This study examined preoperative pain and symptom profiles in 70 children undergoing surgery for idiopathic scoliosis. The researchers found that 30% of children had a high preoperative pain and symptom profile characterized by higher levels of depression, fatigue, pain interference, widespread pain, and pain catastrophizing. Children with this high profile, particularly girls and those with long-standing pain, had worse postoperative pain outcomes including higher reported pain, greater opioid use, and ongoing pain issues up to 6 months later. The findings suggest a subset of children are more vulnerable to pain due to central sensitization, and this vulnerability predicts poorer recovery after spine fusion surgery.

1. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
1594 www.anesthesia-analgesia.org May 2017 • Volume 124 • Number 5
Copyright © 2017 International Anesthesia Research Society
DOI: 10.1213/ANE.0000000000001963
M
any children who undergo surgical correction of
idiopathic scoliosis (IS) are susceptible to poor
pain outcomes, including persistent or recur-
rent musculoskeletal pain. Up to 65% have reported ongo-
ing, unchanged pain 1 to 5 years after spine fusion,1–3
and
a similar number described persistent back pain 23 years
after having undergone spine fusion during adolescence.4
Persistent or recurrent musculoskeletal pain imposes a tre-
mendous burden on the child’s physical and social func-
tioning and quality of life.5
Despite these concerns, we have
limited understanding of the nature and course of musculo-
skeletal pain in children with IS. This lack of understanding
may contribute to suboptimal management and the high
rates of persistent pain currently observed.
The presence of preoperative pain has been found to
strongly predict persistent postoperative pain, whereas sur-
gical and structural nociceptive factors (eg, degree or loca-
tion of spinal curve, surgical instrumentation) have, to date,
failed to do so.1,6
Such findings suggest a baseline pain vul-
nerability that may affect children’s recovery and long-term
surgical outcomes. The pain profile and potential mecha-
nisms for pain vulnerability in children with IS are largely
unknown. However, studies in children and adults with
differing chronic pain conditions reveal symptom patterns
similar to those observed in patients with centralized pain
conditions (ie, fibromyalgia). These include multiple-site
BACKGROUND: Preoperative pain predicts persistent pain after spine fusion, yet little is under-
stood about the nature of that pain, related symptoms, and how these symptoms relate to post-
operative pain outcomes. This prospective study examined children’s baseline pain and symptom
profiles and the association between a high symptom profile and postoperative outcomes.
METHODS: Seventy children (aged 10–17 years) scheduled for correction of idiopathic scoliosis
completed pain and symptom surveys during their preoperative visit (ie,pain intensity [0–10 numeric
rating scores], a pediatric version of the 2011 fibromyalgia survey criteria [including pain locations
and symptom severity scale], neuropathic pain symptoms [painDETECT], and Patient-Reported
Outcome Measurement System measures of fatigue, depression, function, pain interference, and
pain catastrophizing). Pain intensity and total analgesic use were recorded daily postoperatively
and for 2 weeks after discharge. A 2-step cluster analysis differentiated a high and low pain and
symptom profile at baseline, and a multivariate main effects regression model examined the asso-
ciation between pain profile and posthospital discharge pain and analgesic outcomes.
RESULTS: The cluster analysis differentiated 2 groups of children well characterized by their
baseline symptom reporting. Thirty percent (95% confidence interval [CI], 20.2%–41.8%) had
a high symptom profile with higher depression, fatigue, pain interference, a pediatric version
of the fibromyalgia survey criteria symptoms, neuropathic pain, and catastrophizing. Girls were
more likely than boys to be clustered in the high symptom profile (odds ratio [OR], 5.76 [95%
CI, 1.20–27.58]; P = .022) as were those with preoperative pain lasting >3 months (OR, 3.42
[95% CI, 1.21–9.70]; P = .018). Adjusting for sex, age, and total in-hospital opioid consump-
tion, high cluster membership was independently associated with higher self-reported pain after
discharge (mean difference +1.13 point [97.5% CI, 0.09–2.17]; P = .015). Children in the high
symptom cluster were more likely to report ongoing opioid use at 2 weeks compared with the
low symptom group (87% vs 50%; OR, 6.5 [95% CI, 1.30–33.03]; P = .015). At 6 months, high
symptom cluster membership was associated with higher pain intensity, higher pain interfer-
ence, and ongoing analgesic use (P ≤ .018).
CONCLUSIONS: A behavioral pain vulnerable profile was present preoperatively in 30% of chil-
dren with idiopathic scoliosis and was independently associated with poorer and potentially
long-lasting pain outcomes after spine fusion in this setting. This high symptom profile is similar
to that described in children and adults with chronic and centralized pain disorders and was
more prevalent in girls and those with long-standing pain. Further study is needed to elucidate
the potential mechanisms behind our observations. (Anesth Analg 2017;124:1594–602)
From the Departments of *Anesthesiology, †Orthopedic Surgery, ‡Internal
Medicine, and §Psychiatry, University of Michigan Health Systems and
Medical School, Ann Arbor, Michigan.
Accepted for publication January 10, 2017.
Funding: None.
The authors declare no conflicts of interest.
Reprints will not be available from the authors.
Address correspondence to Terri Voepel-Lewis, PhD, RN, Room 4917, Mott
Children’s Hospital, 1540 East Hospital Dr, Ann Arbor, MI 48109. Address
e-mail to terriv@umich.edu.
A High Preoperative Pain and Symptom Profile
Predicts Worse Pain Outcomes for Children After
Spine Fusion Surgery
Terri Voepel-Lewis, PhD, RN,* Michelle S. Caird, MD,† Alan R. Tait, PhD,* Shobha Malviya, MD,*
Frances A. Farley, MD,† Ying Li, MD,† Matthew D. Abbott, MD,† Tara van Veen, BS,*
Afton L. Hassett, PsyD,* and Daniel J. Clauw, MD*‡§

2. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
May 2017 • Volume 124 • Number 5 www.anesthesia-analgesia.org 1595
pain and somatic symptoms such as sleep and memory dif-
ficulties, fatigue, and depression. Higher degrees of these
symptom patterns as measured by the 2011 fibromyalgia
survey criteria (FSC) were recently found to strongly pre-
dict persistent postoperative pain and higher opioid use in
adults who underwent knee or hip arthroplasty.7
Such data
have informed the hypothesis in adult settings that central
pain sensitization may contribute not only to various chronic
pain conditions, but to pain that persists postoperatively.
A growing body of evidence suggests that similar pain
and somatic symptom patterns begin to emerge during child-
hood in those exposed to chronic pain. Longitudinal studies
in children with nonspecific regional or widespread mus-
culoskeletal pain found that comorbid depression, anxiety,
and difficulty sleeping or daytime fatigue were associated
with persistent pain and functional impairments.8,9
Whether
a similar symptom profile is present in children with IS and
if this is associated with long-term outcomes is unknown. A
better understanding of the nature, early course, and poten-
tial mechanisms of childhood pain is therefore imperative to
optimize early pain management and prevention strategies.
The purpose of this prospective study was to examine
the nature and prevalence of preoperative pain and somatic
symptomatology in children presenting for surgical cor-
rection of scoliosis. We hypothesized that a symptom pro-
file, similar to that observed in children with chronic and
centralized pain disorders, could be identified in a subset
of children with scoliosis before surgery and that this pain
vulnerable profile would be associated with worse postop-
erative pain outcomes, including higher pain intensity and
opioid consumption.
METHODS
With approval from the institutional review board at the
University of Michigan and written parental consent and
child assent, we consecutively recruited English-speaking
children scheduled to undergo elective surgical correction
of IS from July 2014 to September 2015. We excluded chil-
dren with significant cognitive impairment who could not
self-report pain or complete surveys independently and
those undergoing a secondary or repeat major orthopedic
procedure. This study meets the Enhancing the QUAlity
and Transparency of health Research (EQUATOR) guideline
for observational studies.
Measurements
Children Completed the Following Surveys at Baseline.
Pain Intensity: Numeric rating scale (NRS) scores (0 = no
pain to 10 = worst possible pain) to document their worst
and average pain over the previous 6 months. The reliabil-
ity and validity of NRS pain scores to describe pain inten-
sity have been demonstrated in children aged ≥8 years.10,11
Children were also asked to tell us how long their pain had
persisted (open-ended).
Fibromyalgia Survey Criteria: This simple, 2-part, self-report
survey was derived from the 2011 FSC to assess the core
symptoms associated with the most common centralized
pain disorder in adults, fibromyalgia. Similar to the Yunis and
Mani criteria for juvenile fibromyalgia syndrome,12
the FSC
combines a pain location body map with a symptom sever-
ity scale to identify widespread pain and symptoms that are
in common with fibromyalgia. The FSC has been used to
diagnose fibromyalgia and also to generate a score represent-
ing a quantitative continuum of pain centralization or fibro-
myalgia-like symptoms.7,13
We revised this survey to lower
the grade reading level and improve its readability (Flesch-
Kincaid Grade Level 5.3 and Reading Ease 78%) so that it
could be independently completed by children as young as 10
years. The pediatric version of the FSC (pFSC) combines the
scores from the body map and the Somatic Symptom Severity
(SSS) assessments to yield a pFSC score ranging from 0 to 35.
Multisite Pain: A 2-sided (front and back) body map was
used to describe pain location(s) because it was found to be
reliable in children as young as 4 years of age.14
Studies have
found 83% to 98% agreement between children’s identified
pain locations and diagnosed conditions or surgical sites,
and experts recommend unassisted use of the body map in
children 8 years or older.14
The number and location of pain
sites were tallied to provide indices of regional pain (upper
body region versus lower body region) and of widespread
pain (upper and lower body involvement).
Comorbid SSS: We revised the grade reading level of the
adult version of the SSS in the adult version FSC criteria to
document the presence and severity of fatigue, sleep prob-
lems, memory, mood, headaches, and irritable abdomen.
This short tool has face validity and contains items similar
to those on validated but more comprehensive pediatric
somatization inventories.
painDETECT: This 9-item survey was modified slightly to
reduce the reading level and improve its relevance to chil-
dren/adolescents and to differentiate neuropathic from noci-
ceptive pain.15
Children as young as 7 years have reliably
described their pain characteristics (eg, hot, sharp, tingling),
and high test–retest reliability of their use of word descrip-
tors has been found among 8 to 17 year olds (ρ = 0.78–0.95).16
Pediatric Patient-Reported Outcome Measurement System
(PROMIS) Short Forms: These tools were developed by the
National Institutes of Health to measure various health
domains across chronic disorders in children. They have
been tested using item bank development and classification,
and the short forms are considered to efficiently and accu-
rately capture the most informative items for each health
outcome. We included the pediatric PROMIS short forms
for fatigue, depression (reliability = 0.85),17
anxiety (reli-
ability = 0.85),17
pain interference (reliability = 0.90),18
and
mobility (reliability = 0.94).19
Pain Catastrophizing Scale: This 13-item instrument reliably
measures “exaggerated negative pain mental set” in children
aged 9 to 15 years (α = 0.87–0.92).20
The instrument has excel-
lent predictive validity for chronic pain disability in children.21
Procedures
After parental consent and child assent in the preoperative
clinic, children completed the baseline surveys using an
iPad that directly entered data using a unique identification

3. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
1596 www.anesthesia-analgesia.org ANESTHESIA ANALGESIA
Pain Vulnerable Profile in Children With Scoliosis
number. All children completed these surveys indepen-
dently while parents were concurrently busy completing
a different batch of surveys for a secondary study. The
child’s relevant demographics, medical, and surgical his-
tory, including recent and past use of prescribed and non-
prescribed medications and analgesics, were recorded from
the orthopedic clinic records and parent interview.
Children repeated the body map, SSS, and painDETECT
surveys when they arrived on the morning of surgery 1 to 2
weeks after the clinic visit to examine short-term test–retest
consistency in their reporting. The PROMIS anxiety survey
was also completed on the morning of surgery. All surgical
procedures were performed by 3 experienced pediatric ortho-
pedic surgeons and involved the posterior approach, seg-
mental instrumentation with hybrid constructs that included
pedicle screws, sublaminar hooks, and wires. Anesthesia
proceeded per routine practice with mask or IV induction of
general anesthesia (8% sevoflurane with nitrous oxide and
oxygen or propofol) followed by an isoflurane maintenance
as permitted by motor-evoked potential monitoring. Depth
of anesthesia was titrated with remifentanil and dexmedeto-
midine. Children received either an intrathecal injection of
Duramorph or an epidural bolus of hydromorphone intra-
operatively. Postoperatively, all children were transitioned
per routine practice from the regional method to oral oxy-
codone (5 mg every 6 hours) and ketorolac around the clock
together with acetaminophen (15 mg/kg every 4–6 hours)
and diazepam (0.5 mg/kg every 6 hours) as needed plus
2.5 mg/kg oxycodone as needed for breakthrough pain. All
intra- and postoperative opioid and nonopioid analgesics
were recorded and opioids (including intravenous, oral,
intrathecal, and epidural) were converted to oral morphine
equivalents per kilogram per hour of hospital stay (including
perioperatively, ie, intraoperatively and immediately post-
operatively at the direction of the anesthesiologist and post-
operatively based on patient demand and around-the-clock
dosing) using standard conversion charts.22,23
Children were followed prospectively over the course of
hospitalization and the following data collected: daily NRS
pain scores (at rest and on movement), adverse events, and
length of stay. At discharge, parents were given a diary to
record all analgesics given over the first 2 weeks at home
together with children’s pain scores and other symptoms. In
a small subset, we collected 6-month follow-up survey data
from children who had been seen as per routine in clinic.
This survey included ongoing pain presence, pain intensity,
frequency and type of analgesic use, PROMIS pain interfer-
ence, function, painDETECT, and pFSC.
Statistical Analyses
All analyses were conducted using SPSS statistical software
(version 22; IBM, New York, NY). Data are summarized
using descriptive statistics and n (%) or means, standard
deviations, and medians with percentiles, where applicable,
and STROBE guidelines were adhered to in manuscript
preparation. Tests of normality were conducted for our con-
tinuous variables (ie, morphine consumption, pain scores)
and nonparametric comparisons performed where applica-
ble. Intraclass correlation coefficients (ICCs) with 95% con-
fidence intervals (CIs) were used to examine the consistency
(ie, short-term retest reliability) of children’s reports using
the body map, SSS, the pFSC as a whole and painDETECT.
Internal consistency of the SSS and painDETECT instru-
ments was also examined because these instruments have
not been examined in children.
A2-step cluster analysis was performed to profile children
based on their self-reported pain descriptors and somatic
symptomatology (ie, baseline measures in clinic). This pro-
cedure was selected because it is fairly robust to violations
in assumptions of variable independence and normality
and uses a log-likelihood distance measure to identify the
optimal number of homogeneous structures suggested by
the data (ie, not defined a priori), maximizing the difference
between groups in an efficient manner. The Akaike informa-
tion criterion was used as the clustering criterion. We ran-
domly ordered the cases before analyses as recommended to
minimize the potential effect of order on the final solution.
We next compared the resultant clusters to describe
other nonsymptom characteristics of the groups (eg, sex,
self-reported pain duration) and to identify potential con-
founders for postoperative pain outcomes (eg, periopera-
tive analgesic management, surgical factors, and in-hospital
analgesic requirements). These univariate analyses were
conducted using independent t tests for parametric data
(eg, age) and χ2
for nonparametric data (eg, sex).
Outcomes and Hypothesis Tests. Our pain outcomes
included postdischarge pain intensity (ie, 2-week average
self-reported NRS) and total opioid consumption (ie, 2-week
oral morphine equivalents per kilogram). To examine
the association between cluster profile and our 2 related
outcome variables, we used a general linear multivariate
multiple regression main effects model. Both outcome
measures were simultaneously regressed onto the main
factor of interest (ie, cluster membership) to test for the
association of our independent variable while controlling
for child sex and total hospital opioid consumption.
Given the unequal distribution of total hospital morphine
consumption, this factor was analyzed and modeled on
the log scale. We acknowledge the complex relationship
between pain intensity and opioid consumption and that
differences in 1 of these outcomes may counter differences
in the other in clinical practice (eg, higher pain intensity
may increase opioid demand or higher opioid use may
lower pain intensity).24
The multivariate model allowed us
to examine whether the mean difference between clusters
on the combination of pain outcomes was likely to have
occurred by chance.25
Furthermore, the multivariate model
permitted us to examine and more clearly describe which
(if either) of these outcomes was associated with our fixed
factor, cluster membership. The Bonferroni correction was
used to conservatively adjust our level of significance for
multiple comparisons (ie, 0.05/2 based on 2 outcomes), and
estimated marginal mean differences (with 97.5% CIs) are
presented.
In a smaller sample for which we had obtained pilot
longitudinal outcomes at 6 months, we compared ongoing
pain, ongoing analgesic use, and PROMIS pain interference
between clusters (identified only in a large sample) for the
purpose of reporting preliminary relationships. Univariate
analyses were used for these comparisons.

4. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
May 2017 • Volume 124 • Number 5 www.anesthesia-analgesia.org 1597
Sample Size. Based on previous reports of chronic
musculoskeletal pain in children,8,26
we estimated that
approximately 20% of children with scoliosis would
present with multisite pain and symptomatology. We
determined a priori that a sample of 61 children would be
sufficient to estimate a similar proportion with a high pain
and symptom profile with sufficient precision (ie, 95% CI
width of approximately 0.20) and to detect at least a 1.5-
point difference between groups in reported pain intensity
(standard deviation = 1.5; 2-sided α = 0.05; β = 0.20). This
pain intensity difference has been previously shown to
reflect the minimal clinically significant difference in
children and adolescents.10,27–29
RESULTS
Seventy-six children were approached for participation,
and 3 declined as a result of inadequate time to complete
the survey and/or follow-up study. Three children did not
complete significant portions of their baseline surveys leav-
ing 70 for cluster analyses and subsequent comparisons.
Overall, 73% of the children were female, aged 10 to 17
years (mean 13.96 ± 1.76), and with an average body mass
index of 22.18 ± 5.47 kg/m2
.
Description of Baseline Pain and Symptoms
Most children (86%) reported having pain within the
recent 6 months, 45% reported preoperative pain lasting 3
months, and 74% had used analgesics (5% opioids). None
of the children were taking opioids immediately before sur-
gery. Sixty-three percent reported experiencing pain at mul-
tiple sites (ie, 3 or more) although back pain was the most
commonly reported site (86%) and a lower extremity the
next most common (23%). Children’s reports of pain loca-
tion were consistent from preoperative clinic to the day of
surgery approximately 2 weeks later (ICC = 0.927 [95% CI,
0.875–0.957]; P .001), showing excellent reporting reliabil-
ity and stability of pain at baseline.
Children also reported a high prevalence of other somatic
symptoms, including fatigue (58%), depression (37%),
abdominal cramping (39%), and headache (62%). Children’s
symptom reporting was consistent from the clinic visit to the
day of surgery (ICC = 0.816 [95% CI, 0.679–0.894]; P .001),
and there was a high degree of internal consistency among
items in the SSS (Cronbach α = 0.791 [95% CI, 0.71–0.858]).
Scores on the pFSC ranged from 0 to 26 (6.97 ± 5.16; median
6) and were consistent over time, supporting reliability (ICC =
0.912 [95% CI, 0.845–0.950]; P .001). Children’s scores on the
painDETECT ranged from 0 to 31 (mean 5.3 ± 5.62; median
4). The internal consistency of the painDETECT was accept-
able (Cronbach α = 0.757 [95% CI, 0.662–0.834]; P .001), and
children were moderately consistent in their reports of neuro-
pathic symptoms (ICC = 0.654 [95% CI, 0.365–0.811]; P .001).
There was a variable presence and degree of baseline neu-
ropathic pain symptoms in this sample with 51% of children
never or hardly ever noticing the symptoms documented
with painDETECT. One-third (33%) reported numbness,
pain to light pressure, or sudden, electric-like pain attacks.
Smaller numbers reported burning sensations (17%), tin-
gling/prickling (26%), pain to light touching from a blanket
or clothing (14%), and less often, pain to heat or cold (7%).
Baseline Symptom Cluster Description
The cluster analysis identified 2 groups of children that
were well characterized by their baseline pain and somatic
symptom reporting (see Silhouette measure of cohesion and
separation, Figure 1A). The high symptom cluster included
30% of the sample (95% CI, 20.2%–41.8%) that were cohe-
sively profiled by high scores on the PROMIS depression,
pain interference, fatigue, pain catastrophizing, pFSC, and
painDETECT instruments (Figure 1B). Figure 2 depicts the
distribution of symptom scores for clusters in relation to the
overall medians and quartile ranges.
The nonsymptom characteristics of these groups are
depicted in Table 1 together with the surgical and periop-
erative (intraoperative and postanesthesia recovery) data.
Children with longer standing pain (ie, self-reported to
have persisted 3 months’ duration) were more likely to
be a member of the high symptom profile (odds ratio, 3.19
[95% CI, 1.12–9.13]; P = .027). Children in the high symptom
profile also reported higher pain intensity in the previous 6
months than those in the low profile group (6.77 ± 2.05 vs
4.02 ± 2.51; P .001).
Associations Between Cluster Membership and
Postoperative Pain Outcomes
Table 2 depicts the in-hospital and 2-week postdischarge
pain and analgesic data for the groups. To test our hypoth-
esis, we used a multivariate main effects regression model
to examine the association between cluster membership
and the combined and individual pain outcomes (average
pain intensity and total opioid consumption after hospi-
tal discharge) controlling for potential confounders of sex
and total opioid consumption in the hospital. This model
supported our hypothesis that high symptom cluster mem-
bership was independently associated with combined pain
outcomes (Pillai’s Trace F = 3.50 [df, 2]; P = .039) explain-
ing 14% of the variation in 2-week pain outcomes (Table 3).
More specifically, our hypothesis tests demonstrated that
high cluster membership was associated with higher self-
reported pain after discharge (mean difference +1.13 NRS
[97.5% CI, 0.09–2.17]; P = .015) but not with 2-week opioid
consumption (mean difference +0.92 morphine equivalents
per kilogram [97.5% CI, –0.75 to 2.59]; P = .208; see Table 3
for the parameter estimates for all covariates). Although
total opioid consumption did not differ between groups in
our corrected model, children in the high symptom cluster
were more likely to report ongoing opioid use at 2 weeks
compared with the low symptom group (87% vs 50%; odds
ratio, 6.5 [95% CI, 1.30–33.03]; P = .015).
In a small subset of 28 children who have completed their
6-month follow-up to date, all children in the high symp-
tom subset reported ongoing analgesic use compared with
only 41% of those in the low symptom profile (P = .018).
Furthermore, these children reported higher pain intensity
(5.86 ± 2.54 vs 1.73 ± 1.67; P .001) and pain interference
(14.33 ± 8.82 vs 2.67 ± 3.94; P .001).
DISCUSSION
In this sample of children and adolescents with IS, we
found a high prevalence of preoperative pain and comorbid
symptoms including diffuse pain and symptoms similar to

5. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
1598 www.anesthesia-analgesia.org ANESTHESIA ANALGESIA
Pain Vulnerable Profile in Children With Scoliosis
those observed in patients with chronic and centralized
pain conditions. Supporting our hypothesis, we differenti-
ated a high pain and symptom profile in 30% of children
(95% CI, 20.7%–41.8%) that was associated with higher
pain scores in the hospital and at home but not indepen-
dently associated with opioid use. Furthermore, children in
our differentiated high symptom profile group were more
likely to report ongoing opioid use at the 2-week clinic visit
and, in a smaller sample, higher pain, pain interference,
and analgesic use 6 months after surgery. These findings
suggest that children with a high pain and somatic symp-
tom profile preoperatively are potentially vulnerable to
poor self-reported postoperative pain outcomes, including
ongoing analgesic use.
Our data suggest that as many as one-third of children
with surgical range IS present for surgery with a symp-
tom profile similar to that observed in individuals with
chronic and centralized pain disorders. Although this
symptom profile was largely differentiated by psychologic
factors such as depression, pFSC scores contributed signifi-
cantly and independently to cluster membership, helping
to identify children at risk for higher pain and opioid use
postoperatively. Similar to our findings, high FSC scores
in adults have helped describe a phenotype, including
higher anxiety, fatigue, depression, and neuropathic pain,
and predicted persistent pain and higher opioid use after
arthroplasty.7,30
FSC scores have been used to identify the
prevalence of widespread pain and SSS in a variety of adult
populations,7
and scores not only differentiate fibromyalgia,
but also depict a quantitative continuum of centralized pain
(ie, “fibromyalgia-like” state) that has been observed across
many differing chronic pain states.13,31,32
Furthermore, data
from subjects with differing chronic pain conditions show
abnormal sensory responses, including hyperalgesia, allo-
dynia, poor pain inhibition, and augmented central ner-
vous system activity, suggesting a fundamental problem
with pain processing that could potentially be explained
by a central mechanism.33,34
It remains unknown whether
a common underlying physiologic mechanism can explain
the symptom profiles observed in our study. In addition, we
used the pFSC to score symptoms that have been described
in children with fibromyalgia; however, this instrument as
used here is not meant to diagnose, but merely to examine
the presence and degree of symptomatology. Despite this
limitation, our findings raise important considerations for
the use of qualitative and quantitative pain and symptom
measures to help identify children who are vulnerable to
worse pain outcomes.
Importantly, 41% of girls in our sample were clustered in
the high symptom group compared with only 11% of boys.
Figure 1. A, Summary of the resultant cluster analysis. B, Depiction of the resultant cluster symptom characteristics.

6. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
May 2017 • Volume 124 • Number 5 www.anesthesia-analgesia.org 1599
This is similar to findings in large healthy cohorts where
girls were at higher risk for persistent multisite pain, mood,
and somatic symptoms.35
In our sample, we did not find an
independent association between sex and pain outcomes in
our corrected model. Sex differences in children’s clinical
and experimental pain responses as well as their response
to analgesia remain poorly understood given mixed
findings across studies. However, a recent meta-analysis
found that younger boys and girls did not differ in their
pain responses, but in those 12 years of age, girls reported
higher pain intensity and lower thresholds than did boys,
suggesting a potential effect of puberty.36,37
Socialization
and parental preferences may play a role in pain and anal-
gesic use since we previously reported that parents of girls
Figure 2. Comparison of characteristic
traits between high and low symptom
clusters. Box represents median (ie, per-
pendicular line) within the inner 25, 75
quartile; horizontal lines depict the median
and inner quartile range for the high and
low symptom groups. Data presented are
the median (intraquartile range [IQR]) of
the groups.
Table 1. Nonsymptom Characteristics and Perioperative Data for the Clusters
High Symptom Cluster (n = 22) Low Symptom Cluster (n = 48) P Value
Age (y) 13.5 ± 1.54 14.13 ± 1.83 .168
Female sex 20 (91%) 30 (65%) .021
Body mass index (kg/m2
) 23.62 ± 5.71 21.58 ± 5.34 .150
ASA I/II/III 5 (23%)/14 (64%)/3 (14%) 9 (19%)/37 (77%)/2 (4%) .303
Spinal segments fused .545
Thoracic and lumbar 17 (77%) 40 (83%)
Thoracic only 5 (23%) 8 (17%)
Number spinal levels fused 10.4 ± 1.7 10.9 ± 1.8 .254
Distribution of surgeons A/B/C 32%/36%/32% 35%/46%/19% .673
Method of opioid delivery intraoperatively
Intrathecal morphine 16 (73%) 31 (65%) .501
Epidural morphine 6 (27%) 12 (25%) .840
Remifentanil infusion 13 (59%) 28 (58%) .952
Dexmedetomidine infusion 11 (50%) 33 (69%) .132
Total analgesic doses intraoperatively and in the postanesthesia care unit
Morphine equivalents per kga
0.32 ± 0.18 0.47 ± 0.31 .033
Diazepam per kg 0.04 ± 0.04 0.05 ± 0.05 .458
Acetaminophen per kg 11.93 ± 8.86 11.92 ± 8.75 .993
Ketorolac per kg 0.29 ± 0.17 0.33 ± 0.20 .491
Anesthesia duration (h) 5.46 ± 1.19 5.90 ± 1.10 .132
Surgical duration (incision to dressing) 3.80 ± 1.01 4.17 ± 1.06 .169
Data are presented as mean ± standard deviation or n (%), as applicable.
Abbreviation: ASA, American Society of Anesthesiologists.
a
Includes all intrathecal, epidural, and intravenous opioids given in the operating room or recovery room.

7. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
1600 www.anesthesia-analgesia.org ANESTHESIA ANALGESIA
Pain Vulnerable Profile in Children With Scoliosis
exhibited a higher likelihood of treating their child with
potent analgesics compared with parents of boys with the
same pain intensity.38
Across studies, observed sex differ-
ences in chronic and centralized pain conditions have been
attributed to various biologic, hormonal, and/or socio-
psychologic factors, yet underlying mechanisms remain
unclear.39,40
Our findings suggest that symptom vulnerabil-
ity in girls may manifest during adolescence; however, fur-
ther study is needed to examine potential mechanisms for
sex-related differences in the pain experience.
Limitations
The cross-sectional nature of our baseline assessments
poses limitations. First, children were asked to report cur-
rent and recent symptoms and, although the associations
between many were high, causality between symptoms
remains unknown. Next, it is possible that a symptom recall
bias may have influenced findings given that recalled pain
was recently found to be greater than concurrent ratings in
adults, particularly in those who were habitual symptom
reporters.41
Recall bias may be somewhat reduced in our
study because children’s reports of pain and symptoms
were consistent from the preoperative clinic to the day of
surgery. Importantly, the nature and prevalence of pain
and symptoms could cluster differently in a larger or sepa-
rate sample given the wide CI in our study (ie, 20%–41%),
warranting further study in a larger sample. The potential
influence of perioperative analgesic technique and early
analgesic management on longer term outcomes cannot be
overlooked. Although our groups were similar with regard
to the use of regional technique and anesthetic agents, they
differed in perioperative opioid consumption. Our standard
analgesic approach after surgery (ie, oxycodone around the
clock) precluded our ability to demonstrate a difference in
opioid consumption in the hospital. However, when cor-
rected for total in-hospital opioid consumption, cluster
membership remained an important factor for 2-week pain
intensity. Whether this relationship will hold for longer
term outcomes and, in particular, for pain interference or
function remains to be tested in a larger sample. Finally, our
data suggest the presence of a pain vulnerable phenotype
based on self-reported symptoms, but the underlying mech-
anism remains unknown and the possibility of preoperative
pain centralization remains only speculative. Further study
incorporating quantitative sensory testing methods to
examine sensory differences is needed to test hypothesized
mechanisms.
CONCLUSIONS AND IMPLICATIONS
This study provides evidence that a behavioral pain vulnera-
ble profile, which is present preoperatively in many children
with operative range IS, predicts poorer and potentially long-
lasting pain outcomes after spine fusion. This complex clini-
cal presentation was particularly prevalent in girls and those
who reported long-standing preoperative pain. The under-
lying mechanism for the observed symptom profile remains
Table 2. Postoperative Analgesic Use and Pain Outcomes Between Clusters
High Symptom Cluster (n = 22) Low Symptom Cluster (n = 48) P Value
Early in-hospital outcomes
Average reported pain (NRS 0–10)a
6.00 ± 1.97 4.56 ± 1.65 .002
Pain on movement (NRS 0–10) 6.93 ± 1.67 5.48 ± 1.80 .003
Morphine equivalents per kg per hb
0.013 ± 0.01 0.015 ± 0.013 .605
Diazepam per kg per h 0.008 ± 0.004 0.008 ± 0.003 .535
Acetaminophen per kg per h 2.15 ± 1.51 1.96 ± 1.14 .571
Ketorolac per kg per h 0.06 ± 0.04 0.07 ± 0.03 .440
Pain and analgesic use first 2 wk at home
Average reported pain (NRS 0–10)c
4.93 ± 1.27 3.70 ± 1.38 .031
Morphine per kg totald
5.27 ± 3.27 3.68 ± 1.84 .006
Diazepam per kg total 1.41 ± 1.16 1.09 ± 0.85 .269
Ibuprofen per kg total 79.39 ± 97.46 90.65 ± 99.41 .707
Acetaminophen per kg total 200.63 ± 172.28 198.54 ± 144.28 .964
Nonparametric comparisons used where applicable.
Abbreviation: NRS, Numeric Rating Scale.
a
Shapiro-Wilk (SW) P = .349.
b
SW P .001.
c
SW P = .467.
d
SW P = .077.
Table 3. Parameter Estimates (Marginal Means) for Each of the Covariates in the Multivariate Model
Factors
Outcome: Average Pain Intensity
β (97.5% CI) P Valueb
Outcome: Opioid Consumptiona
β (97.5% CI) P Valueb
High symptom cluster membership (referent low) 1.13 (0.09 to 2.17) .015 0.92 (−0.75 to 2.59) .208
Female sex (referent male) −0.004 (−1.13 to 1.14) .993 1.39 (−0.43 to 3.20) .083
Age 0.037 (−0.27 to 0.34) .778 −0.02 (−0.47 to 0.50) .935
LN (in-hospital opioid use)c
−0.29 (−0.29 to 0.88) .248 0.57 (−0.36 to 1.51) .162
Abbreviation: CI, confidence interval.
a
Oral morphine equivalents per kilogram total.
b
Conservatively, we used the Bonferroni adjustment for multiple comparisons (ie, 0.05/2, given 2 outcomes of interest); 97.5% confidence intervals are therefore
reported.
c
Log normal transformation of this skewed independent variable was used for this analysis.

8. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
May 2017 • Volume 124 • Number 5 www.anesthesia-analgesia.org 1601
unknown. Further study is warranted to better understand
the nature of pain and symptom profiles and potential mech-
anisms and to facilitate targeted treatments to reduce long-
term negative postoperative outcomes. E
ACKNOWLEDGMENTS
The authors thank the following persons who took part in
recruiting or collecting data for this study: Reilly Philliben,
Tabitha Kalibat, Caroline Hyman, Marianna Dorta, Adam
Niemman, Elizabeth Feenstra, and Kenneth Hayes.
DISCLOSURES
Name: Terri Voepel-Lewis, PhD, RN.
Contribution: This author helped design and conduct the study,
oversee recruitment, analyze and interpret the data, and prepare
and revise the manuscript.
Name: Michelle S. Caird, MD.
Contribution: This author helped design the study, oversee recruit-
ment, interpret the data, and revise the manuscript.
Name: Alan R. Tait, PhD.
Contribution: This author helped analyze and interpret the data,
and revise the manuscript.
Name: Shobha Malviya, MD.
Contribution: This author helped interpret the data and revise the
manuscript.
Name: Frances A. Farley, MD.
Contribution: This author helped oversee recruitment, interpret
the data, and revise the manuscript.
Name: Ying Li, MD.
Contribution: This author helped oversee recruitment, interpret
the data, and revise the manuscript.
Name: Matthew D. Abbott, MD.
Contribution: This author helped oversee recruitment, interpret
the data, and revise the manuscript.
Name: Tara van Veen, BS.
Contribution: This author helped oversee recruitment, collect and
enter the data, and revise the manuscript.
Name: Afton L. Hassett, PsyD.
Contribution: This author helped design the study, interpret the
data, measurements, and revise the manuscript.
Name: Daniel J. Clauw, MD.
Contribution: This author helped design the study, interpret the
data, measurements, and revise the manuscript.
This manuscript was handled by: James A. DiNardo, MD, FAAP.
REFERENCES
1. Landman Z, Oswald T, Sanders J, Diab M; Spinal Deformity
Study Group. Prevalence and predictors of pain in surgical
treatment of adolescent idiopathic scoliosis. Spine (Phila Pa
1976). 2011;36:825–829.
2. Connelly M, Fulmer RD, Prohaska J, et al. Predictors of postop-
erative pain trajectories in adolescent idiopathic scoliosis. Spine
(Phila Pa 1976). 2014;39:E174–E181.
3. Sieberg CB, Simons LE, Edelstein MR, et al. Pain prevalence
and trajectories following pediatric spinal fusion surgery.
J Pain. 2013;14:1694–1702.
4. DanielssonAJ, NachemsonAL. Back pain and function 23 years
after fusion for adolescent idiopathic scoliosis: a case-control
study—part II. Spine (Phila Pa 1976). 2003;28:E373–E383.
5. Rabbitts JA, Palermo TM, Zhou C, Mangione-Smith R. Pain
and health-related quality of life after pediatric inpatient sur-
gery. J Pain. 2015;16:1334–1341.
6. Bastrom TP, Marks MC, Yaszay B, Newton PO; Harms Study
Group. Prevalence of postoperative pain in adolescent idio-
pathic scoliosis and the association with preoperative pain.
Spine (Phila Pa 1976). 2013;38:1848–1852.
7. Brummett CM, Janda AM, Schueller CM, et al. Survey criteria
for fibromyalgia independently predict increased postoperative
opioid consumption after lower-extremity joint arthroplasty:
a prospective, observational cohort study. Anesthesiology.
2013;119:1434–1443.
8. Mikkelsson M, El-Metwally A, Kautiainen H, Auvinen A,
Macfarlane GJ, Salminen JJ. Onset, prognosis and risk factors
for widespread pain in schoolchildren: a prospective 4-year
follow-up study. Pain. 2008;138:681–687.
9. Hoftun GB, Romundstad PR, Rygg M. Factors associated with
adolescent chronic non-specific pain, chronic multisite pain,
and chronic pain with high disability: the Young-HUNT Study
2008. J Pain. 2012;13:874–883.
10. Voepel-Lewis T, Burke CN, Jeffreys N, Malviya S, Tait
AR. Do 0-10 numeric rating scores translate into clinically
meaningful pain measures for children? Anesth Analg.
2011;112:415–421.
11. von Baeyer CL. Numerical rating scale for self-report of pain
intensity in children and adolescents: recent progress and fur-
ther questions. Eur J Pain. 2009;13:1005–1007.
12. Zernikow B, Gerhold K, Bürk G, et al; Arbeitsgemeinschaft
der Wissenschaftlichen Medizinischen Fachgesellschaften.
[Definition, diagnosis and therapy of chronic widespread pain
and so-called fibromyalgia syndrome in children and adoles-
cents. Systematic literature review and guideline] [in German].
Schmerz. 2012;26:318–330.
13. Wolfe F, Brähler E, Hinz A, Häuser W. Fibromyalgia prevalence,
somatic symptom reporting, and the dimensionality of polys-
ymptomatic distress: results from a survey of the general popu-
lation. Arthritis Care Res (Hoboken). 2013;65:777–785.
14. von Baeyer CL, Lin V, Seidman LC, Tsao JC, Zeltzer LK. Pain
charts (body maps or manikins) in assessment of the location of
pediatric pain. Pain Manag. 2011;1:61–68.
15. Freynhagen R, Baron R, Gockel U, Tölle TR. painDETECT:
a new screening questionnaire to identify neuropathic com-
ponents in patients with back pain. Curr Med Res Opin.
2006;22:1911–1920.
16. Wilkie DJ, Holzemer WL, Tesler MD, Ward JA, Paul SM,
Savedra MC. Measuring pain quality: validity and reli-
ability of children’s and adolescents’ pain language. Pain.
1990;41:151–159.
17. Irwin DE, Stucky B, Langer MM, et al. An item response analy-
sis of the pediatric PROMIS anxiety and depressive symptoms
scales. Qual Life Res. 2010;19:595–607.
18. Varni JW, Stucky BD, Thissen D, et al. PROMIS Pediatric Pain
Interference Scale: an item response theory analysis of the pedi-
atric pain item bank. J Pain. 2010;11:1109–1119.
19. Varni JW, Magnus B, Stucky BD, et al. Psychometric properties
of the PROMIS® pediatric scales: precision, stability, and com-
parison of different scoring and administration options. Qual
Life Res. 2014;23:1233–1243.
20. Vervoort T, Goubert L, Eccleston C, Bijttebier P, Crombez G.
Catastrophic thinking about pain is independently associ-
ated with pain severity, disability, and somatic complaints in
school children and children with chronic pain. J Pediatr Psychol.
2006;31:674–683.
21. Guite JW, McCue RL, Sherker JL, Sherry DD, Rose JB.
Relationships among pain, protective parental responses,
and disability for adolescents with chronic musculoskeletal
pain: the mediating role of pain catastrophizing. Clin J Pain.
2011;27:775–781.
22. McPherson M. Demystifying Opioid Conversion Calculations: A
Guide for Effective Dosing. Bethesda, MD: American Society of
Health-System Pharmacists; 2009.
23. International Anesthesia Research Society (IARS). Pain sub-
specialty: pain and regional anesthesia opioid conversion—
iv and it. Open Anesthesia. 2016. Available at: https://www.
openanesthesia.org/aba_opioid_conversion_-_iv_and_it/.
Accessed February 16, 2016.
24. Kissin I. Patient-controlled-analgesia analgesimetry and its
problems. Anesth Analg. 2009;108:1945–1949.
25. Tabachnick B, Fidell L. Multiple regression. Using Multivariate
Statistics. 5th ed. Boston, MA: Pearson Education Inc; 2007; p
243–247.
26. El-Metwally A, Salminen JJ, Auvinen A, Macfarlane G,
Mikkelsson M. Risk factors for development of non-specific
musculoskeletal pain in preteens and early adolescents: a pro-
spective 1-year follow-up study. BMC Musculoskelet Disord.
2007;8:46.

9. Copyright © 2017 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
1602 www.anesthesia-analgesia.org ANESTHESIA ANALGESIA
Pain Vulnerable Profile in Children With Scoliosis
27. Hirschfeld G. Clinically meaningful changes in pain ratings:
why we need special cut points in children and adolescents.
Pain Manag. 2014;4:81–83.
28. Hirschfeld G, Wager J, Schmidt P, Zernikow B. Minimally
clinically significant differences for adolescents with
chronic pain-variability of ROC-based cut points. J Pain.
2014;15:32–39.
29. McConahay T, Bryson M, Bulloch B. Clinically significant
changes in acute pain in a pediatric ED using the color analog
scale. Am J Emerg Med. 2007;25:739–742.
30. Brummett CM, Urquhart AG, Hassett AL, et al. Characteristics
of fibromyalgia independently predict poorer long-term anal-
gesic outcomes following total knee and hip arthroplasty.
Arthritis Rheumatol. 2015;67:1386–1394.
31. Wolfe F, Clauw DJ, Fitzcharles MA, et al. Fibromyalgia criteria
and severity scales for clinical and epidemiological studies: a
modification of the ACR Preliminary Diagnostic Criteria for
Fibromyalgia. J Rheumatol. 2011;38:1113–1122.
32. Clauw DJ. Fibromyalgia: a clinical review. JAMA.
2014;311:1547–1555.
33. Williams DA, Clauw DJ. Understanding fibromyalgia: lessons
fromthebroaderpainresearchcommunity.JPain.2009;10:777–791.
34. Woolf CJ. Central sensitization: implications for the diagnosis
and treatment of pain. Pain. 2011;152:S2–S15.
35. Kröner-Herwig B, Gassmann J, van Gessel H, Vath N. Multiple
pains in children and adolescents: a risk factor analysis in a lon-
gitudinal study. J Pediatr Psychol. 2011;36:420–432.
36. El-Metwally A, Salminen JJ, Auvinen A, Kautiainen H,
Mikkelsson M. Prognosis of non-specific musculoskeletal pain
in preadolescents: a prospective 4-year follow-up study till
adolescence. Pain. 2004;110:550–559.
37. Boerner KE, Birnie KA, Caes L, Schinkel M, Chambers CT. Sex
differences in experimental pain among healthy children: a sys-
tematic review and meta-analysis. Pain. 2014;155:983–993.
38. Voepel-Lewis T, Zikmund-Fisher BJ, Smith EL, Zyzanski S, Tait
AR. Parents’ preferences strongly influence their decisions to
withhold prescribed opioids when faced with analgesic trade-
off dilemmas for children: a prospective observational study.
Int J Nurs Stud. 2015;52:1343–1353.
39. Hassan S, Muere A, Einstein G. Ovarian hormones and chronic
pain: a comprehensive review. Pain. 2014;155:2448–2460.
40. Luntamo T, Sourander A, Rihko M, et al. Psychosocial deter-
minants of headache, abdominal pain, and sleep problems in
a community sample of Finnish adolescents. Eur Child Adolesc
Psychiatry. 2012;21:301–313.
41. Walentynowicz M, Bogaerts K, Van Diest I, Raes F, Van den
Bergh O. Was it so bad? The role of retrospective memory in
symptom reporting. Health Psychol. 2015;34:1166–1174.