This document discusses the Brief Psychiatric Rating Scale for Children (BPRS-C), a 21-item clinician-rated scale used to assess symptoms in youth. It provides evidence that: 1) The BPRS-C demonstrates good reliability and concurrent validity when rated by clinicians for a sample of 547 youth aged 3-18 receiving psychiatric services. Interrater reliability and factor analysis supported the scale's structure. 2) Scores on the BPRS-C were found to be sensitive to symptomatic changes with treatment, such as decreases seen after hospitalization. It has also detected effects of some medications. 3) Further analysis revealed some errors in how items mapped onto scales. The authors propose

1. Mental health program evaluation and treatment out-
come assessment benefits from the availability of objec-
tive quantitative measures of youth adjustment that are
completed by adults (Lyons et al., 1997). This preference
for adult description reflects the concern that children
and adolescents seen for evaluation often present with
motivational, academic, or cognitive deficits that may
limit the validity of self-report of current adjustment.
The diagnostic issue of self-report accuracy, even in older
children and adolescents, has motivated the development
of questionnaire scales to measure the presence of
response sets, such as defensiveness and problem exagger-
ation. The accurate identification of these response sets is
important, as their influence may restrict the validity of
objective measures of child and adolescent adjustment
(cf. Roberts et al., 1989; Stavrakaki et al., 1987b; Wrobel
et al., 1999). Indeed, a comprehensive clinical evaluation
of a child or adolescent often relies on the systematic col-
lection of observations that are normed for parents or
classroom teachers. Several behavior checklists or multi-
dimensional questionnaires to be completed by teachers
or parents are readily available (Lachar, 1998). Clinician
The Brief Psychiatric Rating Scale for Children (BPRS-C):
Validity and Reliability of an Anchored Version
DAVID LACHAR, PH.D., SONJA L. RANDLE, M.D., R. ANDREW HARPER, M.D.,
KATHY C. SCOTT-GURNELL, M.D., KAY R. LEWIS, M.D., CYNTHIA W. SANTOS, M.D.,
ANN E. SAUNDERS, M.D., DEBORAH A. PEARSON, PH.D., KATHERINE A. LOVELAND, PH.D.,
AND SHARON T. MORGAN, PH.D.
ABSTRACT
Objective: Because the accuracy of problems reported by referred children may be compromised by their academic, cog-
nitive, or motivational limitations, clinician rating forms may contribute to the accurate assessment of youth adjustment. One
such measure, the 21-item Brief Psychiatric Rating Scale for Children (BPRS-C), received psychometric study to estimate
its potential contribution to the measurement of symptom dimensions. BPRS-C reliability and concurrent validity were cal-
culated for youths who were receiving psychiatric services within a medical school department. Method: Five hundred
forty-seven children aged 3 to 18 years were rated by faculty or trainees; a subsample of 90 was concurrently rated by two
observers. BPRS-C psychometric performance was demonstrated through interrater agreement, factor analysis, and multi-
variate analyses of variance across seven diagnosis-based groups. Results: Although items and scales demonstrated sub-
stantial reliability and concurrent validity, item factor analysis revealed a few apparent errors in item-to-scale assignment.
These errors were minimized by the use of three new second-order factor-derived scales: Internalization, Developmental
Maladjustment, and Externalization. Conclusions: The BPRS-C can be easily integrated into academic clinical practice
and is a reliable and valid method of child description. Additional study of three new BPRS-C factor scales and the applica-
tion of the BPRS-C to the quantification of clinician observation of child symptomatic status are warranted. J. Am. Acad.
Child Adolesc. Psychiatry, 2001, 40(3):333–340. Key Words: clinician rating form, outcome evaluation, treatment effect,
child psychopathology, Brief Psychiatric Rating Scale for Children.
J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001 333
Accepted October 10, 2000.
Dr. Lachar is Professor and Director of the UTMSI Psychological Assessment
Laboratory; Dr. Randle is Assistant Professor and Director Child/Adolescent Out-
patient Service; Dr. Harper is Assistant Professor and Chief of Child and Adoles-
cent Services, Harris County Psychiatric Center (HCPC); Dr. Scott-Gurnell is
Clinical Assistant Professor and attending psychiatrist, HCPC; Dr. Lewis is
Clinical Associate Professor and Director, Mental Retardation and Developmental
Disabilities Clinic; Dr. Santos is Clinical Associate Professor and Director, Child
Fellowship Program; Dr. Saunders is Associate Professor and Division Chief; Dr.
Pearson is Associate Professor; Dr. Loveland is Professor and Director, Center for
Human Development Research; and Dr. Morgan is Assistant Professor and child
psychologist, HCPC. All from the Division of Child and Adolescent Psychiatry,
Department of Psychiatry and Behavioral Sciences, University of Texas-Houston
Medical School.
Reprint requests to Dr. Lachar, Department of Psychiatry and Behavioral
Sciences, University of Texas-Houston Medical School, P.O. Box 20708, Houston,
TX 77225; e-mail: David.Lachar@uth.tmc.edu.
0890-8567/01/4003-0333᭧2001 by the American Academy of Child
and Adolescent Psychiatry.

2. ratings, however, directly represent the professional inte-
gration and interpretation of such information.
The Brief Psychiatric Rating Scale for Children
Although clinician-completed rating forms can be inte-
grated into contemporary clinical practice, their value in
recording and quantifying trained diagnostic observations
of children has received little attention. The Brief Psychi-
atric Rating Scale for Children (BPRS-C) is not a child
version of the well-known 18-item BPRS clinical rating
form that has often been applied in the study of severely
disturbed adult psychiatric patients (Faustman and
Overall, 1999). The BPRS-C consists of 21 items that are
rated on 7-point scales of severity (not present, very mild,
mild, noderate, moderately severe, severe, extremely
severe). These items were selected to generate seven scales
(Behavioral Problems, Depression, Thinking Distur-
bance, Psychomotor Excitation, Withdrawal-Retardation,
Anxiety, and Organicity) that summarize subsets of three
consecutive items, as well as a summary Total scale. This
rating scale structure and specific content was derived
from a factor analysis of the 63 items of the Children’s
Psychiatric Rating Scale. These data were obtained from
five child and adolescent psychiatrists who completed a
rating for one hypothetical patient for each of 18 DSM-III
diagnoses (Overall and Pfefferbaum, 1982; Pfefferbaum
and Overall, 1982). The approach of generating ratings of
hypothetical patients was selected to ensure that ratings of
a broad spectrum of maladjustment would be obtained
(Pfefferbaum and Overall, 1983).The BPRS-C was devel-
oped to describe major differences among child and
adolescent patients, to characterize them into syndrome
groupings, and to document response to various treat-
ments through repeated BPRS-C completion (Overall
and Pfefferbaum, 1982). The BPRS-C focuses on symp-
tomatic status and thereby complements clinician ratings
of functional impairment, or “caseness,” which is usually
obtained from other clinician ratings forms such as the
Children’s Global Assessment Scale (Bird, 1999).
BPRS-C Literature Review
The BPRS-C has been applied to outcome assessment
(Plante et al., 1995), to psychometric study on the basis of
direct ratings of 48 inpatients and outpatients, and in rat-
ings extracted from the medical records of 40 psychiatric
inpatients (Gale et al., 1986; Mullins et al., 1986). Addi-
tional BPRS-C applications have contributed to the
description or classification of study participants (Emslie
et al., 1990, 1994; McConville et al., 1990; Pfefferbaum
et al., 1987; Seshadri et al., 1989). Studies have validated
the BPRS-C Depression and Anxiety scales by examining
their correlations with the self-report measures of the Chil-
dren’s Depression Inventory and the Revised Children’s
Manifest Anxiety Scale (Stavrakaki et al., 1991), although
these specific results were not replicated in a subsequent
study (Nelson et al., 1995).The intake BPRS-CTotal score
differed between groups of nonpatients, patients diag-
nosed as depressed, and patients diagnosed with an anxiety
disorder (Field et al., 1987). In addition, the BPRS-C
Depression scale was the single most discriminating vari-
able between depressed and anxious patients (Stavrakaki
et al., 1987a).
The BPRS-C has been found to be sensitive to symp-
tomatic change during inpatient treatment. The com-
pletion of the BPRS-C at admission and at discharge
generated significantly lower scores for the Anxiety and
Depression scales after 6 weeks of hospitalization (Nelson
et al., 1995). In another study, depressed and non-
depressed hospitalized children and adolescents did not
differ at admission on the BPRS-CTotal score, although a
second rating at discharge resulted in lower psychopathol-
ogy ratings for both study groups (Benfield et al., 1988).
A major application of the BPRS-C has been in the
quantification of the effects of medication. Simeon and
Ferguson (1987) demonstrated a reduction on the BPRS-
C Anxiety, Depression, and Psychomotor Excitation
scales for overanxious or avoidant outpatients treated with
alprazolam, although a subsequent double-blind study
obtained comparable improvement in ratings in both
placebo and drug conditions (Simeon et al., 1992). Two
studies reported the effect of haloperidol on a group of
children who met the criteria for childhood schizophre-
nia: Spencer et al. (1992) demonstrated a medication-
over-placebo effect by using the BPRS-C Total score.
This finding was retained when the study group was
enlarged from 12 to 16 children (Spencer and Campbell,
1994). Placebo-controlled studies that applied the
BPRS-C Total score demonstrated no effect of bupro-
pion in children with attention-deficit/hyperactivity dis-
order (ADHD) (Casat et al., 1989), clonazepam in
patients with an anxiety disorder (Graae et al., 1994), or
fluoxetine in patients with major depression (Emslie
et al., 1997). One hypothesis to be drawn from these
negative results is that the BPRS-C Total score, a sum-
mary of 21 diverse symptoms, may not be sensitive to
narrowly defined symptomatic improvement. Medication
LACHAR ET AL.
334 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001

3. effects in the form of symptom- or dimension-specific
changes may have been documented if scale or item
analyses had been applied.
Objectives
These initial applications and the stated potential value
of the BPRS-C (Brady and Kendall, 1992; Jackson, 1985;
McReynolds, 1989; Riddle, 1989) suggest that this rating
scale merits additional study within a substantial child
and adolescent psychiatry sample.This current effort eval-
uates the internal structure and reliability of BPRS-C
items and scales and estimates their concurrent validity, as
demonstrated by their performance across DSM-IV-
defined diagnostic samples.
METHOD
Subjects
Children and adolescents were rated by one faculty or trainee clini-
cian within the context of a 6-month survey of all patients seen for
diagnostic or therapeutic services by the Division of Child and Adoles-
cent Psychiatry, Department of Psychiatry and Behavioral Sciences,
University of Texas-Houston Medical School. BPRS-C ratings were
obtained for 547 patients seen within four clinical settings (48% psy-
chiatry inpatient, 8% pediatric hospital consultation, 10% outpatient
clinic for behavioral and developmental disabilities, and 34% child
psychiatry outpatient clinic). This diverse sample was 63% male, 62%
aged 12 to 18 years (mean = 12.3, SD = 3.6; range = 3–18 years), with
substantial representation of white (45%), African-American (34%),
and Hispanic (19%) youth. Family composition or guardianship for
these 547 children was mother and father (39%), mother only (41%),
grandparents (9%), other relatives (4%), or agency placement (7%).
Tabulation of the number of clinician-assigned DSM-IV diagnoses per
child documented a 46% comorbidity rate (1 = 54%, 2 = 37%, 3 =
8%, 4 = 1%). The assignment of these children to a limited number of
relatively large yet symptomatically homogeneous samples on the basis
of diagnosis represented a challenge because of both comorbidity and
the use of a wide variety of separate yet similar diagnoses. Seven sam-
ples that incorporated 87% of the total sample (n = 476) were sequen-
tially developed in the following manner: All children assigned any
psychotic diagnosis were placed into one sample (n = 46, 46% comor-
bid with primarily either a substance abuse or disruptive behavior diag-
nosis) and all children with a pervasive developmental disorder were
placed in another sample (n = 70; 56% comorbid, primarily with a
diagnosis of mental retardation). A third sample consisted of children
with the frequent comorbidity pattern of at least one disruptive behav-
ior diagnosis and a second reflecting the presence of depression (n =
65). The remaining children, who were diagnosed with either one or
more disruptive behavior disorder or one or more depressive disorder,
were assigned to one of four groups: One group included children with
major depression (n = 64, 31% with varied comorbid diagnoses), and
another included children with any remaining diagnosis related to
depression (n = 65, 29% with varied comorbid diagnoses). Children
with conduct disorder formed another group (n = 82, 44% with
comorbid substance abuse, ADHD, or mental retardation), and the
final group included all the remaining children with ADHD combined
type (n = 84, 32% with a comorbid diagnosis representing cognitive or
academic disability).The 71 children who remained unassigned by this
process consisted of five small groups with diagnoses of only cognitive
and academic disability (n = 13), anxiety disorders (n = 12), bipolar dis-
orders (n = 15), adjustment reactions (n = 22), or oppositional defiant
disorder (n = 9).
A subset of 90 children from this total sample were independently
rated by two clinicians (in most cases on the basis of the conjoint
intake, consultation, or admission evaluation completed by one fac-
ulty member and one advanced trainee).
Measure
An anchored version of the 21-item BPRS-C form was completed.
Unlike most applications of the adult BPRS that are based on an inde-
pendent standardized interview, the BPRS-C was completed as a
debriefing instrument. That is, clinicians rated each item on the basis
of all the information known about each patient when the rating was
made. This procedure was selected as being most consistent with the
goal of evaluating the potential contribution of integrating this rating
scale into routine clinical practice to support objectives that require
quantification of patient status, such as program evaluation. This
form provides a definition of the domain for each rated item as well as
examples of behaviors for four of the seven severity ratings (not pre-
sent, mild, moderately severe, and extremely severe), that is, “anchors”
to facilitate selection among severity rating options. Each completed
BPRS-C provides a numeric value for each of 21 items, seven scales,
and a Total scale score. In addition, this current study developed three
second-order factor-derived scales.
Although the original BPRS-C did not provide anchors, their addi-
tion to the current rating form was intended to increase its structure
and thereby increase reliability. Such an effect has been suggested for
the addition of anchors to the adult BPRS (Gabbard et al., 1987;
Woerner et al., 1988). Specific training was not conducted on the use
of this rating form, although faculty occasionally defined rating scale
dimensions and discussed completed ratings with trainees in the con-
text of supervision.
Statistical Analyses
Item Analyses. Data analysis was conducted at both the item and the
scale levels. Interrater reliability, percent descriptive (% endorsement,
a sum of all but the “not present” option), and item dimension struc-
ture as defined by principal components analysis with varimax rota-
tion were obtained for the 21 BPRS-C items. In addition, an estimate
of concurrent validity was obtained through application of multivari-
ate analysis of variance (MANOVA) of items across seven diagnosis-
based groups. Conservative post hoc analysis followed significant
univariate analyses. Each statistically significant item was further eval-
uated by contrasting its performance in each group against all of the
remaining subjects that were not assigned to that specific group.
Scale Analyses. For the eight established BPRS-C scales, coefficient
α, a measure of internal consistency, and interrater reliability were
calculated. In addition, three scales representing second-order factor-
derived dimensions were also constructed. Separate MANOVAs of
two sets of scale scores (seven established, three new) across diagno-
sis-derived groups and a comparable ANOVA for the Total scale
score provided an estimate of concurrent validity. The post hoc anal-
ysis of significant scale ANOVAs identified the specific diagnostic
groups that obtained significantly higher or lower mean scores when
compared with all of the remaining subjects.
BPRS-C VALIDITY AND RELIABILITY
J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001 335

4. RESULTS
BPRS-C Item Performance
Table 1 presents BPRS-C item performance, interrater
reliability, and internal structure as demonstrated by an
exploratory factor analysis. Although our total sample rep-
resented the full range of psychopathology and symptom
intensity, 7 of the 21 BPRS-C items were endorsed
infrequently (i.e., <21%). Peculiar Fantasies, Delusions,
Hallucinations, Speech or Voice Pressure, Disorientation,
Speech Deviance, and Stereotypy showed a mean endorse-
ment of 14% compared with a mean endorsement of 60%
for the remaining 14 items. Interrater reliability estimates of
these 21 items were particularly robust, considering the
attenuated range presented in the correlation of individual
items, with 14 items generating r values greater than 0.69.
Relatively poor interrater agreement was obtained for items
descriptive of covert states and internalizing symptoms
(Peculiar Fantasies, Blunted Affect,Tension, and Anxiety).
The varimax rotated item factor structure presented in
Table 1 and its underlying correlation matrix (available
from senior author) were compared with the previously
established structure of clinician ratings of hypothetical
patients that formed the basis for the original three-item
BPRS-C scales. The items from four BPRS-C scales
(Behavior Problems, Depression, Thinking Disturbance,
and Withdrawal-Retardation) demonstrated the most
robust item-to-item support for their scale placement. In
contrast, the Sleeping Difficulties item from the Anxiety
scale showed a larger correlation with the three items of
the Depression scale than with the other two items of the
Anxiety scale. In addition, the Underproductive Speech
item obtained correlations with the Speech Deviance and
Stereotypy items of the Organicity scale, which were
similar in magnitude to the other two items of the
Withdrawal-Retardation scale. The Speech Deviance and
Stereotypy items also showed substantial correlation with
the Hyperactivity item. Indeed, Table 1 indicates that the
six items that generate the Withdrawal-Retardation and
Organicity scales obtained weights primarily on our fac-
tor dimension I (labeled Developmental Maladjustment),
reducing the seven expected factors to the six that were
actually obtained. In addition, the Speech or Voice Pres-
sure item contributed little to the score of the Psycho-
motor Excitation scale in comparison with this scale’s
two remaining items. (Both Hyperactivity and Dis-
LACHAR ET AL.
336 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001
TABLE 1
Item Analyses of the Anchored Brief Psychiatric Rating Scale for Children (BPRS-C)
% Varimax Principal Components
BPRS-C Item Mean/SD 1–6 r1/r2 I II III IV V VI
1. Uncooperativeness 2.6/1.9 76 0.76 — — 88 — — —
2. Hostility 2.2/2.0 64 0.76 — — 89 — — —
3. Manipulativeness 1.8/1.7 64 0.65 — — 83 — — —
4. Depressed Mood 2.2/2.0 73 0.79 — 86 — — — —
5. Feelings of Inferiority 1.7/1.5 71 0.68 — 57 — — — —
6. Suicidal Ideation 1.4/2.0 40 0.83 — 80 — — — —
7. Peculiar Fantasies 0.3/0.9 12 0.40 — — — — 72 —
8. Delusions 0.3/1.0 9 0.95 — — — — 85 —
9. Hallucinations 0.4/1.2 15 0.83 — 49 — — 63 —
10. Hyperactivity 1.8/1.9 59 0.79 — — — 83 — —
11. Distractibility 2.3/1.9 74 0.79 — — — 83 — —
12. Speech or Voice Pressure 0.3/0.9 12 0.73 — — — 47 46 —
13. Underproductive Speech 1.1/1.8 41 0.63 86 — — — — —
14. Emotional Withdrawal 1.3/1.6 53 0.80 76 — — — — —
15. Blunted Affect 1.2/1.7 44 0.59 73 — — — — —
16. Tension 1.5/1.3 69 0.53 — — — — 77
17. Anxiety 1.5/1.4 67 0.53 — — — — — 83
18. Sleeping Difficulties 1.5/1.9 50 0.81 — 70 — — — —
19. Disorientation 0.2/0.7 17 0.84 57 — — — — —
20. Speech Deviance 0.7/1.6 20 0.90 62 — — 46 — —
21. Stereotypy 0.4/1.2 12 0.83 66 — — 39 — —
% Common variance 15.6 14.5 11.9 10.7 9.6 7.4
Note: Mean/SD, percentage of items rated 1 through 6 (% 1–6) (versus not present), and factor analysis are based on total sample (N = 547).
Interrater reliability subsample (r1/r2) n = 90. I = Development Maladjustment, II = Dysphoria, III = Behavior Problems, IV = Attention Deficit
Hyperactivity, V = Psychosis, VI = Anxiety.

5. tractibility showed a correlation of 0.70 with Psycho-
motor Excitation.)
BPRS-C Scale Performance
The analyses of BPRS-C scales presented in Table 2
demonstrate adequate internal consistency (approx-
imately >.69) for the Total and for six of the three-item
scales.The Anxiety scale, with its misplaced Sleeping Dif-
ficulties item, obtained a coefficient α of only .57. Inter-
rater reliability estimates of all seven three-item scales and
Total scale scores exceeded 0.70 (range = 0.75–0.91,
median = 0.84). The rotated scale factor structure also
presented in Table 2 and the underlying correlation
matrix (available from senior author) were inspected.The
correlation of each scale to the Total score was substantial
(range = 0.44–0.70, median = 0.48), consistent with the
Total scale coefficient α of .79. The principal compo-
nents analysis with varimax rotation of the seven three-
item BPRS-C scales provided substantial support for
three second-order summary dimensions. Scale scores
were subsequently calculated for Internalization (the sum
of Depression, Anxiety, and Thinking Disturbance),
Externalization (the sum of Behavior Problems and Psy-
chomotor Excitation), and Developmental Malad-
justment (the sum of Withdrawal-Retardation and
Organicity). Table 2 also presents substantial coefficient
α and interrater reliability estimates for these three new
second-order scales. All item and scale score means in this
presentation were generated from the conversion of the
seven ranked item rating options to the quantitative val-
ues 0 to 6.
Evidence of Concurrent Validity
The performance of BPRS-C items and scales were
mainly consistent with diagnostic group placement.
MANOVA of the 21 scale items across diagnosis-based
groups was highly significant (Wilks λ = 0.059, F126,2611 =
13.02, p < .0001), as was the performance of the seven
established three-item scales (Wilks λ = 0.107, F42,2175 =
31.59, p < .0001) and the three new factor-derived
broad-band scales across these diagnostic groups (Wilks
λ = 0.152, F3,467 = 865.48, p < .0001). ANOVA of the
Total score across these seven groups was also significant
(F6,469 = 30.85, p < .0001). Subsequent univariate analysis
of the 21 items and 11 scales consistently revealed statis-
tically significant (p < .0001) variability across diagnostic
groups. Instead of a pairwise analysis of diagnostic
groups that would result in 21 comparisons for each item
and scale, a more conservative approach was followed
that resulted in seven comparisons for each item and
scale.To identify the BPRS-C elements significantly asso-
ciated with diagnostic group placement, each group value
was contrasted with the average value of the remaining
six groups. In this manner, scores either significantly
above or significantly below the average of the remaining
diagnostic groups were identified. In addition, the stan-
dardized difference between study group and the sample
remainder was calculated for each statistically significant
BPRS-C VALIDITY AND RELIABILITY
J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001 337
TABLE 2
Scale Analyses of the Anchored Brief Psychiatric Rating Scale for Children (BPRS-C) Within the Total Referred Sample
Varimax Principal Components
BPRS-C Scale α Mean/SD r1/r2 I II III
Behavior Problems 0.87 6.5/5.1 0.79 — — 83
Depression 0.82 5.3/4.7 0.85 82 — —
Thinking Disturbance 0.69 1.0/2.5 0.91 59 — —
Psychomotor Excitation 0.67 4.3/3.8 0.85 — 54 68
Withdrawal-Retardation 0.82 3.6/4.3 0.78 48 71 —
Anxiety 0.57 4.5/3.4 0.75 77 — —
Organicity 0.71 1.3/2.9 0.91 — 91 —
Internalization 0.77 10.8/8.2 0.85
Developmental Maladjustment 0.82 4.9/6.4 0.86
Externalization 0.76 10.9/7.1 0.81
Total 0.79 26.6/14.4 0.83
% Common variance 27.3 26.5 17.9
Note: Mean/SD, percentage of items rated 1 through 6 (% 1–6) and factor analysis are based on the total sample (N = 547).
Factor loadings <0.40 omitted. Largest scale loading in boldface type. I = Internalization, II = Developmental Maladjustment,
III = Externalization.

6. contrast (Cohen, 1988). Statistical significance was estab-
lished at p < .01 and p < .001, and standardized differ-
ences (i.e., mean 1 versus mean 2 divided by the standard
deviation) were classified as either small (<0.40), medium
(0.40–079), or large (>0.79).
Sixty-eight percent of the 147 item analyses (table avail-
able from senior author) and 71% of the 77 scale compar-
isons (Table 3) were significant at p < .01 (all these group
means are followed by either a symbol or letter). Adding
the requirement of a meaningful (medium or large) group
difference to scale classification (i.e., notations a and b)
limited results to 43% of item analyses (63 of 147) and
58% of scale analyses (45 of 77). These analyses can be
reviewed by row to identify the rated symptoms that most
contributed to diagnostic group separation, or by col-
umn to derive a description of rated diagnostic group
characteristics.
In the review of table rows, BPRS-C scales demonstrated
more robust differences among diagnosis-based groups
than did BPRS-C items. Except for the Thinking Distur-
bance scale, which only obtained a meaningful group
difference for patients with a psychotic diagnosis, the
remaining 10 scales obtained either four or five group-to-
remainder differences representing at least a medium effect.
The comparable performance of individual items ranged
from eight items that received only one or two differentiat-
ing values across the seven diagnostic groups to nine items
that received four or five significant group contrasts.
BPRS-C Performance of Diagnosis-Based Groups
A review of table columns demonstrates the descrip-
tive value of the BPRS-C. Children with combined-type
ADHD were more often characterized by hyperactivity,
but they were also defined by lower ratings on nine
BPRS-C items: Depressed Mood, Emotional With-
drawal, Uncooperativeness, Hostility, Suicidal Ideation,
Underproductive Speech, Blunted Affect, Anxiety, and
Sleeping Difficulties. The comparable column of Table
3 simplifies these results by demonstrating a larger scale
score for Psychomotor Excitation and lower scores for
Behavior Problems, Depression, Withdrawal-Retardation,
Anxiety, Internalization, Developmental Maladjust-
ment, and Total pathology. Children who received a pri-
mary diagnosis of conduct disorder or similar disruptive
behavior were more likely to be described by the three
items Uncooperativeness, Hostility, and Manipulative-
ness and less likely to be described by the six items
Depressed Mood, Suicidal Ideation, Underproductive
Speech, Emotional Withdrawal, Blunted Affect, and
Sleeping Difficulties. Comparable scale analyses identified
a higher value for Behavior Problems and lower values
for Depression, Withdrawal-Retardation, Anxiety, Organ-
icity, Internalization, Developmental Maladjustment, and
Total pathology. Patients who received any psychotic
diagnosis were more likely to be characterized by 10
items that were suggestive of a relatively severe disorder—
Depressed Mood, Suicidal Ideation, Peculiar Fantasies,
LACHAR ET AL.
338 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001
TABLE 3
Brief Psychiatric Rating Scale for Children (BPRS-C) Mean Scale Performance Across Seven Diagnosis-Defined Groups
Diagnostic Group
BPRS-C Scale (T) (1) (2) (3) (4) (5) (6) (7)
Behavior Problems 6.9 5.1a
9.5a
7.7 6.0 4.6a
5.9 9.8a
Depression 5.6 2.6a
3.7a
10.1b
1.4b
10.4b
6.9* 7.1*
Thinking Disturbance 1.0 0.4† 0.3† 5.8b
0.7 0.6 0.5 0.5
Psychomotor Excitation 4.5 6.3a
3.4† 4.1 8.5b
2.1a
2.1a
4.2
Withdrawal-Retardation 3.9 1.1a
1.7a
7.1b
8.5b
5.3* 2.9* 2.5†
Anxiety 4.6 2.7a
3.0a
7.1b
5.5 6.4a
4.7 4.1
Organicity 1.4 0.9 0.4a
0.8 6.6b
0.2a
0.1a
0.3a
Internalization 11.2 5.7b
7.0a
23.0b
7.6a
17.4b
12.1 11.7
Developmental Maladjustment 5.2 1.9a
2.1a
7.9a
15.1b
5.4 3.0† 2.8a
Externalization 11.4 11.4 13.0 11.8 14.6a
6.7b
8.0a
14.0a
Total 27.8 19.0a
22.1a
42.7b
37.3a
29.5 23.1† 28.5
Note: (T) = mean value for the total diagnosis-based sample (N = 476). Diagnostic group: (1) = attention-deficit/hyperactivity disorder,
combined type (n = 84); (2) = conduct disorder or other disruptive disorder (n = 82); (3) = any psychotic disorder (n = 46); (4) = pervasive devel-
opmental disorder or autism (n = 70); (5) = major depression (n = 64); (6) = other depressive disorder (n = 65); and (7) = any disruptive dis-
order with any anxiety/depression disorder (n = 65). See “Method” section for additional details. Post hoc scale analyses contrasted each of these
seven diagnostic groups with the sum of the remaining groups; * difference statistically significant at p < .01; † difference statistically significant
at p < .001, although absolute difference represents a small effect (<0.40); a
p < .001, associated with a medium absolute errect (0.40–0.79); and
b
p < .001 associated with a large absolute effect (>0.79).

7. Delusions, Hallucinations, Underproductive Speech,
Emotional Withdrawal, Blunted Affect, Anxiety, and
Sleeping Difficulties—but were not likely to be charac-
terized by the Hyperactivity item. Scale analysis revealed
that this diagnostic sample was described as more de-
pressed, with evidence of Thought Disturbance, Anxiety,
and Withdrawal-Retardation, as well as higher ratings
on Internalization, Developmental Maladjustment, and
Total pathology.
The BPRS-C descriptions of children with either
autism or pervasive developmental disorder included
higher mean severity ratings for 10 rating items (Hyper-
activity, Distractibility, Speech or Voice Pressure, Under-
productive Speech, Emotional Withdrawal, Blunted
Affect, Tension, Disorientation, Speech Deviance, and
Stereotypy) and lower mean intensity ratings for Manip-
ulativeness, Depressed Mood, Feelings of Inferiority, and
Suicidal Ideation. Comparable scale results demonstrated
that these patients obtained higher scores on Psycho-
motor Excitation, Withdrawal-Retardation, Organicity,
Developmental Maladjustment, Externalization, and
Total pathology, and lower scores on Depression and
Internalization.
Children with a primary diagnosis of major depres-
sion, in comparison with the remaining children, were
more often described by the six items Depressed Mood,
Feelings of Inferiority, Suicidal Ideation, Emotional
Withdrawal, Blunted Affect, and Sleeping Difficulties,
and less often described by the items Uncooperativeness,
Hostility, Hyperactivity, Distractibility, and Speech
Deviance. This group obtained higher scale values for
Depression, Anxiety, and Internalization and lower scale
values for Behavior Problems, Psychomotor Excitation,
Organicity, and Externalization.
The remaining two diagnosis-related groups obtained
less differentiation on BPRS-C items and scales. Patients
who received a primary diagnosis of depression, excluding
major depression or any psychotic state, were character-
ized by lower ratings on the items Hyperactivity, Distract-
ibility, Speech Deviance, and Stereotypy and by lower
mean scores on the Psychomotor Excitation, Organicity,
and Externalization scales. Group 7, consisting of children
who were assigned a comorbid pattern of a disruptive
(non-ADHD) diagnosis and either an anxiety diagnosis
or nonpsychotic depression diagnosis other than major
depression, obtained ratings that reported disruptive
behaviors but no indication of internalizing symptoms.
Items that were statistically descriptive of this group were
limited to Uncooperativeness, Hostility, and Manipula-
tiveness, whereas Stereotypy was less descriptive. This sev-
enth diagnostic group obtained higher mean scale scores
on Behavior Problems and Externalization and lower
mean scale scores on Organicity and Developmental
Maladjustment.
DISCUSSION
The anchored BPRS-C was easily integrated by teach-
ing faculty into academic clinical service units, and all
study cases were routinely rated during a 6-month period
by either a faculty or trainee clinician. The estimates of
interrater reliability (mainly from clinician pairs with vary-
ing levels of experience) and scale internal consistency
obtained were, without any specific rating scale training,
equal to or better than the values previously generated by
smaller samples (Gale et al., 1986; Mullins et al., 1986).
Limitations
The factor analysis of BPRS-C items demonstrated
both consistency with the seven established scales and evi-
dence of misplaced or ineffective items (such as Sleeping
Difficulties on the Anxiety scale, and Speech or Voice
Pressure on the Psychomotor Excitation scale). Additional
effort to improve the BPRS-C at the item level could
include the substitution of items (i.e., substitute “Fidget-
ing” for Speech or Voice Pressure) and the transfer of
items to the scale that demonstrated the highest item-to-
scale correlation (i.e., adding Sleeping Difficulties to the
Depression scale). On the other hand, the development of
composite indices, such as the three suggested by the cur-
rent factor analysis of the seven BPRS-C scales, in most
cases either resolves these probable errors or minimizes
their effect. Comparable study of these three broad-band
scales (Internalization, Externalization, Developmental
Maladjustment) demonstrated favorable estimates of
internal consistency, interrater reliability, and concurrent
validity as evidenced by the effective discrimination
among diagnosis-based samples.
Summary and Clinical Implications
Our current effort supports the reliability and validity
of the BPRS-C and suggests that continued study and
application of this 21-item anchored clinician rating is
warranted. BPRS-C items and scales demonstrated ade-
quate interrater reliability, the majority of scales obtained
substantial estimates of internal consistency, and item
BPRS-C VALIDITY AND RELIABILITY
J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001 339

8. and scale concurrent validity was established by perfor-
mance across seven diagnosis-based samples. These sam-
ples incorporated symptomatic heterogeneity because of
both diagnostic comorbidity and the use of clinical ver-
sus standardized assignment of diagnoses; thus, BPRS-C
performance represented evidence of substantial descrip-
tive power. In addition, a factor analysis of the seven
BPRS-C scales suggested a three-dimensional solution
that resolved issues of item misplacement.
Additional effort will be necessary to further establish
the BPRS-C’s sensitivity to symptomatic improvement
by incorporating rating scale administration both before
and after a period of programmatic or specific treatment.
Such efforts should incorporate item and scale analyses
that contrast both different diagnosis-based samples as
well as different treatments applied to similarly diagnosed
children. Concurrent rating of functional impairment or
symptom severity is encouraged in these efforts to facili-
tate examination of the relative contributions of global
ratings of adjustment and ratings of symptom dimen-
sions to this quantification of psychiatric treatment.
REFERENCES
Benfield CY, Palmer DJ, Pfefferbaum B, Stowe ML (1988), A comparison of
depressed and nondepressed disturbed children on measures of attribu-
tional style, hopelessness, life stress, and temperament. J Abnorm Child
Psychol 16:397–410
Bird HR (1999), The assessment of functional impairment. In: Diagnostic
Assessment in Child and Adolescent Psychopathology, Shaffer D, Lucas CP,
Richters JE, eds. New York: Guilford, pp 209–229
Brady EU, Kendall PC (1992), Comorbidity of anxiety and depression in
children and adolescents. Psychol 111:244–255
Casat CD, Pleasants DZ, Schroeder DH, Parler DW (1989), Bupropion in
children with attention deficit disorder. Psychopharmacol Bull 25:198–201
Cohen J (1988), Statistical Power Analysis for the Behavioral Sciences, 2nd ed.
Hillsdale, NJ: Erlbaum
Emslie GJ, Rush AJ, Weinberg WA et al. (1997), A double-blind, random-
ized, placebo-controlled trial of fluoxetine in children and adolescents
with depression. Arch Gen Psychiatry 54:1031–1037
Emslie GJ, Rush AJ, Weinberg WA, Rintelmann JW, Roffwarg HP (1990),
Children with major depression show reduced rapid eye movements. Arch
Gen Psychiatry 47:119–124
Emslie GJ, Rush AJ, Weinberg WA, Rintelmann JW, Roffwarg HP (1994),
Sleep EEG features of adolescents with major depression. Biol Psychiatry
36:573–581
Faustman WO, Overall JE (1999), Brief Psychiatric Rating Scale. In: The Use
of Psychological Testing for Treatment Planning and Outcome Assessment,
2nd ed, Maruish M, ed. Hillsdale, NJ: Erlbaum, pp 791–830
Field TM, Sandberg D, Goldstein S et al. (1987), Play interactions and inter-
views of depressed and conduct disorder children and their mothers.
Child Psychiatry Hum Dev 17:213–234
Gabbard GO, Coyne L, Kennedy LL et al. (1987), Interrater reliability in the
use of the Brief Psychiatric Rating Scale. Bull Menninger Clin 51:519–531
Gale J, Pfefferbaum B, Suhr MA, Overall JE (1986), The Brief Psychiatric
Scale for Children: a reliability study. J Clin Child Psychol 15:341–345
Graae F, Milner J, Rizzotto L, Klein RG (1994), Clonazepam in childhood
anxiety disorders. J Am Acad Child Adolesc Psychiatry 33:372–376
Jackson A (1985), Teaching pediatric psychopharmacology: an interdiscipli-
nary model. J Am Acad Child Psychiatry 24:103–108
Lachar D (1998), Observations of parents, teachers, and children: contribu-
tions to the objective multidimensional assessment of youth. In: Compre-
hensive Clinical Psychology, Bellack AS, Hersen M, eds. Vol 4: Assessment,
Reynolds CR, ed. New York: Pergamon, pp 371–401
Lyons JS, Howard KI, O’Mahoney MT, Lish JD (1997), The Measurement
and Management of Clinical Outcomes in Mental Health. New York: Wiley
McConville BJ, Steichen-Asch P, Harris R et al. (1990), Pediatric bone-marrow
transplants: psychological aspects. Can J Psychiatry 35:769–775
McReynolds P (1989), Diagnosis and clinical assessment: current status and
major issues. Annu Rev Psychol 40:83–108
Mullins D, Pfefferbaum B, Schultz H, Overall JE (1986), Brief Psychiatric
Rating Scale for Children: quantitative scoring of medical records. Psychi-
atry Res, 19:43–49
Nelson WM, Renzenbrink G, Kapp CJ (1995), Sensitivity of clinically hospi-
talized adolescents’ self-report measures to change over time. J Clin Psychol
51:753–760
Overall JE, Pfefferbaum B (1982), Brief reports and reviews: the Brief Psy-
chiatric Rating Scale for Children. Psychopharmacol Bull 18:10–16
Pfefferbaum B, Mullins D, Rhoades HM, McLaughlin EJ (1987), Rorschach
assessment of borderline children. J Psychol 121:219–228
Pfefferbaum B, Overall JE (1982), Diagnostic and Statistical Manual-III and
the Children’s Psychiatric Rating Scale. Psychopharmacol Bull 18:107–109
Pfefferbaum B, Overall JE (1983), Diagnostic structure of the Children’s
Psychiatric Rating Scale (CPRS). J Clin Child Psychol 12:167–173
Plante TG, Couchman CE, Diaz AR (1995), Measuring treatment outcome
and client satisfaction among children and families. J Ment Health Adm
22:261–269
Riddle MA (1989), Research on a children’s psychiatric inpatient service. J Am
Acad Child Adolesc Psychiatry 28:42–46
Roberts N, Vargo B, Ferguson HB (1989), Measurement of anxiety and depres-
sion in children and adolescents. Psychiatr Clin North Am 12:837–860
Seshadri S, Srinath S, Girinaji S (1989), Deliberate self-harm (attempted sui-
cide) in children. Indian J Psychol Med 1:13–15
Simeon JG, Ferguson HB (1987), Alprazolam effects in children with anxiety
disorders. Can J Psychiatry 32:570–574
Simeon JG, Ferguson HB, Knott V et al. (1992), Clinical, cognitive, and neu-
rophysiological effects of alprazolam in children and adolescents with
overanxious and avoidant disorders. J Am Acad Child Adolesc Psychiatry
31:29–33
Spencer EK, Campbell M (1994), Children with schizophrenia: diagnosis,
phenomenology, and pharmacotherapy. Schizophr Bull 20:713–725
Spencer EK, Kafantaris V, Padron-Gayol MV, Rosenberg CR, Campbell M
(1992), Haloperidol in schizophrenic children: early findings from a
study in progress. Psychopharmacol Bull 28:183–186
Stavrakaki C, Vargo B, Boodoosingh L, Roberts N (1987a), The relationship
between anxiety and depression in children: rating scales and clinical var-
iables. Can J Psychiatry 32:433–439
Stavrakaki C, Vargo B, Roberts N, Boodoosingh L (1987b), Concordance
among sources of information for ratings of anxiety and depression in
children. J Am Acad Child Adolesc Psychiatry 26:733–737
Stavrakaki C, Williams EC, Walker S, Roberts N, Kotsopoulos S (1991),
Pilot-study of anxiety and depression in prepubertal children. Can J
Psychiatry 36:332–338
Woerner MG, Mannuzza S, Kane JM (1988), Anchoring the BPRS: an aid to
improved reliability. Psychopharmacol Bull 24:112–117
Wrobel TA, Lachar D, Wrobel NH, Morgan ST, Gruber CP, Neher JA (1999),
Performance of the Personality Inventory for Youth validity scales. Assess
6:367–376
LACHAR ET AL.
340 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 40:3, MARCH 2001