International Journal of Methods in Psychiatric Research

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010)

Published online 19 July 2010 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/mpr.319

ADHD in a representative sample of the

Brazilian population: estimated
prevalence and comparative adequacy
of criteria between adolescents
and adults according to the item
response theory
GUILHERME POLANCZYK,1,2 RONALDO LARANJEIRA,3,4 MARCOS ZALESKI,4,5 ILANA PINSKY,3,4
RAUL CAETANO6 & LUIS AUGUSTO ROHDE2,7

1 Department of Psychiatry, University of São Paulo Medical School, Brazil

2 Institute for Developmental Psychiatry (INCT-CNPq), Brazil
3 Department of Psychiatry, Federal University of Sao Paulo, UNIAD, Brazil
4 Institute for Public Policies on Alcohol and Drugs (INCT-CNPq), Brazil
5 Federal University of Santa Catarina, Santa Catarina, Brazil
6 University of Texas School of Public Health, Houston, TX, USA
7 ADHD Program, Hospital de Clinicas de Porto Alegre, Federal University of Rio Grande do Sul, Brazil

Key words Abstract

attention-deficit/hyperactivity Background: Attention deficit hyperactivity disorder (ADHD) is a clinically sig-
disorder, WHO ASRS screener, nificant disorder in adulthood, but current diagnostic criteria and instruments
adult, prevalence, epidemiology do not seem to adequately capture the complexity of the disorder in this devel-
opmental phase. Accordingly, there are limited data on the proportion of adults
Correspondence affected by the disorder, specially in developing countries.
Professor Luis Augusto Rohde, Method: We assessed a representative household sample of the Brazilian popu-
ADHD Program, Hospital de lation for ADHD with the Adult ADHD Self-report Scale (ASRS) Screener, and
Clinicas de Porto Alegre, Rua evaluated the instrument according to the Rasch model of item response theory.
Ramiro Barcelos, 2350, Porto Results: The sample was comprised by 3007 individuals, and the overal prevalence
Alegre, RS, 90035-003, Brazil. of positive screeners for ADHD was 5.8% [95% confidence interval (CI), 4.8–
Telephone/Fax (+55) 51 3321
7.0]. Rasch analyses revealed the misfitt of the overall sample to expectations of
3946
the model. The evaluation of the sample stratified by age revealed that data for
Email: [email protected]
adolescents showed a signficant fittnes to the model expectations, while items
completed by adults were not adequated.
Received 8 August 2008;
Conclusions: The lack of fitness to the model for adult respondents challenges the
revised 9 January 2009;
accepted 2 February 2009 possibility of a linear transformation of the ordinal data into interval measures
and the utilization of parametric analyses of data. This result suggests that diag-
nostic criteria and instruments for adult ADHD must take into account a

Copyright © 2010 John Wiley & Sons, Ltd 177

ADHD in a representative sample of the Brazilian population Polanczyk et al.

developmental perspective. Moreover, it calls for further evaluation of

currently employed research methods in light of modern theories of
psychometrics. Copyright © 2010 John Wiley & Sons, Ltd.

Introduction symptom was taken into consideration, leading the

authors to propose the use of this scoring method
Attention deficit hyperactivity disorder (ADHD) is a clini- (Kessler et al., 2007). However, the psychometric proper-
cally significant disorder in adulthood associated to several ties of this instrument were neither extensively explored
adverse outcomes in different areas of life (Wilens et al., in samples of different cultural backgrounds nor using
2004). Current diagnostic systems seem not to capture different methodological approaches.
adequately the complexity of the disorder in this develop- It has been proposed that the ASRS Screener has
mental stage (Brown, 2006; McGough and Barkley, 2004; the potential to be a tool for health care professionals
Rohde, 2008; Wilens et al., 2004; Willoughby, 2003). to screen adults for ADHD (Kessler et al., 2005). Inte-
Accordingly, the distribution of adult ADHD in the com- grating this scale into the routine of community
munity is poorly understood (Polanczyk and Rohde, 2007; health care services would be of extreme value to
Polanczyk et al., 2007). identify ADHD cases in adulthood, especially in countries
The current lack of agreement between experts on the where there are scarce resources for mental health care. In
definition of the most appropriate diagnostic criteria for this regard, there is a strong relevance of both generating
adults, and consequently the lack of a ‘gold-standard’ diag- data on the prevalence of ADHD in adulthood in
nostic tool, is a pivotal limitation for the estimation of developing countries, and evaluating the psychometric
ADHD prevalence. Few studies have estimated the preva- properties of the ASRS Screener according to modern
lence of the disorder in community samples of adults (de approaches.
Graaf et al., 2008; Faraone and Biederman, 2005; Fayyad Item response theory (IRT) is a general statistical
et al., 2007; Kessler et al., 2006; Kooij et al., 2005), and they theory about the relationship between items (questions or
used diverse diagnostic definitions and methodological criterion) and subjects’ ability (symptom severity), which
approaches, which yielded estimates that vary from 1% has been increasingly used to assess the adequacy of
(Kooij et al., 2005) to 16.4% (Faraone and Biederman, DSM-IV diagnostic criteria and instruments for several
2005). Only three developing countries (Colombia, disorders, including ADHD (Gomez, 2008; Hartman
Lebanon and Mexico) have their adult population evalu- et al., 2008; Ietsugu et al., 2007). Rasch measurement is an
ated for ADHD (Fayyad et al., 2007). These surveys were IRT model that assumes that the probability that a person
conducted in the context of the World Health Organization will endorse a symptom is a logistic function of the dif-
(WHO) World Mental Health surveys, and the prevalence ference between a person’s ability (intensity of symp-
imputation in these locations were performed after cali- toms) and the difficulty of the item (severity of the
brating the data in accordance to a US clinical sample. In symptom evaluated) (Rasch, 1960). Data is compared to
this sense, much less is known about ADHD in the devel- the expectations of the model and once it fits it, a linear
oping world. transformation of the raw ordinal scale is possible. This
Several instruments have been constructed to evaluate makes possible the analysis of data using parametric tests.
ADHD in adulthood, and the WHO Adult ADHD Self- Moreover, Rasch analyses provides results on the internal
report Scale (ASRS), which was developed by a group of consistency of the instrument, assess the way categories of
experts in ADHD in adulthood, is one of the most prom- an item works, and if items are answered in a different
ising scales for use in clinical and research settings way according to specific characteristics of the subjects
(Kessler et al., 2005). It is composed by six items and was (Pallant and Tennant, 2007).
derived from an 18-item instrument that reflects all Diag- Thus, we aimed to assess the proportion of individuals
nostic and Statistical Manual of Mental Disorders, Fourth in a representative sample of the Brazilian population
Edition (DSM-IV) ADHD symptoms. The abbreviated who have a positive screening for ADHD, and to
version outperformed the original scale in its diagnostic evaluate the scale according to the Rasch model of IRT,
properties (Kessler et al., 2005) and a superior diagnostic comparing the adequacy of criteria between adolescents
accuracy was achieved when the intensity of each and adults.

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

178 Copyright © 2010 John Wiley & Sons, Ltd
Polanczyk et al. ADHD in a representative sample of the Brazilian population

Methods erties (sensitivity: 68.7%, specificity: 99.5%, total classifi-

cation accuracy: 97.9%, and positive predictive value from
Setting
56.8 to 94.7, for a cut-off of four or more positive items)
Brazil constitutes one of the biggest countries of the world (Kessler et al., 2005). The frequency of each symptom is
in territorial extension. With approximately 170 million evaluated in a four-point scale, from never to very often.
inhabitants, it is the most populous country in Latin The Portuguese version of the scale is available at http://
America and ranks sixth in the world. The country is www.hcp.med.harvard.edu/ncs/asrs.php
divided into five geographic regions, which are highly
heterogeneous in terms of socio-economic, ethnic, and
Analysis of data
cultural characteristics.
Based on a previous study in a different population
(Kessler et al., 2007), a positive screening in the ASRS
Sample and data collection
Screener was defined as a score of 14 or higher. Socio-
The first Brazilian National Alcohol Survey (BNAS) is a demographic correlates were evaluated by using logistic
nationally representative sample of 3007 Portuguese- regression analysis, conducted on data weighted for cor-
speaking household residents in urban and rural areas aged recting to the probability of selection and non-response
14 or older (Pechansky et al., 2009). The study was con- rates. Post-stratification weights were calculated to adjust
ducted between November 2005 and April 2006. Institu- the sample to known Census population distributions of
tionalized and indigenous people living in tribes were not socio-demographic variables. Analyses were conducted
included. Respondents were selected through a three-stage with Taylor linearized variance estimation using STATA 9.2
cluster sampling procedure. to account for the complex nature of the sample. Signifi-
The first stage involved the selection of 143 counties, the cance tests of sets of coefficients used Wald c2 tests adjusted
primary sampling unit (PSU), using probability propor- for the design. Statistical significance was evaluated by
tional to size (PPS) methods. Stage 2 involved the selection using two-sided design-based tests with an alpha level of
of two census sectors for each county based on the PPS, 0.05.
with an exception of census sectors in the 14 biggest coun- The fitness of the ASRS Screener was tested in compari-
ties which were included a priori due to the number of son to the expectations of the Rasch model (Rasch, 1960).
inhabitants in these counties, totaling 325 census sectors. An estimate of the internal consistency reliability of the
Stage 3 corresponded to the selection of eight households scale is calculated based on the Person Separation Index
within each census sector by simple random sampling, (PSI). Three overall fit statistics are considered: two item–
followed by the selection of a household member using the person interaction statistics transformed to approximate a
‘the closest future birthday’ technique. A total of 2522 z score, and an item–trait interaction statistic reported as a
interviews were conducted with respondents 18 years of c2. A significant test indicates that the hierarchical ordering
age or older and 485 additional interviews were conducted of the items varies across the trait, thus compromising the
with respondents 14 to 17 years of age. The survey response required property of invariance. In addition to these
rate was 66.4%. The sample was weighted to correct for the overall summary fit statistics, item-fit statistics is presented
probability of selection of respondents into the sample and (misfit is defined by residual ⱖ2.5 or significant c2 test for
non-response rates. Post-stratification weights were calcu- a p value < 0.05 after Bonferroni correction) and the cat-
lated to adjust the sample to known Census population egory ordering is evaluated (the consistency of categories
distributions of socio-demographic variables. All respon- with the level of the trait evaluated), and item bias, or
dents granted their informed consent. The project was differential item functioning (DIF). We evaluated the fol-
approved by the Institutional Review Board (IRB) of the lowing characteristics as ‘person factors’ for DIF analysis:
Federal University of São Paulo. gender, social class according to the Brazilian Association
of Market Research Institutes criteria (Rutter, 2006),
Brazil’s geographic region, and educational level. We used
Instrument
RUMM2020 software to perform these analyses.
The WHO ASRS Screener has been developed in the
context of the WHO World Mental Health Survey Initiative
Results
and was derived from an 18-item questionnaire which mir-
rored all 18 DSM-IV ADHD symptoms. ASRS Screener is Table 1 depicts the distributions of respondents according
composed by six questions, with adequate diagnostic prop- to categories of score and age strata, weighted to represent

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

Copyright © 2010 John Wiley & Sons, Ltd 179
ADHD in a representative sample of the Brazilian population Polanczyk et al.

Table 1 Proportion of respondents stratified by age range Table 2 Socio-demographic distribution of individuals with
and scores strata on the ASRS Screener, weighted to rep- positive screening for ADHD
resent the Brazilian population
Percentage Odds ratio
Age range (years) Characteristic (SE)1 (95% CI)2

ASRS Screener strata <18 18–44 >44 Gender

Male 30.3 (4.0) 1
0–9 66.5 77.3 82.9 Female 69.7 (4.0) 2.03 (1.3–3.2)
10–13 25.9 17.5 11.0 Ethnicity
14–17 6.8 4.5 5.2 White 50.7 (4.9) 1
18–24 0.8 0.7 0.9 Non-white 49.3 (4.9) 0.93 (0.6–1.4)
Age (years)
14–17 13.8 (2.4) 1
18–44 51.9 (4.3) 0.76 (0.3–1.7)
the Brazilian population. According to the dichotomous >44 34.3 (4.4) 0.87 (0.3–2.2)
scoring approach originally proposed (Kessler et al., 2007), Occupational status
5.8% [95% confidence interval (CI), 4.8–7.0] of subjects Employed 49.8 (4.9) 1
screened positive for ADHD. The prevalence of positive Unemployed 3.4 (1.3) 0.58 (0.2–1.4)
screeners stratified by age range was 7.6% (95% CI, 5.4– Student 13.5 (2.7) 1.41 (0.5–3.7)
10.7) for respondents younger than 18 years, 5.2% Homemaker 19.5 (3.1) 1.3 (0.8–2.2)
(95% CI, 4.0–6.8) for respondents 18 to 44 years old Retired 13.8 (3.0) 1.05 (0.5–2.1)
and 6.1% (95% CI, 4.5–8.3) for respondents older than 44 Educational status
years of age. No significant differences were detected in the (years)
rates between the three age groups (F1.82,294.7 = 1.37, 0–5 35.0 (4.5) 1
p = 0.2). 6–8 27.3 (3.9) 0.88 (0.5–1.5)
In regard to demographic characteristics, women pre- 9–11 30.0 (4.0) 0.98 (0.6–1.6)
sented a higher prevalence of positive screening than men >12 8.1 (3.2) 0.94 (0.4–2.4)
[odds ration (OR) = 2.03, 95% CI, 1.3–3.2]. No significant Marital status
differences on rates of positive versus negative screeners Single 35.3 (4.2) 1
Married 51.2 (4.4) 0.91 (0.5–1.6)
were found concerning ethnicity, occupational, educa-
Divorced 6.0 (1.6) 0.97 (0.5–2.1)
tional, and marital status, social class and geographic
Widow/er 7.5 (1.9) 1.07 (0.5–2.6)
region in Brazil (Table 2). Correlates were also analyzed
Social class
within each age group (14 to 17, 18 to 44, and older than
A 0.8 (0.6) 1
44). Among subjects 18–44 years of age, women had a
B/C 53.2 (4.7) 4.45 (0.9–20)
higher prevalence of screening positive than men D/E 46.0 (4.7) 3.89 (0.8–18.6)
(OR = 2.04, 95% CI, 1.0–4.1). There was a significant lower Geographic region
rate of positive screeners among respondents 25 to 34 North 6.2 (1.2) 1
(OR = 0.32, 95% CI, 0.2–0.7) and 35 to 44 (OR = 0.29, 95% South 7.9 (2.8) 0.77 (0.4–1.6)
CI, 0.1–0.6) years of age in comparison to those aged 18 to Southeast 23.4 (3.9) 1.6 (0.9–2.8)
24. Further analyzes on the rates of positive screeners Northeast 53.3 (4.8) 1.2 (0.6–2.1)
according to ethnicity, occupational, educational, and Central-west 9.2 (2.3) 1.74 (0.7–4.2)
marital status, social class and geographic region in Brazil,
within the three age strata, yielded non-significant differ- 1
Weighted percentages for the oversample of adolescents;
ences (available upon request). SE: standard error.
2
The fitness of the ASRS Screener to the Rasch model Multivariate logistic regression analysis.
was initially assessed in the total sample (483 cases were
excluded from Rasch analysis due to extreme scores). It was
detected a significant item–trait interaction (c2 271.700, items, or both. The residual mean value for items was 1.01
df = 30; p < 0.001), suggesting that there is some degree of with a standard deviation (SD) of 3.06, much higher than
misfit between the data and the model, which could be the expected value of 1, given adequate fit to the model.
caused by misfit to model expectations of respondents or The residual mean value for persons was -0.25 with a SD of

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

180 Copyright © 2010 John Wiley & Sons, Ltd
Polanczyk et al. ADHD in a representative sample of the Brazilian population

Table 3 Fit of the ASRS Screeners to the Rasch model

Item Location SE Fit residual df c2 df Probability

Item 1 0.471 0.023 -0.76 2099 59.437 5 0.000006

Item 2 0.316 0.022 -1.778 2099 76.276 5 0.000001
Item 3 -0.037 0.02 1.745 2099 11.451 5 0.043139
Item 4 0.107 0.021 -1.573 2099 59.55 5 0.000001
Item 5 -0.463 0.018 2.221 2099 6.437 5 0.266015
Item 6 -0.393 0.018 6.228 2099 58.549 5 0.000001

Note: Misfit values are in italic typeface (after Bonferroni correction).

1.17, indicating a reasonable fit among the respondents in cational level. Items 1, 2, 3 and 5 showed altered thresholds.
the sample (Table 3). With respect to reliability, the PSI Altered thresholds were reordered in the three sets of analy-
statistic was 0.808, which indicates that the ASRS Screener ses and did not result in a significant improvement of the
has good person separation reliability. The same pattern overall performance of the scale.
of results was confirmed in a random sample of 1000
subjects.
Discussion
Subsequently, we stratified the sample in three catego-
ries according to age range (14 to 17, 18 to 44, and older We have evaluated the prevalence of individuals with posi-
than 44) to test the hypothesis that the misfit of the data to tive screening for ADHD in a representative sample of
the model was related to age. Data from each age group household Brazilian population. It was estimated that
were tested against the Rasch model separately. The PSI approximately 6% of the Brazilian population screens
statistic varied from 0.80 to 0.82 for the three groups, indi- positive for ADHD based on the ASRS Screener. This is the
cating that ASRS Screener has good internal consistency first representative survey conducted in Brazil that evalu-
reliability. In the group of adolescents, item–trait ated adult ADHD and the fourth conducted in a develop-
interaction showed borderline significance (c2 = 44.6; ing country to date.
p = 0.04) indicating a slight variability on hierarchical We found a significant higher prevalence of positive
ordering of items across the trait. The residual mean value screeners among women, a finding also detected in the
for items was 0.53 (SD 0.68) and the residual mean value for subgroup of 18 to 44 years of age. It is well established the
persons was -0.29 (SD 1.16). There were no items display- higher prevalence of ADHD among boys, specially in clini-
ing DIF or presenting misfit. Items 1, 2, 3 and 4 showed cal samples (Pliszka, 2007). However, in community
altered thresholds. In the strata composed by individuals samples, the difference of ADHD rates between boys and
age 18 to 44, item–trait interaction was highly significant girls is less prominent (Polanczyk and Rohde, 2007). A
(c2 = 185.3; p < 0.001) indicating a lack of hierarchical scarcity of data had been published on this matter among
ordering of items across the trait. The residual mean value adults. Kessler et al. (2006) and Fayyad et al. (2007) found
for items was 0.74 (SD 2.44) and the residual mean value for a higher prevalence of ADHD among men in their sample,
persons was -0.27 (SD 1.14). Items 1, 2, 4 and 6 showed with modest OR (1.6 and 1.5, respectively). Kooij et al.
misfit to model expectation and items 4 and 6 showed some (2005) explored the validity of two different cut-off points
degree of uniform DIF regarding social class. Moreover, (six and four symptoms) for the diagnosis of ADHD and
item 6 showed DIF regarding demographic region in Brazil. found no gender effect when applying the cut-off of six
Items 1 and 3 showed altered thresholds. In the group of symptoms but a higher prevalence of females when apply-
subjects older than 44 years of age, item–trait interaction ing the cut-off of four symptoms. The authors ruled out a
was also highly significant (c2 = 90,3; p < 0.001) indicating confounding effect of comorbidities and hypothesized that
an altered hierarchical ordering of items across the trait. women during adulthood may be more sensitive than men
The residual mean value for items was 0.42 (SD 1.79) and to the identification of symptoms. An alternative explana-
the residual mean value for persons was -0.3 (SD 1.26). tion is that the stability of ADHD symptoms during the
Items 1, 2 and 6 showed misfit to model expectation and lifetime might be higher in women. We cannot exclude that
item 6 showed uniform DIF regarding social class and edu- the presence of other mental disorders (e.g depression or

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

Copyright © 2010 John Wiley & Sons, Ltd 181
ADHD in a representative sample of the Brazilian population Polanczyk et al.

anxiety) have a role in the higher prevalence of ADHD adolescents did not. In this regard, our results, along with
among women in our sample. Our results, in conjunction previous studies, may indicate the existence of problems in
with those from Kooj’s investigation (Kooij et al., 2005), the construct validity of ADHD diagnostic criteria for
points to this intriguing hypothesis, which deserves further adults, since this diagnosis was more extensively validated
study. However, the fact that a higher prevalence of women in children and adolescents (Pliszka, 2007; Rohde, 2008;
screened positive on ASRS Screener may not necessarily Wilens et al., 2004).
reflect a higher prevalence of ADHD in this group. The The results from the first study to assess ADHD symp-
instrument is a screening tool and, although presented toms in a representative sample of the Brazilian population
adequate sensitivity (68.7%), specificity (99.5%), and total indicated that these are common problems in all age ranges
classification accuracy (97.9%), there are uncertainties in this country. Furthermore, our results point to the rel-
about the appropriate scoring method (Kessler et al., evance of evaluating both the validity of constructs which
2005). investigators intend to measure, and the properties of the
Aiming to assess the adequacy of the ADHD construct scales used in this field. The phenomenology of ADHD in
and ASRS Screener, we evaluated the data in relation to the adults seem to be specific and a list of symptoms that takes
Rasch model (Chachamovich et al., 2008; Hartman et al., into account this developmental perspective must be devel-
2008; Tennant et al., 2004). It is important to note that oped and validated in this age group before any effort to
previous epidemiological investigations in the adult integrate adult ADHD diagnosis in future classificatory
ADHD have not assessed the psychometric properties of systems.
the instruments used in light of more recent approaches
like IRT. Our results from Rasch analyses provides several
Acknowledgements
interesting insights to the ADHD and to the epidemiology
field in general. We showed that data pertaining the overall The authors thank Marcos Sanches and Eduardo Chachamov-
sample did not fit the model. Since conceptual assump- ich for their valuable contributions to this study.
tions of the Rasch model support its independence of the
individuals who answer the items (i.e. sample indepen- Funding source
dent) (Andrich, 1988), this finding should call attention
to the necessity of the evaluation of psychometric Funding for this study was provided by Secretária Nacional
properties of instruments in different populations with Antidrogas, Federal Government of Brazil, which had no
further role in study design; in the collection, analysis and
complementary methodological approaches. The absence
interpretation of data; in the writing of the report; and in the
of fitness to the model challenges the possibility of a linear
decision to submit the paper for publication.
transformation of the ordinal data from the ASRS into
interval measures, as originally proposed (Kessler et al.,
2007). This has an important practical implication: since Declaration of interest statement
the interval between two scores may not be the same along Dr Polanczyk has served as a speaker for Novartis and is a
the entire continuum, mathematical operations and para- recipient of a 2008 National Alliance of Research on Schizo-
metric analysis of data may not be possible (Andrich, 1988; phrenia and Depression (NARSAD) Young Investigator
Pallant and Tennant, 2007). Thus, any epidemiological data Award. Dr Rohde was on the speakers’ bureau and/or acted as
based on cut-off scores derived from models assuming consultant for Eli-Lilly, Janssen-Cilag, and Novartis in the last
interval properties of the data is questionable without the three years. Currently, his only industry related activity is
appropriate testing. taking part of the advisory board/speakers bureau for Eli Lilly,
A relevant finding is the superior fitness to the Rasch Novartis, and Shire (less than U$10,000 per year and reflecting
less than 5% of his gross income per year). The ADHD and
model for data provided by adolescents in relation to data
Juvenile Bipolar Disorder Outpatient Programs chaired by
answered by adults (18 to 44 years old and 45 years old or
him received unrestricted educational and research support
more). This very interesting result is in complete agree- from the following pharmaceutical companies in the last three
ment with previous studies (Merrell and Tymms, 2005; years: Abbott, Bristol-Myers Squibb, Eli-Lilly, Janssen-Cilag,
Smith and Johnson, 2000) that evaluated the ADHD crite- Novartis, and Shire.
ria with the Rasch model, indicating that the same behav-
ioral scale for ADHD could not be used by young children
References
and college students. ASRS Screener was constructed after
the adaptation of each ADHD criterion to adulthood. Even Andrich D. (1988) Rasch Models for Measurement, Sage
so, data from adults misfit the model, while data from Publications.

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

182 Copyright © 2010 John Wiley & Sons, Ltd
Polanczyk et al. ADHD in a representative sample of the Brazilian population

Brown T.E. (2006) Toward an adequate understanding of Screener in a representative sample of health plan
attention deficit disorders. Revista Brasileira de Psiquiatria, members. International Journal of Methods in Psychiatric
28, 261–262. Research, 16, 52–65.
Chachamovich E., Fleck M., Trentini C.M., Power M. (2008) Kooij J.J., Buitelaar J.K., van den Oord E.J., Furer J.W., Rijnders
The Brazilian WHOQOL-OLD Module version: a Rasch C.A., Hodiamont P.P. (2005) Internal and external validity
analysis of a new measure. Revista de Saúde Pública, 42, of attention-deficit hyperactivity disorder in a population-
308–316. based sample of adults. Psychological Medicine, 35, 817–
de Graaf R., Kessler R.C., Fayyad J., Ten Have M., Alonso J., 827.
Angermeyer M., Borges G., Demyttenaere K., Gasquet I., McGough J.J., Barkley R.A. (2004) Diagnostic controversies in
de Girolamo G., Haro J.M., Jin R., Karam E.G., Ormel J., adult attention deficit hyperactivity disorder. American
Posada-Villa J. (2008) The prevalence and effects of Adult Journal of Psychiatry, 161, 1948–1956.
Attention-Deficit/hyperactivity Disorder (ADHD) on the Merrell C., Tymms P. (2005) Rasch analysis of inattentive,
performance of workers: results from the WHO World hyperactive and impulsive behaviour in young children
Mental Health Survey Initiative. Occupational and and the link with academic achievement. Journal of Applied
Environmental Medicine, 65, 835–842. DOI: 10.1136/ Measurement, 6, 1–18.
oem.2007.038448 Pallant J.F., Tennant A. (2007) An introduction to the Rasch
Faraone S.V., Biederman J. (2005) What is the prevalence of measurement model: an example using the Hospital
adult ADHD? Results of a population screen of 966 adults. Anxiety and Depression Scale (HADS). British Journal of
Journal of Attention Disorders, 9, 384–391. Clinical Psychology, 46, 1–18.
Fayyad J., De Graaf R., Kessler R., Alonso J., Angermeyer M., Pechansky F., De Boni R., Diemen L.V., Bumaguin D., Pinsky
Demyttenaere K., De Girolamo G., Haro J.M., Karam E.G., I., Zaleski M., Caetano R., Laranjeira R. (2009) Highly
Lara C., Lepine J.P., Ormel J., Posada-Villa J., Zaslavsky reported prevalence of drinking and driving in Brazil: data
A.M., Jin R. (2007) Cross-national prevalence and corre- from the first representative household study. Revista
lates of adult attention-deficit hyperactivity disorder. Brasileira de Psiquiatria, 31, 125–130.
British Journal of Psychiatry, 190, 402–409. Pliszka S. (2007) Practice parameter for the assessment and
Gomez R. (2008) Item response theory analyses of the treatment of children and adolescents with attention-
parent and teacher ratings of the DSM-IV ADHD rating deficit/hyperactivity disorder. Journal of the American
scale. Journal of Abnormal Child Psychology, 36, 865– Academy of Child and Adolescent Psychiatry, 46, 894–
885. 921.
Hartman C.A., Gelhorn H., Crowley T.J., Sakai J.T., Stallings Polanczyk G., Lima M.S., Horta B.L., Biederman J., Rohde L.A.
M., Young S.E., Rhee S.H., Corley R., Hewitt J.K., Hopfer (2007) The worldwide prevalence of ADHD: a systematic
C.J. (2008) Item response theory analysis of DSM-IV can- review and metaregression analysis. American Journal of
nabis abuse and dependence criteria in adolescents. Journal Psychiatry, 164, 942–948.
of the American Academy of Child and Adolescent Psychia- Polanczyk G., Rohde L.A. (2007) Epidemiology of attention-
try, 47, 165–173. deficit/hyperactivity disorder across the lifespan. Current
Ietsugu T., Sukigara M., Furukawa T.A. (2007) Evaluation of Opinion in Psychiatry, 20, 386–392.
diagnostic criteria for panic attack using item response Rasch G. (1960) Probabilistic Models for Some Intelligence and
theory: findings from the National Comorbidity Survey in Attainment Tests, University of Chicago Press.
USA. Journal of Affective Disorders, 104, 197–201. Rohde L.A. (2008) Is there a need to reformulate attention
Kessler R.C., Adler L., Ames M., Demler O., Faraone S., Hiripi deficit hyperactivity disorder criteria in future nosologic
E., Howes M.J., Jin R., Secnik K., Spencer T., Ustun T.B., classifications? Child & Adolescent Psychiatric Clinics of
Walters E.E. (2005) The World Health Organization Adult North America, 17, 405–420.
ADHD Self-Report Scale (ASRS): a short screening scale Rutter M. (2006) Pesquisa de mercado, Ática.
for use in the general population. Psychological Medicine, Smith E.V.J., Johnson B.D. (2000) Attention deficit hyperac-
35, 245–256. tivity disorder: scaling and standard setting using Rasch
Kessler R.C., Adler L., Barkley R., Biederman J., Conners C.K., measurement. Journal of Applied Measurement, 1, 3–
Demler O., Faraone S.V., Greenhill L.L., Howes M.J., Secnik 24.
K., Spencer T., Ustun T.B., Walters E.E., Zaslavsky A.M. Tennant A., Penta M., Tesio L., Grimby G., Thonnard J.L.,
(2006) The prevalence and correlates of adult ADHD in the Slade A., Lawton G., Simone A., Carter J., Lundgren-
United States: results from the National Comorbidity Nilsson A., Tripolski M., Ring H., Biering-Sorensen F.,
Survey Replication. American Journal of Psychiatry, 163, Marincek C., Burger H., Phillips S. (2004) Assessing and
716–723. adjusting for cross-cultural validity of impairment and
Kessler R.C., Adler L.A., Gruber M.J., Sarawate C.A., Spencer activity limitation scales through differential item func-
T., Van Brunt D.L. (2007) Validity of the World Health tioning within the framework of the Rasch model: the
Organization Adult ADHD Self-Report Scale (ASRS) PRO-ESOR project. Medical Care, 42, I37–I48.

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

Copyright © 2010 John Wiley & Sons, Ltd 183
ADHD in a representative sample of the Brazilian population Polanczyk et al.

Wilens T.E., Faraone S.V., Biederman J. (2004) Attention- Willoughby M.T. (2003) Developmental course of ADHD
deficit/hyperactivity disorder in adults. Journal of the symptomatology during the transition from childhood to
American Medical Association, 292, 619–623. adolescence: a review with recommendations. Journal of
Child Psychology and Psychiatry, 44, 88–106.

Int. J. Methods Psychiatr. Res. 19(3): 177–184 (2010). DOI: 10.1002/mpr

184 Copyright © 2010 John Wiley & Sons, Ltd