IJMPR 14.

4 crc 11/21/05 3:21 PM Page 186

International Journal of Methods in Psychiatric Research

14(4): 186–201 (2005)
DOI: 10.1002/mpr.7

ADHD Rating Scale IV: psychometric

properties from a multinational study as a
clinician-administered instrument
S. ZHANG,1 D.E. FARIES,1 M. VOWLES,2 D. MICHELSON1,3

1 Lilly Research Laboratories, Indianapolis, USA

2 Lilly Research Centre, Surrey, UK
3 Indiana University School of Medicine, Indianapolis, USA

ABSTRACT
The development of rating scales for attention-deficit/hyperactivity disorder (ADHD) has traditionally focused on parent-
or teacher-rated scales. However, clinician-based instruments are valuable tools for assessing ADHD symptom severity.
The ADHD Rating Scale IV (ADHD RS), clinician administered and scored, has been validated as a useful instrument to
assess ADHD symptoms among American children and adolescents. In this study, we assessed the psychometric properties
of the scale in a recent clinical trial conducted mainly in Europe with over 600 children and adolescents diagnosed with
ADHD. The trial was conducted in 11 European countries plus Australia, Israel, and South Africa.
Results based on data in the study indicate that this version of the scale has acceptable psychometric properties including
inter-rater reliability, test-retest reliability, internal consistency, factor structure, convergent and divergent validity, discrim-
inant validity, and responsiveness. There were low-to-moderate ceiling and floor effects. The psychometric properties were
comparable with other validated scales for assessing ADHD symptom severity. These results were consistent across the 14
countries participating in this trial. Overall, the data from this study support the use of the ADHD RS as a clinician-rated
instrument for assessing the severity of ADHD symptoms in children and adolescents in Europe. Copyright © 2005 John
Wiley & Sons, Ltd.

Key words: Attention Deficit Hyperactivity Disorder, Rating Scale-IV-Parent, psychometric properties
Information about the scale rating procedure is given in the method section of this paper.

Introduction children (National Institute of Clinical Excellence,

Attention-deficit/hyperactivity disorder (ADHD) is a 2000). In the Netherlands, a prevalence of 2% to 4%
neurobiological disorder characterized by symptoms of for children between the ages of 5 and 14 years has
inattention and/or hyperactivity/impulsivity with an been reported (Health Council of the Netherlands,
onset during childhood (Pliszka et al., 1996; Faraone 2000), while Baumgaertel, Wolraich, and Dietrich
and Biederman, 1998). It is one of the most common (1995) found a prevalence rate of 17.8% for attention
psychiatric disorders of childhood and adolescence, and deficit disorders in German school-aged children.
occurs in 3% to 7% of school-aged children in the US These differences, however, appear to be primarily due
(American Psychiatric Association, 2000). Outside of to different diagnostic criteria and methodology rather
the US and Canada, the reported prevalence rates for than true differences in prevalence, as evidenced when
ADHD vary from country to country (Mueller et al., consistent criteria are used to assess the presence of the
1995; Reid et al., 1998; Livingston, 1999; Graetz at al., disorder (Prendergast et al., 1988).
2001). In the UK, the estimated prevalence rate of The standard instruments for assessing ADHD
ADHD has been reported to be 5% for school-aged symptom severity have traditionally been parent- or

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 187

ADHD Rating Scale IV 187

teacher-rated scales, which have been well validated Parent-, teacher-, and clinician-scored versions of
and in use for many years (Barkley, 1990; Conners, the ADHD Rating Scale-IV (ADHD RS, DuPaul et
1997; DuPaul et al., 1998). Cohen et al. (1990) con- al., 1998; Faries, 2001) have been validated in North
cluded that the low correlation between parent- and American populations. Magnusson et al. (1999) stud-
teacher-scored scales was not due to low reliability of ied the validity of an Icelandic version of ADHD RS
the scales but suggested the need for multiple methods rated by parents and teachers in over 400 Icelandic
for assessing symptom severity. Brown et al. (2001) schoolchildren. However, the validity of this instru-
found that the teacher rating scale had a lower effect ment has not been previously studied in majority
size when compared with the parent- and clinician- non-North American populations, especially as a clin-
rated scales. Recently, Beiderman et al. (2004) ician-rated instrument.
performed a literature search and summarized clinical The objective is to examine the psychometric
trials, which contained both parent- and teacher- properties of the ADHDRS-IV-Parent: Clinician
reported measures. Their results showed similar administrated and scored (ADHDRS-PI), scored by
sensitivity for parent and teacher ratings. trained clinicians based on parent interviews in chil-
Parent- and teacher-rated scales provide important dren with a diagnosis of ADHD who reside outside of
information for assessing ADHD symptom severity but North America. Specifically, the assessment includes
a clinician-rated assessment may be preferable as a pri- the inter-rater reliability, internal consistency, factor
mary outcome measure in a clinical trial designed to structure, test-retest reliability, convergent and diver-
assess treatment efficacy. First of all, clinicians are gent validity, discriminant validity, responsiveness,
trained to assess the impact of interventions already and ceiling and floor effects. These psychometric
implemented to cope with ongoing behaviour prob- properties are assessed relative to Clinical
lems (for example, special classroom methods), an Global Impressions-ADHD-Severity (CGI-ADHD-S),
important factor in achieving an accurate assessment Conners’ Parent Rating Scales (CPRS), and Conners’
of the severity of the underlying disorder. Secondly, use Teacher Rating Scales (CTRS).
of a clinician-rated scale avoids the problems of need-
ing to obtain rating for children from multiple teachers Materials and methods
and issues of school vacations. Thirdly, inclusion of
clinician assessments is consistent with the DSM-IV Study design and rating scale
criteria for the diagnosis of ADHD, which requires The study of the validity and reliability of the
that the ADHD symptoms are present to a degree of ADHDRS-PI was assessed as part of a large clinical trial
impairment in at least 2 settings, which cannot be sat- of atomoxetine. The study was conducted at 33 sites in
isfied by parent or teacher ratings alone. Furthermore, the UK, France, Spain, Italy, Belgium, the Netherlands,
in the clinical trial setting, trained clinician raters can Germany, Poland, Hungary, Sweden, Norway, Israel,
apply standardized, symptom-severity inclusion crite- South Africa, and Australia, and enrollment took place
ria, which lead to the selection of a more homogenous over approximately 11 months. Principle investigators
patient population and a reduction in variability that at each site were physicians with specialty training in
can be important in detecting true drug effects. psychiatry or paediatrics and psychologists with experi-
Additionally, application of standardized inclusion cri- ence diagnosing and treating children and adolescents
teria (DSM-IV) by the trained clinicians facilitates the with ADHD. After description of the procedures, pur-
incorporation of data collected from multiple sources pose of the study, and prior to the administration of any
and settings into a single score, thereby reducing the study procedure or dispensing of study medication, writ-
statistical multiplicity that occurs when data are col- ten informed consent was obtained from each patient’s
lected inconsistently using separate measures. And parent or guardian and written assent was obtained from
finally, in neuroscience clinical trials, using trained each patient. This study was reviewed by each site’s
clinician raters can reduce variability by having them institutional review board and was conducted in accor-
apply consistent judgements on severity ratings across dance with the ethical standards of the Declaration of
patients; therefore, clinician-rated instruments of ill- Helsinki 1975, as revised in 2000.
ness severity are typically required as the primary In this trial, 604 children and adolescents, aged 6
efficacy outcome measures by regulatory authorities. through 15 years, who met the Diagnostic and

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 188

188 Zhang et al.

Statistical Manual of Mental Disorders, Fourth Edition diagnosis of ADHD. Each item is scored on a 0 to 3
(DSM-IV) criteria for a diagnosis of ADHD, were scale: 0 = none (never or rarely); 1 = mild (some-
enrolled at 33 investigational sites in 14 countries. times); 2 = moderate (often); 3 = severe (very often).
The study design included a 1-week assessment and The total score is computed as the sum of the scores on
drug washout phase for those children and adolescents each of the 18 items. In addition to the total score, the
taking any medication excluded by the protocol, fol- scores from the inattention and hyperactivity/impul-
lowed by a 10-week, open-label, treatment period sivity subscales were computed. The inattention
during which atomoxetine was administrated twice a subscale score is the sum of the scores on the odd-num-
day (the subsequent extension phase of the study is bered items and the hyperactivity/impulsivity subscale
ongoing). Detailed information regarding the efficacy is the sum of the scores on the even-numbered items.
and safety of the use of atomoxetine has been previ- A clinician-rated symptom severity for each item was
ously published (Allen et al., 2001; Michelson et al., based on his or her interview with the child’s parent or
2001; Spencer et al., 2001; Michelson et al., in press). primary caretaker. The clinicians were trained at the
Only data from the acute phase are presented here. time of study start-up to rate scores based on the fre-
Patients were children and adolescents 6 through quency of the behaviour (across multiple settings) and
15 years of age at the time of study entry. For each the degree of impairment, and to use developmental
patient, the clinical diagnosis of ADHD was confirmed comparison (to consider the age appropriateness of
using the Kiddie Schedule of Affective Disorders and behaviour in rating each item).
Schizophrenia for School-Aged Children Present and The Clinical Global Impressions-ADHD-Severity
Lifetime Version (KSADS-PL) (Kaufman et al., 1996). (CGI-ADHD-S) (Guy, 1976) is a single-item clini-
Patients must have had an ADHDRS-PI score of at cian rating of the clinician’s assessment of the
least 1.5 standard deviations above the age and gender severity of the ADHD symptoms in relation to the
norm for their diagnostic subtype using published US clinician’s total experience with patients with
norms for the ADHDRS-PI. Generally, the same ADHD. Severity is rated on a 7-point scale: 1 =
investigator completed both KSADS-PL and the normal, not ill; 2 = minimally ill; 3 = mildly ill; 4 =
ADHD RS. Patients who were taking psychotropic moderately ill; 5 = markedly ill; 6 = severely ill; and 7
medication at study entry had a washout equal to a = very severely ill. It was completed at each visit.
minimum of five half-lives of the psychotropic medica- Other scales used in this study included the Conners’
tion prior to obtaining the baseline severity assessment Parent Rating Scale – Revised: Short Form (CPRS)
and starting treatment with atomoxetine. Patients and Conners’ Teacher Rating Scale-Revised: Short
were seen at approximately weekly or biweekly visits Form (CTRS) (Conners, 1997). Both the CPRS and
during the 10-week, open-label period, and no other CTRS have 4 subscales and were collected at Visit 1
psychotropic medications were allowed during the and endpoint: ADHD Index, Hyperactivity,
study. Cognitive, and Oppositional.
Parent-, teacher-, and clinician-scored instruments
were used in this study to assess the severity of ADHD. Reliability and validity methods
Since clinicians at each investigational site were
fluent in English, the English version of clinician-rated Inter-rater reliability
scales was used without translation into the native lan- Inter-rater reliability refers to a scale’s ability to
guages. The intention of this paper is simply validating achieve similar ratings by different raters assessing the
the use of the English clinician-rated ADHDRS-PI same patient. The Kappa statistic and average squared
scale in countries outside the US. The translation and deviation from the mode were used to assess inter-rater
validation of self and investigator forms is the subject reliability. Kappa statistic has a range from 0 to 1 and is
of future work. commonly used to assess inter-rater reliability when
The primary efficacy measure was the ADHDRS- observing categorical variables. The Kappa statistic is
PI. It was completed at each visit to assess ADHD the proportion of agreeing pairs out of all possible
symptom severity over the past week or past 2 weeks. pairs, adjusted for chance agreement (Fleiss, 1971).
The ADHDRS-PI is an 18-item scale with one item The average squared deviation was computed for each
for each of the 18 symptoms contained in the DSM-IV rater by averaging (over the 18 items) the squared

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 189

ADHD Rating Scale IV 189

difference between their ratings and the mode rating Test-retest reliability
for each item from the entire group (Channon and A good rating scale should be able to reproduce the
Butler, 1998); values from all raters were then aver- same score for the same individual at different times
aged. Unlike the Kappa statistic, the average squared while in the same disease condition. Test-retest relia-
deviation statistic more severely penalizes ratings that bility indicates such stability of a rating scale.
are more than 1 point from the mode. The intra-class correlation coefficient (ICC) is the
For this study, all clinical personnel who would be recommended measure to assess the test-retest reliabil-
using the ADHDRS-PI were trained and assessed in 2 ity (Deyo et al., 1991). The ICC is computed as the
rater-training sessions prior to the study. The rater- variability due to patients divided by the total variabil-
training session began with a presentation and ity (included variability due to patients, time, and
discussion of the rating scale. Raters then watched a other factors). Since Pearson’s correlation coefficient
videotaped interview between a clinician and a parent is the most commonly used measure for assessing the
of a child with ADHD, and independently completed strength of the relationship between the scores, both
the ADHDRS-PI after the video. The results were coefficients were computed to assess the correlation
then presented and points of agreement and disagree- between the ADHD-PI total scores at Visits 1 and 2.
ment were discussed to help gain consistency in future To show the variability of correlation, the confidence
scoring. The raters then watched a second videotaped intervals for ICC and Pearson’s correlation coefficients
interview and completed the ADHDRS-PI for the were computed based on Fisher’s Z transformation
second time. Data from 1 of the 2 rater-training ses- (Anderson, 1990). To further quantify shifts over time,
sions were collected and available for the analysis in t-test was used to test the null hypothesis that the
this manuscript. mean ADHD total scores obtained were similar at
Visits 1 and 2.
Factor structure In this study, ADHDRS-PI and CGI-ADHD-S were
Factor analysis is a powerful tool used to uncover the completed at both Visits 1 and 2, which were sched-
latent structure (dimensions) of a set of variables. It uled approximately 1 week apart. Since no study drug
can be used to validate a scale by demonstrating that was dispensed at these two visits, the changes in
its constituent items load on the same factor, and to ADHDRS-PI and CGI-ADHD-S scores from Visits 1
drop proposed scale items that cross-load on more than to 2 can be used to assess the stability of scores.
1 factor. Patients who were taking medication for ADHD other
In this paper, a principal-components factor analy- than the study drug at Visit 1 were excluded from the
sis with Varimax rotation (Reid, 1995) was performed analysis group due to the possible changes in ADHD
to examine the factor structure of ADHD RS. symptoms from Visits 1 to 2 after discontinuing the
drug. Test-retest reliability was not assessed for the
Internal consistency CPRS and CTRS as these scales were only adminis-
Cronbach’s alpha was used to evaluate internal consis- tered at Visit 1.
tency (Cronbach, 1951). It assesses the degree to In addition to the correlation analysis, a graphical
which each item of a rating scale measures the same approach, Bland and Altman Plot (Bland and Altman,
construct based upon all possible correlations between 1986), which plots the difference between the 2 mea-
2 sets of items within a rating scale. The range of the sures plot against the average score, was also used to
statistic is from 0 to 1. The accepted minimal standard assess the test-retest reliability.
to claim internal consistency is 0.65 (Nunnally, 1994). Starting from this section, the imputed scores for
The scores for all 18 items in ADHDRS-PI at Visit 1 total and subscales are used for all analyses. The
for all enrolled patients were used to compute imputed score for subscale was computed as follows: if
Cronbach’s alpha for ADHDRS-PI total score. Scores only 1 single item was missing in the subscale, the mean
from all odd- and even-numbered items were used to score for all other items in the subscale was imputed as
compute Cronbach’s alpha for ADHDRS-PI the score for the missing score. The total score was
inattention subscale and hyperactive/impulsive sub- computed as the sum of the imputed subscale scores. If
scale. Similar analysis was also done for ADHDRS-PI more than 1 item was missing in the subscale then the
at Visit 2. subscale score and total score would be missing.

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 190

190 Zhang et al.

Convergent and divergent validity impulsive, and combined (inattentive plus hyperactive
Convergent and divergent validity were utilized to /impulsive) was conducted. As noted previously,
establish the construct validity of the scale. ADHD subtype was assessed at Visit 1 using the
Convergent/divergent validity indicates a relationship KSADS-PL. Patients with an inattentive subtype did
between the scale under review and other scales not have sufficient hyperactive/impulsive symptoma-
thought to measure the same/different construct. In tology to meet the full combined ADHD subtype. This
this manuscript, the scale under review is ADHDRS- provides an opportunity to assess whether the hyperac-
PI; other validated scales for comparison were the tive/impulsive subscale of the ADHDRS-PI was able
CPRS (parent scored), CTRS (teacher scored), and to distinguish between the ADHD subtypes.
CGI-ADHD-S (investigator scored). All of these Therefore, we think the comparison between subtypes
scales were used to measure the severity of ADHD can be used to assess the discriminant validity of a sub-
symptoms. scale of the ADHDRS-PI.
To assess convergent validity, Pearson’s correlation Secondly, although patients without ADHD symp-
coefficients were computed between the following toms were not recruited, with active treatment,
measures (measures of the same construct): ADHDRS- severity levels of the symptoms decrease to about
PI Total with CGI-ADHD-S, CPRS ADHD Index and normal (as indicated by CGI-ADHD-S score = 1 or 2)
CTRS ADHD Index, ADHDRS-PI Inattention at endpoint for some patients. Thus, an analysis of
Subscale with CPRS Cognitive and CTRS Cognitive variance can be used to compare mean ADHDRS-PI
Subscale, and ADHDRS-PI hyperactive/impulsive total scores with CGI-ADHD-S scores. Comparison
subscale with CPRS hyperactive and CTRS hyperac- with the CGI-ADHD-S scores also provides additional
tive subscale. Correlations were computed for both information regarding the clinical significance of spe-
baseline and change-from-baseline-to-endpoint scores. cific ADHDRS-PI total scores.
To assess divergent validity, Pearson’s correlation Third, Pearson’s correlation coefficient between the
coefficients were computed between the following ADHDRS-PI and other measures that should not be
measures (measures of different construct): ADHDRS- logically related were computed. Several measurements
PI Inattention Subscale with CPRS hyperactive and included in the study were the Children’s Depression
CTRS hyperactive subscale, and ADHDRS-PI hyper- Inventory (CDI) and Children’s Depression Rating
active/impulsive subscale with CPRS cognitive and Scale-Revised (CDRS) to measure presence and sever-
CTRS cognitive subscale. Correlations were computed ity of depression, and the Multidimensional Anxiety
for baseline scores. Scale for Children (MASC) to assess anxiety. The cor-
To show the variability of correlation, the confi- relation of scores between ADHDRS-PI and these
dence intervals for Pearson’s correlation coefficients measurements should be low compared with that
were computed based on Fisher’s Z transformation between ADHDRS-PI and CPRS or CTRS.
(Anderson, 1990).
Responsiveness
Discriminant validity Responsiveness indicates the ability of a scale to detect
The discriminant validity of a scale measures the scales small but clinically significant changes in the patient’s
ability to distinguish between different groups of sub- symptom severity when a change has occurred. The
jects. An instrument for assessing ADHD symptom standardized response mean (SRM) is a commonly
severity with discriminant validity should distinguish used statistic to assess responsiveness (Stratford et al.,
between patients with and without a diagnosis of 1996). The SRM is defined as the mean change-from-
ADHD, or between patients with and without signifi- baseline score divided by the standard deviation of the
cant hyperactive symptoms. In this study, only patients changes scores. The SRM based on the ADHDRS-PI
with ADHD symptoms were recruited. Thus, instead was compared with the SRM from other validated
of a comparison between ADHD patients and a con- scales (CGI-ADHD-S, CPRS, and CTRS).
trol group, 3 alternative approaches were used to assess
the discriminant validity. Minimal clinically important differences
First, a comparison between patients with different Another important need is to determine the between-
ADHD subtypes, namely inattentive, hyperactive/ and within-treatment minimum clinically important

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 191

ADHD Rating Scale IV 191

differences (MCID) for an instrument. The MCID change scores for the instrument. They become critical
helps clinicians interpret the relevance of changes in when statistically significant differences needed to be
the instrument scores. The within-treatment MCID is justified as clinically relevant.
defined as the improvement in a score with treatment at
which a patient recognizes that she/he is improved. The Ceiling and floor effects
between-treatment MCID is the minimum difference Ceiling and floor effects exist if a substantial percent-
between 2 treatments that can be considered clinically age of the patient scores are at the ends of the scales –
relevant. One widely accepted way to determine the then the scale may not be able to accurately capture
MCIDs is to anchor the scale to a global rating scale change scores or differentiate among patients near the
such as CGI-Improvement (CGI-I). Unfortunately, ceiling or floor. A scale with floor effect lacks the abil-
CGI-I was not collected in this study; as an alternative, ity to detect minor disease symptoms, while a scale
CGI-ADHD-S was used. First, the LOCF change from with ceiling effect would be less sensitive to changes in
baseline to endpoint of CGI-ADHD-S was calculated the more serious symptoms (Stucki and Michel, 1995;
for each patient. If the change score is 0, then the Herrmann et al., 1997). Percentages of lowest and
patient was rated as having ‘no change’; if the change highest possible scores at baseline and endpoint were
score is 1 (the smallest change detectable by the CGI- calculated to assess ceiling and floor effects.
ADHD-S), then the patient was rated as ‘a little better’.
The mean change in the ADHDRS for those subjects Results
who rated as ‘a little better’ could be considered as the
within-treatment MCID. The difference in the mean Subjects
changes for subjects who rated as ‘a little better’ and Six-hundred-and-four patients enrolled in this study
who rated ‘no change’ could be considered as the (14 countries, 33 investigational sites). Table 1 sum-
between-treatment MCID. The between-treatment marizes the patient characteristics for this group as
MCID can be a sound choice for the treatment differ- well as the patient characteristics for a similar US-
ence in order to power the clinical studies. These two based study. The mean (SD) age for the group was
MCIDs provide guidance to researchers to interpret the 10.24 (2.25) years.

Table 1. Summary of baseline patient characteristics

Current multinational study (N=604) Previous US study (Faries, 2001) (N=228)

Variable n (%) n (%)

Gender
Male 541 (89.6) 211 (92.5)

Origin
Caucasian 583 (96.5) 175 (76.8)

ADHD subtype
Hyper/impulsive 30 (5.0) 3 (1.6)
Inattentive 124 (20.5) 52 (22.9)
Combined 450 (74.5) 172 (75.8)

Previous stimulant 341 (56.5) 125 (54.8)

Age (yrs.)
5–7 44 (7.3)
8–9 155 (25.7) 69 (30.3)
10–11 184 (30.5) 72 (31.6)
12–13 140 (23.2) 51 (22.4)
14–15 81 (13.4) 36 (15.8)

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 192

192 Zhang et al.

The diagnosis of ADHD and comorbid diagnoses The pattern of loadings also showed a clear hyperac-
were assessed by clinical interview and confirmed tivity-impulsivity factor and a clear inattention factor.
using the KSADS-PL semi-structured interview. In Odd-numbered items (reflecting inattention) loaded
addition to the diagnosis of ADHD, 45.5% of the on Factor 2 and even-numbered items (reflecting
patients also had a diagnosis of oppositional defiant hyperactivity) loaded on Factor 1.
disorder, 5.6% had conduct disorder, and 1.5% had
depression. Internal consistency
Cronbach’s alpha for the ADHDRS-PI total score was
Inter-rater reliability 0.795 based on Visit 1 data and 0.838 based on Visit 2
Rater training data from 41 raters were collected for data from the 604 enrolled patients (for the inatten-
analysis prior to the start of the trial. The Kappa statis- tion subscale, 0.724 for Visit 1 and 0.770 for Visit 2; for
tics for the first and second tapes were 0.58 and 0.63, the hyperactivity-impulsivity subscale, 0.825 for Visit
respectively. The average squared deviations from the 1 and 0.848 for Visit 2). The item-to-total correlations
mode for the first and second tapes for the rating scale range from 0.25 to 0.51 at Visit 1, and from 0.29 to
were 0.38 and 0.30, respectively. Agreement was simi- 0.57 at Visit 2.
lar for inattention and hyperactive/impulsive items.
Approximately 80% of the raters independently gave a Test-retest reliability
total score in the range of 33 through 39. Test-retest assessments included the 565 patients who
were not taking stimulant medication for ADHD at
Factor structure Visit 1 and who had an efficacy measurement
The first three eigenvalues in the solution were 5.99 (ADHDRS-PI) at both Visits 1 and 2. Table 3 reports
(55%), 2.80 (26%), and 0.80 (7%). Kaiser’s criterion the ICC, the Pearson’s correlation coefficients, the
of an eigenvalue greater than 1 indicated that two fac- mean changes and their associated confidence inter-
tors could be extracted for ADHDRS-PI. Table 2 vals for ADHDRS-PI total score, ADHDRS-PI
shows the structure matrix for the two-factor solution. inattentive subscale score, ADHDRS-PI hyperactive/

Table 2. Factor pattern matrix of a Varimax rotation for the ADHD RS

Item # Item short description Factor 1 Factor 2 Communality

Hyperactivity
10 On the go 0.71 0.02 0.50
06 Runs about 0.68 0.09 0.48
16 Difficult waiting turn 0.63 0.09 0.41
18 Interrupts 0.62 0.08 0.39
08 Difficult playing 0.61 0.16 0.39
14 Blurts out answers 0.54 0.13 0.31
04 Leaves seat 0.53 0.06 0.29
12 Talks excessively 0.51 0.04 0.26
02 Fidgets 0.46 0.12 0.22

Inattentive
07 No follow-through 0.04 0.60 0.36
17 Forgetful 0.06 0.57 0.33
09 Difficult organizing 0.00 0.56 0.31
11 Avoids tasks 0.04 0.54 0.29
01 Close attention 0.05 0.52 0.27
15 Easily distracted 0.13 0.48 0.25
03 Sustaining attention 0.14 0.44 0.22
13 Loses things 0.10 0.42 0.19
05 Does not listen 0.27 0.35 0.19

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 193

ADHD Rating Scale IV 193

Table 3. Test-retest reliability of ADHDRS-PI and CGI-ADHD-S

Variable N ICC Pearson’s correlation Mean difference

(LCI, UCI)* (LCI, UCI)** (LCI, UCI)***

ADHDRS-PI Total score 565 .84 (.82, .87) .85 (.82, .87) .09 (–.25, .44)
ADHDRS-PI Hyper/imp Subscale 565 .89 (.87, .90) .89 (.87, .90) .19 (–.03, .40)
ADHDRS-PI Inattentive Subscale 565 .77 (.74, .80) .78 (.74, .81) –.09 (–.31, .12)
CGI-ADHD-S score 566 .85 (.83, .87) .86 (.83, .88) .02 (–.01, .06)

ICC: Intra-class coefficient from ANOVA model with term of PATIENT and VISIT; mean difference: mean of change
score from Visits 1 to 2.
LCI – lower bound of 95% confidence interval.
UCI – upper bound of 95% confidence interval.
* 95% confidence interval of ICC based on Fisher's Z transformation.
** 95% confidence interval of Pearson’s correlation between Visit 1 and Visit 2 scores. Based on Fisher’s Z transformation.
*** 95% confidence interval of mean difference based on normal distribution.

impulsive subscale score, and CGI-ADHD-S score. ADHDRS-PI total scores at those 2 visits. The mean
The correlation coefficients for the ADHDRS-PI scale difference was –0.14, and the limits of agreement were
and the CGI-ADHD-S scale were both high (range –8.9 to 8.6.
from 0.78 to 0.89), and the mean differences in scores
from Visits 1 to 2 were both very low (range from Convergent and divergent validity
–0.09 to 0.19). Pearson’s correlation coefficients between ADHDRS-
Figure 1 plots the difference of ADHD RS total PI scores and scores from other scales thought to
score between Visits 1 and 2 against the average of measure the same construct for both baseline and

Figure 1. ADHDRS-PI: difference vs. average of total scores measured at Visits 1 and 2 with 95% limits of agreement.

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 194

194 Zhang et al.

change-from-baseline-to-endpoint measurements are Table 5. As expected, the ADHDRS-PI inattentive

presented in Table 4. Correlations between the subscale had a much lower correlation with other
ADHDRS-PI scores and other parent (CPRS) and hyperactivity subscales than with assessments of other
clinician (CGI-ADHD-S) measures of ADHD symp- cognitive subscales. The corresponding trend was
tom severity for both baseline and change score were observed for the ADHDRS-PI hyperactive/impulsive
moderate to high. Correlations with teacher-rated subscale.
measures were low to moderate. All correlations were
statistically different from 0. Discriminant validity
Pearson’s correlation coefficients between The ADHDRS-PI total score and subscale scores for
ADHDRS-PI scores and scores from other scales patients with each ADHD subtype were summarized at
thought to measure a different construct at baseline, as baseline and reported in Figure 2. A statistically signif-
well as the correlations between scales thought to icant difference (P < 0.001) was noted between
measure the same set of symptoms, are presented in inattentive subtype patients and hyperactive/impul-

Table 4. Convergent validity of ADHDRS-PI based on score at baseline and change score from baseline to endpoint

Variable 1 Variable 2 Baseline Change

N Correlationa (LCIb, UCIc) N Correlationa (LCIb, UCIc)

ADHDRS-PI Total CGI-ADHD-S 604 0.56 (0.50, 0.61) 603 0.77 (0.74, 0.80)
CPRS ADHD Index 602 0.49 (0.42, 0.54) 566 0.71 (0.67, 0.75)
CTRS ADHD Index 535 0.21 (0.13, 0.29) 457 0.18 (0.09, 0.27)
ADHDRS-PI CPRS Cognitive 602 0.53 (0.47, 0.58) 566 0.67 (0.63, 0.72)
Inattentive CTRS Cognitive 529 0.15 (0.07, 0.24) 452 0.14 (0.05, 0.23)
ADHDRS-PI CPRS Hyperactive 603 0.73 (0.69, 0.76) 572 0.72 (0.68, 0.76)
Hyper/imp CTRS Hyperactive 536 0.36 (0.28, 0.43) 461 0.21 (0.12, 0.29)

a
Correlation is assessed by Pearson’s correlation coefficient between variable 1 and variable 2.
b
LCI – lower bound of 95% confidence interval based on Fisher's Z transformation.
c
UCI – upper bound of 95% confidence interval based on Fisher's Z transformation.

Table 5. Correlations between the ADHDRS-PI subscales and corresponding CPRS and CTRS subscales at
baseline

Variable 1 Variable 2

ADHDRS-PI CPRS CTRS

Cognitive Hyperactive Cognitive Hyp/impulsive

subscale subscale subscale subscale
correlationa correlationa correlationa correlationa
(LCIb, UCIc) (LCIb, UCIc) (LCIb, UCIc) (LCIb, UCIc)
N = 602 N = 603 N = 529 N=536

Inattentive subscale 0.53 (0.47, 0.58) 0.15 (0.07, 0.23) 0.15 (0.07, 0.24) 0.02 (–0.07, 0.10)
Hyper/imp subscale 0.06 (–0.02, 0.14) 0.73 (0.69, 0.76) –0.09 (–0.17, –0.01) 0.36 (0.28, 0.43)
a
Correlation is assessed by Pearson’s correlation coefficient between variable 1 and variable 2.
b
LCI – lower bound of 95% confidence interval based on Fisher’s Z transformation.
c
UCI – upper bound of 95% confidence interval based on Fisher’s Z transformation.

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 195

ADHD Rating Scale IV 195

Figure 2. ADHDRS-PI baseline total and subscale scores by subtype.

sive subtype patients on both subscales of the –0.05 (P = 0.183) for CDRS, 0.016 (P = 0.709) for
ADHDRS-PI. In addition, statistically significant dif- CDI, and 0.002 (P = 0.956) for MASC. Thus, correla-
ferences were noted between the inattentive-subtype tions between the ADHDRS-PI total score and
patients and combined-subtype patients on the measurements of comorbid disease severity scores were
ADHDRS-PI hyperactive/impulsive subscale, and very low and were not statistically significant.
between the hyperactive/impulsive-subtype and
combined-subtype patients on the ADHDRS-PI Responsiveness
inattention subscale. For patients who received at least one dose of atomoxe-
Figure 3 summarizes the mean ADHDRS-PI total tine, the mean baseline, mean change from baseline to
scores by CGI-ADHD-S score at endpoint. Analyses of endpoint, standard deviation in change scores, and the
variance (ANOVA) on endpoint data indicate statisti- standardized response mean (SRM) for ADHDRS-PI,
cally significant differences in mean ADHDRS-PI total CGI-ADHD-S, CPRS, and CTRS are presented in
scores between each CGI-ADHD-S level (results not Table 6. The SRM produced by the ADHDRS-PI
shown). In the total sample of patients, CGI-ADHD-S scores was similar or numerically higher than the SRM
scores of ‘minimally ill’, ‘mildly ill’, ‘moderately ill’, and using other parent and clinician measures of ADHD.
‘markedly ill’ corresponded to mean ADHDRS-PI total All scales demonstrated a statistically significant
t-score 52.3, 60.0, 68.6, and 78.1, respectively. In a change from baseline. SRM for ADHDRS-PI was also
population sample, a t-score of 50 represents the mean calculated for each ADHD subtype, and is included in
raw score for a child of a given age and gender, and Table 6. The scores were consistent across 3 subtypes,
each change of 10 points in t-score corresponding to 1 with the hyperactive/impulsive subtype having a
SD from the mean for each patient. In this study, nor- slightly lower score, which might be due, in part, to the
mative data based on a sample of 2000 US children much smaller sample size. In addition, patients were
have been used to compute t-scores (transformations of divided into 3 groups according to their baseline
raw scores based on normative data). ADHD RS total score, and the SRM for ADHDRS-PI
The Pearson correlation coefficients for total scores was computed for each subgroup. The results are as fol-
at baseline between ADHDRS-PI total score were lows: for Group 1 (ADHD RS total score (≤36), SRM =

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 196

196 Zhang et al.

Figure 3. Mean ADHDRS-PI total scores by CGI-ADHD-severity score.

Table 6. Responsiveness of ADHDRS-PI based on change score from baseline to endpoint in the atomoxetine group

Baseline Change

Population Variable N Mean SD Mean SD P Value* SRM**

All patients ADHDRS-PI total 603 41.32 7.85 –23.37 12.68 <.001 1.84
Mixed subtype ADHDRS-PI total 449 43.52 6.80 –24.57 12.90 <.001 1.90
Inatt. subtype ADHDRS-PI total 124 34.89 7.33 –20.54 11.44 <.001 1.80
Hyper. subtype ADHDRS-PI total 30 34.90 6.40 –17.13 10.64 <.001 1.61
All patients CGI-ADHD-S 603 5.21 0.79 –2.65 1.46 <.001 1.82
All patients CPRS ADHD index 572 28.37 5.70 –12.83 8.91 <.001 1.44
All patients CTRS ADHD index 458 24.04 8.02 –5.98 7.70 <.001 0.78

*Within-treatment group P values are from Wilcoxon’s Signed Rank Test.

**SRM (Standardized Response Mean) is computed as the mean change from baseline to endpoint divided by the stan-
dard deviation of change from baseline to endpoint scores.

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 197

ADHD Rating Scale IV 197

1.67; for Group 2 (ADHD RS total score of 37 to 44), decrease from baseline) for Group 3. The between-
SRM = 1.96; and for Group 3 (ADHD RS total score treatment MCID for the ADHD RS total score is 7.7
(≤45), SRM = 2.21. It seems that SRM is affected by for Group 1, 7.6 for Group 2, and 5.2 for Group 3.
baseline severity; the higher the baseline ADHD RS
total score, the greater the change from baseline to Ceiling and floor effects
endpoint (P < 0.001). The frequency of the highest and lowest possible
scores is reported at both baseline and endpoint (Table
Minimal clinically important differences 7). The percentage of patients achieving the highest
The within-treatment MCID for ADHD RS total and lowest possible scores was very small at both base-
score is 10.2 (or a 27% decrease from baseline). The line and endpoint (less than 7%).
between-treatment MCID is 6.6 (or a 19% decrease
from baseline). If grouping the patients into approxi- Validation of ADHDRS-PI by country
mately three equal groups based on their baseline This was a multicountry study so the psychometric
ADHD RS total score (Group 1: (≤36, Group 2: 37–44, properties of ADHDRS-PI were also investigated for
Group 3: (≤45), then the within-treatment MCID for individual countries. Table 8 summarizes the key results.
the ADHD RS total score is 9.7 (or a 35% decrease Across all 14 countries, the range of Cronbach’s
from baseline) for Group 1, 9.6 (or a 24% decrease alpha for the ADHDRS-PI total score at Visit 1 was
from baseline) for Group 2, and 11.4 (or a 23% 0.718 to 0.844. Intra-class correlation coefficients

Table 7. Frequency of best and worst possible scores for ADHDRS-PI at baseline and endpoint

Baseline Endpoint

Frequency Best score Worst score Best score Worst score

n (%) n (%) n (%) n (%)

ADHDRS-PI total 0 (0.0) 12 (1.86) 5 (0.78) 0 (0.0)

ADHDRS-PI inattentive 0 (0.0) 44 (6.82) 16 (2.48) 1 (0.16)
ADHDRS-PI hyper/imp 1 (0.16) 42 (6.51) 43 (6.67) 1 (0.16)

Table 8. Psychometric properties of ADHDRS-PI by country

Internal Consistency Test-retest reliability Responsiveness

Country N (Cronbach’s Alpha) (ICC) (SRM)

Australia 38 0.844 0.817 1.53

Belgium 36 0.815 0.890 1.31
Germany 26 0.718 0.899 1.73
Spain 36 0.765 0.865 2.42
France 35 0.736 0.760 1.82
UK 29 0.788 0.733 2.92
Hungary 77 0.741 0.768 2.68
Israel 34 0.815 0.871 1.15
Italy 31 0.718 0.804 2.37
Holland 31 0.798 0.731 1.33
Norway 24 0.731 0.946 1.74
Poland 62 0.739 0.868 2.03
Sweden 37 0.756 0.940 3.15
South Africa 68 0.815 0.800 2.02

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 198

198 Zhang et al.

between ADHDRS-PI total score for Visits 1 and 2 either too short or the items included in the scale have
ranged from 0.731 to 0.946, respectively. The SRM very little in common. Conversely, a very high value
produced for the ADHDRS-PI total scores had a mini- suggests some redundancy in the scale. Therefore, the
mum of 1.15 (Israel) and a maximum of 3.15 internal consistency of the ADHDRS-PI scale
(Sweden). Positive correlations were found between (Cronbach’s alpha = 0.795 at Visit 1 and 0.838 at Visit
ADHDRS-PI and other ADHD scales (CGI-ADHD-S 2) is satisfied and is not sufficiently high to suggest
and CPRS) for all 14 countries with the exception of redundancy in items high enough to suggest accept-
Norway, where the correlation between ADHDRS-PI able internal consistency. The modest item-to-total
total score and CPRS ADHD Index subscale score was correlation also indicated that there were no items
–0.08 at baseline. The number of patients with a within the scale that showed either a non-acceptable
hyperactive/impulsive subtype was too small to help correlation, or high colinearity. Note that Cronbach’s
assess discriminant validity within each country, a sta- alpha is a little higher at Visit 2 compared with Visit 1,
tistically significant difference (P <0.05) was noted which is consistent with the previous finding that the
between inattentive-subtype patients and combined Cronbach’s alpha value increases once raters have
subtype patients on the ADHDRS-PI hyperactive/ more experience with the scale. Also noted that this
impulsive subscale. Correlations with the ADHDRS- Cronbach’s alpha is slightly lower than the target
PI inattentive subscale were much lower with CPRS (0.90) suggested for use of the scale for individual
Hyperactivity subscales than with assessments of rather than group comparisons (Perrin et al., 1997).
CPRS cognitive subscales, and vice versa for the In this study, ICCs for ADHDRS-PI total and sub-
ADHDRS-PI hyperactive/impulsive subscale (results scale scores range from 0.773 to 0.887, respectively,
not shown here). (see Table 2). Landis and Koch (1977) suggested that
ICCs above 0.60 indicate satisfactory test-retest relia-
Discussion bility and ICCs greater than 0.80 are excellent. The
In this study, the psychometric properties of the score of 0.773 to 0.887 indicated excellent test-retest
ADHD RS-PI, clinician administered and scored, were reliability of ADHDRS-PI. Bland and Altman plot
assessed in a group of over 600 patients from Europe also showed a good agreement between scores
(about 500 patients), Australia, Israel, and South obtained at two different time points, which suggested
Africa (106 patients) with a diagnosis of ADHD. there is no potential systematic difference.
Generally, a Kappa statistic greater than 0.6 sug- In general, correlations between the ADHDRS-PI
gested good agreement, and a larger value indicated and other measures of the same set of ADHD symp-
greater strength of agreement (Landis and Koch, toms were high except for correlations with CTRS,
1977). There are no published guidelines that define especially for the ADHDRS-PI Inattentive subscale
an acceptable level of average squared deviation to and CTRS cognitive subscale (see Table 3). At the
assess inter-rater reliability. However, average squared same time, correlations between scales thought to
deviation from a similar training tape for a US trial measure different symptom groups were low. These
using ADHDRS-PI was 0.28 (Faries, 2001). These results suggest adequate content validity for the scale.
results suggested that the inter-rater reliability of The low correlation between ADHDRS-PI and CTRS
ADHDRS-PI (Kappa = 0.63, average squared devia- is consistent with other researches comparing parent
tion = 0.30) is satisfactory for a multicentre clinical and teacher scales (Faries, 2001). This may be due to
trial. multiple factors, including the fact that the scales are
The results of exploratory factor analysis of the scale not completed at the same time (due to teacher sched-
indicate that a two-factor solution would best repre- ules), there is limited contact or time to develop
sent the structure of this scale, which is also consistent relationships between teacher and child, and the
with the DSM-IV two-dimensional diagnostic criteria. CTRS cognitive subscale assesses a slightly broader set
To assess internal consistency, the literature suggests of symptoms (academic performance) than just the
that 0.65 to 0.70 is an acceptable minimal standard ADHD symptom list. Previous studies also suggest that
(Nunnally, 1994; Perrin et al., 1997). The higher there is a low correlation between teachers’ ratings and
Cronbach’s alpha, the greater the internal consistency. ratings made by trained classroom observers using the
A very low value indicates that the rating scale is CTRS (Conger et al., 1983; Kazdin et al., 1983). This

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 199

ADHD Rating Scale IV 199

may be in part due to the observation that teachers ADHDRS-PI can be successfully used to identify
tend to put excessive emphasis on children’s academic patients with ADHD symptoms, and to detect the
excellence (Lessing et al., 1974). change of symptom severity over time under treatment
There were statistically significant differences (P < in both the US and outside the US.
0.001) among different ADHD-subtype patients (see The ADHDRS-PI demonstrated acceptable inter-
Figure 1) indicating that the ADHDRS-PI discrimi- nal consistency across all 14 countries (minimum
nates between patients whose diagnosis suggested the Cronbach’s alpha was 0.718, see Table 8). ADHDRS-
presence of clinically significant hyperactive/impulsive PI was also found to have acceptable levels of
or inattentive symptoms and those whose diagnosis did test-retest reliability for all 14 countries as ICCs
not. The statistically significant differences in ranged from 0.731 to 0.946. Note correlations of at
ADHDRS-PI mean total scores between each CGI- least 0.60 are considered satisfactory while those
Severity score suggest that ADHDRS-PI had the greater than 0.80 are considered excellent (Landis and
ability to discriminate between groups of patients at Koch, 1977). As with findings from the overall popu-
different severity levels (see Figure 2). These two lation, acceptable levels of convergent/divergent
results indicate the satisfactory discriminant validity of validity were found between ADHDRS-PI and CPRS
the ADHDRS-PI scale. Patients with ADHD showed for all 14 countries. Acceptable discriminant validity
large improvement in ADHDRS-PI total score over was observed within each country. The SRM produced
time (SRM ranged from 1.61 to 1.90 for patients with for the ADHDRS-PI total scores by countries had the
different ADHD subtype). There are no published minimum as 1.15, which indicated acceptable respon-
guidelines that define an acceptable level of SRM. siveness of ADHDRS-PI for all countries. Consistent
However, this result is comparable with CGI-ADHD- results of the psychometric properties across patient
S (SRM = 1.82), which is a well-accepted clinician groups from multiple countries in this study indicate
rating scale. This is also consistent with the result this rating scale is reasonable for multinational prac-
(SRM = 1.21) from a US trial with a similar design tice.
using the ADHDRS-PI (Faries, 2001). These results There are several limitations to this study. It has
indicated acceptable responsiveness of the ADHDRS- been noted that culture may play a role in ADHD
PI scale. Note that the SRM score in this study is assessment. Though Magnusson et al. (1999) showed
relatively high. We think the reason is that most that the factor structures of this rating scale were
patients entered the study with severe ADHD symp- highly similar across cultures and the norm scores were
toms (mean ADHD RS total t-score 3 SD above similar to those found in American studies when rated
norms), and thus had bigger opportunity to change. by parents, that study used an Icelandic version of the
The higher SRM for patients with a higher baseline scale and only applied it to Icelandic schoolchildren.
score supports our assumption. In this study, the English version of the rating scale
For ADHDRS-PI total score, the percentages of was utilized in countries where English was not the
patients given the best and worst possible scores were native language and the US norms were applied to
very small (see Table 7), thus suggesting that ceiling populations outside the US. Thus, the ability of the
and floor effects are not an issue for this scale in this clinicians to translate to non-English-speaking
patient population. patients was an additional source of variability and
The patient characteristics and baseline ADHD results may not extend to other or future translations.
severity scores in this study were very similar to those Nonetheless, Buitelaar et al. (2004) reported that
enrolled in a similar US study used to assess the valid- while there were several differences between the inter-
ity of ADHDRS-PI (Faries, 2001). Table 1 summarizes national and North American study populations, the
the patient baseline characteristics for both studies. two groups were very similar in most respects.
Moreover, the psychometric properties of the scale for We also realized that because of the small sample
both patient populations were similar. This similarity size in each country (604 participants spread over 14
allows us to address the interpretation of data from US countries), it is difficult to interpret the validity of this
studies relative to other countries. The consistency of rating scale for each country.
the results of this study with the results of the atomox- In this study, children without ADHD symptoms
etine study conducted in the US suggests that were not recruited for ethical reasons. Therefore, a

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 200

200 Zhang et al.

comparison between patients with ADHD and con- Anderson TW. An Introduction to Multivariate Statistical
trols was not possible. Analysis. 2 edn. Beijing People’s Republic of.China:
Although discriminant validity could be assessed by John Wiley & Sons, 1990, pp. 120–5.
Barkley RA. Attention Deficit Hyperactivity Disorder: A
comparing patients with a diagnosis of ADHD inat-
Handbook for Diagnosis and Treatment. New York:
tentive subtype and a diagnosis of ADHD combined Guilford, 1990.
subtype (defined elsewhere in text), the inference was Baumgaertel A, Wolraich ML, Dietrich M. Comparison of
somewhat limited. Additionally, there was neither a diagnostic criteria for attention deficit disorders in a
placebo-control arm nor an active comparator German elementary school sample. J Am Acad Child
(methylphenidate) in this analysis. Therefore, we Adolesc Psychiatry 1995; 34: 629–38.
could not assess responsiveness of the scale using effect Biederman J, Faraone SV, Monuteaux MC, Grossbard JR.
How informative are parent reports of attention-
sizes based on treatment differences from placebo, nor
deficit/hyperactivity disorder symptoms for assessing
compare the standardized response mean from atom- outcome in clinical trials of long-acting treatments? A
oxetine with another proven efficacious compound. pooled analysis of parents’ and teachers’ reports.
Another limitation is that more than 50% of Pediatrics 2004; 113: 1667–71.
patients had previous drug therapy for ADHD, and we Bland JM, Altman DG. Statistical methods for assessing
speculate that the parents of those patients with previ- agreement between two methods of clinical measure-
ous medication might have significantly more ment. Lancet 1986; I: 307–10.
knowledge about ADHD than parents of children Brown, RT, Freeman WS, Perrin JM, Stein MT, Amler
RW, Feldman HM, Pierce K, Buitelaar JK, Danckaerts
without previous medication. Clinician-scored
M, Gillberg C, Zuddas A, Becker K, Bouvard M, Fagan
ADHDRS-PI is based on interviews with parents, so J, Gadoros J, Harpin V, Hazell P, Johnsson M, Lerman-
there is potential bias toward either higher or lower Sagie T, Soutullo C, Wolanczyk T, Zeiner P, Fouche DS,
scores for patients with previous experience compared Krikke-Workel J, Zhang S, Michelson. A prospective,
with patients who were treatment naïve. This aspect is multicenter, open-label assessment of atomoxetine in
not covered in this paper and might be worth further non-North American children and adolescents with
research. ADHD. Eur Child Adolesc Psychiatry 2004; 13(4):
249–57.
In conclusion, our results support the validity and
Campbell DT, Fiske DW. Convergent and discriminant
reliability of the ADHD RS as a clinician-adminis- validation by the multitrait-multimethod matrix.
tered and clinician-scored tool for assessing the Psychological Bulletin. 1959; 56: 81–105.
severity of ADHD symptoms in children and adoles- Channon E, Butler A. Comparing investigators’ use of
cents under a research setting in Europe. The ADHD rating scales such as PANSS in muti-investigator
RS as a clinician-rated tool could be used for screening studies of schizophrenia. Poster presentation at the
patients for baseline symptom severity to determine Eleventh European College of
Neuropsychopharmacology (ECNP) Congress, 1998.
whether they meet a threshold condition for study par-
Cohen M, Becker MG, Campbell R. Relationships among
ticipation and for following the course of symptom four methods of assessment of children with attention
change over a clinical trial. deficit-hyperactivity disorder. J School Psychol 1990;
28: 189–202.
Acknowledgement Conger AJ, Conger JC, Wallander J, Wark D, Dygdon J. A
This research was funded by Eli Lilly and Company. generalizability study of the Conners’ Teacher Rating
Scale-Revised. Educ Psychol Meas 1983; 43: 1019–31.
References Conners CK. Conners’ Rating Scales: Revised Technical
Allen AJ, Spencer TJ, Heiligenstein JH, Faries DE, Manual. North Towanda (New York): Multi-Health
Kelsey DK, Laws HF, Wernicke J, Kendrick KL, Systems, 1997.
Michelson D. Safety and efficacy of atomoxetine Cronbach LF. Coefficient alpha and the internal structure
for ADHD in two double-blind, placebo-controlled of tests. Psychometrika 1951; 16: 297–334.
trials. Biological Psychiatry, Society of Biological Deyo RA, Diehr P, Patrick KL. Reproducibility and
Psychiatry 56th Annual Convention and Scientific responsiveness of health status measures: Statistics and
Program, 3–5 May 2001; New Orleans LA. Dallas, TX: stratifies for evaluation. Control Clin Trials 1991; 12:
Elsevier. 142S–158S.
American Psychiatric Association. Diagnostic and DuPaul GJ, Power TJ, Anastopoulos AD, Reid R. ADHD
Statistical Manual of Mental Disorders: DSM-IV-TR. Rating Scale IV: checklists, norms, and clinical inter-
Washington DC: American Psychiatric Press, 2000. pretation. New York: Guilford, 1998.

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)
IJMPR 14.4 crc 11/21/05 3:21 PM Page 201

ADHD Rating Scale IV 201

Faraone SV, Biederman J. Neurobiology of attention- Deficit/Hyperactivity disorder treated with atomoxe-
deficit hyperactivity disorder. Biol Psychiatry 1998; 10: tine: a randomized, double-blind, placebo-controlled
951–8. study. J Am Acad Child Adolesc Psychiatry 2004;
Faries DE, Yalcin I, Harder D, Heiligenstein JH. Validation 43(7): 896–904.
of the ADHD Rating Scale as a clinician administered Mueller CW, Mann EM, Thanapum S, Humris E, Ikeda Y,
and scored instrument. J Atten Disord 2001; 5(2): Takahashi A, Tao KT, Li BL. Teachers’ ratings of disrup-
107–15. tive behaviour in five countries. J Clin Child Psychol
Fleiss JL. Measuring nominal scale agreement among many 1995; 24(4): 434–42.
raters. Psychol Bull 1971; 76(5): 378–82. National Institute of Clinical Excellence. Guidance on the
Graetz BW, Sawyer MG, Hazell PL, Arney F, Baghurst P. Use of Methylphenidate (Ritalin, Equasym) for
Validity of DSM-IV ADHD subtypes in a nationally Attention Deficit/Hyperactivity Disorder (ADHD) in
representative sample of Australian children and Childhood. Technology Appraisal Guidance no. 13.
adolescents. J Am Acad Child Adolesc Psychiatry London: NICE, 2000.
2001; 40: 1410–17. Nunnally, JC. Psychometric Theory. 3 edn. New York:
Guy W. ECDEU assessment manual for psychopharma- McGraw-Hill, 1994.
cology, revised. Bethesda (MD): US Department of Perrin ES, Aaronson NK, Alonso J, Burnam A, Lohr K,
Health, Education, and Welfare, 1976. Patrick KL. Instrument Review Criteria. Medical
Health Council of the Netherlands. Diagnosis and treat- Outcomes Trust 1997: 1–5.
ment of ADHD. Publication no. 2000/24. The Hague: Pliszka SR, McCracken JT, Maas JW. Catecholamines in
Health Council of the Netherlands, 2000. attention-deficit hyperactivity disorder: current
Herrmann C. International experiences with the Hospital perspectives. J Am Acad Child Adolesc Psychiatry
Anxiety and Depression Scale - a review of validation 1996; 35(3): 264–72.
data and clinical results. J Psychosom Res 1997; 42: Prendergast M, Taylor E, Rapoport JL, Bartko J, Donnelly
17–41. M, Zametkin A, Ahearn MB, Dunn G, Wieselberg HM.
Kaufman J, Birmaher B, Brent D, Rao U, Ryan N. Kiddie- The diagnosis of childhood hyperactivity. A US-UK
Sads-Present and Lifetime Version (K-SADS-PL). cross-national study of DSM-III and ICD-9. J Child
Version 1.0. Pittsburgh: Department of Psychiatry, Psychol Psychiatry 1988; 29(3): 289–300.
University of Pittsburgh School of Medicine, 1996. Reid, R. Assessment of ADHD with culturally different
Kazdin AE, Esveldt-Dawson K, Loar LL. Correspondence groups: the use of behaviour rating scales. School
of teacher ratings and direct observations of classroom Psychology Review 1995; 24: 537–60.
behaviour of psychiatric inpatient children. J Abnorm Reid R, DuPaul GJ, Power TJ, Anastopoulos AD, Rogers-
Psychol 1983; 10; 483–95. Adkinson D, Noll M, Riccio C. Assessing culturally
Landis JR, Koch GG. The measurement of observer agree- different students for attention deficit hyperactivity
ment for categorical data. Biometrics 1977; 33: 159–74. disorder using behaviour rating scales. J Abnorm Child
Lessing E, Oberlander MI, Barbera L. Convergent validity Psychol 1998; 26(3): 187–98.
of the IPAT Children’s Personality Questionnaire and Spencer F, Biederman J, Heiligenstein J, Wilens T, Faries
teachers’ ratings of the adjustment of elementary school D, Prince J. An open-label, dose ranging study of
children. Soc Behav Pers 1974; 2: 222–9. tomoxetine in children with attention deficit hyperac-
Magnusson P, Smart J, Gretarsdottir H, Prandardottir H. tivity disorder. J Child Adolesc Psychopharmacol 2001;
Attention-Deficit/Hyperactivitity symptoms in 11(3): 251–65.
Icelandic schoolchildren: assessment with the Stratford PW, Binkley JM, Riddle DL. Health status
Attention Deficit/Hyperactivity Rating Scale-IV. measures: strategies and analytic methods for assessing
Scand J Psychol 1999; 40: 301–6. change scores. Phys Ther 1996; 76(10): 109–23.
Livingston R. Cultural issues in Diagnosis and treatment of Stucki G, Michel BA. How to measure improvement: rules
ADHD. J Am Acad Child Adolesc Psychiatry 1999; and fallacies. Rheumatol Eur Supp 1995; 2: 107–11.
38(12): 1591–4.
Michelson D, Faries DE, Wernicke J, Kelsey DK, Kendrick Correspondence: Shuyu Zhang, Lilly Research
KL, Sallee FR, Spencer T, and the Atomoxetine ADHD Laboratories, Eli Lilly and Company, Drop Code 6161,
Study Group. Atomoxetine in the treatment of chil-
Indianapolis, IN 46285.
dren and adolescents with ADHD: A randomized,
placebo-controlled dose-response study. Pediatrics
2001; 108 (5): 1–9. Telephone (+1) 317-2763455.
Michelson D, Buitelaar JK, Danckaerts MJ, Gillberg C, Fax (+1) 317-4336590.
Spencer T, Zuddas A, Faries D, Zhang S, Biederman J. Email: shuyu_zhang@lilly.com.
Relapse prevention in Patients with Attention-

Copyright © 2005 John Wiley & Sons, Ltd. 14: 186–201 (2005)