The Trichotillomania Scale for Children: Development
David F. Tolin Æ Æ Gretchen J. Diefenbach Æ Æ Christopher A. Flessner Æ Æ
Martin E. Franklin Æ Æ Nancy J. Keuthen Æ Æ Phoebe Moore Æ Æ John Piacentini Æ Æ
Dan J. Stein Æ Æ Douglas W. Woods Æ Æ
Trichotillomania Learning Center Scientific Advisory Board
Published online: 8 January 2008
? Springer Science+Business Media, LLC 2008
repetitive hair-pulling resulting in alopecia. Although this condition is frequently observed
in children and adolescents, research on pediatric TTM has been hampered by the absence
of validated measures. The aim of the present study was to develop and test a new self-
report measure of pediatric TTM, the Trichotillomania Scale for Children (TSC), a
measure that can be completed by children and/or their parents. One hundred thirteen
Trichotillomania (TTM) is a chronic impulse control disorder characterized by
D. F. Tolin (&) ? G. J. Diefenbach
The Institute of Living, Anxiety Disorders Center, 200 Retreat Avenue, Hartford, CT 06106, USA
D. F. Tolin
Yale University School of Medicine, New Haven, CT, USA
C. A. Flessner ? D. W. Woods
University of Wisconsin-Milwaukee, Milwaukee, WI, USA
M. E. Franklin
University of Pennsylvania School of Medicine, Philadelphia, PA, USA
N. J. Keuthen
Harvard University/Massachusetts General Hospital, Boston, MA, USA
Duke University Medical Center, Durham, NC, USA
University of California at Los Angeles, Los Angeles, CA, USA
D. J. Stein
University of Cape Town, Cape Town, South Africa
D. J. Stein
Mt. Sinai School of Medicine, New York, NY, USA
Trichotillomania Learning Center Scientific Advisory Board,
Santa Cruz, CA, USA
Child Psychiatry Hum Dev (2008) 39:331–349
children meeting self-report criteria for TTM, and 132 parents, provided data over the
internet. An additional 41 child–parent dyads from an outpatient clinic also provided data.
Replicated principal components analysis, with elimination of poorly-loading items,
yielded two components, which we labeled Severity (five items) and Distress/Impairment
(seven items). The TSC total score and subscales showed adequate internal consistency and
test-retest reliability. Parent–child agreement was good in the internet sample, but more
modest in the clinic sample. Children’s TSC scores correlated significantly with other
measures of TTM severity, although some exceptions were noted. Parents’ TSC scores
correlated significantly with other measures of parent-rated TTM severity in the internet
sample, but showed more attenuated relationships with child- and interviewer-rated TTM
severity in the clinic sample. The present results suggest that the TSC may be a useful
measure of TTM for child and adolescent samples, although additional clarification of
convergent validity is needed.
Hair-pulling ? Questionnaire ? Impulse control disorders ? Child disorders
Trichotillomania (TTM), a chronic impulse control disorder characterized by repetitive
pulling out of one’s own hair and resulting alopecia, appears to be more common in young
people than was previously believed. Although early reports  described child and
adolescent TTM as a very rare condition, more recent studies using epidemiological and
student samples of older adolescents and young adults estimate the prevalence of clinically
significant hairpulling to range between 1% and 3.5% [2–4]. The prevalence of TTM
among younger children remains unknown.
Psychiatric comorbidity appears to be quite common among adults with TTM, partic-
ularly mood, anxiety, substance use, and personality disorders [5–8]. In children and
adolescents, approximately one third to two thirds of TTM patients meet criteria for at least
one comorbid Axis I disorder, with a preponderance of anxiety and internalizing disorders
TTM appears to be associated with substantial impairment and reductions in quality of
life. The majority of adult TTM patients report problems such as impaired social func-
tioning, negative affect, interference with grooming behaviors, impaired recreational
activity, work productivity, and physical illness or symptoms caused by pulling [12, 13].
Children and adolescents with TTM report spending an average 30–60 min per day pulling
hair, and report experiencing significant distress about their symptoms .
Research on pediatric TTM has been hampered by the absence of validated measures for
use with children and adolescents. A self-report measure, the Massachusetts General
Hospital Hairpulling Scale (MGH-HPS)  has demonstrated good psychometric prop-
erties in adult samples [14, 15], although correlations with global clinician ratings are low
. Furthermore, the language in the MGH-HPS might not be appropriate for younger
children, TTM severity is represented by a single score rather than by specific factors, and
there is no means of assessing parents’ perceptions of TTM severity. Clinician-rated scales
such as the Psychiatric Institute Trichotillomania Scale  and the NIMH Trichotillo-
mania Severity Scale  have shown poor internal consistency in small (N’s = 22–28)
adult samples [16, 19], perhaps due in part to the conceptual differences across the specific
items. The lack of validated measures is of particular concern for clinical trials with
pediatric samples. One trial  found that the NIMH Trichotillomania Severity Scale was
sensitive to the effects of treatment; however, the poor internal consistency of that scale
may render findings somewhat unstable.
332Child Psychiatry Hum Dev (2008) 39:331–349
The aim of the present study was to develop and test a new self-report measure of
pediatric TTM, the Trichotillomania Scale for Children (TSC). Because of possible dis-
crepancies between child- and parent-report of symptom severity [20–22], a child version
(TSC-C) and parent version (TSC-P) were developed. It was predicted that the TSC-C and
TSC-P would demonstrate adequate internal consistency, test-retest reliability, and con-
vergent and divergent validity.
Data were collected as part of the Trichotillomania Impact Project for Children (TIP-C),
the details of which are reported elsewhere . Participants were children aged 10–17
who met self-reported modified diagnostic criteria for TTM and one of their parents or
legal guardian. Modified diagnostic criteria required that the child (1) pulls his/her hair
resulting in noticeable hair loss (both parent and child report); (2) ‘‘never/almost never’’
(0–10% of the time) pulls his/her hair because voices tell him/her to pull hair (both parent
and child report); (3) ‘‘never/almost never’’ (0–10% of the time) pulls because he/she
believes small bugs are crawling on him/her (child report); (4) has not always pulled as the
result of physical causes (e.g., skin conditions, physical illness, or injury) or the use of
medications, drugs, or alcohol (parent report); and (5) hair pulling results in at least ‘‘mild
to moderate’’ impairment (a score of 3 or greater on a 9-point Likert scale) in day-to-day,
social, interpersonal, or academic functioning (parent or child report).
Three hundred thirty six children and their parents participated in the TIP-C, and of
these, 133 of the children met the modified diagnostic criteria for TTM. Only those
participants who completed the TSC portion of the TIP-C (n = 113 children, n = 132
parents) are included in the current study. As shown in Table 1, this was a primarily female
and Caucasian sample, with an average age of 14 years. Per parental report, 67.3%
(n = 76) had previously been diagnosed with TTM and 37.2% (n = 42) had been diag-
nosed with another psychiatric disorder instead of or in addition to TTM. Thirty-nine
children, or 34.8% of the sample, had never been formally diagnosed with any psychiatric
condition, including TTM. The most common categories of diagnosed comorbid disorders
according to parent report were anxiety disorder, mood disorder, and attention deficit-
hyperactivity disorder (ADHD).
Forty-one child–parent dyads participated in the present study as part of an open trial and
randomized controlled trial of CBT for pediatric TTM . Data were included when
either the child (n = 38) or parent (n = 39) completed the measure. Inclusion criteria were
age 8–17, primary diagnosis of TTM as assessed using the Trichotillomania Diagnostic
Interview (TDI) , and minimum symptom duration of 6 months. Exclusion criteria
were a primary diagnosis other than TTM; current bipolar illness, developmental disorder,
thought disorder; or current psychotherapy. As shown in Table 1, this sample was sig-
nificantly younger than the internet sample, with a trend (p = .06) toward a greater
Child Psychiatry Hum Dev (2008) 39:331–349333
proportion of minority participants. As is customary in TTM studies , criteria B and C
(increasing and decreasing tension) of the DSM-IV-TR criteria for TTM were deemed
optional, as these criteria have been found to exclude patients with clearly significant hair
pulling [6, 8, 26]. Comorbid disorders were diagnosed for 36.8% (n = 14) of the sample
using the Anxiety Disorders Interview Schedule for DSM-IV, Child Version (ADIS-C)
, with the most common comorbid conditions anxiety disorders and disruptive
Trichotillomania Scale for Children
The first two authors (DFT and GJD), psychologists with extensive experience in the
assessment of TTM, created the initial item pool for the Trichotillomania Scale for
Children (TSC). Fifteen items were created to reflect three a priori domains of TTM
psychopathology: severity, distress, and impairment (five items per domain). The items
were also reviewed and modified by two additional psychologists with expertise in TTM.
Items created for the severity scale assessed frequency of urges/pulling, duration of pulling
episode, number of hairs pulled, and controllability of pulling. Distress scale items
assessed emotional responses associated with pulling (e.g., guilt, embarrassment, sadness,
self-reproach). Impairment items assessed interference with peer/family relationships,
schoolwork, and grooming routines. Each item consisted of three or more sentences,
reflecting varying levels of severity [e.g., (0) I did not pull any hair at all, (1) I pulled out
between 1 and 10 hairs on most days, (2) I pulled out more than 10 hairs on most days].
Items were scaled from 0 to 2, with higher scores indicating more severe symptoms. In
Table 1 Sample descriptions=
Internet sample Clinic sample
Child age 14.38 (2.33)12.58 (2.87) 3.88**
Child female81 (71.7%)28 (73.7%)1.00
Child White/Non-Hispanic101 (89.4%)29 (76.3%) .06
Child Hispanic/Latino6 (5.5%) 2 (5.3%)
Child African-American1 (0.9%) 6 (15.8%)
Child multiracial3 (2.7%) 0 (0.0%)
Child anxiety disorder 26 (23.0%)11 (28.9%) .51
Child mood disorder21 (18.6%)3 (7.9%).20
Child ADHD17 (15.0%) NA
Child tic disorder6 (5.3%)1 (2.6%).68
Child eating disorder5 (3.8%) NA
Child disruptive behavior disorder4 (3.5%)4 (10.5%) .11
Child Asperger’s disorder 1 (0.9%) NA
Child PTSD1 (0.9%) 0 (0.0%)
Parent age44.14 (6.75) NA
NA = Not Assessed, FET = Fisher’s Exact Test, ** p\001
334Child Psychiatry Hum Dev (2008) 39:331–349
cases where multiple answers were selected, the highest (i.e., most severe) selected value
was scored. Once the initial item pool was completed a parallel parent version of the TSC
was created by modifying the administration instructions.
Measures of TTM (Internet Sample)
Children rated [from 1 (mild) to 9 (severe)] the degree to which TTM interferes with their
social life, ability to make friends or get closer to friends, and school or school work.
Parents rated [from 1 (mild) to 9 (severe)] the degree to which TTM interferes with their
child’s social life, ability to form and maintain close relationships, ability to work, and
academic life. They also provided a numeric rating (up to 10+) for the number of family
vacations, social events, and days of school missed in the past 12 months due to TTM.
Measure of Comorbid Psychopathology (Internet Sample)
Children completed the Multidimensional Anxiety Scale for Children (MASC) , a self-
report measure of anxiety symptoms. The MASC demonstrates excellent internal consis-
tency and adequate convergent and divergent validity . Parents completed a parallel
version of this measure, the Parent Report on Child’s Anxiety Symptoms (PROCAS) .
Items constituting the PROCAS are identical to MASC items except that nouns and pro-
nouns are altered to match the parent’s perspective (e.g., ‘‘My child…’’ rather than ‘‘I…’’).
March et al.  found that parent–child agreement ranged from r = 0.18 (father–child,
MASC total score) to r = 0.71 (mother–child, Physical Symptom subscale). Subsequent
research has demonstrated acceptable to very good internal consistency for parent-report of
the Harm/Avoidance (a = 0.68), Separation Anxiety (a = 0.72), Physical Symptom
(a = 0.81), and Social Anxiety (a = 0.85) subscales . The Children’s Depression
Inventory (CDI)  was administered to assess self-reported depressive symptoms in the
child sample. The CDI demonstrates strong internal consistency , acceptable test-retest
reliability , and acceptable convergent validity .
Clinician Measures of TTM (Clinic Sample)
The NIMH Trichotillomania Severity Scale (NIMH-TSS)  is a semi-structured clini-
cian-rated scale comprised of five items that assess time spent pulling in the past week,
time spent pulling the previous day, resistance to pulling, distress, and interference.
Resistance is rated on a scale ranging from 0 to 4, and the four other items are rated on a
scale ranging from 0 to 5, with higher scores indicating greater symptom severity. The total
severity score is calculated by summing the five items. The NIMH-TSS demonstrates
adequate internal consistency, excellent inter-rater agreement, and adequate correlations
with other TTM interviews, but shows poor correspondence with self-reported TTM
severity and degree of alopecia . Additional information about TTM severity was
obtained using the Psychiatric Institute Trichotillomania Scale (PITS) , a clinician-
rated measure that assesses various aspects of TTM severity. PITS items are scaled from 0
to 7, with higher scores indicating more severe symptoms. The PITS shows excellent inter-
rater reliability but rather low internal consistency . Therefore, emphasis was placed on
item scores collected via the PITS rather than the total score. The items used for the present
analyses were Severity, Impairment, and Distress.
Child Psychiatry Hum Dev (2008) 39:331–349335
Measures of Comorbid Psychopathology (Clinic Sample)
The MASC  was administered to children to assess self-reported anxiety symptoms.
The CDI  was administered to children to assess self-reported depressive symptoms.
The TIP-C study was approved by the University of Wisconsin–Milwaukee’s Institutional
Review Board and the survey was linked to the website of the Trichotillomania Learning
Center (TLC), a consumer organization, from May through July 2006. Participants were
recruited through an e-mail distribution to TLC members. Respondents were informed that
submission of the survey was an indication of consent to participate in research. The
child’s parent was asked to complete his/her section of the survey first then to leave the
room and allow his/her child to complete the survey on his/her own unless the child
specifically requested assistance with completing the survey.
Thirty-one participants were assessed at the University of Pennsylvania School of Medi-
cine in Philadelphia, PA; 10 were assessed at the Institute of Living in Hartford, CT.
Participants at the two sites did not differ in terms of age or TSC scores. There was,
however, a significant sex difference between the two sites, with girls representing 100% of
the IOL participants and only 61.2% of the University of Pennsylvania participants
(Fisher’s Exact Test = .012). Children were accompanied by at least one parent, and
written assent (children) and consent (parents) were obtained. Assessments were completed
as part of the pretreatment evaluation for a treatment-outcome study, which was approved
by both the University of Pennsylvania and Hartford Hospital Institutional Review Boards
. Doctoral-level psychologists or postdoctoral fellows trained and supervised by a
licensed psychologist completed diagnostic assessments and clinician-rated measures. A
subsample of children (n = 9) and parents (n = 5) completed the TSC a second time at the
beginning of the first therapy session to determine test-retest reliability. The mean inter-
vening interval was 16.22 days (SD = 10.18) for children and 12.60 days (SD = 5.68) for
To examine whether the variability in TSC item scores could be explained by underlying
factors, we used principal components analysis with oblimin rotation for the TSC-C in the
internet sample. Number of components was determined by examining eigenvalues greater
than 1.0, visual examination of the scree plot, and parallel analysis [33, 34] using Monte
Carlo simulation software . In parallel analysis, the pre-rotation eigenvalues are
compared to those from a matrix of random values using the same N and number of
variables; eigenvalues from the data set greater than those from the random matrix are
retained. The factor structure was then replicated in the parent sample using the TSC-P.
336 Child Psychiatry Hum Dev (2008) 39:331–349
Internal consistency of the TSC-C and TSC-P, as well as their subscales, was calculated
using Cronbach’s a. Test-retest reliability was examined in the clinic subsample using
Pearson’s r. Parent–child agreement was examined using Pearson’s r as well as inde-
pendent-samples t-tests. Convergent validity was examined using Pearson’s r.
The KMO index of sampling adequacy for the TSC-C in the internet sample was .78,
indicating that the correlation matrix was suitable for factor analysis . The analysis
yielded four components with eigenvalues greater than 1.0. Examination of the scree plot
suggested that a solution with no more than three components was appropriate. Parallel
analysis revealed that two eigenvalues from the present data set exceeded those from the
simulation. Therefore, a 2-component solution was used (eigenvalues 4.58 and 2.12); these
components accounted for 44.73% of the variance. The correlation between the two
components was .35. Next, only those items that loaded C.40 on one component and B.30
on the other were retained. This resulted in the removal of three items, leaving two
components which we labeled Severity (five items) and Distress/Impairment (seven items).
Examinations of skewness and kurtosis indicated that both scales were approximately
normally distributed. Item loadings are shown in Table 2. Each of the two subscales was
scored by taking the mean of its responses; a total score was derived by summing the two
Table 2 Pattern matrix for the Trichotillomania Scale for Children (child and parent versions)=
Sad or depressed because of hair pulling or bald patches
Upset at self because of hair pulling or bald patches
Embarrassed about hair pulling or bald patches
How has hair pulling affected the way child looks
How much longer did it take to get ready for school or go
out because of bald patches
How guilty about hair pulling
Avoid activities because of hair pulling or bald patches
Frequency of hair pulling-.048
How many hairs pulled out-.138
Duration of hair pulling episodes .021
Frequency of urges to pull-.017
Control over urges to pull .149
Teased by others because of hair pulling or bald spots.318.104.22.1685
Hair pulling causes problems with family members .329.079-.027 .440
Interference with school or chores.257.323.182.506
Child Psychiatry Hum Dev (2008) 39:331–349 337
A similar analysis was conducted for the TSC-P. The KMO index of sampling adequacy
was .829, indicating that the correlation matrix was suitable for factor analysis . A 2-
component solution yielded eigenvalues of 4.89 and 2.08, accounting for 46.44% of the
variance. The correlation between the two components was .37. As shown in Table 2,
loading patterns were similar to those obtained for the TSC-C.
The TSC-C and TSC-P showed adequate to good internal consistency (a) in the internet
sample. For the TSC-C, Severity a = .76, item-total correlations .36–.64; Distress/
Impairment a = .84, item-total correlations .43–.70; total score a = .83, item-total cor-
relations .38–.62. For the TSC-P, Severity a = .76, item-total correlations .43–.60;
Distress/Impairment a = .85, item-total correlations .48–.74; total score a = .84, item-
total correlations .32–.70. Internal consistency was also adequate to good in the clinic
sample: For the TSC-C, Severity a = .72, item-total correlations .26–.69; Distress/
Impairment a = .83, item-total correlations .41–.66; total a = .82, item-total correlations
.28–.64. For the TSC-P, Severity a = .71, item-total correlations .29–.64; Distress/
Impairment a = .76, item-total correlations .32–.66; total score a = .70, item-total cor-
The TSC-C and TSC-P showed adequate test-retest reliability in the clinic sample. For the
TSC-C, Severity r = .81, Distress/Impairment r = .84, total r = .89. For the TSC-P,
Severity r = .70, Distress/Impairment r = .97, total r = .90.
Table 3 shows Pearson correlations between TSC-C and TSC-P scores. Scale-specific
correlations were adequate to high in the internet sample (Table 3, top), with the strongest
Table 3 Correlations between parent and child scores on the Trichotillomania Scale for Children=
Total .291 .354*
* p\.05. ** p\.01
338Child Psychiatry Hum Dev (2008) 39:331–349
correlation for Distress/Impairment. The correlation between the TSC-C and TSC-P total
scores was high. Independent-samples t-tests did not indicate significant differences
between children and parents on Severity (t243= 0.21, p = .83), Distress/Impairment
(t244= 0.26, p = .79), or total score (t243= 0.29, p = .77). Mean scores on the TSC-C
were: Severity M = 1.49, SD = 0.39; Distress/Impairment M = 1.20, SD = 0.47; total
M = 2.70, SD = 0.70. Scores on the TSC-P were: Severity M = 1.48, SD = 0.38; Dis-
tress/Impairment M = 1.19, SD = 0.52; total M = 2.67, SD = 0.75.
Scale-specific agreement was moderate but significant in the clinic sample (Table 3,
bottom), with the strongest agreement again seen for Distress/Impairment. Independent-
samples t-tests did not indicate significant differences between children and parents on
Severity (t73= 0.38, p = .70), Distress/Impairment (t73= -1.50, p = .14), or total score
(t73= -0.79, p = .43). Mean (SD) scores on the TSC-C were: Severity 1.31 (0.42),
Distress/Impairment 0.80 (0.47), total 2.11 (0.73). Mean (SD) scores on the TSC-P were:
Severity 1.28 (0.38), Distress/Impairment 0.95 (0.43), total 2.23 (0.59).
Tables 4 and 5 show Pearson correlations between impairment ratings and scores on the
TSC-C and TSC-P, respectively, in the internet (top) and clinic (bottom) samples. On the
TSC-C, higher TTM Severity, Distress/Impairment, and total scores were associated with
significantly greater self-reported interference with social functioning, making friends,
and school work, as well as greater self-reported depression severity on the CDI.
Table 4 Correlations between Impairment ratings and scores on the Trichotillomania Scale for Children-
Child Version (TSC-C)=
Interferes with social lifeChild.306**.618**.585**
Interferes with making new friends or getting
closer to friends
Child .298**.623** .584**
Interferes with school or school work Child.334**.362**.428**
MASC total scoreChild.181 .366**.345**
CDI total scoreChild.369**.596** .605**
NIMH-TSS Interviewer.352*.311 .398*
MASC total scoreChild.334 .310.388*
CDI total scoreChild.387*.606** .607**
Note: MASC = Multidimensional Anxiety Scale for Children, CDI = Children’s Depression Inventory,
CGI-S = Clinician’s Global Impression-Severity, NIMH-TSS = NIMH Trichotillomania Severity Scale
* p\.05. ** p\.01
Child Psychiatry Hum Dev (2008) 39:331–349339
Distress/Impairment and total score, but not Severity, were significantly associated with
greater self-reported anxiety on the MASC. On the TSC-P, higher TTM Severity, Dis-
tress/Impairment, and total scores were associated with significantly greater self-reported
interference with social life, ability to form and maintain close relationships, ability to
work, and anxiety as measured by the PROCAS. Distress/Impairment and total score, but
not Severity, were associated with greater impairment in academic life, missing events,
and school absence. Distress/Impairment was associated with more frequent school
In the clinic sample, there were no additional parent ratings of impairment for exam-
ination of convergent validity. Therefore, TSC-P scales were compared to child and
interviewer ratings. Scores on the TSC-C showed moderate but significant correlations
with TTM severity as measured by the NIMH-TSS and CGI-S. The TSC-C subscales
showed a specific relationship with PITS Severity, Interference, and Distress scores. TSC-
C scores also correlated significantly with depression and anxiety as measured by the CDI
and MASC. Scores on the TSC-P did not correspond as well to TTM severity on the
NIMH-TSS, although the TSC-P total score was moderately and significantly correlated
with TTM severity on the CGI-S and the TSC-P Severity and Distress/Impairment were
significantly correlated with PITS Severity and Distress, respectively. TSC-P scores were
not significantly related to child-reported distress on the CDI or MASC.
Table 5 Correlations between Impairment Ratings and Scores on the Trichotillomania Scale for Children-
Parent Version (TSC-P)=
Variable Rater TSC-P scale
Interferes with social life Parent.203* .443** .414**
Interferes with ability to form and maintain close
Parent .179* .396**.369**
Interferes with ability to workParent .421* .435*.516**
Interferes with academic lifeParent .079.298** .246*
How many events missed in past 12 mo. Parent.152.548** .466**
How many days of school missed in past 12 mo. Parent.141.416**.364**
PROCAS Total Score Parent .223*.430**.415**
CGI-S Interviewer .329.236 .360*
NIMH-TSSInterviewer .178 .194.256
PITS severityInterviewer .325*-.061.163
PITS interferenceInterviewer .122 .192.218
MASC total scoreChild-.137 .040-.054
CDI total scoreChild .131.296.299
Note: PROCAS = Parent Report on Child’s Anxiety Symptoms, MASC = Multidimensional Anxiety Scale
for Children, CDI = Children’s Depression Inventory, CGI-S = Clinician’s Global Impression-Severity,
NIMH-TSS = NIMH Trichotillomania Severity Scale
* p\.05. ** p\.01
340 Child Psychiatry Hum Dev (2008) 39:331–349
To date, no valid measures of pediatric TTM have been developed. Even in adult samples,
most measures such as the PITS and NIMH-TSS show substantial psychometric limitations
[16, 19] and therefore the utility of these measures for evaluating TTM in children and
adolescents is questionable. The most clearly reliable measure of adult TTM is the Mas-
sachusetts General Hospital Hairpulling Scale [14, 15]; however, even this measure shows
poor correlations with global clinician ratings of TTM severity , and the language
might not be appropriate for children. Development of reliable and valid measures of
pediatric TTM, therefore, is critical for both clinical and research purposes. In the present
study, the TSC-C and TSC-P showed a replicable factor structure and acceptable internal
consistency and test-retest reliability, although the small sample size for the test-retest
sample renders this result tentative.
The present results suggest that the TSC may be a useful measure of TTM for child and
adolescent samples. One advantage of the measure is its ability to collect data from children
and their parents separately. Children and parents frequently provide differing estimates of
severity of child psychopathology [20–22], and therefore it is important to solicit separate
reports on the child’s TTM symptoms and associated impairment. In the internet sample,
children and parents showed good agreement about TTM severity. However, parent–child
agreement was weaker in the clinic sample, and this might reflect a limitation of the
measure. The reasons for the difference are not clear, although the much smaller sample size
for the clinic sample could have contributed. Children in the clinic sample were also two
years younger on average than were those in the internet sample, with six children (16% of
the sample) younger than age 10, the minimum age for the internet study. Results might be
more reliable with older children and adolescents. Consistent with this notion, Keuthen
et al.  found that adolescent hair-pullers showed better concordance with parents than
did younger children on variables such as awareness of hair-pulling, interference with
academic functioning, and anxiety. Finally, for reasons that are not clear, the clinic sample
showed lower mean TSC scores than did the internet sample.
It is not entirely surprising that children and their parents would offer different per-
spectives on the child’s TTM symptoms. Hair pulling is frequently performed covertly, and
children often go to great lengths to conceal their pulling and resulting alopecia [11, 13,
38]. In such cases, parents might not have accurate information about the frequency and
duration of pulling episodes. Conversely, many children (as well as adults) engage in
‘‘unfocused’’ pulling behaviors outside of awareness [e.g., 2, 8, 39], in which case pulling
severity might be better estimated by an observer. Although distress is inherently sub-
jective and presumably more easily detected by the child, impairment in many cases might
be more obvious to the parent. We would suggest that a thorough assessment of pediatric
TTM involve both child- and parent-report.
Another potential advantage of the TSC is its ability to assess both the severity of TTM
and resulting distress and impairment. These two scales were replicated in children and
parents, and show good internal consistency. Various aspects of TTM are not always
strongly correlated, as suggested by the psychometric properties of measures such as the
PITS, in which a single item is used for each of several facets of TTM including hair
pulling frequency and duration, interference, distress, and alopecia severity. The PITS
shows excellent inter-rater reliability but low internal consistency , suggesting that
although the items are reliable, they do not interrelate strongly. Therefore, different
dimensions of TTM are best considered separately, and degree of distress/Impairment
cannot necessarily be inferred from severity of pulling or alopecia.
Child Psychiatry Hum Dev (2008) 39:331–349341
The convergent validity of the TSC is less clear. The TSC-C correlated somewhat more
strongly with depression (as measured by the CDI) than it did with other measures of TTM
severity (CGI-S, NIMH-TSS, PITS items). Thus, some of the variance in TSC-C scores
might be attributable to negative affectivity. It bears mentioning again that the NIMH-TSS
and PITS suffer from substantial psychometric limitations , and therefore it is not
entirely surprising that correlations with the TSC would be attenuated. Another possible
contributor is method variance: the TSC and CDI are both self-report instruments, whereas
the CGI-S, NIMH-TSS, and PITS are all rated by interviewers. The TSC-P in general
showed rather modest correlations against child- and interviewer-rated impairment.
Interpretation of these findings is complicated somewhat by the use of different raters: as
described above, parents’ and children’s’ impressions of psychiatric impairment often
differ, and method variance is a potential limitation. Nevertheless, additional research is
needed to examine more carefully the extent to which the TSC-C and TSC-P correspond to
other indices of TTM severity, distress, and impairment. Reassuringly, correlations tended
to be particularly high for associations between the distress/Impairment subscale of the
TSC-C and other scales of impairment, and between the TSC-C total score and other scales
Another critical question for clinical practice as well as clinical trials is the extent to
which the TSC is sensitive to treatment outcome. The sole published treatment outcome
study of pediatric TTM of which we are aware  used the NIMH-TSS and CGI-S, and
found that both of these measures were sensitive to the effects of cognitive-behavioral
therapy. Outcome research is needed to determine whether the TSC can be used in a
similar fashion. If so, the TSC could be a useful addition to clinical research by allowing
for more frequent self-report assessments of children and their parents.
It might be argued that the use of the internet for data collection is a limitation of the
present study. The internet is increasingly being used for mental health research , and
several studies indicate that web-based data collection results in greater sample diversity,
generalizes across presentation formats, and findings are consistent with data collected
using more traditional means . Equivalence of internet and paper- and pencil mea-
surement has been established in clinical disorders, including anxiety  and obsessive–
compulsive disorder . The present clinic and internet samples were reasonably similar
in terms of basic demographics, and when similar measures were used (e.g., MASC),
convergent scale validity was comparable across the two samples.
Trichotillomania (TTM) is a chronic impulse control disorder characterized by repetitive
pulling out of one’s own hair and resulting alopecia. TTM is associated with high rates of
psychiatric comorbidity and functional impairment, indicating the need for additional
research. However, such research has been hampered by the lack of psychometrically
validated measures of TTM in children and adolescents. Although several TTM self-report
and interviewer-rated scales exist, most either show poor psychometric properties in adults
or use language that may not be appropriate for children and adolescents. The aim of the
present study was to develop and test a new self-report measure of pediatric TTM, the
Trichotillomania Scale for Children (TSC). A child version (TSC-C) and parent version
(TSC-P) were developed. Initial psychometric analyses were conducted using a large
internet-based sample of children with TTM (n = 113) and their parents (n = 132). As
expected, rates of parent-reported psychiatric comorbidity were high. We also conducted
342Child Psychiatry Hum Dev (2008) 39:331–349
additional analyses in a sample of 41 child–parent dyads seeking outpatient treatment for
pediatric TTM. As was the case with the internet sample, rates of diagnosed comorbid
psychiatric disorders were high. The TSC was developed by psychologists experienced
with pediatric TTM. From an initial pool of 15 items (scaled 0–2), principal components
analysis in the internet sample of children revealed a 2-factor solution with 12 adequately-
loading items. The two resulting subscales were labeled Severity and Distress/Impairment.
The factor structure was replicated in the parent internet sample. The TSC-C and TSC-P
showed adequate to good internal consistency in all samples. Test-retest reliability in the
clinic sample was adequate. Parent–child agreement was adequate to high in the internet
sample, but more modest in the clinic sample. In the internet sample, the TSC-C and TSC-
P were significantly correlated with other ratings of TTM-related impairment, suggesting
adequate convergent validity. In the clinic sample, the TSC-C and TSC-P showed more
attenuated correlations with other measures; however, method variance (e.g., comparing
parent, child, and interviewer measures) may have obscured the findings. The present
results suggest that the TSC may be a useful measure of TTM for child and adolescent
samples. Additional research is needed to examine more carefully the extent to which the
TSC corresponds to other indices of TTM severity, distress, and impairment, and to
determine the extent to which the TSC is sensitive to treatment outcome.
of Mental Health grant # R21MH061457, and a research grant by the Trichotillomania Learning Center. In
addition to 4 of the authors (MEF, DWW, NJK, and JP), the Trichotillomania Learning Center Scientific
Advisory Board includes Carol Novak, M.D., Fred Penzel, Ph.D., and Harry H. Wright, M.D., M.B.A.
This research was supported by Hartford Hospital grant #126083, National Institute
Child Psychiatry Hum Dev (2008) 39:331–349343
2. On most days in the last week, how often did you actually pull your hair? This question
means how many times you had a period of pulling—not how many hairs you pulled.
(0) I did not pull my hair at all.
(1) I pulled my hair between 1 and 5 times a day.
(2) I pulled my hair more than 5 times a day.
3. On most days in the last week, how much time did each period of hair pulling last?
(0) I did not pull my hair at all.
(1) A hair-pulling period lasted between 1 second and 5 minutes.
(2) A hair-pulling period lasted more than 5 minutes.
4. On most days in the last week, how many hairs did you pull out?
(0) I did not pull any hair at all.
(1) I pulled out between 1 and 10 hairs on most days.
(2) I pulled out more than 10 hairs on most days.
5. On most days in the last week, how much control did you have over urges to pull your hair?
(You can check more than one thing if you want to).
(0) I did not feel like pulling my hair at all.
(0) I felt like pulling, but I never actually pulled my hair.
(1) I could stop myself from pulling some of the time.
(2) I could not stop myself from pulling at all, even when I really wanted to stop.
(1) Some times I did not want to stop myself from pulling.
(2) I did not want to stop myself from pulling at all.
Note to scorer: Score item 0-2 based on highest rating
Name _________________ Date _________________
TRICHOTILLOMANIA SCALE FOR CHILDREN, CHILD VERSION (TSC-C)
These questions are about your hair pulling. There are no right or wrong answers. Each item is made up
of three sentences. Your job is to pick the sentence that comes closest to describing how things have
been for you in the past week. When you have picked the best sentence, put a check mark in the box
next to it. If you can’t decide which sentence is the best one, it’s OK to check more than one. Try it now
How much do you like homework?
(0) I do not like homework at all.
(1) I like homework a little bit.
(2) I like homework a lot.
Now the questions will be about hair pulling. When we talk about hair pulling, it doesn’t have to be on
your head—it could also include eyebrows, eyelashes, or anywhere else on your body. Remember, your
job is to pick the sentence that best describes how your hair pulling has been in the last week.
1. On most days in the last week, how often did you feel like pulling your hair?
(0) I did not feel like pulling my hair at all.
(1) I felt like pulling my hair once in a while.
(2) I felt like pulling my hair very often.
For office use only: Sum (items 1-5) _____ ÷ 5 = ________ Severity score
6. During the last week, how much longer did it take for you to get ready for school, or get ready
to go out, because you had to hide bald patches?
(0) I didn’t have to hide any bald patches.
(1) It took me a little longer to get ready than other kids, because I had to be careful about how I
styled my hair or put on my makeup.
(2) It took me a whole lot longer to get ready than other kids, because I had to be careful about how I
styled my hair or put on my makeup.
(2) I decided not to go somewhere this week, because it seemed like too much effort to style my hair
or put on my makeup.
Note to scorer: Score item 0-2 based on highest rating
344Child Psychiatry Hum Dev (2008) 39:331–349
10. During the last week, how embarrassed were you about hair pulling or bald patches?
(0) I was not embarrassed at all about hair pulling or bald patches.
(1) I was a little embarrassed about hair pulling or bald patches.
(2) I was extremely embarrassed about hair pulling or bald patches.
11. During the last week, how upset at yourself did you get because of hair pulling or bald
(0) I did not get upset with myself at all because of hair pulling or bald patches.
(1) I got a little upset with myself because of hair pulling or bald patches.
(2) I got extremely upset with myself because of hair pulling or bald patches.
12. During the last week, how sad or depressed did you get because of hair pulling or bald
(0) I did not get sad or depressed at all because of hair pulling or bald patches.
(1) I got a little sad or depressed because of hair pulling or bald patches.
(2) I got extremely sad or depressed because of hair pulling or bald patches.
For office use only: Sum (items 6-12) _____ ÷ 7 = ________ Distress/Impairment score
Severity Score ______ + Distress/Impairment Score ______ = Total Score
7. How much would you avoid the activities listed below because you were embarrassed about
hair pulling or bald patches? (You can check more than one thing if you want to).
(0) I would not avoid any of these activities because of hair pulling or bald patches.
(1) I might not go swimming because I’d be embarrassed about bald patches.
(2) I definitely would not go swimming because I’d be embarrassed about bald patches.
(1) I might not go outside on a windy day because people might see my bald patches.
(2) I definitely would not go outside on a windy day because people might see my bald patches.
(1) I might not go outside on a sunny day because people might see my bald patches.
(2) I definitely would not go outside on a sunny day because people might see my bald patches.
(1) I might not go to a school dance or a party because people might see my bald patches.
(2) I definitely would not go to a school dance or a party because people might see my bald patches.
(1) I might not go on a field trip because people might see my bald patches.
(2) I definitely would not go on a field trip because people might see my bald patches.
(1) I might not hang out with friends or classmates because they might see my bald patches.
(2) I definitely would not hang out with friends or classmates because they might see my bald
Note to scorer: Score item 0-2 based on highest rating
8. During the last week, how do you think hair pulling has affected the way you look?
(0) Hair pulling has not affected the way I look.
(1) I don’t look as good as I would if I didn’t pull my hair.
(2) Hair pulling has made me look really ugly.
9. During the last week, how guilty do you feel about hair pulling?
(0) I did not feel guilty about hair pulling.
(1) I felt a little guilty about hair pulling.
(2) I felt extremely guilty about hair pulling.
Child Psychiatry Hum Dev (2008) 39:331–349 345