ORIGINAL ARTICLE
The Wong-Baker Pain FACES Scale
Measures Pain, Not Fear
Gregory Garra, DO, Adam J. Singer, MD, Anna Domingo, BS, and Henry C. Thode, Jr, PhD
Objective: The Wong-Baker FACES Pain Rating Scale (WBS) is pre-
ferred by parents and patients for reporting pain severity. However, it is
speculated that the ‘‘no hurt’’ and ‘‘hurts worst’’ anchors confound pain
measurement with nonnociceptive states. The objective of our study was
to determine if fear confounds reporting of pain severity on the WBS. We
hypothesized that the WBS would correlate with a psychometrically dif-
ferent pain severity scale (the visual analog scale [VAS]) and not correlate
with a fear measure, the Child Medical Fear Scale (CMFS).
Methods: This was a prospective observational study of children 7 to
12 years presenting to a university-based suburban pediatric ED with
acute pain. Patients rated pain severity on the WBS ordinal scale and a
100-mm unhatched VAS with marked end points of ‘‘no pain’’ and
‘‘worse pain ever.’’ Patients also completed a 26-item CMFS. Correlations
between the WBS and VAS with the CMFS total score were assessed with
Spearman correlation and exploratory factor analysis.
Results: All 3 scales were completed in 197 children. Correlation between
the severity scales (WBS-VAS) was moderate: 0.63 (95% confidence in-
terval [CI], 0.54Y0.71). However, correlations between the WBS-CMFS and
VAS-CMFS were poor: j0.02 (95% CI, j0.16 to j0.12) and 0.01 (95%
CI, j0.13 to 0.15), respectively. Correlations did not differ by sex, grade,
pain location, or cause of pain (traumatic vs atraumatic). Exploratory factor
analysis demonstrated excellent loadings within 2 factors: pain and fear.
Conclusions: The WBS demonstrates moderate correlation with another
measure of pain (VAS) and is not mistaken for fear among school-aged
patients presenting to the ED with pain.
Key Words: pain measurement, pain severity, medical fear, validity
(Pediatr Emer Care 2013;29: 17Y20)
P ain is a multidimensional experience. Measurement of pain
is complicated by issues such as age, developmental level,
cognitive and communication skills, prior pain experiences,
cultural beliefs and norms, fear, and anxiety.1 In the emergency
department (ED) setting, children may experience as much fear
and anxiety as they do pain.2 Distinguishing between these
sensations can be challenging and is important for treatment.
Self-report scales are commonly used to measure pain in
children. These scales are dependent on the sensory, emotional,
and contextual nature of the noxious stimuli.3 Facial expression
drawings (‘‘faces scales’’) are among many tools available to
assess pain severity in children. The Wong-Baker FACES Pain
Scale (WBS) is preferred by children and parents for reporting
pain severity.4,5 However, there is controversy over the affect-
laden anchors of the WBS versus neutral anchors of the Faces
Pain ScaleYRevised.4,6,7
From the Department of Emergency Medicine, Stony Brook University
Hospital, Stony Brook, NY 11794.
Disclosure: The authors declare no conflict of interest.
Reprints: Gregory Garra, DO, Stony Brook University Medical Center,
Emergency Medicine, Z=8350, Stony Brook, NY 11794
(e-mail: Gregory.Garra@stonybrook.edu).
Copyright * 2013 by Lippincott Williams & Wilkins
ISSN: 0749-5161
Pediatric Emergency Care & Volume 29, Number 1, January 2013
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Given the multidimensional aspect of pain, an appropriate
measurement instrument must demonstrate the ability to measure
pain while discriminating from other confounders (discriminant
validity). A recently published systematic review on the validity
and reliability of faces scales identified only a few studies
addressing discriminant validity of faces scales.7 The objective
of our study was to determine if the WBS is mistaken for fear.
We hypothesized that pain severity reporting on the WBS
would correlate with scores on a psychometrically different
severity scale (100-mm visual analog scale [VAS]) and would
not correlate with scores on a medical fearfulness scale (Child
Medical Fear Scale [CMFS]).
METHODS
Study Design
This was a single-center, observational cohort of patients,
aged 7 to 12 years, presenting to the ED with pain as a chief
complaint. The study was approved by our Office for Research
Compliance. Informed written consent was obtained from all
parents and assent was obtained from patients older than 10 years
before study enrollment.
Setting
The study was conducted in a university-based pediatric
ED with 28,000 visits per year from September 2010 to
December 2010.
Methods for Selecting Participants
A convenience sample of ED patients were identified by
research assistants who were in-serviced on the study criteria
and objectives.8 Research assistants were present in the ED
for patient enrollment Mondays through Saturdays from 10 AM
until 8 PM. Research assistants identified patients from ages 7 to
12 years at presentation to the ED by reviewing intake forms.
Patients were eligible for the study if pain was reported in the
chief complaint. Patients were excluded from the study if they
possessed any disabilities (auditory, visual, physical, or mental)
that interfered with their ability to comprehend instructions
for completing the various pain assessment scales or marking a
response on the various pain assessment scales.
Study Protocol
Research assistants approached eligible patients for enroll-
ment before physician evaluation. Following informed written
consent, collection of demographic and clinical data was per-
formed using a standardized data collection form. Data collected
included demographic information such as age, sex, ethnicity,
and level of education. Specific historical information included
the location of pain, etiology of pain (traumatic vs atraumatic),
mechanism of injury (if traumatic), and duration of pain.
Following collection of demographic data, pain measure-
ments and the CMFS were administered by research assistants.
Patients were asked to rate their pain on the WBS and VAS.
Research assistants pointed to each of the 6 faces and described
each face using the ‘‘brief word instructions’’ provided with the
www.pec-online.com 17
Garra et al
FIGURE 1. Child Medical Fear Scale distribution by WBS
categories.
scale.9 Patients were asked to circle the face that best repre-
sented their level of pain severity. Immediately after obtaining
WBS scores, patients were asked to rate their pain severity on a
horizontally positioned, 100-mm VAS with marked end points
of ‘‘no pain’’ and ‘‘worst possible pain.’’ Research assistants were
instructed to ask, ‘‘Show me on the line, the amount of pain you
have; here is no pain; there is the worst possible pain.’’ Patients
were instructed to draw a mark on the line to indicate the level of
pain. The data collection instrument was designed to prevent the
patient from visualizing their response on the WBS when
marking the VAS. Following completion of the WBS, research
assistants administered the CMFS.
The CMFS was utilized to assess the degree of fearfulness
about medical experiences.10 The scale consists of questions
pertaining to 4 categories: procedural, environmental, intraper-
sonal, and interpersonal fears. The items are answered with a
structured, 3-point forced-choice format. The child is asked to
indicate whether he/she is not at all afraid (1), a little afraid (2),
or a lot afraid (3) for each of the 29 statements. The points are
added up to provide the total fear score for the child.
TABLE 1. Mean VAS and CMFS for WBS Categories
n 3 12
Mean VAS, mm 15 (0Y69) 28.6 (17Y40) 37.6 (31Y44)
(95% CI)
Mean CMFS 40 (35Y46) 44 (38Y50)
(95% CI)
* 1983 Wong-Baker FACES Foundation. Used with permission.
18 www.pec-online.com
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care & Volume 29, Number 1, January 2013
Three questions were eliminated from the questionnaire
because of concerns that they would frighten children or require
further explanation: ‘‘I am afraid I might die if I go to the
hospital,’’ ‘‘I am afraid the nurse will tell me something is
wrong,’’ and ‘‘I am afraid the nurse will not tell me what to do.’’
The revised scale scores range from 26 to 78, with low scores
reflecting children who have low fears of medical experiences
and high scores indicating greater fearfulness.
All scales were administered in an interview format, thus
allowing opportunities to clarify questions or concepts that were
confusing. Parents were encouraged not to tutor, coerce, assist,
or influence their child’s responses.
Data Analysis
All data were entered into PASW Statistics18.0 for Windows
(SPSS Inc, Chicago, Ill). Categorical variables are presented as the
percentage frequency of occurrence. Continuous variables are
presented as means and SDs. Correlations between the WBS and
VAS with the CMFS total score and subscale (procedural, envi-
ronmental, intrapersonal, and interpersonal fears) scores were
assessed with Spearman correlation.
Visual analog scale, WBS, and the total CMFS score were
entered into a factor analysis. Because WBS is an ordinal-level
variable, a categorical principal components analysis was per-
formed using the SPSS procedure CATPCA. Visual analog scale
and the CMFS total score were used as numeric variables be-
cause VAS is continuous, and the CMFS variable is a sum of
scores. There was no prior assumption of the number of latent
factors underlying the measurements. Rotation of factors was
not performed. Loadings greater than or equal to 0.7 were con-
sidered excellent.
RESULTS
Two hundred forty-six children were enrolled. All 3 scales
were completed in 196 patients after excluding 1 subject whose
CMFS was determined to be a high outlier (Fig. 1). Excluding
this subject had no effect on the results. Of the population
studied, 44% were female; 78% were white. The median dura-
tion of symptoms was 8 hours (interquartile range, 3Y48 hours).
The most common pain locations were the upper extremity
(29%), lower extremity (26%), head/ scalp (16%), abdomen
(14%), and other (15%). Trauma/injury was the reason for pain
42 72 54 14
53.1 (48Y58) 71.8 (67Y76) 76.9 (65Y89)
40 (37Y42) 40 (38Y42) 40 (38Y42) 41 (39Y44)
* 2013 Lippincott Williams & Wilkins
Pediatric Emergency Care & Volume 29, Number 1, January 2013
in 65%. The mean VAS was 55 (SD, 24.3) mm. The mean
CMFS score was 40 (SD, 8).
Mean VAS scores and mean CMFS for each of the WBS
pain categories are provided in Table 1. Analysis of variance
demonstrated a significant difference in the mean VAS between
WBS categories; however, tests for homogeneity of variance
across categories were not significant. There was no significant
difference in mean CMFS between the WBS categories. Corre-
lation between WBS and VAS was moderate, 0.63 (95% confi-
dence interval [CI], 0.54Y0.71). Correlations between the total
CMFS and the WBS and VAS pain scales were poor: j0.02
(95% CI, j0.16 to 0.12) and 0.01 (95% CI, j0.13 to 0.15),
respectively. Correlations between the CMFS subscales (proce-
dural, interpersonal, intrapersonal, and environmental fears) and
the WBS and VAS were also poor. Correlations did not differ by
sex, grade, pain location, or cause of pain (traumatic vs atraumatic).
Results of the exploratory factor analysis are listed
in Table 2. Factor loadings can be considered correlations
between observed variables and underlying latent factors that
the variables represent, with the latent factors being indepen-
dent. Factor 1 can be interpreted as the pain factor because of
the high correlation with the pain measurements (WBS loading
0.90 and VAS loading 0.90) but is not related to fear (CMFS
loading j0.05). Factor 2 is only correlated with fear.
DISCUSSION
This study examined the performance of the WBS, a faces
pain scale with affect-laden anchors, on school-aged children’s
reporting of acute painful conditions. We found that each WBS
face category was associated with progressively higher mean
VAS scores. The most commonly selected face was the ‘‘hurts
even more’’ face, which correlated with a mean VAS of 55 mm.
There were few patients who selected the smiling, ‘‘no hurt’’
face. However, the inclusion criterion was a chief complaint
of pain. On the opposite end of the spectrum, there were only
14 children (7%) who selected the tearful, ‘‘worst hurt’’ face. As
a result, CIs for the mean VAS within this category were wide.
Taken from another perspective, nearly a fifth of the study
population reported a VAS greater than 80 mm. The majority
(18/35) of those patients did not select the tearful, ‘‘worst hurt’’
face. This finding supports notion that emotionally laden
anchors are mistaken for nonnociceptive emotions and bias se-
verity reporting.4Y6,11 In a study of pain ratings using hypo-
thetical vignettes, Chambers and Craig6 suggested that the
presence of a smiling face biased pain ratings away from the
‘‘no pain’’ end of spectrum. A similar conclusion was reported
in a study of pain ratings among children after venipuncture.4
It is speculated that children younger than 10 years have
difficulty differentiating simultaneous states (pain and anxi-
ety).12 Robertson11 reported in a study that included 23 children
undergoing elective surgery that ‘‘on more than 1 occasion’’ a
face other than the smiling face was chosen as part of a preop-
erative pain assessment tool. In the same study, children who
were being discharged after abdominal surgery sometimes chose
the smiling faces, suggesting that the term ‘‘hurting’’ or ‘‘feeling’’
is confused. The aim of our study was to ascertain whether fear
confounds pain reporting on the WBS. The study population
appeared to have a baseline level of fear, with an overall mean
CMFS of 40 (SD, 8). However, the mean CMFS scores between
the WBS categories were not statistically different. Correlations
between pain severity reporting (WBS and VAS) and CMFS
scores were poor. Further attempts to discern discrimination be-
tween pain and fear were statistically analyzed using exploratory
factor analysis. Factor analysis tells us what variables group
* 2013 Lippincott Williams & Wilkins
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Wong-Baker Pain FACES Scale Measures Pain
together. We assume that observed variables are correlated or go
together because they share 1 or more underlying causes (fac-
tors). In our analysis, the VAS and WBS group together. We can
therefore assume that WBS and VAS are correlated because they
share 1 or more underlying factors (ie, pain measurement).
Likewise, neither the WBS nor the VAS correlated with CMFS.
Hence, they did not share underlying factors with each other.
Although our study methodology does not provide strong
proof of discriminant validity, the results suggest that fear does
not appear to play a role in pain severity reporting among
school-aged children. This distinction is important if the WBS
is utilized for measuring severity and treating pain in emergency
settings. A pain severity scale that fails to distinguish fear/
anxiety from pain may result in children receiving inappropriate
medications for a fearful state. Likewise, a scale that confuses
pain for fear/anxiety may decrease the likelihood of receiving
appropriate pain treatment, a problem that is all too common in
the ED setting.13 We believe the results of our study provide
clinicians with increased confidence that the WBS is providing
information on pain severity and not pain confounders.
Limitations
As with any survey research, this study has its limitations.
This was a single-center study enrolling a convenience sample
of predominantly white patients. Convenience sampling may
result in erroneous data. Because we did not collect data on
patients presenting to the ED on overnight shifts, it is possible
that we may have missed an entirely different population of
patients. Furthermore, it is possible that the WBS may not work
as well in other ethnic groups. Beyer and Knott14 demonstrated
low and nonsignificant correlations between the Oucher Scale
and the CMFS among African American children. However,
correlations were not low enough to discriminate pain and fear
among older Hispanic children.
We used a revised version of the CMFS. The revised ver-
sion has not been tested elsewhere. However, our revision
contained the same exact language as that which was used with
the original scale but contained 3 fewer questions. We chose to
eliminate the question on ‘‘I am afraid I will die’’ because our
plan was to interview young children who might not understand
the concept of death. We further chose to eliminate questions
that seemed redundant as they were asked in a physician con-
text: ‘‘I am afraid the nurse will tell me something is wrong with
me’’ and ‘‘I am afraid the nurse will not tell me what he/she is
going to do to me.’’ In addition, the CMFS was administered as
an interview in the presence of caregivers. The original study on
the CMFS was conducted in a school setting.10 It is possible that
subtle interviewer biases, the presence of caregivers, and utilizing
a revised version of the scale reduced the overall validity and
reliability of the CMFS.
Order effects may confound the results of the study. The pain
scales were always administered in the same order and before the
CMFS. Counterbalancing the order of administration or random-
izing the order of scale collection may have provided different
TABLE 2. Factor Loadings From Categorical Factor Analysis
Using WBS, Pain VAS, and CMFS (n = 196)
Factor 1 Factor 2
WBS 0.90 j0.004
VAS 0.90 0.06
CMFS j0.05 0.998
www.pec-online.com 19
Garra et al
results. Furthermore, CMFS and VAS were not completed by all
patients. Thirty-one patients did not complete the CMFS.
According to our research associates, there were a number of
patients who simply lost interest in answering the 26-question
survey. Likewise, there were 11 patients who, despite instruc-
tions, could not grasp the concept of the VAS. However, results
of WBS/CMFS correlation would not have changed with in-
clusion of these subjects.
The CMFS is a multi-item scale. The VAS and WBS are
single-item scales. By virtue of the shared method variance of the
single-item scales, one would expect a higher correlation (and
factor loading) between these scales than in comparison with the
multi-item scale. We could have developed a single-item visual
analog, numeric rating, or Likert fear scale asking, ‘‘How scared
are you right now?’’ However, it was our intent to evaluate pre-
viously studied scales.
In essence, we attempted to determine discriminant validity
of the WBS. The most appropriate statistical method for deter-
mining discriminant validity is debatable. A simplistic method
for demonstrating that measures are dissimilar is to perform a
correlation coefficient. Theoretically, low cross-construct corre-
lations provide evidence that 2 sets of measures are discriminated
from each other. More sophisticated methods of demonstrating
discriminant validity utilize multitrait-multimodel matrix. The
multitrait-multimodel uses more than 1 measure of constructs
and more than 1 method to measure the constructs to assess
convergent and discriminant validity.15 We did not use other
sophisticated analytical methods for proving that the WBS
discriminates pain from fear.
CONCLUSIONS
The concern over the impact of affect-laden anchors in a
pain faces scales is established and understandable. An appro-
priate tool should neither overstate nor underreport pain severity.
Likewise, an appropriate measurement tool should not report
nonnociceptive experiences. In our study, the WBS correlated
reasonably well with a VAS and did not appear to be mistaken for
fear. Despite the concern for underreporting of pain severity, the
WBS appears to provide valid estimates of pain when concur-
rently measured with a VAS among school-aged children. Within
the limits of our study, there is no reason to believe that severity
reporting on the WBS is confounded by fear. Our results suggest
that school-aged children are able to discriminate pain from fear
when completing pain severity scales.
20 www.pec-online.com
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care & Volume 29, Number 1, January 2013
ACKNOWLEDGMENTS
The authors thank the research associates who were vital
in collecting data.
REFERENCES
1. O’Rourke D. The measurement of pain in infants, children and
adolescents: from policy to practice. Phys Ther. 2004;84:560Y570.
2. Magaret ND, Clark TA, Warden CR, et al. Patient satisfaction in the
emergency department: a survey of pediatric patients and their parents.
Acad Emerg Med. 2002;9:1379Y1388.
3. Howard RF. Current status of pain management in children. JAMA.
2003;290:2464Y2469.
4. Chamber CT, Giesbrecht K, Craig KD, et al. A comparison of faces
scales for the measurement of pediatric pain: children’s and parents’
ratings. Pain. 1999;83:25Y35.
5. Chambers CT, Hardial J, Craig KD, et al. Faces scales for the
measurement of postoperative pain intensity in children following
minor surgery. Pain. 2005;21:277Y285.
6. Chambers CT, Craig KD. An intrusive impact of anchors in children’s
faces pain scales. Pain. 1998;787:27Y37.
7. Tomlinson D, von Baeyer CL, Stinson JN, et al. A systematic review
of faces scales for the self-report of pain intensity in children.
Pediatrics. 2010;126:e1168Ye1198.
8. Hollander JE, Singer AJ. An innovative strategy for conducting clinical
research: the academic associate program. Acad Emerg Med. 2002;9:
134Y137.
9. Jacob, E. Pain assessment and management in children. In:
Hockenberry. MJ, Wilson D, eds. Wong’s Essentials of Pediatric
Nursing. 8th ed. St Louis, MO: Mosby/ Elsevier; 2009:162.
10. Broome M, Hellier A. School-age children’s fears of medical
experiences. Issues Compr Pediatr Nurs. 1987;10:77Y86.
11. Robertson J. Pediatric pain assessment: validation of a multidimensional
tool. Pediatr Nurs. 1993;19:209Y213.
12. Gross AL, Ballif B. Children’s understanding of emotion from facial
expression and situations: a review. Dev Rev. 1991;11:368Y398.
13. Kaplan CP, Sison C, Platt SL. Does a pan scale improve pain
assessment in the pediatric emergency department? Pediatr Emerg
Care. 2008;24:605Y608.
14. Beyer JE, Knott CB. Construct validity estimation for the
African-American and Hispanic versions of the Oucher Scale.
J Pediatr Nurs. 1998;13:20Y31.
15. Bollen KA. Structural Equation Models With Latent Variables.
New York: John Wiley & Sons; 1989.
* 2013 Lippincott Williams & Wilkins