Adolescent idiopathic scoliosis and back pain

This broad narrative review addresses the relationship between adolescent idiopathic scoliosis (AIS) and back pain. AIS can be responsible for low back pain, particularly major cases. However, a linear relationship between back pain and the magnitude of the deformity cannot be expected for any individual patient. A large number of juvenile patients can remain pain-free. The long-term prognosis is rather benign for many cases and thus a tailored approach to the individual patient seems mandatory. The level of evidence available does not allow stringent recommendations for any of the disorders included in this review.


Background
Scoliosis is a frequent pathology in adolescents. Young patients and their families frequently blame (minor) deformities, considered to be the cause of back pain. In our opinion, the relationship between the two is not so clear based on scientific evidence. For this reason, we here review this topic in depth.
A high prevalence of low back pain (LBP) among children and adolescents has been identified in the last few decades. A recent systematic review and meta-analysis [1] has reported a figure of 39.9 % (95 % CI ranging from 34.2 to 45.9 %) for lifetime prevalence. In terms of aetiologies, a retrospective study of almost 2000 patients less than 21 years-old referred for a spine evaluation reported that when a pathology is identified, the most frequent diagnosis was scoliosis (1439/1953), followed by Scheuermann's kyphosis (163/1953) and spondylolisthesis (154/1953) [2]. Other series of cases have also shown similar findings [3,4]. However, the role of spinal deformities in LBP in the general population is not clear. For this reason, we here review the recent literature on the associations between pain and the most common adolescent spinal deformity, which is adolescent idiopathic scoliosis.
Trying to establish a causal relationship between deformities and back pain is not an easy task. Since Sir Austin Bradford Hill [5] published his important paper in 1965, the difficulties and limitations in drawing conclusions on the causal relationship between two variables have been the focus of several publications [6][7][8][9].
Our aim is to review the literature to evaluate the association in adolescents between back pain and idiopathic scoliosis, reported in studies done in different settings and/or with different perspectives.

Methodology
This paper is the product of the collaboration between 2 clinicians: an orthopaedic surgeon highly focused on deformities (FP) and a rheumatologist with a special interest in juvenile spinal disorders (FB). We have tried to find clinically relevant information that might answer some of the questions the practitioners have. Therefore this article does not aim to be a systematic review as we have not applied strict methodology as recommended, for example, by the Cochrane collaboration, for such reviews. We began with a bibliography search limited to MEDLINE and expanded this body of literature with a search of the publications cited in the selected articles. The search was done using several key words for back pain (backache OR back pain OR low back pain OR lumbar pain OR vertebral pain OR spinal pain) as well as for the age category of interest (adolescent OR teenager OR juvenile OR paediatric OR infant OR children). We then limited the search to metaanalyses and expanded it to include "systematic reviews", and "cohort or longitudinal studies", if the information gathered was too limited with the most stringent criteria. Our focus was on back pain and consequently the articles not including data on symptoms were excluded except for the Epidemiology section.

Adolescent idiopathic scoliosis (AIS) and back pain
According to the glossary of the Scoliosis Research Society (SRS) (http://www.srs.org/professionals/online-educationand-resources/glossary/revised-glossary-of-terms), idiopathic scoliosis can be defined radiographically as a lateral curvature of the spine greater than or equal to 10°Cobb with rotation, of unknown aetiology.
The possible relationship between scoliosis and back pain can be approached from different angles. Some examples may be: a) comparing the epidemiological data of the two diagnostic entities; b) examining the prevalence of back pain in adolescents with scoliosis or the prevalence of scoliosis among back pain sufferers; c) examining the prevalence of back pain over time in scoliosis left untreated; or d) examining the impact of treatment of the deformity on back pain.

Prevalence of scoliosis
Adolescent idiopathic scoliosis is considered a quite common disorder, with a pooled referral rate for radiography of 5.0 %, according to a meta-analysis of 36 studies looking at the effectiveness of school scoliosis screening [10]. According to the definition used, the overall prevalence of AIS has been reported to range between 0.47 and 5.2 % in a recent review [11]. There are cases of scoliosis secondary to other pathologies, but those idiopathic (i.e. AIS) are by far the most frequent. In the series published by Rogala et al. [12], all but 9 of 1231 cases of structural scoliosis were idiopathic. We have limited this section of the review to adolescents or young adults with AIS.
There are other studies that have prospectively evaluated the epidemiology of scoliosis and its natural history in large samples of adolescents, but the most frequently referenced do not include any information on symptoms [11][12][13][14][15]. Even the recent meta-analysis previously mentioned did not report data on back pain [10].
Scoliosis can be associated with other deviations from the normal spinal morphology. For example, in the series reported by Deacon et al. [16], 35 out of 50 adolescents were diagnosed with both Scheuermann's disease and scoliotic curves. Overall, 43 curves were identified in these patients, which were divided in 2 types: 13 were apical at the same level as the kyphosis, and 30 occurred in regions of compensatory lordosis. Finally, in the small series reported by Blumenthal et al. [17], 4 out of 13 cases of lumbar Scheuermann's also had scoliosis but none required treatment.

Pain associated with scoliosis
Regarding associated symptoms, a German study evaluated the data of more than 640,000 youths included in an insurance database. ICD diagnosis codes M40-43 would be relevant for the purpose of our review (M40: kyphosis and lordosis; M41: scoliosis; M42: spinal osteochondrosis; and M43: other deforming dorsopathies). For scoliosis, data from 2002 showed the following means in % (girls/ boys) for prevalence at 0-14 years of age and at 15-24 years: M41: 2.31 (2.51/2.12) and 3.44 (3.80/3.07) [18]. Ramirez et al. [19] reported on more than 2400 subjects with AIS. Of these, 23 % reported back pain at the time of diagnosis. An additional 9 %, initially free of pain and managed with observation alone, developed pain during follow-up. A study from Japan including more than 30,000 adolescents concluded that the subgroup with scoliosis had an approximately 3 to 5 fold increased point and lifetime prevalence of backache in the upper and middle right part of the back [20]. In a prospective multicentre study, Lonner et al. [21] compared three groups of adolescents including 894 with AIS and 31 healthy controls. The pain scores on the specific subdomain of the SRS-22 questionnaire were 4.15 in the scoliosis group and 4.24 among the controls, which is not a significant difference.
A retrospective chart review of a random sample of 310 individuals (10-17 years) selected from all cases of AIS referred to a Canadian university hospital has been published recently [22]. The authors concluded that the prevalence of back pain was "moderately high" but the reported data on pain do not seem very homogeneous (gathered by the attending physicians, or reported by medical references, or by letters from the parents, narrative or through a pain score, with or without a specific topography of pain recorded, etc.). Of note, severe pain was documented in only 1 % of the charts [22].
A prospective multicentre study including 744 patients (621 females) surgically treated for AIS addressed the differences between genders in functional outcome. Before surgery, males were aged an average of 15.2 years and females 14.0 years (p < 0.001) with no significant differences in maximum Cobb angles (F: 53.3°and M: 55.9°) or Risser grades (F: 3.2 and M: 3.5) [23]. At baseline, the scores on the pain domain of the SRS-30 were 4.1 among girls and 4.3 in boys, below the statistical significance threshold [23].
A recent meta-analysis has been published comparing selective versus non-selective thoracic fusion in Lenke 1C curves [24]. Preoperative data on pain using the SRS-22 has shown slightly different scores between the two groups that is 4.13 (0.77) in the selective fusion group versus 3.92 (0.79) in those undergoing non-selective procedures. The difference is statistically significant (p = 0.038), but is at the limit of the minimum clinically important difference (MCID), reported to be 0.20 [21,25,26]. As any other tool, the questionnaires used to gather information on patients with scoliosis have some limitations. An American study evaluated SRS-22 performance in 450 healthy adolescents (mean age: 16 years; range: 9.3 to 21.8 years). Concerning specifically pain, the mean score was 4.3 ± 0.6 with males scoring a bit higher (actual figures by gender not reported but r = 0.103, p < 0.05) [27]. Moreover, ethnicity was also a significant factor with African Americans scoring significantly higher at 4.5 (i.e. less pain) than Hispanics (4.3). Other sociodemographic variables were significantly associated with different domains of the questionnaire [27].
Other studies have also shown that culture and ethnicity have an influence on outcomes with Caucasians reporting more pain than East Asians on the SRS-30 [28,29]. Finally, in a Polish study, the living environment has also been reported to influence the results of the SRS-24 questionnaire, with rural patients reporting more pain than those from an urban environment [30].
The Pediatric Outcomes Data Collection Instrument (PODCI), a multidimensional tool developed in North America (also known as POSNA for Pediatric Orthopedic Society of North America) [31], was used to evaluate 102 patients with AIS (as well as other groups of patients with different pathologies) [32]. Scores of these patients were compared to those of a small group of 27 "healthy" adolescents evaluated in a different study [33]. Of these 102 patients, 95 (86 girls) filled in the patient questionnaire. The scores of the two groups for the comfort/pain scale were 86.7 ± 14.5 for the healthy group versus 75.2 ± 22.4 for the AIS group. Each dimension is scaled from 0 to 100, with 100 the most favourable outcome. The result is statistically significant at p < 0.05; however, the differences between the two groups do not seem clinically relevant when compared with the properties of the same tool described elsewhere [34].
Considering another perspective on the association between symptoms and scoliosis, a study carried out on 1743 men in the military (range: 18-30 years) found a prevalence of idiopathic scoliosis of 6.65 % among those with no symptoms and free of any lytic or olisthetic lesions [35]. Of note, none had a Cobb angle >20°. Prevalence ranged from 13.3 to 23.8 % in the symptomatic and asymptomatic subgroups with uni-or bilateral pars break and 18.3 % among those without any lesion of the posterior arch but reporting back pain.
Clark et al. [36] have just published the results of a prospective, population-based, birth cohort study with complete data on 3184 participants. Subjects were evaluated at age 15 for scoliosis using total-body dualenergy X-ray absorptiometry (DXA) and surveyed at age 18 for pain and function. A multivariable analysis shows a significant association between small spinal curves (≥6°) at age 15 and self-reported back pain at age 18. Spinal curvature is also associated with days off school and avoidance of activities.
It appears quite clear from comparing the prevalence of back pain and scoliosis that the latter cannot be the main explanation for LBP for a majority of adolescents reporting such symptoms, although scoliosis playing a role in some patients cannot be ruled out. However, comparing the results of different studies does not allow any firm conclusions to be drawn on the possible causal relationship between different variables. Nevertheless, the compelling and overwhelming predominance of girls in the adolescent cohorts with idiopathic scoliosis (up to almost 90 % in some series, such as that published by Théroux et al. [22]) casts doubt on the aetiological role of scoliosis on back pain in the general population, at least among boys. Another piece of information that suggests a limited role of scoliosis in back pain is the weak or even absent correlation between the magnitude of the curves measured by the Cobb angle and the presence of pain [22,28,37,38].
The main references quoted in this section are summarized in Table 1.

Effect of treatments on back pain
In this section, the effects of different treatments on the magnitude of scoliotic curves have not been included. Such an analysis, as well as the impact of treatment in cosmetic, psychological or other issues, would be beyond the scope of this review. Further, long-term follow-up studies in middle-age adults have also not been included, as the role of degenerative changes cannot be identified.

Conservative management
Numerous conservative techniques have been used to treat AIS, such as acupuncture, braces, electrical stimulation, exercises, foot orthosis, osteopathy, and yoga. The most recent systematic review on braces for AIS reported long-term stability in terms of back pain (very low quality evidence) in addition to a similar outcome for quality of life and psychological and cosmetic issues [39]. The authors highlighted that performing a meta-analysis was not possible due to differences among the studies.
In a systematic review by Maruyama et al. [40], bracing was compared to observation, other conservative treatments, and surgery. Considering the effect on QoL, the evidence was considered "conflicting" for each of the 3 comparisons. As such, the authors prudently concluded only that braces "may not have a negative impact on patients' QOL".   The authors of the most recent Cochrane review comparing surgical and conservative management of adolescent idiopathic scoliosis did not find any papers that would allow drawing conclusions regarding back pain [41]. Several other recent review articles have not specifically included information on back pain [42][43][44][45][46].
Osteopathic manipulative treatments in children were reviewed by Posadzki et al. [45]. Only one study of the included references concerned the treatment of scoliosis. There is currently no evidence to support such an approach for the treatment of AIS.
In his recent review paper, Kim [47] found no evidence available for yoga, Pilates, foot orthosis or acupuncture. On the contrary, he found level I and II evidence for bracing and level II evidence in favour of scoliosis specific exercises (despite some concerns such as the difficulty of attributing all the benefit to exercises due to frequent associated bracing, the order of magnitude of the correction and some doubts concerning the level of evidence of the studies) and against electrical stimulation. Indeed, only one well-designed study published in 1995 is specifically quoted in Kim's review. In their prospective controlled trial Nachemson et al. [48] compared 3 groups: 1 braced, 1 treated by electrical stimulation, 1 observed. The success rates were 74 % in braced, 33 % in electrical stimulation and 34 % in observed. Thus the conclusion was that electrical stimulation is comparable to natural history. Kim [47] highlights the fact that the concept of correcting scoliosis by stimulating to muscles has almost disappeared but animal studies are being done to evaluate the possibilities to influence asymmetric vertebral growth by means of electrical stimulation. Only one study cited by this review addressed the impact on pain, but this study assessed exclusively non-structural scoliosis with a mean Cobb angle < 8° [49].
A recent systematic review on the effects of exercise in patients with AIS concluded that exercises improve quality of life [50]. The data on pain is clearly reported in only two of the nine studies included in the review, of which both used the SR-22. In the RCT by Monticone et al. [51], at baseline, the scores in the pain dimension were 3.8 (0.4) in the exercise group and 3.9 (0.5) among the controls. At 1-year follow-up, the same values were 4.7 (0.2) and 4.2 (0.4), respectively. In the retrospective study by Noh et al. [52] (method of randomisation not clearly mentioned if any), the same scores were 4.5 (2.4) in the corrective spinal technique (combining the Schroth concepts and core stabilization techniques) and 3.8 (1.6) in the control group (conventional exercise) at the beginning of the trial and 4.9 (1) and 4.6 (2.4), respectively, at 3 months. Notably, both groups began with 16 patients each, but at follow-up, data were reported for only 8 subjects in the experimental group and 4 patients in the control group.
A recent trial compared an 8-week program of weeklysupervised spinal stabilisation exercises to unsupervised exercises [53]. Pain, evaluated by numerical rating scale, scored 5.4 on average in each group. After 8 weeks, the improvement was 3.9 (1.8) in the supervised group versus 2.2 (2.0) in the unsupervised group. The study has clear limitations, such as the small numbers and the dropout rate in the unsupervised group, as highlighted by the authors.
The beneficial effects of bracing in terms of pain are influenced by compliance with the SOSORT Brace Treatment Management Guidelines (SBTMG) [54].

Surgical treatment
A recent review of the literature with statistical analysis included 16 cohorts with data on pain. Of these, 81 % reported a statistically significant postoperative (2 years) improvement in pain, however, in only one study were the improvements clinically important [55]. Similarly, another systematic review concluded in favour of surgery to reduce the magnitude of spinal curves but stated that evidence supporting the correlation of this result with reduced pain is lacking [56].
A multicentre, prospective, consecutive clinical series reported on the prevalence and predictors of pain in AIS treated surgically [57]. Preoperative data from 1433 patients (80.4 % girls; mean curve: 56.7°) and changes in pain based on a subset of 295 cases with complete data and 2 years follow-up were reported. Pain was evaluated using the SRS-22 questionnaire. Preoperatively, 77.9 % of patients reported some back pain, mainly mild to severe. The prevalence of mild to severe pain in the past 6 months improved with surgery from 78.3 % of the subset (N = 295) preoperatively, to 68.8 % at 1 year and to 64.4 % at 2 years. Indeed, at follow-up, 40 % of the 295 patients reported a decrease in pain, 38.6 % no change and another 20.3 % reported an increase in pain. Of note, while reported pain decreased with treatment, analgesic use remained unchanged. Reduction in pain and absolute pain at 2 years were both correlated with patients' perception of deformity, evaluated by means of the SAQ (Spinal Appearance Questionnaire). Such an association might be interpreted as decreasing the likelihood of a major biomechanical role of the deformity to explain pain.
The previously cited multicentre study that included 744 cases treated surgically reported outcomes at 2-years follow-up [23]. The SRS-30 pain scores improved to 4.3 in females and 4.5 among males, which was a significant difference (p = 0.003). Of note, although in absolute values the differences between genders were the same pre-and post-operatively (i.e. 0.2), the pre-operative difference was not statistically significantly [23].
Akazawa et al. [58] reported the results of a survey on health-related quality of life and LBP in a group of 80 patients with AIS (mean age at follow-up: 47.4 ± 6.8 years) and a matched control group (mean age: 46.7 ± 6.3 years). There was no statistical difference in terms of pain (4.3 ± 0.6 in the idiopathic scoliosis group versus 4.2 ± 0.5 in the controls on the SRS-22 pain score) and a small difference of 1 point on the Roland & Morris questionnaire (2.4 ± 4.1 in the scoliosis group versus 1.4 ± 3.1 in the control group), which is below the MCID [59].
A prospective registry based study (N = 584 patients) recently reported on the prevalence of postoperative pain and its relationship with preoperative pain [60]. Patients reporting postoperative pain had significantly worse preoperative pain scores (SRS-22) with a mean of 3.8 ± 0.8 versus 4.2 ± 0.7, which is both statistically (p < 0.001) and clinically significant.
Another multi-centred registry study of AIS patients treated surgically (n = 190 with 2-year follow-up and n = 77 with 5 years follow-up) reported on pain prevalence and trajectories after spinal fusion. Patients were evaluated by means of the SRS-30. Moderate to severe pain in the past month was reported by 35 % of patients preoperatively. The same figure was 11 % at 1-year follow-up and 15 % at 2-and 5-year follow-up. Curiously, medication usage did not change significantly: daily opioid use was reported by 1 % across all time points and weekly or less frequent use of non-opioid medication also remained stable (23 % preoperatively, 25 % at 2-year follow-up and 25 % at 5-year follow-up) [61].
A longitudinal cohort of 745 patients with AIS treated surgically was followed for 2 years and the results of the pre-and postoperative SRS-22 compared [62]. The pre-and postoperative pain scores were 4.16 ± 0.71 and 4.31 ± 0.72, respectively. The difference is statistically significant (P < 0.0001) but below the MCID. At 2 years, the correlation of the reported pain with the satisfaction score was 0.260, slightly lower than for appearance (0.280) and for the total SRS-22 score (0.398) but higher than for activity (0.172) or mental (0.202) scores.
We have identified two further studies comparing the effects of bracing and surgery for the treatment of AIS. A short-term study using the SRS-22 questionnaire evaluated the outcome at age 16.3 ± 1.6 years of 109 Dutch adolescents with AIS treated either by brace, surgery, or brace followed by surgery [63]. The group treated by brace only reported significantly less pain at follow-up than the two groups that received surgery (4.5 vs 4.1). Moreover, there was no interaction between pain and the time passed since the end of treatment (i.e. interval at which the questionnaire was completed) for any of the groups [63]. A Danish group had reported on the outcome at 10 years of 181 patients with AIS treated either by brace or surgery [64]. Pain was assessed by means of 6 questions using VAS. The intensity of back and leg pain was mild, with values of 2.5 (0-10 scale) for the most severe back pain within the last 2 weeks for the braced patients and 2.1 in those operated. The scores of the SF-36 were compared with a Danish control cohort. The bodily pain scores were 74.6 (95 % CI.: 68.7-80.5) in the braced subjects, and 71.4 (66.1-76.8) in the surgical cohort, which is not a significant difference. However, both values were significantly below the scores for the reference population (79.8 (77.6-82.0)) [64]. The authors of the Ste-Justine cohort did not find any major difference in self-reported pain based on preoperative characteristics, the degree of surgical correction or the distal level of fusion [65].
A recent study has shown that both the SRS-22 and SRS-24 questionnaires are able to discriminate between AIS patients with preoperative curves >80°and those <45°. Specifically concerning pain, evaluated by the SRS-22 tool, the scores were 4.2 ± 0.7 for small curves versus 3.8 ± 0.9 for large (p = 0.003). The corresponding values obtained with the SRS-24 were 3.8 ± 0.6 and 3.5 ± 0.7 (p = 0.002). Postoperatively, the same questionnaires did not show any significant difference in pain according to neither the percent correction (<40 % Vs ≥80 %) nor when stratifying into post-operative curves <11°or ≥29° [66].
A relationship between scoliosis and pain might be mediated by variables other than the curve itself. It has been shown that compared with asymptomatic volunteers, baseline data on lumbar stiffness evaluated by means of Lumbar Stiffness Disability Index (LSDI) correlates with pain (Pearson correlation "r" LSDI versus SRS-22 Pain: −0-749) and functional limitations (LSDI vs SRS-22 Function: −0.760) in patients with spinal deformity [67].

Untreated scoliosis
The literature comparing untreated AIS with normal controls was recently reviewed and statistically analysed [68]. Twenty-one cohort studies were included that used different questionnaires (SRS-24/23/22/22r/30). Of these, 81 % reported statistically worse pain than unaffected adolescents; however, in only 5 % of cohorts was the difference clinically important. In comparison, in patients' self-image, almost three-quarters of the studies with a statistically significant difference were also clinically significant.
A few studies on the conservative management of adolescents with idiopathic scoliosis have included control groups not braced. In the series reported by Pham et al. [69], a subgroup of 32 patients (30 girls, age 12.5 ± 1.4 years) with Cobb's angles 26.5°± 2.4°was left non braced until the next follow-up visit at least 3 months later and compared with patients wearing a specific brace (Chênau brace) full-time or during the night (weaning period). No differences in terms of pain were found.
Another study compared 78 patients with AIS being braced with 136 patients being observed. The two groups PedsQL (only 1 item in one of the 4 dimensions focus specifically on pain).
Back pain did not change in the long term (very low quality evidence) The authors highlight that it was not possible to perform a metaanalysis due to important clinical differences among studies.  The reduction in OSW was similar in both groups and the improvement in PSFS was higher in the supervised group but the difference between the 2 groups < MCID in adults.
All the patients included in the trial had low back pain at baseline.   were very similar except for the magnitude of the curves (24.6°for the observed subjects and 34.5°for those being braced, p < 0.0001). The authors used the parents' forms for the Child Health Questionnaire (CHQ) and the American Academy of Orthopaedic Surgeons Pediatric Outcomes Data Collection Instrument (PODCI). National normative values were available for the CHQ but not for the second instrument. No significant differences between the two groups of patients (observed vs braced) were reported for any of the pain dimensions of the two questionnaires. Moreover, when compared with the scores of healthy children, those with AIS showed similar scores on the bodily pain domain of the CHQ [70].
The main references quoted in this section are summarized in Table 2.

Conclusions
Back pain in adolescents is quite common, especially in girls. There is no doubt that some AIS patients are back pain sufferers; however, pain does not seem to be a major problem for the vast majority of adolescents with an idiopathic form of this type of deformity. Some elements from the literature suggest that the link between pain and scoliosis is not strongly linked with a biomechanical problem. In most studies, pain has not shown a strong correlation with the Cobb angle, untreated cases fare reasonably well from the perspective of back pain, there is a much greater difference between genders in the incidence of scoliosis compared with the epidemiological data on back pain, and patients' self-perception of their image correlates with pain. All these findings argue against a major aetiological role of the idiopathic scoliotic deformity of adolescents on back pain. However, the impact of pain in adults' scoliosis is entirely different [71,72] and out of the scope of this review.