted with posterior Harrington instrumentation but it did categorize what we were doing in a simple and memorable way. For instance, type 3 is a straightforward single right thoracic curve, say no more. Then again for type 1 curves one obviously cannot leave out the bigger lumbar curve. Then for type 4 curves treated posteriorly we have always known that we have to go down to the lower lumbar spine and for type 5 Moe double thoracic majors we have always known that we have to go up to the cervicothoracic junction or above to treat what is called the signe d’épaule. 44 We always thought this phrase was coined by Pierre Stagnara but Jean Dubousset (personal communication) reassures us that the sign was taught to him by Pierre Queneau, although it may have originally emanated from Pierre Stagnara. Notwithstanding, the King classification was a review of a huge number of patients (more than 400) from the major center at the time. We rather liked this classification although it was how thoracic curves were always approached from the back. We could also see clearly the important role of anterior surgery.
The next most important determinant of fusion levels was contained in the paper describing the Lenke classification (▶ Table 4.9) in 2001 which is in common use today. 52 Again it focuses on posterior surgery only. It also tries to incorporate the sagittal profile using the lateral radiograph of the patient which of course is not valid as explained in ▶ 2 (▶ Fig. 2.10). With a three-dimensional deformity each vertebra occupies a different position in space as regards inclination and rotation and therefore there is no one plane in which the whole deformity can be visualized. 57– 61 Stagnara’s plan d’election views are with reference to the curve apex but they are only true planar views of the apical area 62 (the apical body and probably one or two above and below). They are obviously better than AP and lateral radiographs of the patient but it simply isn’t in any way meaningful to try and register the sagittal profile with lateral views of the patient—so why take them? Let alone then use strange terms like hypokyphosis and hyperkyphosis when they’re all lordotic! One of us (J.H.) didn’t like the sagittal modifiers for the above reasons and the other (R.A.D.), having spent most of his career showing that kyphosis doesn’t exist with idiopathic scoliosis, could not have faith in a system devised by a group most of whom didn’t appear to appreciate the real three-dimensional deformity of structural scoliosis.
These Lenke curve patterns, as shown in ▶ Table 4.9, are the same established curve patterns going back to the days of Goldstein and Waugh. 2 The curves themselves do not change, just the classifications! The lumbar modifier for type 1 main thoracic curves is thought to be useful to help to decide if the thoracolumbar or lumbar curve is likely to be structural or not (▶ Fig. 4.8). Of course these are not really structural curves but represent what is happening below the main thoracic curve—either the lumbar spine is straight (A), or there is a small left lumbar curve (B), or the lumbar curve is a bit bigger (C). The CSVL is the same as the CSL of King based upon Harrington. If this line runs up the middle of the lumbar spine then it is a lumbar modifier A, if it still lies within the lumbar spine it is a B and if it is on the concave side of the lumbar spine it is a C. Therefore, initially for diagnostic registration of curve pattern you have 1 to 6 then for type 1 curves you add on A, B, or C according to the lumbar spine. Then you add on the thoracic sagittal profile measured from T5 to T12 as being N for normal (10 to 40 degrees, – (hypo) less than 10 degrees, and then + (hyper) if it is more than 40 degrees. Therefore, a 1AN is a main thoracic curve with a straight thoracolumbar spine below and a lateral X-ray of the patient showing a sagittal profile of 10 to 40 degrees. Incidentally T5 cannot be easily identified on a lateral because of shoulder overlap. Furthermore, thoracic kyphosis changes during adolescence (▶ Fig. 3.11) so there isn’t a normal thoracic kyphosis as a basic reference. God knows what the true intraobserver error is let alone the interobserver error, never mind what they tell you, 52 but do scoliosis surgeons want to be so regimented and more importantly do they find such a busy classification (invalid in part) really helpful? For instance, a type 1AN is a simple single thoracic curve, why say anything else? The lumbar curve is purely compensatory and doesn’t need a lumbar modifier. Type 1B is always going to have a sufficiently mobile lumbar spine below while a 1C has always had a lumbar curve below that needs its mobility checked out.
Fig. 4.8 Radiographic PA images with outlined vertebral bodies, CSVL, and L4 tilt for (a) a Lenke type 1AR curve (L4 tilts to the right); (b) a Lenke type 1 AL curve (L4 tilts to the left); (c) a Lenke type 1B curve; and (d) a Lenke type 1C curve. (Reproduced with permission from Newton P, O’Brien M, Shufflebarger H, et al. Idiopathic Scoliosis: The Harms Study Group Treatment Guide. Stuttgart/New York: Thieme; 2010: 202.)
While this thoracic sagittal profile assessment is spurious based on the lateral view of the patient and invalid as regards this classification it is still worth thinking again about what this pseudokyphosis really means. 58– 61 As the primary lordosis rotates out of the sagittal plane so of course it is less well seen on the lateral view of the patient which shows a progressively greater amount of pseudokyphosis. Thus what this pseudokyphosis really registers is the amount by which the lordosis has rotated. Therefore, if, in Lenke language, the sagittal modifier is + (i.e., the pseudokyphosis has got bigger than 40 degrees), then this means that this particular curvature is big and has a lot of rotation. This is particularly well seen in scolioses in dystrophic neurofibromatosis. With a lot of structural skeletal problems, including dystrophic spinal bone producing a very sharp angular scoliosis, the lateral view of the patient may register a pseudokyphosis of 50 degrees or more. This led to the recommendation that if there was 50 degrees of kyphosis on the lateral view then this warranted anterior surgery, as well as posterior surgery. 63 It was not of course kyphosis at all but a scoliosis with a lot of rotation particularly at the apex and clearly these very big deformities manifestly cannot be managed by posterior surgery only so the call to add an anterior stage while being entirely correct was based on a misunderstanding of what the lateral X-ray of the patient really shows. In simple terms you cannot specify a particular angle but if the pseudokyphosis is particularly big or the patient very young then watch out for the need to do an anterior procedure as well, if not before.
Then going on to pseudokyphoses of less than 10 degrees, these are particularly well seen in double-structural curves, thoracic, and thoracolumbar or lumbar, where the whole spine is lordotic throughout (▶ Fig. 2.21) and that is all that minus in the thoracic sagittal modifier really means. This is where the expression junctional kyphosis between two curves is again spurious. If this junctional vertebra lies between two structural lordoscolioses then the junctional vertebra cannot be kyphotic (▶ Fig. 2.22) but it may look as if it is on the lateral view of the patient 64 because the junctional area is less lordotic than the curves above and below.
For a classification to be useful it has to be straightforward and readily applicable and because the lateral view of the patient doesn’t mean anything then it is better to leave out the sagittal modifier altogether and then the Lenke classification can have some use (for posterior surgery only), but then it isn’t much different from King.
In any case, it does not really matter which classification you use because they are all trying to describe the same original curve patterns. 2 ▶ Table 4.7 is the original SRS classification slightly amended and highly recommended. It is more important to understand what, say, pseudokyphosis really means. In addition, the Lenke classification addresses posterior surgery only whereas many of these curves can be treated anteriorly with a much shorter fusion and just as good a result, if not better. For instance, all Lenke type 6 have a major structural curve in the thoracolumbar or lumbar region eminently, and we would say only, treatable anteriorly. There has been a marked trend in recent years, probably because of these classifications, to go with posterior surgery only—or the trend to do posterior surgery only has generated these classifications—and as a result anterior surgery seems to be less frequently carried out although it has been shown to be superior than equivalent posterior surgery. 65, 66 We have enormous anterior experience being disciples of Leatherman and Zielke, and we feel it right and proper to guide learning scoliosis surgeons through the front of the spine as well rather than just the back.
Going back to our simple treatment philosophy we have asked the patient and family if they find the deformity acceptable as it is and if they do so then ideally we want a nonoperative treatment modality that can do just that—namely, prevent progression and maintain acceptability.
It would be marvellous if we had a bracing or casting system that really was effective. Unfortunately, there is no evidence of such nor does it seem possible to conceive of an effective orthosis (see ▶ 3). The great problem is the Hueter-Volkmann law because as soon as the lordosis buckles out of the sagittal plane, the vertebrae, particularly around the apical region, become progressively more three-dimensionally deformed and thus progressively more unstackable in a straight position (▶ Fig. 4.9). The best way of seeing this in one’s mind is to consider blocks of Lego that progressively become more deformed and less rectangular as one proceeds through the curve to the apex. It might be possible to consider some form of corrective treatment if the blocks of Lego remained rectangular but when deformed three-dimensionally they can only be restacked in the original deformed scoliotic position (▶ Fig. 4.9), unless space is made available between adjacent vertebral bodies or the end plates are refashioned perpendicular (▶ Fig. 3.9). This is only necessary around the apical region (4 or 5 vertebrae) where three-dimensional deformation is maximal (the Leeds procedure) 67– 69 (see surgical treatment section). What orthosis on this earth, or in any other galaxy, could possibly redress that progressive growth deformity and imbalance? Not surprisingly side-bending radiographs show progressively less flexibility as Cobb angle increases. There is no added fibrosis or tightness or whatever to produce lack of flexibility, rather it is simply a question of unstackability, and if it wasn’t then it wouldn’t be idiopathic scoliosis!
Fig. 4.9 Progressive asymmetrical wedging with growth. True lateral X-ray films of the apical vertebrae of different size curves. (a) Curve of 20 degrees with reasonable sagittal shape. (b) Curve of 40 degrees, showing that the back and inferior surfaces of the vertebral body are ellipsoid. (c) Curve of 80 degrees, showing marked asymmetrical 3D vertebral wedging. These vertebrae will not stack straight unless something is done about it. (Reproduced with permission from Newton P, O’Brien M, Shufflebarger H, et al. Idiopathic Scoliosis: The Harms Study Group Treatment Guide. Stuttgart/New York: Thieme; 2010: 95.)
By contrast an idiopathic spinal deformity without significant vertebral deformation, at least in three dimensions, is the opposite of idiopathic scoliosis, namely Scheuermann’s thoracic hyperkyphosis. 59 This progresses in the sagittal plane because it is always behind the axis of spinal column rotation and, being under tension, will not buckle. 70 Certainly as Scheuermann’s disease progresses the vertebrae become more wedged with anterior compression but not in three dimensions. Not surprisingly therefore when an extension orthosis is prescribed or extension casting the deformity is correctable over time provided there is sufficient growth remaining. 71, 72 After about, say, 60 degrees, the deformity of Scheuermann’s disease is less responsive to extension bracing or casting and there may not be enough growth left. This is where surgical treatment may be required, only of course if the patient finds the deformity unacceptable and nothing to do with any X-ray measurements suggested by the surgeon. As hyperkyphosis needs extension then lordoscoliosis needs flexion and that of course is precisely what causes it to buckle further 19 (▶ Fig. 4.1). The condition of LOIS is therefore not readily treatable externally although one hears through the grapevine that yet another brace trial is being considered, 73 surely not.
Of course for a prospective controlled trial to have any substance then there should already be very good evidence that the treatment works and, not surprisingly, such evidence is conspicuous by its absence; two retrospective trials 74, 75 did not show any benefit for brace treatment while a prospective nonrandomized study had far too many progressive thoracic curves in the nontreatment group, 76, 77 thus favoring brace treatment.
Since the time of Hippocrates there have been innumerable contraptions applied to scoliosis patients with no effect whatsoever. 78 Even the diligent work of Cotrel and his colleagues in Berck-Plage with adolescent girls wandering about the town with their elongation derotation flexion (EDF) casts trying to stand and walk upright balancing a book on their heads had no beneficial effect on curbing the progression of their deformities. 79
What happened in idiopathic scoliosis was that the Milwaukee brace was empirically prescribed. It is difficult to understand why this should happen because the Milwaukee brace was specifically designed to prop up the spine with a poliomyelitis scoliosis after surgery and had nothing at all to do with idiopathic scoliosis let alone as a preventative measure. 80 It was primarily designed to distract the spine and not surprisingly with the initial mandibular counter-distraction piece significant dental problems arose. 81 Then this was exchanged for a choker round the neck to actively maintain the patient upright. It was felt in the initial phases of treatment that the lumbar spine should be flattened by the orthosis which seemed to provide an improvement in Cobb angle when brace wearing. Of course this is easily understandable because with the lumbar spine effectively flexed this caused active hyperextension of the thoracic spine above and led to movement of the thoracic lordosis toward the midline. 19 However, the improvement in the thoracic spine Cobb angle was only about half of that which would be achieved by natural daily side bending and if the brace was worn for the prescribed 23 hours a day then the spine was never able to achieve its maximum side bending position with ordinary activities of daily living that would happen with any normal adolescent with their sporting and other physical activities. However, the proponents dictated that it must be worn for curves beyond, say, 25 degrees without any objective data to support their insistence.
Evidence-based medicine has been “in” for many years now. It means “the integration of individual clinical expertise with the best available external evidence from systematic research, particularly concerning the scientific principles governing treatment.” 82 It is not possible to identify any criteria by which the nonoperative treatment of LOIS adheres to the principles of evidence-based medicine. Not only was the orthosis used for treating LOIS devised for a completely different type of scoliosis under entirely different circumstances but the early experience with this orthosis was reported at a time when it was not conventional to apply statistical methods or any other rigorous analytical approach to the problem. This is what we have done, this is what has happened, and because we are the senior scoliologists of the day, you’d better do the same. Dissent or challenge did not seem to be the most prudent route to a successful career in spinal surgery! In the 1970s there were various retrospective reports with poor numbers followed up, usually less than half, suggesting benefit from brace wearing. 9, 10, 83 However even these papers did highlight their problem: “there have been no published data with regard to long-term end results comparable to the available follow-up studies of untreated patients.” 9 Another commented “the role of the Milwaukee brace in the treatment of idiopathic scoliosis is still unclear” … “what then is the proper role of the Milwaukee brace in scoliosis treatment … further follow-up must be obtained in these patients.” 9
Then some common sense came into the situation and in 1984 the Gothenburg Group carried out a retrospective comparison of 144 braced patients versus 111 untreated, both groups with a mean deformity of less than 30 degrees. There was no statistical benefit for the braced patients. 74
Caroline Goldberg then assessed the efficacy of the Boston brace and, like Gothenburg, no difference was reported between the braced and unbraced controls. She quite rightly stated “this raises very seriously the question of whether bracing can be considered an effective way of altering the natural history of late-onset idiopathic scoliosis.” 75 Notwithstanding, bracers continued unabated. In 1994 the Minneapolis Group published their results of more than 1,000 patients treated with a brace between 1954 and 1979. 83 The braced patients were compared with 727 patients who were not braced but just followed up. Failure was defined as an increase in Cobb angle of at least 5 degrees or surgical intervention. The results would suggest that failure rates were lower in the braced group although of course no statistical evidence was produced. Despite the fact that it had been previously stated that “the role of the Milwaukee brace in the treatment of idiopathic scoliosis is still unclear,” 10 academic criticism was not well tolerated 84 and it was further stated “in the 1980s, a negative attitude about bracing existed, it was so extreme that Professor Robert Dickson of Leeds, England, stated that ‘there was no place at all for brace treatment.’” 19 It is always advisable before you go into print to make sure that what you are quoting is correct. RAD never stated that there was no place at all for brace treatment, rather he chose his words more precisely: “set against the background of natural history there is no evidence that Milwaukee brace treatment alters the course of scoliosis”—a quite different statement. Noonan and Weinstein expressed concern about the large number of patients being excluded from the Minneapolis study and that only 28% of the study questionnaires had been completed. 85 Then the Puerto Rico Group suggested better results from bracing but there were far more thoracic curves in the untreated group compared to the treated group—thoracic curves well known to have a greater progression potential. 86 They also talked about the patients fully complying with treatment whereas Houghton in Oxford with hidden compliance meters showed that adolescent children in Oxford only wore the brace 10% of the prescribed time. 87 RAD’s great friend, the late Greg Houghton’s (he was knocked off his bicycle by a drunk driver and suffered severe and fatal head injuries) results have been challenged at international meetings but RAD knew this study from his alma mater and it was impeccable.
Not surprisingly with so many vested interests and so many stake holders a prospective randomized controlled trial was recommended. Those carrying out this trial quite rightly stated as regards previous reports “none of these studies met the stringent criteria for scientific evidence that must be used to prove the effectiveness of treatment.” They added “a well-designed study must include a large cohort of similar patients with similar patterns and sizes of deformity and that they should be randomized to different treatment methods and followed until at least skeletal maturity.” 76 Not surprisingly randomization was impossible because bracers wouldn’t stop bracing and nonbracers wouldn’t start. Notwithstanding and for very good reasons, a nonrandomized study was carried out of 286 patients comprising 129 observed patients, 111 braced patients, and, for whatever reason, 46 electrically stimulated although this latter treatment method had been discarded years earlier. Perhaps they thought that the electrically stimulated group might form another control group. The results are shown in ▶ Table 4.10 and it can be seen that failure (6 degrees or more of progression) occurred most in the electrically stimulated and least in the braced group. A 6-degree threshold might be statistically significant but can a change in Cobb angle of 6 degrees really be regarded as clinically significant?
No. of subjects
Note: Failure = 6 degrees of progression.
The next article in the Journal of Bone and Joint Surgery concerned the proportions of thoracic and thoracolumbar curves in these groups as is shown in ▶ Table 4.11. Dangerous thoracic curves 32 were much less prevalent in the braced group than the observed group while the electrical group had many more thoracic curves. 77 You simply couldn’t stack the odds better! The only conclusion that can be drawn from this trial was that no benefit could be attributed to bracing for scoliosis. Or alternatively that the differences between the three groups actually was measuring not failure but the difference in the proportions of significantly progressive thoracic curves among the three groups. Indeed, at this point, Stuart Weinstein and RAD surveyed the world literature on bracing for idiopathic scoliosis and we found no evidence base for brace wearing. 88
Bracers apparently cannot be subdued and another trial is planned. 73 Prospective trials are designed to follow on from good retrospective or other data that suggest that treatment is beneficial. The prospective trial then looks at different categories of patients, such as, 20 to 30 degrees, 30 to 40 degrees, thoracic vs. lumbar, to see in which way the treatment is beneficial and not to assess whether it is beneficial in the first place. There is, therefore, on strict research principles, no evidence to support any further trials on the nonoperative treatment for idiopathic scoliosis. A silver bullet might however be required.
Bracers have published some papers beyond the Nachemson trial paper suggesting again that bracing might be beneficial. Katz et al in 2010 published a study of 126 patients with curves between 25 and 45 degrees who were braced with a heat sensor measuring the exact number of hours of brace wear. 89 A total of 100 patients completed the study and were managed with a Boston brace; 82% of patients who wore the brace for more than 12 hours had a successful outcome (less than 6 degrees of progression). For those who wore the brace 7 to 12 hours the success rate was 61% and for those who wore it less than 7 hours, 31%. Again, like other studies, the main endpoint criterion was a Risser scale of 4 or 5, although they did measure height at the same time. In patients who were Risser 0 or 1 with curves of more than 35 degrees there was only a barely significant difference between brace wear times (p = 0.05). Of course what happens a year or two or three after the cessation of brace treatment, bearing in mind that the spine is still growing, is uncertain. However, once again one can question the significance of 6 degrees of progression, or indeed of 6 degrees of improvement. 89
In 2011 Negrini et al in Italy published an extraordinary paper about brace treatment for individuals whose curves were in excess of 45 degrees and who had steadfastly refused surgery. 48 First, a one-off operation to make a straight spine nowadays is a very safe and reliable procedure. Whether or not that was a properly and fully discussed option for them is unclear but in any event they declined surgery and were “treated” with a brace. There were 28 subjects whose age at the start of treatment was 14.2 years on average and whose Cobb angle measured on average almost 50 degrees (49.4 degrees), range 45 to 58 degrees. Reported compliance was allegedly 94%! Six patients (21%) finished with a curve of between 30 and 35 degrees and 12 patients (45%) finished up between 36 degrees and 40 degrees. Improvements were found in 71% of patients overall. Their alleged improvements were an average of 8 degrees for thoracic curves and 16 degrees for lumbar curves. One finds these figures unbelievable simply because with curves of more than 45 degrees at the age of more than 14 years then the apical three-dimensionally wedged vertebrae can simply not be corrected by any form of external orthosis.
While we were writing this last section on bracing it has now become quite clear that another trial has been set up and indeed is under way. 90 This paper was published to describe the design and development of the brace trial, referred to acronymously as BrAIST (Bracing in Adolescent Idiopathic Scoliosis Trial). We were interested to read that one of us (RAD) was part of the brace protocol development committee. A Cobb angle of 50 degrees was regarded as being the surgical threshold and the question that the trial was primarily addressed to was “do braces lower the risk of curve progression to a surgical threshold in patients with adolescent idiopathic scoliosis relative to watchful waiting alone?” The endpoint was either skeletal maturity or the surgical threshold. There seem to be two main problems here one of which was the surgical threshold of 50 degrees Cobb angle and the other the taking of lateral X-rays evaluating them as showing either kyphosis or lordosis when of course all idiopathic scolioses are lordotic. Indeed they are lordotic to begin with before the spinal column turns out of the sagittal plane to become more and more visible in the frontal plane as the scoliosis with rotation. In this regard lateral X-rays are meaningless. While 10 to 40 degrees might be a reasonable range of normality for kyphosis on a lateral X-ray of normal children, such appearances of kyphosis are spurious in the presence of idiopathic scoliosis and are better described as “pseudokyphoses” (see ▶ 3 for more detail). Then the Cobb angle of 50 degrees has not as far as we know been validated as a threshold for surgery. Rather it is the patient and family finding the deformity unacceptable, regardless of Cobb angle, which is the essential driver for surgical treatment. There are lots more criticisms one can level at this trial design but there is one good thing about it; namely, the lead investigator is Stuart Weinstein. It is a pity the reference list does not include any of the publications from the Leeds Group. They did not, for instance, include the publication by Weinstein and Dickson on “Bracing and Screening – Yes or No.” 87
We do hope the proposed trial 90 is carried out properly with rigorous separation and stratification for the important variables such as curve pattern and compliance. It certainly should be with Stuart Weinstein at the helm and then it might lead to some definitive results. Although it is difficult to conceive why an external orthosis should work in this progressive three-dimensional deformity, we do not want to know about preventing a few degrees of progression. What we do want to know is that if the brace does work then what are the real benefits so that we can take these to our patient population in a risks and rewards equation. With excellent results with modern instrumentation and surgery being an ever more safe procedure it is going to take some persuasion for a teenager to wear a brace for a longish time each day for years without a goal at the end.
In guiding the patient and family as regards surgical treatment it is important to understand that the only criterion that has to be fulfilled for surgery to proceed is that the deformity has become unacceptable to the patient to the point where they are prepared to undergo surgery in an attempt to restore acceptability. If they have reached general skeletal maturity, usually about 15 years of age in girls, then the deformity should not substantially progress although we have already seen that the vertebral epiphyses do not fuse until the early 20s. 35 Some degree of progression may therefore occur up until this age but usually not of any great clinical significance. Therefore, if the deformity is still acceptable at the age of 15 it probably will continue to be so and can of course still be observed serially until the early 20s to make sure that unacceptability is not reached in the interim. The idea that it will progress a degree or so throughout the rest of life is a possibility although quite why is uncertain. 41 A few degrees, say 10 degrees at most, may occur in terms of progression up until the early 20s when the spine finally stops growing 35 but there is no evidence to support the notion that any further progression is due to pregnancy. 37, 38
If of course unacceptability has been achieved by the age of 15 then that is the essential indication for surgical intervention and it is extraordinary to see that when acceptability has been restored by surgical intervention how they rejoin their happy and healthy peers, a real sensation of achievement for both patient and surgeon.
Once surgery has been decided upon, only now do we restart taking X-rays. Hitherto we have taken only one diagnostic PA X-ray and one lateral, or should have done, when we first saw the patient. Until unacceptability is reached there is absolutely no need to take any further X-rays. Why should we? The lateral is generally misunderstood and we don’t want or need to see radiographic progression when it can be seen by the patient who is the chief and only determinant of whether the deformity is acceptable or not. Far too many X-rays are taken of these growing children, usually girls, and this practice must be stopped. The only requirement to take further X-rays of the spine is when unacceptability has been reached and surgery is being planned, other than harmless X-rays of the left hand and wrist for bone age as and when necessary.
4.3.4 Radiological Evaluation
The surgeon should already know the curve pattern from the initial PA X-ray as well as visual clinical inspection. The flexibility of the curve should also be known by clinical examination and so these preoperative films are important for precision in choosing the length of the curve and for deciding overall surgical strategy. It is standard to take PA and lateral standing X-rays against a grid (▶ Fig. 4.5). The following statement might be contentious but if the lateral X-ray is only for adding +, N, or – to so-called kyphosis then why should a lateral X-ray be taken at all?
Of course it has been traditional in orthopaedic surgery to take AP and lateral X-rays so that we have information about the problem in two planes at rectangles. This is clearly very important for fractures, tumors, and many other pathologies but it is less useful for looking at deformities. The main reason for taking X-rays in the typical idiopathic case is to make sure there isn’t some form of subtle congenital anomaly producing or contributing to the deformity and in this regard lateral films are certainly beneficial to pick up for example minor degrees of failure of segmentation which may induce a nearby idiopathic-type lordoscoliosis.
As regards assessing the spinal deformity itself the lateral view in particular has been confusing showing the spurious appearance of kyphosis when all these idiopathic deformities are lordoscolioses. If you do however know the importance of lordosis in the three-dimensional deformity of idiopathic scoliosis, then the lateral X-ray can give you lots of useful information. For relatively small Cobb angles the lateral X-ray does show a flat thoracic spine and you may still see the original lordosis over the T7, 8, 9, and 10 levels. This is because with smaller curves the lordosis has not rotated much from the sagittal plane but as it does so and with bigger Cobb angles then the more it moves away from the sagittal plane into the frontal plane the more it will wrongly register a thoracic kyphosis. The bigger the pseudokyphosis, the bigger the scoliosis and the bigger the amount of rotation. Not only does this mean a worse progression potential in terms of prognosis but also that the more the need is for anterior surgery to shorten the front of the spine and make space available for the deformed vertebrae to collapse into. This is particularly well seen with the short sharp angular curves of dystrophic von Recklinghausen’s disease where rotation is excessive. However, do not take a lateral X-ray and assign it (–, N, +) to fit into the Lenke classification because that is misleading and means that you have not understood the contents of ▶ 3.
Maximum lateral bending films to right and left can then be taken and these should be with maximum patient effort plus a pull from the arms of a radiographer or resident or over the fulcrum of a cylindrical bolster at the curve apex. Interpretation of maximum side bending films is not straightforward. If you select the end-vertebrae on the PA erect film then sometimes with a flexible curve the side bending film will register a Cobb angle of 0 degrees or even a negative one. However, what this is doing is really testing the flexibility of the upper and lower compensatory curves and not of the apical region (▶ Fig. 4.10). An alternative strategy would be to take the maximum side bending film and select the end-vertebrae on this film and compare it with the same end-vertebrae on the PA erect film. Then what you are really looking at is the degree of flexibility over the stiffer apical region, which is much more important than the overall flexibility of the curve from end-vertebra to end-vertebra on the erect film. Supine maximum traction films are also commonly taken but if such views are not going to be acted upon then don’t take them. We quite like these stretch films as they can provide additional information when the flexibility on side-bending films is not clear. If you are concerned about pedicular size over the apical region, then a PA Stagnara plan d’election view can be taken to see the pedicles en face. 62 You might well be surprised how small your target is!
Fig. 4.10 This diagram is taken from Roaf’s 1955 paper on wedge resection. It demonstrates the spurious correction of Cobb angle that can be obtained by in Roaf’s case a Risser corrective jacket which could just as well be a lateral bending film or a maximum traction stretch film. (a) Curve with a Cobb angle of 143 degrees. (b) The “corrected” position where the Cobb angle now measures 65 degrees. Of course the fixed apical region doesn’t change at all.(Reproduced with permission and copyright © of the British Editorial Society of Bone and Joint Surgery, Roaf R. Wedge Resection for Scoliosis. J Bone Joint Surg (Br) 1955;1:97–101).
As MRI is so available and safe, this should be performed preoperatively just to make sure there is not some hidden problem in the neuraxis such as a syrinx, bearing in mind the high rate of neurological complications with scoliosis correction in the presence of a syrinx. 91 Huebert and MacKinnons’ two cases developed paraplegia following Harrington instrumentation and subsequently died. This publication was of course more than 40 years ago when distraction was the essential force vector and a more recent publication from Auckland New Zealand was much less concerning. 92 They looked at 13 patients with thoracic curves who had undergone neurosurgical decompression for a syrinx. The degree of so-called thoracic kyphosis was much greater than comparable patients with uncomplicated idiopathic thoracic curves. Curve correction was almost 50% and no unfavorable neurological problems arose—all patients undergoing spinal cord monitoring and wake-up tests. Interestingly syrinx decompression did not lead to an improvement in the degree of scoliosis. Of course the difference between Huebert and MacKinnon’s two cases and the New Zealand ones is that the former were undetected syrinxes whereas in the New Zealand cases the syrinx was decompressed before surgical treatment, presumably the critical factor. You might wish to ask a neurosurgical colleague to have a look at the images if the syrinx is unusually sizeable. You will almost certainly get the answer back that the syrinx does not demand neurosurgical action in its own right. But that is not the question. The question is “if I possibly distract the spinal cord or otherwise change its shape could I cause neurological problems?” If the syrinx is sizeable (we know of no particular dimensions) then it would be advisable to avoid a scoliosis lengthening procedure and thus to adopt an anterior approach with preliminary disk removal. In the presence of a syrinx paralysis is an unacceptable risk without apical shortening.
It is traditional to measure the Cobb angle using a protractor 93 (▶ Fig. 2.8) or Cobbometer 94 (▶ Fig. 2.9) in the clinic. Now there are tools on the Picture Archiving and Communications System (PACS) system to draw the lines and measure it for you. Again using the PACS system a vertical plumb line can be dropped from the spinous process of the vertebra prominens downward to check spinal balance. Spinal balance can also be assessed using a plumb line clinically, but also by looking at the size of the compensatory
It is important to measure apical rotation which may vary several degrees for a given Cobb angle. The first reliable measure of vertebral rotation introduced was that of Nash and Moe 96 (▶ Fig. 2.12). This is still the best method of measuring apical rotation and, like Cobb angle, can be used in the clinic. Now new three-dimensional CT and MRI techniques have been developed that are not such a radiation hazard although they are supine images (▶ Fig. 2.14).
Then a radiograph of the left hand and wrist should be taken to check on maturity status unless one taken on a previous occasion has already shown skeletal maturity. Measuring Risser sign 33 is not really appropriate because maturity of the pelvis is not the same as maturity of the spine which may occur in girls 5 years or more later when the vertebral epiphyses themselves fuse. 35 However, the Risser sign is in such common usage and of course it does have value in gauging where patients are as they approach maturity (▶ Fig. 2.23). Clearly, as you move onward from fusion of the pelvis, growth diminishes appreciably but notwithstanding growth still does occur and while that may be of no importance at all to a straight spine it may be very important indeed for a deformed one.
4.3.5 Specific Curve Patterns
Surgical treatment differs in terms of the six curve patterns in our amended original classification (▶ Table 4.7) and the six described by Lenke (▶ Table 4.9). 52, 97 Both anterior and posterior approaches are appropriate for most of these curve types. It is crucially important to select the right fusion levels so as to obtain and maintain the optimal correction and preserve spinal balance. Strategies additional to simple anterior or posterior instrumentation are required for rigid late-onset idiopathic cases which would be big very stiff curves, with less than 20% peri-apical correction on bending films if you want to try to put a figure on it. 98 These latter cases are complex ones that, more often than not, may require osteotomies.
Meanwhile there are a number of important decisions that need to be made relating to surgical treatment. 99 Which if any minor curves need to be included in the fusion? What vertebral levels should be included in the fusion? What is the best approach? Is an anterior “release” indicated? Selecting the lowest instrumented vertebra (LIV) particularly with posterior instrumentation is crucial (▶ Fig. 4.8).
Recently the concept of selective versus nonselective surgery for adolescent idiopathic scoliosis has come more into focus although it really has always been a problem and a challenge to us. Although this concept is attributable to Moe 56 it really has always been in the minds of all scoliosis surgeons because, as Sucato says, “the goals of surgical treatment in late-onset idiopathic scoliosis are to prevent progression of the curve and to correct the spinal deformity while maintaining overall coronal and sagittal balance. These goals should be achieved with fusion of as few spine motion segments as possible.” 100 Selective posterior fusion really applies only to thoracic curves. Whether or not a selective thoracic fusion is carried out depends upon assessment of the flexibility of the lower curve so that the mobility and health of the lumbar spine can, if possible, be preserved. For thoracic and thoracolumbar/lumbar double patterns (Lenke 3 curves) selective fusion is often all that is required but that is nothing new and has been challenging scoliosis surgeons ever since Harrington introduced his instrumentation. 56 If the lower curve straightens out or restacks in balance below the expected thoracic curve correction then selective fusion is the treatment of choice.
There are a number of controversial matters to be taken into consideration in deciding site and length of fusion as well as balance and the question of future pain and degeneration. It would seem obvious that if you just do a selective fusion leaving the lumbar spine free then the lower spine would have more movement to compensate for the fusion higher up. Indeed that seems to almost be the perceived wisdom about selective fusion until you start looking at the matter in detail as Engsberg et al did when they looked at 30 patients with adolescent idiopathic scoliosis undergoing instrumented fusion assessed objectively by videography with reflective surface markers. 101 Whereas range of motion was obviously reduced in the fused regions of the spine it was also reduced in the unfused regions and the lack of compensatory increase in motion at unfused regions contradicted the current theory addressing the need for early postoperative range of motion therapy to facilitate motion in unfused segments. Meanwhile Cochrane worked with Alf Nachemson in Gothenburg and demonstrated that with the old fashioned Harrington instrumentation if the lower hook went down to L4 or 5 then retrolisthesis was locally produced along with significant low back pain as well as degenerative facet joint changes and disk space narrowing in 11% of patients. 102 Certainly there would appear to be mechanical problems and pain at the bottom of a long Harrington fusion but the question of degenerative changes is quite unclear as there is a natural trend for degenerative disk disease to be located at L4/5 and L5/S1 and a natural trend for primary facet joint osteoarthritis to occur at L3/4 and L4/5. A lot more work needs to be done to sort out who really is going to be disadvantaged by way of degeneration caused or accelerated by scoliosis surgery. Furthermore, degenerative disk disease is quite different from degenerative posterior facet joint primary osteoarthritis. The pathologies are different and the processes are different, with degenerative disk disease tending to affect a younger age group than degenerative posterior facet joint osteoarthritis—the latter occurs much more commonly in older individuals and females who more often have the gene and also of course progressive idiopathic scoliosis.
The term adjacent segment degeneration is used to describe radiographic changes seen at levels adjacent to a previous spinal fusion procedure but do not necessarily correlate with any clinical findings. 103 The term adjacent segment disease refers to those who have new clinical symptoms that do correspond with new radiographic changes adjacent to a previous spinal fusion. This matter of adjacent segment problems seems to have attracted more than considerable attention in recent years to the point where there is perhaps a general notion amongst spinal surgeons that both adjacent segment degeneration and disease do pose considerable problems. But where is the evidence?
Two earlier reports indicate that 25% of patients undergoing anterior cervical fusion for degenerative disease or cervical myelopathy over the postoperative 5 to 9 years went on to develop adjacent segment degeneration with an average prevalence rate of 2 to 3%. 104, 105 Herkowitz 106 studied 44 patients randomized to either anterior cervical diskectomy and fusion or posterior foraminotomy without fusion and treated for cervical radiculopathy. More patients undergoing posterior foraminotomy without fusion developed adjacent level degeneration but there was no correlation with symptoms. Bohlman 107 reviewed more than 100 patients undergoing anterior diskectomy and fusion for radiculopathy finding that 9% went on to develop adjacent segment disease requiring additional surgery while Gore 105 described 14% of their patients requiring additional surgery for adjacent segment disease. The annual incidence of adjacent segment disease requiring surgery appeared to be between 1.5% and 4%. Lunsford 108 reported on more than 300 patients undergoing anterior cervical diskectomy some without fusion and found an annual incidence of adjacent segment disease of about 2.5% with no difference between those who were fused or not. Henderson 109 found an average annual incidence of about 3% in those undergoing posterior foraminotomy without fusion. These clinical observations suggest that anterior surgery with fusion and posterior surgery without lead to similar rates of adjacent segment disease. Hilibrand 110 reported on more than 400 anterior cervical decompression and fusion procedures for radiculopathy and/or myelopathy and looked for the development of any new neurological symptoms referable to adjacent levels calculating a Kaplan-Meier survivorship analysis. There was an overall annual incidence of approximately 3% developing adjacent segment disease but the Kaplan-Meier survivorship analysis suggested a much higher likelihood up to 25% at 10 years. Risk factors included neurological problems from adjacent levels at the time of initial surgery as well as surgery performed next to C5/6 or C6/7, the levels by far the most common for the development of natural degenerative disease anyway. Interestingly anterior fusions at more than one level significantly lowered the rate of adjacent segment disease. The conclusions of all this were that adjacent segment disease was a common problem that may reflect the natural history of the underlying cervical spondylosis and that fusion per se may not be the culprit originally thought.
In the lumbar spine Lehmann 111 and Luk 112 demonstrated instability at the segment above lumbar or lumbosacral fusions, with no correlation with the patient’s symptoms. Penta 113 compared the natural history of lumbar segments in fused and nonfused patients and interestingly found no difference in the rates of adjacent segment degeneration with a third in both groups developing degenerative changes at the level above the fusion. Again increasing length of fusion did not appear to increase the extent of adjacent segment degeneration. Rahm 114 found that a third developed adjacent segment degeneration and did have worse clinical results but interestingly the development of a pseudarthrosis appeared to be a protective factor against the development of adjacent segment degeneration. Whitecloud 115 found it much more difficult to obtain a solid fusion when operating the level adjacent to a previously operated level with an 80% pseudarthrosis rate. Etebar 116 found a 14% rate of symptomatic adjacent degenerative disease while Ghiselli 117 found that only 1 of their 32 patients developed symptoms following a lumbosacral fusion. Interestingly Throckmorton 118 looked at patients whose surgery was adjacent to a degenerated disk or a normal disk and found the worse clinical outcomes occurred with normal adjacent disks. Ghiselli 119 looked at more than 200 patients with lumbar fusions and found that 37% would be expected to require additional surgery for adjacent segment disease, with highest incidences in floating lumbar fusions.
Based on the present scientific literature it is still not clear whether these radiographic and clinical findings are the result of the spinal fusion within the iatrogenic production of a rigid motion segment or whether they represent the progression of the natural history of the underlying degenerative disease. It would be therefore very unwise to blame previous surgery for the “adjacent segment degeneration and disease” and in particular additional fusion which, if anything, seems to have a protective effect. The very name “adjacent segment degeneration” implies that there is a tacit belief in trouble adjacent to an operated area and it is that belief that is so infectious amongst spinal surgeons that carrying out adjacent level fusions is endemic while there is no evidence base to support blaming the previous surgery over and above the natural and constitutional process of degeneration.
If there is no way of confirming that a degenerative motion segment is symptomatic and thus the source of pain, then spinal fusion or whatever has no particular target to aim at. It can hardly therefore be the raison d’être for concern about adjacent level disease when there is no way of knowing whether the incriminated level is the real source of symptomatic concern. It certainly makes the concept of reducing movement but not abolishing it (e.g., ligament support or disk replacement) as being beneficial as naively fanciful at best. Orthopaedic associations around the world made the problem of back pain a key issue for the “millennium” and this is well articulated by Nachemson. 120, 121 Diagnosing low back pain is verging on the impossible with one level disk degeneration, spondylosis, mild scoliosis, or low grade spondylolisthesis all being shown to have no predictive value as they are as common in people with pain as in people without pain. The radiologists Roland and van Tulder have already told us to stop using unproven labels. 122 The ability to diagnose facet syndrome has been disproven in several randomized trials and degenerative disk disease, isolated disk resorption, and segmental instability have all been described as “waste baskets” 120 (▶ Table 4.12). Diagnosing these can lead some patients into a sick role behavior.
Almost half of patients with this finding on radiography do not have back pain, so this finding may not be related to the patient’s pain
Advanced disk degeneration
Roughly 40% of patients with this finding do not have back pain, so this finding may be unrelated
Roughly half of patients with this finding do not have back pain, so this finding may beunrelated
Roughly half of patients with this finding do not have back pain, so this finding may be unrelated
Almost half of patients with this finding do not have back pain, so this finding may be unrelated
Almost half of patients with this finding do not have back pain, so this finding may be unrelated
More than 40% of patients with this finding do not have back pain, so this finding may be unrelated
Source: Reproduced with permission from Roland M, van Tulder M. Should radiologists change the way they report plain radiography of the spine? Viewpoint, Lancet 1998;352:229–230.
The worst culprits in all this surgical treatment of low back pain are the spine surgeons ourselves. Indeed, one of RAD’s old mentors, Alf Nachemson, with whom he did a fellowship in 1974, was never backward in coming forward and entitled one of his last papers “Failed back surgery syndrome is syndrome of failed back surgeons” 123—a paper that all back surgeons should read and particularly opinionated ones. He described our shortcomings (▶ Table 4.13) and concluded by saying that the overwhelming evidence of our failures explains the poor results of revision surgery for failed back surgery syndrome. Of course we need to pay attention to not only red flags but also yellow flags 123 (▶ Table 4.14), rid ourselves of matters that have no evidence base, and give up for good the idea that for the chronic back pain patient at the end of the line there is always a man/woman with a white coat and a scalpel in their pocket ready to do a spinal fusion. Ciol 124 found that in the United States 20% of patients over 65 who had undergone lumbar spine operations had one or more reoperations within 4 years and others have found higher rates. 125 Nachemson stated, not wholly with tongue in cheek, that at a number of meetings held in 1998 in the United States, the UK, and Sweden, orthopaedic surgeons were questioned on whether they would send a chronic low back patient to a fusion procedure and also if they would undergo the procedure themselves. ▶ Table 4.15 shows the results which speak for themselves about spinal surgeons having serious delusions of adequacy!
In the absence of any scientifically admissible evidence of efficacy or efficiency for revision back surgery, the syndrome of failed back surgery is described by the following combination of back surgeons’ shortcomings:
The overwhelming evidence of our failures explains the poor results of revision surgery for failed back surgery syndrome.
Would undergo fusion themselves (%)
Anyway, this textbook is not about back pain (thank goodness) but is about the treatment of spinal deformities and so be reassured by Weinstein 41 that “the incidence of back pain in patients with scoliosis is comparable to the incidence of back pain in the general population.” In the Iowa long-term follow-up study of 161 living patients with LOIS, 80% reported some backache and there were more, 86% in the control group of 100 with no deformity. ▶ Table 4.16 summarizes the position perfectly.
Scoliotic patients* (%)
(n = 161)
Scoliotic Patients† (%)
(n = 106)
(n = 100)
Never have pain
Rare pain (one to five times in life)
Occasional pain (few days per year)
Frequent pain (few days per month)
*Data adapted from Weinstein SL, Zavala DC, Ponseti IV. Idiopathic scoliosis: Long-term follow-up and prognosis in untreated patients. J Bone Joint Surg (Am) 1981;53:702–712.
†Adapted from Weinstein SL, Dolan LA, Spratt KF, Peterson K, Spoonamore M. Natural History of Adolescent Idiopathic Scoliosis: Back Pain at 50 years. Presented at the annual meeting of the Scoliosis Research Society, September 1998; New York, NY.
Thoracic Curves (Lenke 1, King 3) 125
You shouldn’t have forgotten by now that all idiopathic curves are lordotic including thoracic ones and so a crucial part of the correction strategy is to restore thoracic kyphosis (▶ 3). This can be carried out anteriorly (▶ Fig. 4.11 and ▶ Fig. 4.12) or posteriorly (▶ Fig. 4.13, ▶ Fig. 4.14, and ▶ Fig. 4.15). Thoracic curves tend to be stiffer than curves lower down because of the attached chest wall and because disk height is less. Of course if the curve is not too big and remains nice and flexible then it is easy to obtain correction by instrumentation only. However, for bigger curves that do not adequately correct on side-bending then something else needs to be done first to render the curve flexible enough to be taken up by instrumentation although one more often sees these big stiff curves perhaps unwisely being dragged into place by single posterior transpedicular instrumentation only. As a general rule if the thoracic curve does not correct by at least 50% then consideration should be given to a preliminary anterior multiple diskectomy if posterior instrumentation is your decision. 125 This so-called anterior release is the fundamental part of the Leeds procedure originally described all of 30 years ago and for precisely this reason for making curves flexible enough to be taken up by instrumentation (▶ Table 4.17). 67– 69 At that time it may have appeared somewhat aggressive to be going into the chest with someone who only had idiopathic scoliosis. Of course then it was generally perceived that adolescent idiopathic scoliosis did cause irreversible chest dysfunction but we knew the Brompton experience of the difference between EOIS and LOIS 16– 19 and we soon published our first few cases to show that a thoracotomy was in no way dangerous or detrimental to our patients. 68 There never is a problem with doing a thoracotomy for thoracic scoliosis and we have done hundreds and the results of subsequent instrumentation have been spectacularly good. The main purpose of the procedure was to provide space for the deformity to collapse into (now reinvented and called an anterior release) and therefore to correct spontaneously as postoperative X-rays clearly indicated with more than half of the ultimate correction being achieved before the metalwork was put in. And so we have never regretted carrying out an anterior multiple diskectomy. Four or five disks, however, are quite sufficient for the adolescent with a stiffer thoracic curve (▶ Fig. 3.27) so that a posterior second stage, if chosen, can effect an excellent correction (▶ Fig. 4.16). Looking at the lateral radiographs it can be seen how a nice physiological kyphosis is naturally restored after anterior multiple diskectomy with shortening of the front of the spine and this of course forms the other essential part of the Leeds procedure. Pari-passu with that naturally comes some correction in the frontal plane as well. It does seem rather excessive and traumatic to have done all the necessary preparatory flexibility work anteriorly to have another anaesthetic and big operation from the back when the front of the spine is asking for instrumentation then and there (▶ Fig. 4.11).
Fig. 4.11 Anterior only—thoracic curves. (a) PA radiograph of a Lenke 1AL right thoracic curve. (b) Lateral radiograph showing the typical thoracic flat back. (c) PA radiograph after anterior instrumentation showing an excellent correction going no lower than T12. (d) Lateral radiograph showing restoration of a natural thoracic kyphosis. (e,f) PA and lateral radiographs at follow-up 2 years later showing a solid anterior fusion with maintenance of the corrected position. A perfect result from one procedure and a very happy girl.