European Spine Journal 2016 (Sep); 25 (9): 2809–2821 ~ FULL TEXT
Mieke Dolphens, Stijn Vansteelandt, Barbara Cagnie, Andry Vleeming, Jo Nijs, Guy Vanderstraeten, Lieven Danneels
Department of Rehabilitation Sciences and Physiotherapy,
Faculty of Medicine and Health Sciences,
Ghent University, Campus Heymans (UZ, 3B3),
De Pintelaan 185, 9000, Ghent, Belgium.
PURPOSE: To investigate the factors related to the 1-month period prevalence of low back pain (LBP), neck pain (NP) and thoracic spine pain (TSP) in young adolescents, thereby considering potential correlates from the physical, sociodemographic, lifestyle, psychosocial and comorbid pain domains.
METHODS: In this cross-sectional baseline study, 69 factors potentially associated with spinal pain were assessed among 842 healthy adolescents before pubertal peak growth. With consideration for possible sex differences in associations, multivariable analysis was used to simultaneously evaluate contributions of all variables collected in the five domains.
RESULTS: A significantly higher odds of LBP was shown for having high levels of psychosomatic complaints (odds ratio: 4.4; 95 % confidence interval: 1.6-11.9), a high lumbar lordotic apex, retroversed pelvis, introverted personality, and high levels of negative over positive affect. Associations with a higher prevalence and odds of NP were found for psychosomatic complaints (7.8; 2.5-23.9), TSP in the last month (4.9; 2.2-10.8), backward trunk lean, high levels of negative over positive affect and depressed mood. Having experienced LBP (2.7; 1.3-5.7) or NP (5.5; 2.6-11.8) in the preceding month was associated with a higher odds of TSP, as were low self-esteem, excessive physical activity, sedentarism and not achieving the Fit-norm.
CONCLUSIONS: Psychosomatic symptoms and pain comorbidities had the strongest association with 1-month period prevalence of spinal pain in young adolescents, followed by factors from the physical and psychosocial domains. The role that "physical factors" play in non-adult spinal pain may have been underestimated by previous studies.
KEYWORDS: Adolescent; Back pain; Multivariate analysis; Neck pain; Spinal pain
From the Full-Text Article:
On a global basis, spinal pain represents a common and
significant condition with a tremendous social and economic
impact [1–5]. Since adult spinal pain problems
might originate in childhood and young adolescence
[6–13], thorough understanding of spinal pain at young age
is needed for designing effective prevention and management
Idiopathic adolescent spinal pain (IASP) cannot be
explained straightforwardly in simple models as it has
many aspects. Accordingly, a multidimensional approach
to the understanding of IASP is currently accepted as the
most appropriate perspective. Such an approach should
incorporate factors from all domains of the biopsychosocial
model, including developmental, educational and
cultural background [14–18]. Earlier studies [9, 13,
17–34] and reviews [15, 35–41] identifying risk factors
for IASP have attempted to evaluate the contribution of
multiple different physical, psychosocial, demographic,
environmental and socioeconomic factors to the development
of spinal pain at young age, mostly in relation to
low back pain (LBP) [9, 19, 20, 22–26, 35–38].
However, interpreting the literature on associated and
risk factors is complicated by 4 factors. First, methodological
approaches differ substantially across studies,
including inconsistency and lack of standardization of
spinal pain definitions. Second, most previous studies (e.g.,
refs. [10, 18–22, 42, 43]) relied on statistical procedures
such as multivariate logistic regression, whether or not
preceded by univariate logistic regression analysis, that are
less well suited to deal with large numbers of correlated
factors relative to the sample size . As a consequence,
some risk factors may have been obscured or inflated .
Furthermore, such analyses do not allow weighing the
importance of clusters of variables constituting (putative)
Third, despite accruing evidence for genderspecificity
in spinal pain figures and/or pain sensitivity
[8–10, 15, 17, 21–25, 27–32, 35, 40, 42, 45–55] as well as
in associative or risk factors for IASP [8, 21, 25–27,
42, 54–58], the potentially profound influence of gender on
factors related to the development and maintenance of
IASP has not yet been deeply scrutinized. Fourth, previous
research has identified biological maturation or pubertal
development as a better predictor for IASP than age
[59–62]. Still, the maturational or developmental age of
non-adult study populations has rarely been considered
and, importantly, the vast majority of research overlooks
the well-known gender difference in timing of biological
maturation with girls being advanced, on average, about
2 years when compared to boys . As a result, there is
debate as to the nature of and relationships among the
factors that underlie IASP.
The aim of the current study was to investigate the
(phenotypic) factors related to spinal pain prevalence in
young adolescents before pubertal peak growth using a
comprehensive multivariable approach that assessed the
contribution of factors from multiple domains.
Materials and methods
A cross-sectional population-based study was conducted
from September 2008 to February 2009 in Flanders, Belgium.
64 schools were selected to represent educational
networks and levels within Flemish mainstream education.
From these schools, girls in year 5 of primary education
(age: 10.6 ± 0.47 years) and boys in year 1 of secondary
education (age: 12.6 ± 0.54 years)  were eligible to
participate. These gender-specific age groups were chosen
as the mean age of peak linear growth is known to range
from 11.6 to 12.1 years in girls and from 13.8 to 14.1 years
in boys . The study was restricted to children without
neurological conditions, rheumatic disorders, metabolic or
endocrine diseases, major congenital anomalies, skeletal
disorders, connective tissue disorders, previous spinal
fracture or previous spinal surgery. Parental/guardian as
well as child consent, was obtained before enrolment.
Ethical approval was granted by the ethics committee of
the Ghent University Hospital.
In the class room or at the local pupil guidance center,
with the investigator present, children were asked to
complete a questionnaire on spinal pain and its potential
associated factors. Presumed associated factors included
physical, sociodemographic and lifestyle characteristics,
psychosocial factors, and measures of comorbid pain conditions.
Physical measurements took place between 3 and
5 days after questionnaire assessment. Sociodemographic
data with regard to parental education and employment
were collected through a parental questionnaire.
Outcome measures: spinal pain
The 1-month period prevalence of LBP, neck pain (NP)
and thoracic spine pain (TSP) was determined by selfcomplete
questions including a preshaded manikin
[58, 64, 65]. The questions relevant to this study included
the following: “Has your low back/neck/upper back been
painful in the last 4 weeks?” The 3 categories listed were
not mutually exclusive.
Spinal pain was defined as follows: a discomfort or pain
in the back or neck that is considered to be a local,
uncomfortable feeling in the back or neck, with the possibility
of radiation to other parts of the body. Problems
due to fatigue related to a single exercise are not considered
as back or neck problems. The discomfort or pain can
be intermittent or constant, gradually developed or with a
sudden onset. Spinal pain due to menstruation is not taken
into account. The definition as such was not presented to
the adolescents, but it was orally “translated” in a language
that could be understood by the adolescents. This was done
during the instructions before completion. During
completion, an examiner blinded to the results of spinal
measurements was present to provide assistance if needed.
Collected parameters covered a broad range of measurements
from 5 domains. The variables included in this study
are listed in Online Resource 1.
The study investigators measured body height and weight
using a standardized procedure . Body mass index was
calculated as the ratio of weight to square height and was
transformed into 3 categories (thin, normal, and overweight
or obese) using the cut-off points for age and gender
defined by Cole and colleagues [66, 67]. To determine the
proportion of trunk length to body length, sitting height
was measured . Habitual standing posture in the
sagittal plane was quantified as described previously
[56, 64]. More specifically, data were collected regarding
gross body segment orientations (i.e., trunk lean, body lean,
and anteroposition of the head), specific spinopelvic characteristics
(i.e., pelvic orientation (pelvic tilt, sacral inclination),
spinopelvic extensiveness parameters (number of
vertebrae included in the thoracic kyphosis and lumbar
lordosis, vertebral level of the thoracic apex, lumbar apex,
and intercristal line), magnitude of spinal curves (thoracic
kyphosis, lumbar lordosis) and knee alignment. Furthermore,
postural subgrouping according to global body
alignment (neutral, sway-back, and leaning-forward) was
applied using the categorization proposed by Dolphens
et al. [58, 68]. In full flexion and extension position, the
thoracic and lumbar spine, and trunk and sacral inclination
were recorded using a skin-surface electromechanical
device, the Spinal Mouse (Idiag; Voletswil, Switzerland).
Based on these data combined with the data obtained in
upright standing, the ranges of flexion and extension
motion were determined for the thoracic and lumbar spine.
For spinopelvic range of motion assessment, all motions
and subsequent measurements were performed according
to the manufacturer’s specifications. The presence of a
gibbus deformity was evaluated using the forward-bending
test. To assess generalized joint laxity, the Beighton score
was determined . A participant was classified as
hypermobile when a Beighton score of C4/9 was obtained
. Hand dominance was recorded.
Data on gender, chronological age, and rough indications of
biological age (years from age at peak height velocity (PHV),
maturity status classification, percentage of adult stature and
predicted growth remaining) [64, 70–72] were collected. In
girls, information was gathered on whether they had started
menstruating. The respondent’s level of education (only in
boys) and educational network were recorded. The proxy
measures of family characteristics were family composition,
family size, parental educational attainment, parental
employment status, and parental social class as obtained via
coding occupational information [International Standard
Classification of Occupations (ISCO88) 
For assessing physical activity levels and sedentary
behavior, the participants completed a questionnaire based
on the Flemish Physical Activity Questionnaire . The
pen and paper version of this questionnaire has been shown
to be a reliable and reasonably valid instrument in the
appropriate age group . Physical activity levels were
assessed by combining the amount of active transportation,
physical activity at school and physical activity during
leisure time. The amount of screen behavior, time spent on
homework and time spent reading outside of school hours
were recorded to compute a composite index of sedentary
behavior. Furthermore, according to the Dutch physical
activity guidelines for subjects younger than 18 years, the
children were asked whether they achieved the Dutch
Norm for Health-enhancing Physical Activity (i.e., 1 h or
more of at least moderate-intensity physical activity each
day) and Fit-norm (i.e., 20 min or more of vigorous-intensity
physical activity on at least 3 days each week).
Participants who met at least 1 of the 2 previous mentioned
norms adhered the so-called “Combi-norm” [74, 76, 77].
Finally, participation in physical activity in leisure time
versus not was recorded as a categorical lifestyle marker.
Psychosocial factors were recorded using the Self-Perception
Profile for Adolescents [78, 79] with the less cumbersome
question format proposed by WichstrØm , the
Rosenberg Self-Esteem Scale [81, 82] the Short Amsterdam
Biographical Questionnaire for Children designed to
assess personality traits , Positive And Negative Affect
Schedule [84, 85], the short version of the Depression
Questionnaire for Children , the Satisfaction With Life
Scale [87, 88], the Subjective Vitality Scale , and the
Inventory of Parent and Peer Attachment [90, 91]. For
detailed information see Online Resource 1.
Comorbid (psycho) somatic complaints
To ascertain the participants’ experience of spinal pain, the
1-month period prevalence of pain in each of the 3 spinal
regions was asked (LBP, NP, TSP) [58, 64, 65, 68]. For
assessing the prevalence of other common somatic symptoms
(e.g., headache, abdominal pain, sore throats, etc.), a
psychosomatic complaints list was used with 26 items and
five response categories:
“seldom or never”, “almost every
month”, “almost every week”, “more than once a week”
and “almost every day”. Scores range from 26 to 130, with
high scores reflecting more frequent psychosomatic complaints
For each of the considered outcomes, a logistic regression
model was fitted using bi-level selection methods [93, 94]
on all the variables collected in the 5 domains of interest
described above. These methods are specifically designed
for predicting an outcome measure in the presence of a
large number of correlated factors, thereby taking advantage
of the grouping structure in the domains of interest to
estimate regression coefficients and to select variables by
making use of a group-structured penalty. The resulting
odds ratios (ORs) and associated 95 % confidence intervals
(CIs) were reported. Level of significance was set at
α <0.05. The relative importance of each variable was
evaluated using variable importance scores (VISs). For
each domain, the average VIS was calculated as was the
average group ranking. Further insight into the relative
importance of the different domains was obtained by
contrasting the Nagelkerke R2 between the final selected
model and submodels obtained upon applying bi-level
selection methods on subsets of the 5 domains.
See Online Resource 2 for a more detailed description.
All statistical analysis was performed using the libraries
grpreg and logistf in RStudio Version 0.97.320 statistical
software (RStudio, Inc., 2009–2012).
Any variable with more than 85 missing values (8.0 %)
was excluded from the analysis and 354 incomplete cases
(29.6 %) were discarded, thus leaving a total of 69 variables
collected on 842 participants of whom 385 were girls
and 461 boys. Dummy variables were created for categorical
variables having more than 2 levels. Logarithmic
and square root transformations were used where appropriate.
See Online Resource 1 for more detail.
LBP, NP and TSP were present in 102 (12.1 %), 77 (9.1 %)
and 43 (5.1 %) of the 842 pre-PHV subjects, respectively.
On the domain level, the set of variables comprising the
comorbid pain domain were most important in explaining
1-month period prevalence of spinal pain as can be concluded
from this domain’s low mean rank and high average
importance score (Table 1). Based on the average group
ranking, this domain was followed by the physical and
psychosocial domains, respectively, in each of the 3 spinal
pain sites, whereas the sociodemographic and lifestyle
domains were relatively less important.
The ORs, 95 % CIs and p values estimated for the
factors that were retained using bi-level selection in the
final regression model are shown in Table 2. All variables
whose coefficients are not included in Table 2 were
removed in the variable selection process.
For LBP, 7 variables had VIS values of 0.80 or greater
(Figure 1; Table 2), four of which reached statistical significance.
In particular, the adjusted odds of reporting LBP in
the past month was more than 4 times higher in children
who reported high levels of psychosomatic complaints.
Having a lumbar apex that is located at a higher vertebral
level, introverted personality and high levels of negative
affect over positive affect were associated with LBP
prevalence. Having more posterior tilt of the dorsal surface
of the sacrum in habitual standing was significantly associated
with a higher odds of LBP, but corresponded to a
VIS of 0.71. Three out of the top 7 factors based on VIS did
not reach statistical significance, implying marginal evidence
that children who had experienced TSP or NP in the
preceding month were more likely to report LBP, as were
children who attended general as opposed to vocational
For NP, four variables had VIS values of 0.80 or greater
(Fig. 1; Table 2) of which 3 reached statistical significance:
psychosomatic complaints and 1-month period
prevalence of TSP from the comorbid pain domain, and the
affect-balance score from the psychosocial domain. The
presence of high levels of psychosomatic symptoms and
the presence of TSP in the last month were associated with
a nearly 8 (95 % CI 2.5–23.9) and nearly 5 (95 % CI
2.2–10.8) times higher odds of NP. Having high levels of
negative over positive affect significantly increased the
odds of NP. In contrast, the 1-month period prevalence of
LBP from the comorbid pain domain, a top clinical predictor
based on VIS, was only marginally associated with a
higher odds of NP. Statistically significant associations
were also found for 6 factors with a VIS below 0.80.
Higher odds of NP were shown for backward trunk lean in
habitual standing and high scores on the depression questionnaire.
Though weak, statistically significant associations
with an increased odds of NP were also observed for
high levels of pelvic retroversion in full extension, high
extension motion levels of the thoracic and lumbar spine in
standing posture, and attending subsidized publicly run
schools compared to subsidized privately run schools.
For TSP, both factors with VIS values of 0.80 or greater
(Fig. 1; Table 2) reached statistical significance: the presence
of pain in spinal regions adjacent to the thoracic spine
(NP, LBP) in the preceding month was associated with a
higher odds of reporting TSP. Although not reaching VIS
scores of 0.80 or higher, having a high extent of sedentary
behavior was significantly associated with an increased
probability of reporting TSP, as were not achieving the Fitnorm,
showing high levels of physical activity, and having
Pseudo R2 values displayed in Table 3 for the variables
included in the present study, equaled 20, 33 and 23 % for
LBP, NP and TSP prevalence, respectively. We evaluated
how much these are affected by the exclusion of certain
domains, and found that the comorbid pain domain has the
biggest impact on predictive ability. This suggests that if
only 1 domain (or factor) could be investigated, future
researchers in this area should choose (a variable from) the
comorbid pain domain. On the other hand, when a slenderized
model should be pursued in terms of number of
included domains (i.e., a 4-domain model instead of a
5-domain model), omitting the physical domain would lead
to the greatest information loss in LBP and NP, and to
limited loss of information in TSP. Comparatively less
predictive ability would be lost when omitting the set of
variables constituting the comorbid pain domain (LBP and
NP) or the psychosocial domain (TSP).
For each of the considered outcomes, the contribution of
the sociodemographic and lifestyle domain on predictive
ability is small (Table 3).
This paper reports the results of the first population-based
study among pre-PHV subjects evaluating 69 putative risk
factors for spinal pain, with the goal to ascertain which
variables and domains were related to the 1-month period
prevalence of LBP, NP and TSP in this group of study
participants. Greater numbers of psychosomatic complaints
and pain in another spinal region were the most important
factors associated with elevated spinal pain prevalence.
Specific physical characteristics were related to spinal pain
at young age, as were adverse psychosocial factors. In
contrast, the role of sociodemographic and lifestyle factors
was limited. This is the first study to show that physical
factors, including several biometric and postural measurements,
are crucial for understanding spinal pain at young
age and that the impact of physical factors in IASP may
have been underestimated in earlier studies.
Out of the extensive list of investigated variables, only a
limited number showed a significant association with the
1-month period prevalence of spinal pain. In the more mobile
regions of the spine (i.e., the lumbar and cervical areas), one
of the strongest associations with pain was to the frequency
and number of psychosomatic symptoms reported by the
participant. More specifically, it was found that the presence
of high levels of psychosomatic symptoms was associated
with more than 4 times higher odds of LBP and nearly 8 times
higher odds of NP. The increased likelihood of reporting
LBP or NP with frequent and high numbers of psychosomatic
complaints agrees with previous prospective
[8, 9, 13, 46, 53] and cross-sectional [18, 19, 27, 48] studies in
adolescent [9, 13, 18, 19, 27, 40, 46, 48, 53] and young adult
 study populations.
Interestingly, unlike the models for LBP and NP, the cooccurrence
of psychosomatic complaints appeared to play
only a minor role in 1-month period prevalence of TSP.
Instead, strong links were found between TSP and pain in
the adjacent areas of the spine (i.e., the odds of TSP was
more than 5.5 and 2.6 times higher when NP and LBP,
respectively, was experienced in the preceding month),
whereas co-complaints at other sites in the spine were not
significantly associated with LBP and NP. One exception
to this pattern was the 1-month period prevalence of NP
where TSP was a significant factor: the presence of TSP in
the last month was associated with nearly 5 times higher
odds of NP. The fact that other axial pains were not systematically
significant, independent factors associated with
LBP and NP was a rather surprising finding considering
previous study results [46, 50, 95, 96].
While our data
might reflect that the origin and mechanisms behind pain in
diverse spinal regions may differ in certain respects, care is
warranted when attempting to draw generalized conclusions.
Lack of power cannot be rejected as an alternative
explanation given that high VISs were obtained for all of
the comorbid (psycho) somatic (pain) complaints (Table 2,
Fig. 1). The frequent use in the literature of composite
variables made up of responses to TSP, NP and/or LBP and
of p value based variable selection may also add to the
difficulties in comparing our results with previous research
findings. In fact, replication is required in larger, longer
studies and future research should penetrate the mechanisms
behind the complex relationships between spinal
pain and other (pain) complaints.
The literature is replete with studies examining the
relationships between physical factors and spinal pain at
young age [9, 13, 19–21, 24–26, 28–30, 54]. While the
indexes of physical characteristics strongly vary across
studies, the preponderance of evidence showed that at best
weak associations exist between physical features and
spinal pain. Nevertheless, it is these authors’ contention
that the role that “physical factors” may play in non-adult
spinal pain might have been underestimated by previous
studies (see also below). Based on the present data, 2 of the
most important physical factors for elevated LBP prevalence
were higher lumbar lordotic apex and more retroversed
pelvis in customary standing. In this respect, one
might draw attention to the fact that an increased proportion
of adults with chronic LBP stand with a sagittal
lumbopelvic alignment that is similar to the postural pattern
described above, when compared to healthy referents
[97, 98], yet caution remains warranted in drawing premature
conclusions. Regarding the 1-month period prevalence
of NP in our pre-PHV cohort, the most important
associated factor from the physical domain was backward
trunk lean (i.e., a greater angle subtended between the
vertical and a line joining C7 to the greater trochanter) in
habitual standing. Furthermore, weak effects were seen for
a more retroversed pelvis in full trunk extension and for
higher extension motion levels of the thoracic and lumbar
spine in standing posture, suggesting that movement patterns
associated with spinal extension in stance may play a
role in NP. With respect to TSP, no statistically significant
factors could be disclosed from the physical domain.
Though the effects were more modest (VISs ≤ 0.80),
tendencies towards an increased odds for TSP were found
for overweight and obese subjects when compared to
normal weight subjects and for adolescents displaying low
extension mobility in the lumbar spine. Comparison with
existing literature is not easily done, since this is the first
study to analyze a wide array of putative risk factors for
spinal pain in competition with each other, both within and
between domains. On the other hand, the associations
found in this study do not lack biological plausibility.
Based upon our results regarding the psychosocial factors,
one might argue that aspects of subjective well-being such as
affect, personality dispositions, depressed mood and selfesteem
may contribute to spinal pain at young age. More
specifically, the present study revealed that high levels of
negative affect over positive affect were associated with an
increased risk of pain in the lumbar and cervical areas. Some
evidence was also found for “pain-prone personalities”,
since an increased likelihood of LBP was observed for more
introverted personalities. The results obtained further
demonstrated that depressed mood was significantly associated
with an increased odds of NP, whereas high self-esteem
was accompanied by a decreased odds of TSP, although
these effects were weak. The present study results appear to
corroborate the multiple findings from previous research
highlighting (potentially reciprocal) relationships between
poor psychological health and spinal pain in adolescents
[9, 15, 17–22, 27, 34, 35, 38, 40, 49, 53, 55, 99] and adults
[43, 100–102] albeit a wide variety of “psychosocial measures”
and statistical procedures has been used. The exact
mechanism whereby spinal pain and psychosocial factors are
associated with each other — a matter which is not within the
scope of the present study — is currently not understood, yet
might be related to neuroendocrine, neurological, biomechanical
and/or behavioural pathways [51, 99, 103–109].
Factors from the lifestyle and sociodemographic
domains were least important in explaining spinal pain
prevalence. Apparently, this lack of strong associations is
in accordance with the available literature which generally
points towards no evidence of an association between IASP
and various lifestyle variables related to physical (in)activity
[8, 9, 13, 17–22, 29, 32, 34–36, 38, 41, 49] or
sociodemographic factors [22, 29, 38, 49], apart from a few
exceptions [10, 24, 40]. Nonetheless, our results indicate
that some lifestyle factors might be implicated in childhood
TSP: both extremes in activity level and extremes of
sedentary behavior were significantly associated with an
increased 1-month period prevalence of TSP, whereas
achieving 20 min or more of vigorous-intensity physical
activity on at least 3 days each week (i.e., the Dutch “Fitnorm”)
might have a protective effect. Though these
associations were weak (Table 2), they might suggest that
an inactive spine as well as a spine subjected to high
physical demands may entail a disadvantage for the thoracic
spine in terms of pain at young age while a certain
envelope of vigorous physical activity might imply benefits.
Further investigation will be needed to truly understand
this relationship, as high-quality research focusing on
TSP as a separate entity is scarce. With respect to the 2
most important sociodemographic measures (secondary
education level for LBP and educational network for NP),
one might argue that none of these variables are likely to
contribute directly to the 1-month period pain prevalence.
Instead, other variables associated with these sociodemographics
that were not depicted in the present model of
spinal pain might be related to the variability in pain
prevalence. Further exploration is needed in this regard.
On the risk domain level, i.e., when considering the
(pre-defined) groups of factors potentially associated with
spinal pain, a prominent contribution of comorbid pain
symptoms to the 1-month period prevalence of spinal pain
was demonstrated (see the average importance scores and
group ranking per domain in Table 1). Based on the mean
rank (Table 1), the comorbid pain domain was followed by
the physical domain, for which somewhat smaller—yet still
pronounced—effects were found in each of the three spinal
pain sites. The psychosocial domain was less important in
explaining spinal pain prevalence whereas the lifestyle and
sociodemographic domains did not turn out to be important
domains, except maybe for TSP where the lifestyle domain
reached a relatively high average importance score.
Going deeper into the domain level, two intriguing
observations emerged from our results. First, a low mean
rank together with a low average VIS was observed for the
physical domain, indicating that the physical domain was
retained in nearly all bootstrap analyses because of a small
number of important factors (having a (relatively) high
VIS) besides a large number of surrogate—or unimportant—
factors (with low VIS) in that domain. A conscientious
selection of predictor variables thus appears to be of
overriding importance not to misjudge the role that
“physical factors” may play. This puts into perspective
previous research concluding that psychosocial factors
rather than physical or mechanical factors may be more
important in spinal pain occurring in youths [e.g., 9,19,53],
a view that is not supported by our results. Second, the
domains’ change in the order of merit between the 2 “% of
reduction in total explained variance” conditions per outcome
(Table 3) shows that omitting the physical domain
would imply most information loss, at least for LBP and
NP, whereas omitting the set of variables comprising the
comorbid pain domain would imply less loss of information.
This suggests that the physical variables contain
unique information about the outcome while the information
constituting the comorbid pain domain is captured by
variables in the other domains of interest. It is thus conceivable
that comorbid pain conditions themselves reflect a
range of biopsychosocial processes that have emerged over
time to manifest as one or more comorbid pain conditions
and that, therefore, other (pain) complaints may serve to
predict the risk of spinal pain. A full understanding of such
factors and processes, however, requires further analysis
and research. Furthermore, some caution is warranted when
interpreting Nagelkerke R2 measures since the number of
factors included in a domain may influence these values.
Nonetheless, in case only 1 domain (or factor) could be
investigated, one might propose that future researchers in
this area should choose (a variable from) the comorbid pain
domain in terms of predictive ability.
This study purposefully measured a large number of
variables from multiple domains, recognizing that spinal
pain may be influenced by a multitude of factors. A novel
multivariable analysis method was used that is well suited
to deal with problems created by the number, density and
correlation of data collected in this study cohort. With the
aim to investigate a homogenous male and female population
in terms of growth phase, research subjects were
intentionally recruited according to a maturational benchmark
[i.e., (predicted) APHV] and within a narrow age
range. Taking a developmental age baseline as opposed to
a chronological one as a base for recruitment may be a
promising novel approach, indeed, as studies have indicated
that IASP is associated with pubertal development
rather than with chronological age [59, 60, 62].
we limited the study recall period for IASP to
1 month, since longer time recall periods may result in
unreliable information [59, 65, 110]. Recall bias, however,
cannot be excluded. The value of this study is in understanding
the relative importance and predictive ability of
(pre-defined groups of) factors associated with spinal pain
in otherwise healthy pre-/early adolescents. It confirms that
spinal pain is a complex disorder that is associated with
multiple and overlapping factors, consistent with a
biopsychosocial model of illness. It further underlines the
importance of a broad perspective when studying, preventing
and treating IASP and suggests that the specific
factors behind pain might vary according to the spinal
region. This information can be harnessed to delineate
future research and evidence-based management of spinal
Several limitations should be acknowledged. The main
limitation of this study was its cross-sectional nature at this
phase, which restricts us from drawing any conclusions
regarding temporal or causal relationships between IASP
and putative risk factors. Prospective follow-up evaluations
are needed in this respect. Another concern in this study
was that only 842 of the 1196 adolescents who had data on
the 1-month prevalence of spinal pain (70.4 %)  had no
missing data on any of the independent variables. A multiple
imputation analysis requires modeling of the distribution
of the covariates, which may result in bias if the
model is misspecified, and moreover relies on a subtle
missing at random assumption which essentially states that
missingness may only be selective w.r.t. the observed data.
We chose to use a complete case analysis instead, because
this avoids modeling assumptions on the covariate distribution,
and moreover allows for covariate missingness to
be selective in terms of the (observed and unobserved)
Its main disadvantage relative to a multiple
imputation analysis, however, may be a loss of
information as a result of discarding partially observed
records. The resulting complete-case sample was furthermore
found to be similar to the complete sample with
regard to pain prevalence. We therefore believe this study
may be considered representative of the prevalence of
spinal pain and the associations found. Because of the
many examined correlates of spinal pain and the relatively
low pain prevalence, we could not investigate potential
differences in associations between genders. Therefore,
only the results for both genders combined, albeit adjusted
for gender, have been reported.
Third, there was a substantial
proportion of unexplained variation in all considered
outcomes. Indeed, the variables included in the
present models accounted only for 21, 33 and 34 % of the
variance in LBP, NP, and TSP, respectively. Other factors
or domains — not measured here — may thus be involved,
such as inherited predisposition, dysfunction in central pain
regulation, (spinopelvic) anatomy factors, movement and
motor patterns, lifestyle factors other than those related to
physical activity/sedentary behavior, environmental circumstances,
the cognitive-evaluative component concerning
spinal pain, etc. No conclusions can be drawn about
potential contributions from factors or domains that were
not depicted in this model of spinal pain etiology. Fourth,
no questions were included focusing on duration, frequency,
and severity of pain periods.
We are indebted to the pupils, parents and staff
of the schools and pupil guidance centers for taking part in this study
and to Mr. Roel De Ridder, Ms. Gizem Iˆrem Guvendik, and Ms. Heidi
Bauters for their assistance in data collection.
Conflict of interest
The authors declare no conflict of interest. The
authors had full control of all primary data. The authors agree to allow
the journal to review our data if requested.
Vos T, Flaxman AD, Naghavi M, et al.
Years Lived with Disability (YLDs) for 1160 Sequelae of 289 Diseases and Injuries 1990-2010:
A Systematic Analysis for the Global Burden of Disease Study 2010
Lancet. 2012 (Dec 15); 380 (9859): 2163–2196
Dagenais S, Caro J, Haldeman S (2008)
A systematic review of low back pain cost of illness studies in the United States and internationally.
Spine J 8:8–20
Maniadakis N, Gray A (2000)
The economic burden of back pain in the UK.
Borghouts JA, Koes BW, Vondeling H, Bouter LM (1999)
Cost-of-illness of neck pain in The Netherlands in 1996.
Roth-Isigkeit A, Thyen U, Stoven H, Schwarzenberger J, Schmucker
Pain among children and adolescents: restrictions in daily living and triggering factors.
Hestbaek L, Leboeuf-Yde C, Kyvik KO, Manniche C:
The Course of Low Back Pain from Adolescence to Adulthood: Eight-year Follow-up of 9600 Twins
Spine (Phila Pa 1976) 2006 (Feb 15); 31 (4): 468–472
Harreby M, Neergaard K, Hesselsøe G, Kjer J (1995)
Are radiologic changes in the thoracic and lumbar spine of adolescents risk factors for low back pain in adults? A 25-year prospective cohort study of 640 school children.
Siivola SM, Levoska S, Latvala K, Hoskio E, Vanharanta H,
Keinanen-Kiukaanniemi S (2004)
Predictive factors for neck and shoulder pain: a longitudinal study in young adults.
Jones GT, Watson KD, Silman AJ, Symmons DP, Macfarlane
Predictors of low back pain in British schoolchildren: a population-based prospective cohort study.
Pediatrics 111: 822–828
Hanvold TN, Veiersted KB, Waersted M (2010)
A prospective study of neck, shoulder, and upper back pain among technical school students entering working life.
J Adolesc Health 46:488–494
Brattberg G (2004)
Do pain problems in young school children persist into early adulthood? A 13-year follow-up.
Eur J Pain 8:187–199
Sjolie AN (2004)
Persistence and change in nonspecific low back pain among adolescents: a 3-year prospective study.
Stahl M, Kautiainen H, El-Metwally A, Hakkinen A, Ylinen J, Salminen JJ, Mikkelsson M.
Non-specific Neck Pain in Schoolchildren: Prognosis and Risk Factors for Occurrence and Persistence.
A 4-year Follow-up Study
Pain. 2008 (Jul 15); 137 (2): 316–322
Balague F, Dudler J, Nordin M (2003)
Low-back pain in children.
Briggs AM, Smith AJ, Straker LM, Bragge P (2009)
Thoracic spine pain in the general population: prevalence, incidence and associated factors in children, adolescents and adults. A systematic review.
BMC Musculoskelet Disord 10:77
Petersen S, Brulin C, Bergstrom E (2006)
Recurrent pain symptoms in young schoolchildren are often multiple.
Myrtveit SM, Sivertsen B, Skogen JC, Frostholm L, Stormark
KM, Hysing M (2014)
Adolescent neck and shoulder pain-the association with depression, physical activity, screen-based activities, and use of health care services.
J Adolesc Health 55:366–372
Murphy S, Buckle P, Stubbs D (2007)
A cross-sectional study of self-reported back and neck pain among English schoolchildren and associated physical and psychological risk factors.
Appl Ergon 38:797–804
Watson KD, Papageorgiou AC, Jones GT, Taylor S, Symmons
DP, Silman AJ et al (2003)
Low back pain in schoolchildren: the role of mechanical and psychosocial factors.
Arch Dis Child 88:12–17
Szpalski M, Gunzburg R, Balague F, Nordin M, Melot C (2002)
A 2-year prospective longitudinal study on low back pain in primary school children.
Eur Spine J 11:459–464
O’Sullivan PB, Smith AJ, Beales DJ, Straker LM (2011)
Association of biopsychosocial factors with degree of slump in sitting posture and self-report of back pain in adolescents: a cross-sectional study.
Phys Ther 91:470–483
Balague F, Skovron ML, Nordin M, Dutoit G, Pol LR, Waldburger
Low back pain in schoolchildren. A study of familial and psychological factors.
El-Metwally A, Mikkelsson M, Stahl M, Macfarlane GJ, Jones GT, Pulkkinen L et al (2008)
Genetic and environmental influences on non-specific low back pain in children: a twin study.
Eur Spine J 17:502–508
Kovacs FM, Gestoso M, Gil del Real MT, Lopez J, Mufraggi N, Mendez JI (2003)
Risk factors for non-specific low back pain in schoolchildren and their parents: a population based study.
Steele S, Grimmer K, Williams M, Gill T (2001)
Vertical anthropometric measures and low back pain in adolescents.
Physiother Res Int 6:94–105
Poussa MS, Heliovaara MM, Seitsamo JT, Kononen MH, Hurmerinta KA, Nissinen MJ (2005)
Anthropometric measurements and growth as predictors of low-back pain: a cohort study of children followed up from the age of 11–22 years.
Eur Spine J 14:595–598
Rees CS, Smith AJ, O’Sullivan PB, Kendall GE, Straker LM (2011)
Back and neck pain are related to mental health problems in adolescence.
BMC Public Health 11:382
Poussa MS, Heliovaara MM, Seitsamo JT, Kononen MH, Hurmerinta KA, Nissinen MJ (2005)
Predictors of neck pain: a cohort study of children followed up from the age of 11–22 years.
Eur Spine J 14:1033–1036
Salminen JJ (1984)
The adolescent back. A field survey of 370 Finnish school children.
Acta Paediatr Scand 73(Suppl 315): 1–122
Hertzberg A (1985)
Prediction of cervical and low-back pain
based on routine school health examinations. A nine- to twelve-year follow-up study.
Scand J Prim Health Care 3:247–253
Shan Z,Deng G, Li J, Li Y,ZhangY,ZhaoQ (2014)
How schooling and lifestyle factors effect neck and shoulder pain? A cross-sectional survey of adolescents in China.
Mogensen AM, Gausel AM, Wedderkopp N, Kjaer P, Leboeuf-Yde C (2007)
Is active participation in specific sport activities linked with back pain?
Scand J Med Sci Sports 17:680–686
Lynch AM, Kashikar-Zuck S, Goldschneider KR, Jones BA (2006)
Psychosocial risks for disability in children with chronic back pain.
J Pain 7:244–251
Feldman ED, Shrier I, Rossignol M, Abenhaim L (2002)
Risk factors for the development of neck and upper limb pain in adolescents.
Balague F, Troussier B, Salminen JJ (1999)
Non-specific low back pain in children and adolescents: risk factors.
Eur Spine J 8:429–438
Chen SM, Liu MF, Cook J, Bass S, Lo SK (2009)
Sedentary lifestyle as a risk factor for low back pain: a systematic review.
Int Arch Occup Environ Health 82:797–806
Dunn KM, Hestbaek L, Cassidy JD (2013)
Low back pain across the life course.
Best Pract Res Clin Rheumatol 27:591–600
Leboeuf-Yde C (2004)
Back pain—individual and genetic factors.
J Electromyogr Kinesiol 14:129–133
Hill JJ, Keating JL (2010)
Risk factors for the first episode of low back pain in children are infrequently validated across samples and conditions: a systematic review.
J Physiother 56:237–244
Prins Y, Crous L, Louw QA (2008)
A systematic review of posture and psychosocial factors as contributors to upper quadrant musculoskeletal pain in children and adolescents.
Physiother Theory Pract 24:221–242
Sitthipornvorakul E, Janwantanakul P, Purepong N, Pensri P, van der Beek AJ (2011)
The association between physical activity and neck and low back pain: a systematic review.
Eur Spine J 20:677–689
Pollock CM, Harries RL, Smith AJ, Straker LM, Kendall GE, O’Sullivan PB (2011)
Neck/shoulder pain is more strongly related to depressed mood in adolescent girls than in boys.
Man Ther 16:246–251
Gilkey DP, Keefe TJ, Peel JL, Kassab OM, Kennedy CA (2010)
Risk factors associated with back pain: a cross-sectional study of 963 college students.
J Manip Physiol Ther 33:88–95
Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR (1996)
A simulation study of the number of events per variable in logistic regression analysis.
J Clin Epidemiol 49:1373–1379
Stahl M, Mikkelsson M, Kautiainen H, Hakkinen A, Ylinen J, Salminen JJ (2004)
Neck pain in adolescence. A 4-year followup of pain-free preadolescents.
Mikkelsson M, Salminen JJ, Kautiainen H (1997)
Non-specific musculoskeletal pain in preadolescents. Prevalence and 1-year persistence.
Hoftun GB, Romundstad PR, Zwart JA, Rygg M (2011)
Chronic idiopathic pain in adolescence—high prevalence and disability: the young HUNT Study 2008.
Vikat A, Rimpela M, Salminen JJ, Rimpela A, Savolainen A, Virtanen SM (2000)
Neck or shoulder pain and low back pain in Finnish adolescents.
Scand J Public Health 28:164–173
Diepenmaat AC, van der Wal MF, de Vet HC, Hirasing RA (2006)
Neck/shoulder, low back, and arm pain in relation to computer use, physical activity, stress, and depression among Dutch adolescents.
Hogg-Johnson, S, van der Velde, G, Carroll, LJ et al.
The Burden and Determinants of Neck Pain in the General Population:
Results of the Bone and Joint Decade 2000–2010 Task Force
on Neck Pain and Its Associated Disorders
Spine (Phila Pa 1976). 2008 (Feb 15); 33 (4 Suppl): S39–51
Fillingim RB, Hastie BA, Ness TJ, Glover TL, Campbell CM, Staud R (2005)
Sex-related psychological predictors of baseline pain perception and analgesic responses to pentazocine.
Biol Psychol 69:97–112
Kjaer P, Wedderkopp N, Korsholm L, Leboeuf-Yde C.
Prevalence and Tracking of Back Pain From Childhood to Adolescence
BMC Musculoskelet Disord. 2011 (May 16); 12: 98
Brattberg G (1994)
The incidence of back pain and headache among Swedish school children.
Qual Life Res 3:S27–S31
Brink Y, Crous LC, Louw QA, Grimmer-Somers K, Schreve K (2009)
The association between postural alignment and psychosocial factors to upper quadrant pain in high school students: a prospective study.
Man Ther 14:647–653
Niemi SM, Levoska S, Rekola KE, Keinanen-Kiukaanniemi SM (1997)
Neck and shoulder symptoms of high school students and associated psychosocial factors.
J Adolesc Health 20:238–242
Dolphens M, Cagnie B, Vleeming A, Vanderstraeten G, Danneels L (2013)
Gender differences in sagittal standing alignment before pubertal peak growth: the importance of subclassification and implications for spinopelvic loading.
J Anat 223:629–640
Brink Y, Louw QA (2013)
A systematic review of the relationship between sitting and upper quadrant musculoskeletal pain in children and adolescents.
Man Ther 18:281–288
Dolphens M, Cagnie B, Coorevits P, Vleeming A, Vanderstraeten G, Danneels L(2014)
Classification system of the sagittal standing alignment in young adolescent girls.
Eur Spine J 23:216–225
Jeffries, LJ, Milanese, SF, and Grimer-Somers, KA.
Epidemiology of Adolescent Spinal Pain: A Systematic Overview of the Research Literature
Spine (Phila Pa 1976). 2007 (Nov 1); 32 (23): 2630–2637
LeResche L, Mancl LA, Drangsholt MT, Saunders K, Korff MV (2005)
Relationship of pain and symptoms to pubertal development in adolescents.
Patton GC, Viner R (2007)
Pubertal transitions in health.
Wedderkopp N, Andersen LB, Froberg K, Leboeuf-Yde C (2005)
Back pain reporting in young girls appears to be pubertyrelated.
BMC Musculoskelet Disord 6:52
Malina RM, Bouchard C, Bar-Or O (2004)
Growth, maturation, and physical activity.
Human Kinetics, Champaign
Dolphens M, Cagnie B, Coorevits P, Vanderstraeten G, Cardon G, D’Hooge R et al (2012)
Sagittal standing posture and its association with spinal pain: a school-based epidemiological study of 1196 Flemish adolescents before age at peak height velocity.
Staes F, Stappaerts K, Vertommen H, Everaert D, Coppieters M (1999)
Reproducibility of a survey questionnaire for the investigation of low back problems in adolescents.
Acta Paediatr 88:1269–1273
Cole TJ, Bellizzi MC, Flegal KM, Dietz WH (2000)
Establishing a standard definition for child overweight and obesity worldwide: international survey.
Cole TJ, Flegal KM, Nicholls D, Jackson AA (2007)
Body mass index cut offs to define thinness in children and adolescents: international survey.
Dolphens M, Cagnie B, Coorevits P, Vleeming A, Danneels L (2013)
Classification system of the normal variation in sagittal standing plane alignment: a study among young adolescent boys.
Beighton PH, Grahame R, Bird HA (1999)
Hypermobility of joints.
Mirwald RL, Baxter-Jones AD, Bailey DA, Beunen GP (2002)
An assessment of maturity from anthropometric measurements.
Med Sci Sports Exerc 34:689–694
Saskatchewan Childhood Growth and Development (SCGD) Research group.
Childhood growth utility programs.
University of Saskatchewan, College of Kinesiology, Saskatoon, Saskatchewan, Canada.
Accessed 30 March 2013
Sherar LB, Mirwald RL, Baxter-Jones AD, Thomis M (2005)
Prediction of adult height using maturity-based cumulative height velocity curves.
J Pediatr 147:508–514
Ganzeboom HBG, Treiman DJ (1996)
Internationally comparable measures of occupational status for the 1988 International Standard Classification of Occupations.
Soc Sci Res 25:201–239
Philippaerts RM, Matton L, Wijndaele K, Balduck AL, De Bourdeaudhuij I, Lefevre J (2006)
Validity of a physical activity computer questionnaire in 12- to 18-year-old boys and girls.
Int J Sports Med 27:131–136
Verstraete SJM, Cardon GM, Trost SG, De Bourdeaudhuij IMM (2006)
Reliability and validity of a questionnaire to measure usual physical activity in children with and without parental assistance. In: Verstraete SJM. The effectiveness of an intervention promoting physical activity in elementary school children.
Dissertation, Ghent University
Hildebrandt VH, Chorus AMJ, Stubbe JH (2010)
Trend report of physical activity and health 2008/2009.
TNO Quality of Life, Leiden
Biddle S, Sallis J, Cavill N (1998)
Young and active? Young people and health-enhancing physical activity: evidence and implications.
Health Education Authority, London
Harter S (1988)
Manual for the Self-Perception Profile for Adolescents.
University of Denver, Denver
Treffers PDA, Goedhart AW, Veerman JW, Van den Bergh BRH, Ackaert L, de Rycke L (2002)
Handleiding Competentiebelevingsschaal voor Adolescenten (CBSA) [Manual Self-Perception Profile for Adolescents (SPPA)].
Swets and Zeitlinger, Lisse
Wichstrøm L (1995)
Harter’s Self-Perception Profile for Adolescents: reliability, validity, and evaluation of the question format.
J Pers Assess 65:100–116
Rosenberg M (1965)
Society and the adolescent self-image.
Princeton University Press, Princeton
Franck E, De Raedt R, Barbez C, Rosseel Y (2008)
Psychometric properties of the Dutch Rosenberg Self-Esteem Scale.
Psychol Belg 48:25–35
Van Dijl H, Wilde G (1982)
Handleiding bij de Amsterdamse Biografische Vragenlijst voor Kinderen, ABVK, en de Korte Amsterdamse Biografische vragenlijst voor Kinderen, KABVK [Manual Amsterdam Biographical Questionnaire for Children and Short Amsterdam Biographical Questionnaire for Children].
Van Rossen, Amsterdam
Watson D, Clark LA, Tellegen A (1988)
Development and validation of brief measures of positive and negative affect: the PANAS scales.
J Pers Soc Psychol 54:1063–1070
Bradburn NM (1969)
The structure of psychological well-being.
De Wit CAM (1987)
Depressie Vragenlijst voor Kinderen, DVK en KDVK [Depression Questionnaire for Children, DQC and SDQC].
Diener E, Emmons RA, Larsen RJ, Griffin S (1985)
The Satisfaction With Life Scale.
J Pers Assess 49:71–75
Pavot W, Diener E, Colvin CR, Sandvik E (1991)
Further validation of the Satisfaction with Life Scale: evidence for the cross-method convergence of well-being measures.
J Pers Assess 57:149–161
Ryan RM, Frederick C (1997)
On energy, personality, and health: subjective vitality as a dynamic reflection of well-being.
J Pers 65:529–565
Armsden GC, Greenberg MT (1987)
The inventory of parent and peer attachment: individual differences and their relationship to psychological well-being in adolescence.
J Youth Adolesc 16:427–454
Noom MJ, Dekovic M, Meeus WH (1999)
Autonomy, attachment and psychosocial adjustment during adolescence: a doubleedged sword?
J Adolesc 22:771–783
Victoir A (2006)
DIGGing into adolescents’ health behaviours: From measurement to targeting.
Huang J, Breheny P, Ma S (2012)
A selective review of group selection in high-dimensional models.
Stat Sci 27:481–499
Firth D (1993)
Bias reduction of maximum likelihood estimates.
Hartvigsen J, Davidsen M, Hestbaek L, Sogaard K, Roos EM (2013)
Patterns of musculoskeletal pain in the population: a latent class analysis using a nationally representative interviewer-based survey of 4817 Danes.
Eur J Pain 17:452–460
Beales DJ, Smith AJ, O’Sullivan PB, Straker LM (2012)
Low back pain and comorbidity clusters at 17 years of age: a crosssectional examination of health-related quality of life and specific low back pain impacts.
J Adolesc Health 50:509–516
Chaleat-Valayer E, Mac-Thiong JM, Paquet J, Berthonnaud E, Siani F, Roussouly P (2011)
Sagittal spino-pelvic alignment in chronic low back pain.
Eur Spine J 20:634–640
Jackson RP, McManus AC (1994)
Radiographic analysis of sagittal plane alignment and balance in standing volunteers and patients with low back pain matched for age, sex, and size. A prospective controlled clinical study.
Egger HL, Costello EJ, Erkanli A, Angold A (1999)
Somatic complaints and psychopathology in children and adolescents: stomach aches, musculoskeletal pains, and headaches.
J Am Acad Child Adolesc Psychiatry 38:852–860
Linton SJ (2000)
A review of psychological risk factors in back and neck pain.
Pincus T, McCracken LM (2013)
Psychological factors and treatment opportunities in low back pain.
Best Pract Res Clin Rheumatol 27:625–635
Power C, Frank J, Hertzman C, Schierhout G, Li L (2001)
Predictors of low back pain onset in a prospective British study.
Am J Public Health 91:1671–1678
Marras WS, Davis KG, Heaney CA, Maronitis AB, Allread WG (2000)
The influence of psychosocial stress, gender, and personality on mechanical loading of the lumbar spine.
Hanna T (1988)
Somatics: Reawakening the mind’s control of movement, flexibility, and health.
Addison-Wesley Publishing Company, Rochester
Trivedi MH (2004)
The link between depression and physical symptoms.
Prim Care Companion J Clin Psychiatry 6:S12–S16
Von Korff M, Simon G (1996)
The relationship between pain and depression.
Br J Psychiatry 168:101–108
Lu F, Huo Y, Li M, Chen H, Liu F, Wang Y et al (2014)
Relationship between personality and gray matter volume in healthy young adults: a voxel-based morphometric study.
PLoS One 9:e88763
Gatchel RJ, Peng YB, Peters ML, Fuchs PN, Turk DC (2007)
The biopsychosocial approach to chronic pain: scientific advances and future directions.
Psychol Bull 133:581–624
Bair MJ, Robinson RL, Katon W, Kroenke K (2003)
Depression and pain comorbidity: a literature review.
Arch Intern Med 163:2433–2445
Staes F, Stappaerts K, Lesaffre E, Vertommen H (2003)
Low back pain in Flemish adolescents and the role of perceived social support and effect on the perception of back pain.
Acta Paediatr 92:444–451
Bartlett JW, Carpenter JR, Tilling K, Vansteelandt S (2014)
Improving upon the efficiency of complete case analysis when covariates are MNAR.
Return to PEDIATRICS
Return to BIOPSYCHOSOCIAL MODEL
Return to SPINAL PAIN MANAGEMENT