Journal of Child Psychology and Psychiatry 48:8 (2007), pp 831–839 doi:10.1111/j.1469-7610.2007.01750.

ADHD and gender: are risks and sequela

of ADHD the same for boys and girls?
José J. Bauermeister,1 Patrick E. Shrout,2 Ligia Chávez,1 Maritza
Rubio-Stipec,1 Rafael Ramı́rez,1 Lymaries Padilla,3 Adrianne Anderson,1
Pedro Garcı́a,1 and Glorisa Canino1
1
Behavioral Sciences Research Institute, University of Puerto Rico, San Juan, Puerto Rico; 2Department of
Psychology, New York University, USA; 3Carlos Albizu University, Puerto Rico

Background: Research comparing treatment-referred boys and girls with attention-deficit/hyper-

activity disorder (ADHD) has yielded equivocal results. Contradictory findings may be associated with
differential referral practices or unexplored interactions of gender with ADHD subtypes. Method: We
examined possible gender differences in ADHD and its subtypes among children aged 4 to 17 in a
representative community sample (N ¼ 1896) in Puerto Rico. Caretakers provided information through
the Diagnostic Interview Schedule for Children (version IV) and a battery of impairment, family rela-
tions, child problems, comorbidity and treatment measures. Results: ADHD was 2.3 times more
common in boys than girls, but with one exception there was little evidence that the patterns of asso-
ciations of ADHD with correlates were different for boys and girls. The exception was school suspension,
which was more common among ADHD boys than girls. Additional gender interactions were found when
ADHD subtypes were considered. Among those with combined type (n ¼ 50), boys were more likely to be
comorbid with mood disorders than girls. For those with the inattentive type (n ¼ 47), girls were more
likely to be comorbid with anxiety disorders than boys. Conclusions: Our findings lend cross-cultural
generalizability to recent reports that gender does not interact with correlates for ADHD overall, but that
it may play a role in subtypes. Keywords: Attention deficit/hyperactivity impulsivity disorder, gender
differences, Latino/Hispanics, ADHD subtypes. Abbreviations: CT: combined type; HIT: hyperactive-
impulsive type; IT: predominantly inattentive type; ICC: intraclass correlation coefficient.

The research literature on attention-deficit/hyper- analysis and critical review of the published research
activity disorder (ADHD) reports that individuals literature found no differences in girls and boys on
with this disorder present diverse family back- impulsivity, academic performance, social function-
grounds, patterns of comorbidity, and impairment ing, fine motor skills, parental education or parental
profiles. The heterogeneous nature of ADHD has led depression (Gaub & Carlson,1997), but other stud-
to its subdivision into more homogeneous subtypes ies published after the meta analysis continue to
based on the predominance of symptoms of in- report inconsistent results (see Gershon, 2002;
attention (predominantly inattentive subtype or IT), Heptinsall & Taylor, 2002).
of hyperactivity-impulsivity symptoms (hyperactive- The contradictory results may be associated with
impulsive subtype or HIT) or on both sets of differential referral practices for boys and girls that
symptoms (combined subtype or CT) (American can be related to different impairment or severity
Psychiatric Association, 1994). Research has levels of treated populations. Unexplored inter-
generally supported the validity of the DSM-IV actions of gender with ADHD subtypes may also
subtypes (e.g., Carlson, Shin, & Booth, 1999; explain contradictory findings. Studies based on
Bauermeister et al., 2005; Graetz, Sawyer, Hazell, treated samples may be appropriate when general-
Arney, & Baghurst, 2001). The CT and IT may be izing findings to children who receive services in
stable enough over time to segregate groups for clinics. However, even this generalization may be
research (Lahey, Pelham, Loney, Lee, & Willcutt, limited by the fact that girls with ADHD are under-
2005). represented in these settings (Gaub & Carlson,
The vast majority of the initial papers about ADHD 1997). In addition, treated cases are likely to be more
were based on studies of treated males who mostly impaired than children from the general population
had CT. Recently there has been more attention (Angold, Costello, & Erkanli, 1999). If, as suggested
placed on to what extent males and females with by some researchers, girls with ADHD in clinical
ADHD show different manifestations (e.g., Arcia & settings are more severely affected, comparison be-
Conners, 1998; Arnold, 1996; Gaub & Carlson, tween boys and girls with ADHD in these settings
1997). Like the initial papers, most of the studies of can mask potential sex differences and lead to erro-
gender are also based on treatment samples. These neous conclusions (Gaub & Carlson, 1997).
results have generally been equivocal. A meta In contrast to the studies of treatment samples,
studies of non-referred samples have generally been
Conflict of interest statement: No conflicts declared. consistent in showing no differences of risks for
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
832 José J. Bauermeister et al.

ADHD by gender. Recently, Biederman and col- treated samples. Based on the community sample,
leagues (2005) examined gender effects in a non- we reported that ADHD was 2.2 times more prevalent
referred sample of siblings of probands with ADHD in boys than girls and that it was significantly
and non-ADHD comparison children. No significant associated with child, family, and school variables.
gender differences were reported in psychiatric co- Because we did not examine whether these associ-
morbidity, treatment history, and psychosocial var- ations were different for boys and for girls, we now
iables. The authors concluded that boys and girls report tests of the interactions of the risk variables
with ADHD present similar clinical profiles and with gender for both ADHD and its subtypes.
suggested that gender differences reported in treat- This study provides an important test of the gen-
ment samples may be related to referral biases. eralizability of the existing literature to another cul-
Two other studies of children from the general tural context, namely island-residing Puerto Rican
population found similar results. Graetz, Sawyer, youth. Insofar as that literature generalizes to a
and Baghurst (2005) reported that among children community-based Puerto Rican sample, we expect
who were identified from a nationally representative no gender differences in the risk factors for ADHD,
sample of Australian children, boys and girls who with the possible exception of poorer school adap-
met symptom criteria for DSM-IV ADHD did not tation for boys. We do not expect significant gender
differ on core symptoms of the disorder, comorbidity, differences among the ADHD types in comorbidity
or impairment. The only exceptions were that girls with disruptive disorders. However, we expect that
with ADHD received higher ratings of somatic com- gender will moderate the association between CT and
plaints and better school functioning. However, on IT, and the internalizing disorders.
impairment measures of social problems, school-
work difficulties, and self-esteem, girls were less
impaired than boys in the CT and HIT groups but
equally or more impaired than boys in the IT group.
The second study of children from the general
Method
population (Levy, Hay, Bennett, & McStephen, 2005) Participants
examined whether gender moderated the association
The community sample was an island-wide probability
of ADHD types with comorbidity in a large sample of
household sample of children aged 4–17 years. This
Australian twins and their siblings. They did not find sample and the methods of the study have been de-
significant gender differences among ADHD types in scribed in detail elsewhere (Canino et al., 2004). Briefly,
terms of comorbidity for externalizing disorders, but the sample was stratified by Puerto Rico’s (P.R.) health
they did find that gender interacted with comorbidity reform regions, urban and rural areas, child’s age and
of the IT group with separation anxiety disorder (SAD) gender, using U.S. 1990 Census block groups as prim-
and of the CT group with generalized anxiety disorder ary sampling units. One child was selected at random
(GAD). These disorders were higher in females. from each household adjusted for age and gender. Out of
In summary, the few studies on gender differences 2,102 eligible households, 1,886 parent–child dyads
carried out with non-referred children suggest that were interviewed (completion rate of 90.1%).
the clinical correlates of ADHD are not moderated by
gender, but that gender differences may be found Instruments and measures
among the ADHD types. Although these studies are
suggestive, they are not definitive with respect to Diagnostic Interview Schedule for Children.
Presence of DSM-IV disorders in children during the
findings in the general population. Two of the three
last year was assessed using the latest translation into
studies with non-referred samples were not formally
Spanish of the Computerized Diagnostic Interview
representative of the general population (Biederman Schedule for Children, version IV (DISC-IV; Bravo et al.,
et al., 2005; Levy et al., 2005). Furthermore, the 2001). The DISC-IV is a structured instrument
studies that examined gender differences for the designed to be administered by lay interviewers for the
diagnosis of ADHD used DSM-III criteria (Biederman assessment of psychiatric and substance use disorders
et al., 2005) or used symptom criteria to approxim- in children and adolescents.
ate DSM-IV diagnosis (Graetz et al., 2005). Finally, We defined caseness as presence of DISC-IV criteria
the studies did not adjust for the effects of the based on parental reports including at least one mod-
presence of non-ADHD diagnoses on the gender erate level of impairment or distress linked to each of
differences reported. the psychiatric disorders ascertained. Data have sug-
gested that parents (vs. youths) are more reliable
In this paper, we examine whether boys and girls
informants for ascertaining the presence of ADHD
exhibit different risk factors and correlates for the
(Jensen et al., 1999). ADHD subtypes were identified
diagnosis of ADHD, as well as for ADHD types in a using DISC-IV symptom criteria: IT group, presence of 6
population-based representative sample of Puerto or more symptoms of inattention but 5 or less of
Rican children 4 to 17 years. This article extends hyperactivity-impulsivity; HIT group, 6 or more symp-
analyses of this community sample that were previ- toms of hyperactivity-impulsivity but 5 or less of in-
ously presented (Bauermeister et al., in press) on attention; and CT group, 6 or more symptoms on both
overall correlates of ADHD in both community and symptom dimensions.
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
 ADHD and gender 833

Brief Impairment Scale (BIS). The Spanish BIS pro- and informed consent was obtained from the child’s
vides a measure of the caretaker’s report of a child’s primary caretaker and youths aged 11 and older.
global impairment along interpersonal, school/work,
and self-fulfillment dimensions of functioning. Each
subscale provides reliable and valid measures of Analytic strategy
impairment (Bird et al., 2005). The sample was weighted to represent the population of
children ages 4 to 17 in P.R. Sampling weights reflect
Parent Interviewer Children’s Global Assessment differences in selection probability due to the complex
Scale (PICGAS). The latter yields a global measure of sample design. All statistical analyses were conducted
the child’s impairment on adaptive functioning as using SUDAAN software (release 8.0) (Research Triangle
scored by lay interviewers. The Spanish PICGAS has Institute, 2001).
adequate reliability and validity (Bird et al., 1996). Regression analyses were specified with the corre-
lates as outcome variables and presence of ADHD and
Developmental history. We developed four parent- gender as main predictors. Age and number of non-
reported measures: speech or language problems ADHD disorders were used as covariates in all regres-
identified by a health professional; grade failure; school sion where these factors were not the outcome of
suspension or expulsion; and problems sleeping at interest. In the analyses where the predictors were the
night during the last year. ADHD subtypes, the three IT, HIT and CT groups were
simultaneously included in the regression analyses. To
Parent–Child Attachment Scale. This scale (10 test whether gender moderated the association between
items) is adapted from Hudson’s Index of Parental ADHD and each correlate, the interaction between the
Attitudes and the Child’s Attitude towards Mother/ two (ADHD/gender) was included. The reference group
Father Scale (Hudson, 1982), and assesses the primary in these regressions was girls with non-ADHD dis-
caretaker’s perception about the quality of the parent– orders.
child relationship. We examined the internal consis-
tency (alpha) of this and other scales in our community
sample and the test–retest reliability (intraclass corre- Results
lation coefficient: ICC) in a clinical sample described
elsewhere (Bravo et al., 2001). Alpha is .76 and ICC is Table 1 shows the gender differences in ADHD,
.72. ADHD types and the range of demographic, family,
child, clinical and treatment variables that have
Family Care Burden Scale. This is a short 7-item been shown to be correlated with ADHD. Boys are
scale based on a family burden scale developed by significantly more likely to be diagnosed with ADHD
Messer, Angold, Costello, and Burns (1996). It meas- in the past year than girls, with an age adjusted odds
ures feelings of personal well-being, family relation- ratio of 2.3. This odds ratio was slightly different
ships, activities, and responsibilities (alpha ¼ .72; from the unadjusted 2.2 ratio reported previously
ICC ¼ .50). (Bauermeister et al., in press). Similar ratios were
found for the ADHD types, but only IT was signific-
Parental Discipline Scale. We used the negative dis- antly different from 1.0. For IT the ratio was 2.6, for
ciplinary practices section of the original 8-item scale,
HIT the ratio was 1.4 and for CT the ratio was 2.1.
which include physical punishment, yelling, and emo-
tional detachment (Goodman et al., 1998) (alpha ¼ .62;
Table 1 also shows a number of other gender differ-
ICC ¼ .67). ences. Although the boys and girls in the sample did
not differ by age, overall impairment as measured by
Parents Attitude towards Medication Scale (PAT- PICGAS, family burden, parent–child attachment,
MS). This is a one-item scale designed to assess atti- rates of mood and anxiety disorders, and caretaker’s
tudes towards medication used in the treatment of attitudes toward medication; they did differ on a
behavior problems in children (ICC ¼ .60). number of other variables. Boys have higher scores
on the BIS School scale; receive more negative dis-
Service Assessment for Children and Adolescents cipline practices at home; and present more speech
(SACA). The Spanish version of the SACA (Canino and sleep problems, grade failure and school sus-
et al., 2002) was used to ascertain last-year use of pension, and higher levels of comorbid disruptive
services and treatments by children for emotional, disorders than girls. Boys are also more likely to
alcohol, and drug problems. It has shown fair to sub- receive treatment services than girls.
stantial reliability for most services and for last-year
use of any psychoactive medication. Gender differences in correlates of ADHD. Table 2
shows the average level of each correlate broken
down by ADHD (present/absent) and gender (girls/
Procedures boys). It also shows results of tests of interactions
The procedures of the study have been described in between ADHD and gender in regression models that
detail elsewhere (Canino et al., 2004). The survey was predict each of the correlates. These interactions test
carried out from January 1999 through December whether the strength of association of each correlate
2000. Assent was obtained for children 6 to 10 years with ADHD differs between boys and girls. Although
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
834 José J. Bauermeister et al.

Table 1 Regression analyses for continuous (age, impairment, family) and categorical (ADHD prevalence, child, comorbidity,
treatment) variables among girls and boys

Girls (n ¼ 915) Boys (n ¼ 982) Gender effects

Measures Mean (SE) Mean (SE) b SE p

Continuous
Age (mean years) 10.5 (.2) 10.5 (.1) ).0 .2 .98
Impairment
PICGAS 84.0 (1.0) 82.8 (.8) )1.2 1.2 .30
BIS total 7.3 (.3) 9.1 (.3) 1.8 .4 .00
Interpersonal 1.2 (.1) 1.4 (.1) .2 .1 .13
School 2.2 (.1) 3.5 (.2) 1.4 .2 .00
Self-fulfillment 3.8 (.1) 4.0 (.1) .2 .2 .22
Family variables
Family burden 8.7 (.1) 8.6 (.1) .1 .1 .35
Parent–child attachment 35.4 (.2) 35.0 (.2) ).4 .2 .11
Negative discipline 5.6 (.1) 5.8 (.1) .2 .1 .02
% (n) % (n) OR 95% CI p

Categorical
ADHD 4.7 (43) 10.3 (100) 2.3 1.4–4.0 .01
IT 1.3 (12) 3.6 (35) 2.6 1.1–6.5 .03
HIT 4.6 (42) 6.1 (60) 1.4 .8–2.5 .20
CT 1.5 (14) 3.7 (36) 2.1 .9–4.7 .06
Child problems
Speech problems 4.8 (47) 11.0 (118) 2.4 1.6–3.7 .00
Sleep problems 8.4 (81) 11.5 (109) 1.4 1.0–2.0 .04
Grade failure 10.1 (85) 23.0 (203) 2.8 1.9–4.0 .00
School suspension 4.2 (32) 11.1 (105) 3.0 1.8–4.8 .00
Comorbidity
Any disruptive disorder 4.4 (38) 6.9 (70) 1.6 1.0–2.6 .05
Any mood disorder 2.5 (18) 1.5 (12) .6 .3–1.3 .18
Any anxiety disorder 5.1 (56) 4.7 (41) 1.1 .6–1.9 .75
Treatment
School services 6.0 (53) 11.5 (108) 2.0 1.3–3.1 .00
Outpatient services 3.7 (30) 8.9 (94) 2.5 1.5–4.4 .00
Medication use .24 (3) 2.4 (26) 10.6 2.9–38.0 .00
Medication attitude 35.6 (316) 36.5 (363) 1.0 .8–1.4 .79

Note: Weighted estimates of population means (standard error in parentheses) or percentages (n in parentheses) and unstandarized
regression coefficients (b) or odds ratios (OR). ADHD ¼ Attention-Deficit Hyperactivity Disorder, PICGAS ¼ Parent Interviewer
Children’s Global Assessment Scale; BIS ¼ Brief Impairment Scale; IT ¼ Inattentive Type; HIT ¼ Hyperactive Impulsive Type; CT ¼
Combined Type; CI ¼ 95% Confidence Interval. Age was used as covariate in all regressions where this factor was not the outcome of
interest.

there are more than twice as many boys as girls in from school and this resulted in an odds ratio (OR)
the survey who were diagnosed as having ADHD on for this interaction that was undefined (approaching
the basis of their parents’ report, and although infinity). To get a conservative lower bound for the
ADHD was associated with higher levels of impair- interaction, we carried out an analysis in which a
ment, poor family relations, higher number of child hypothetical female with both ADHD and a school
sleep problems, higher levels of comorbidity, and suspension was introduced into the sample. The
increased use of services, there was little evidence single hypothetical data point reduced the point
that these associations differed by gender. As shown estimate of the interaction from infinity to 124.7. The
in the columns on the right in Table 2, almost none lower 95% bound on the interaction in this analysis
of the ADHD by gender interactions were significant. was 11.3.
With the exception of school suspensions, the rela-
tion of risk variables and correlations to ADHD Gender differences in correlates of ADHD types.
appeared to be similar for boys and girls. As a final analysis we examined gender interactions
School suspensions appeared to be more strongly with ADHD types. Some of the children who were
related to ADHD in boys than girls. Among boys, the classified within an ADHD subtype did not meet full
odds of school suspension with ADHD were 2.3 times criteria for the disorder, and hence the number of
larger than for those without the disorder. Nearly a children with no diagnosis is slightly smaller in
quarter of boys with ADHD (22.7%) in our sample Table 3 than in Table 2. This can be explained by the
reported having been suspended. In contrast, there fact that ADHD subtypes were solely based on
were no girls with ADHD who had been suspended symptom requirement on the DSM IV.
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
 ADHD and gender 835

Table 2 Regression analyses for continuous (age, impairment, family) and categorical (child, comorbidity, treatment) variables
among girls and boys with and without ADHD

ADHD Non-ADHD Interaction Gender and

ADHD

Girls (n ¼ 43) Boys (n ¼ 100) Girls (n ¼ 869) Boys (n ¼ 878)

Measures Mean (SE) Mean (SE) Mean (SE) Mean (SE) b SE p

Continuous
Age (mean years) 8.8 (.7) 9.3 (.5) 10.6 (.2) 10.7 (.2) .5 .8 .54
Impairment
PICGAS 60.8 (5.0) 60.4 (2.8) 85.2 (1.1) 85.3 (.8) )2.4 6.3 .71
BIS total 13.9 (1.9) 16.6 (1.4) 6.9 (.3) 8.2 (.3) 2.0 2.2 .34
Interpersonal 3.1 (.70) 3.5 (.6) 1.1 (.1) 1.1 (.1) .6 1.0 .52
School 4.8 (.8) 7.0 (.5) 2.1 (.1) 3.2 (.2) 1.2 1.0 .25
Self-fulfillment 5.8 (.9) 5.9 (.5) 3.7 (.1) 3.8 (.1) .3 .8 .77
Family variables
Family burden 11.5 (.8) 11.2 (.3) 8.4 (.11) 8.5 (.1) ).1 .9 .91
Parent-child attachment 30.9 (1.1) 30.8 (.5) 35.6 (.2) 35.5 (.2) ).2 1.3 .85
Negative discipline 7.4 (.4) 7.2 (.3) 5.5 (.1) 5.7 (.1) ).4 .5 .46
% (n) % (n) % (n) % (n) OR 95% CI p

Categorical
Child problems
Speech problems 4.9 (3) 23.6 (27) 4.8 (44) 9.6 (90) 3.1 .5–19.5 .23
Sleep problems 25.2 (15) 32.0 (30) 7.6 (66) 9.3 (79) 1.4 .4–4.7 .58
Grade failure 6.8 (3) 27.6 (29) 10.2 (81) 22.3 (173) 1.9 .4–8.6 .42
School suspensiona 0 (0) 22.7 (21) 4.4 (32) 9.8 (84) 124.7 11.3–1378.2 <.01
Comorbidity
Any disruptive disorder 37.9 (16) 43.2 (42) 2.8 (22) 2.8 (28) 1.2 .3–5.4 .78
Any mood disorder 11.3 (4) 8.1 (6) 2.0 (13) .7 (6) 1.8 .2–20.3 .63
Any anxiety disorder 26.7 (10) 23.2 (24) 3.6 (30) 3.0 (31) 1.0 .3–3.7 .98
Treatment
School services 22.2 (8) 28.4 (27) 5.2 (45) 9.6 (81) .7 .2–2.7 .65
Outpatient services 21.2 (6) 27.9 (29) 2.8 (24) 6.7 (64) .6 .2–2.5 .51
Medication use 1.8 (1) 9.6 (11) .2 (2) 1.6 (15) .7 .0–11.3 .80
Medication attitude 48.8 (24) 65.0 (66) 65.1 (568) 63.3 (545) 2.1 .8–5.1 .12

Note: Weighted estimates of population means (standard error in parentheses) or percentages (n in parentheses) and unstandarized
regression coefficients (b) or odds ratios (OR). ADHD ¼ Attention-Deficit Hyperactivity Disorder, PICGAS ¼ Parent Interviewer
Children’s Global Assessment Scale; BIS ¼ Brief Impairment Scale; CI ¼ 95% Confidence Interval. Age and number of non-ADHD
diagnoses were used as covariates in all regressions where these factors were not the outcome of interest.
a
Usual interaction could not be tested because one cell in the analysis had a zero and the estimate of the interaction was undefined.
Instead, interaction test is based on conservative analysis in which a single observation was created to obtain a lower bound on the
association, as described in the text.

Table 3 shows descriptive statistics for the three bound of the interaction, we carried out special
ADHD types, as well as the children with no ADHD analysis similar to what we reported for school sus-
type diagnosis. Interactions of gender with ADHD pension for ADHD. We created a single hypothetical
subtype for age, impairment, family, child, comor- girl with CT and with IT and coded this fictitious
bidity, and treatment variables with gender are person as having a school suspension (or failed
shown in Table 4. To guard against Type I errors in grade for the CT association). Omnibus tests of the
carrying out the multiple tests of the four groups, we interaction based on these analyses were not sig-
carried out an omnibus test for each variable. This nificant for either school suspension or grade failure.
test was significant for comorbidity with mood dis-
orders (Wald F ¼ 3.8; p < .01) and with anxiety dis-
orders (Wald F ¼ 3.0; p < .05). Boys with CT were at
Discussion
greater risk for comorbidity with mood disorders
than girls with the same subtype (30.0% vs. 4.9%). Boys are more than twice as likely to be diagnosed
Girls with IT were at a greater risk for any anxiety with ADHD as girls, and they also tend to show
disorder than boys with IT (58.3% vs. 12.2%). Be- higher levels of conditions that are associated with
cause there were no girls with CT who failed a grade, ADHD and its types. This has raised the possibility
and also no girls with IT or CT who were suspended that boys might be more susceptible to both risks for
from school, it was not possible to estimate and test ADHD and its sequela. Using a representative sam-
the effects for these combinations through weighted ple of children from the general population of
logistic regression. To obtain a conservative lower Puerto Rico we found only slight evidence of gender
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
836 José J. Bauermeister et al.

Table 3 Age, impairment, family, child and comorbidity variables among girls and boys with and without ADHD types in the
community population

IT HIT CT No ADHD types

Girls Boys Girls Boys Girls Boys Girls Boys

(n ¼ 12) (n ¼ 35) (n ¼ 40) (n ¼ 59) (n ¼ 14) (n ¼ 36) (n ¼ 847) (n ¼ 849)

Measures Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE)

Continuous
Age (years) 12.3 (.7) 11.6 (.6) 9.0 (.7) 8.4 (.7) 8.2 (.8) 10.4 (.7) 10.6 (.2) 10.6 (.2)
Impairment
PICGAS 68.6 (5.2) 64.9 (4.2) 63.5 (4.8) 66.7 (2.8) 50.6 (8.9) 51.3 (3.6) 85.8 (1.0) 85.8 (.8)
BIS total 13.3 (2.1) 18.7 (1.7) 11.3 (1.0) 12.7 (1.4) 18.5 (3.2) 22.7 (1.9) 6.8 (.3) 7.9 (.3)
Interpersonal 1.9 (.7) 3.5 (.5) 2.2 (.4) 2.4 (.6) 4.6 (1.3) 5.2 (1.0) 1.1 (.1) 1.1 (.1)
School 4.5 (1.6) 9.7 (.8) 3.8 (.3) 5.1 (.7) 7.6 (1.4) 9.2 (.6) 2.0 (.1) 3.0 (.2)
Self-fulfillment 6.8 (1.1) 5.4 (.7) 5.0 (.8) 5.0 (.5) 6.2 (.9) 8.0 (.7) 3.7 (.1) 3.7 (.1)
Family variables
Family burden 10.4 (.8) 11.0 (.5) 10.8 (.9) 10.3 (.4) 12.7 (.7) 12.4 (.5) 8.4 (.1) 8.4 (.1)
Parent–child attachment 32.4 (1.4) 31.2 (1.1) 32.5 (.5) 31.8 (.8) 27.6 (2.0) 29.6 (.7) 35.7 (.2) 35.6 (.2)
Negative discipline 7.2 (.6) 6.8 (.4) 6.3 (.2) 6.9 (.3) 8.4 (.8) 7.5 (.4) 5.5 (.1) 5.6 (.1)
% (n) % (n) % (n) % (n) % (n) % (n) % (n) % (n)

Categorical
Child problems
Speech problems 5.5 (2) 10.2 (5) 8.6 (5) 25.7 (16) 11.5 (1) 30.4 (13) 4.5 (39) 9.3 (84)
Sleep problems 5.5 (2) 24.3 (9) 32.0 (11) 31.9 (19) 29.1 (6) 42.4 (14) 7.0 (62) 8.5 (67)
Grade failure 24.1 (5) 29.5 (9) 21.7 (7) 24.4 (15) .0 (0) 33.8 (14) 9.6 (73) 22.0 (164)
School suspension .0 (0) 26.9 (10) 6.3 (2) 19.7 (9) .0 (0) 24.0 (8) 4.2 (30) 9.4 (78)
Comorbidity
Any disruptive disorder 20.6 (4) 39.6 (12) 21.6 (11) 30.5 (16) 53.3 (6) 60.0 (23) 2.5 (17) 1.9 (19)
Any mood disorder 39.3 (4) 2.1 (2) 3.0 (3) 0 (0) 4.9 (1) 30.0 (6) 1.9 (10) .5 (4)
Any anxiety disorder 58.3 (5) 12.2 (5) 14.6 (8) 19.7 (14) 11.4 (2) 27.5 (9) 3.4 (26) 3.0 (28)

Note: Weighted estimates of population means (standard error in parenthesis) or percentages (n in parenthesis). ADHD ¼ Attention
Deficit Hyperactivity Disorder; IT ¼ Inattentive type; HIT ¼ Hyperactive-Impulsive type; CT ¼ Combined type; PICGAS ¼ Parent
Interviewer Children’s Global Assessment Scale; BIS ¼ Brief Impairment Scale.

differences in the strength of associations with cor- means that the association is more than 11 times
relates of ADHD. stronger for boys than girls. This finding is intriguing
because, while school suspension or expulsion is
Similarity of correlates of ADHD over gender. Our usually explained by the presence of externalizing
results argue against the hypothesis that the sequela behaviors, both genders presented similar rates of
and risks for ADHD are systematically different for disruptive behavior disorders. Consistent with
boys and girls. Previously we had reported that Graetz et al.’s (2005) findings, it is possible that boys
ADHD became less common as children grow older with ADHD present higher rates of annoyance or
(Bauermeister et al., in press), and we now conclude distress to teachers and problems with schoolwork
that this pattern is the same for boys and girls. Other relative to girls. This hypothesis, and an additional
correlates may be both a combination of risk and one of less tolerance of these behaviors on the part of
sequela of ADHD. These include family burden, school administrators towards boys with the dis-
negative discipline, and quality of parent–child order, can explain our school suspension finding and
relationship. It appears that these variables have deserve further investigation.
similar degrees of association in boys and girls.
Patterns of comorbidity are another example of vari- Comorbidity, ADHD types, and gender. Consistent
ables that are strongly related to ADHD, but which with our expectations and recent research findings
are not different for girls and boys. These findings (Graetz et al., 2005; Levy et al., 2005), we found no
are similar to those reported by Biederman et al. gender differences among the ADHD types in co-
(2005) in a sample of non-referred children. morbidity with disruptive disorders. We identified a
different pattern of comorbidity for boys and girls
School suspension, ADHD, and gender. One with the combined type. Boys in the CT group were at
exception to our general findings of similar associ- a greater risk for depression than their female
ations of ADHD with the outcome variables examined counterparts. This interaction could be associated
is school suspension or expulsion. We conservatively with the findings that males with CT presented more
estimate that the magnitude of the ratio of odds ratios problems at school than girls. For example, whereas
for this interaction was greater than 11.3, which about a third of the boys with CT had failed a grade
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
 ADHD and gender 837

Table 4 Regression analyses for continuous (age, impairment, family) and categorical (child, comorbidity) variables for girls and
boys with and without ADHD types

Interaction, Gender and ADHD types

IT Male HIT Male CT Male

Omnibus
Wald F b SE p b SE p b SE p

Continuous
Age 1.7 ).9 1.0 .37 ).6 1.1 .58 2.2 1.2 .06
Impairment
PICGAS .9 )11.6 9.3 .21 4.8 5.8 .41 3.0 8.20 .71
BIS total 1.6 7.2 3.5 .04 ).2 1.4 .86 1.7 2.9 .55
Interpersonal 1.5 2.9 1.4 .04 ).1 .7 .92 .2 1.3 .89
School 2.0 5.2 2.1 .02 .2 .7 .81 ).1 1.2 .92
Self-fulfillment .9 ).7 1.0 .49 ).2 .8 .80 1.5 1.1 .18
Family variables
Family burden 1.5 1.4 .8 .08 ).7 1.0 .52 ).6 .8 .44
Parent–child attachment 1.0 )2.2 2.1 .31 ).5 1.0 .65 2.5 2.0 .21
Negative discipline .9 .3 .6 .68 .4 .4 .38 )1.2 .9 .20
OR 95% CI p OR 95% CI p OR 95% CI p

Categorical
Child problems
Speech problems .2 .9 .1–6.3 .93 1.6 .3–7.9 .54 1.7 .2–16.3 .65
Sleep problems 1.9 9.5 1.1–79.5 .04 .7 .2–2.3 .53 1.5 .3–7.4 .64
Grade failure 1.2a .5 .1–3.2 .49 .4 .1–1.8 .24 5.0a .5–3.0 .18
School suspension 1.5a 16.8*a 1.0–277.0 .05 1.6 .2–11.3 .64 2.6 a .2–37.1 .47
Comorbidity
Any disruptive disorder .6 3.3 .5–20.0 .20 2.1 .4–11.8 .42 1.8 .2–13.5 .57
Any mood disorder 3.8**a .1 .0–2.1 .16 .6a .0–9.6 .71 32.6 1.7–623.8 .02
Any anxiety disorder 3.0* .1 .0–.8 .03 1.6 .4–6.4 .49 3.5 .5–23.3 .19

Note: Weighted estimates of population unstandarized regression coefficients (b) or odds ratios (OR).
ADHD ¼ attention-deficit hyperactivity disorder; IT ¼ inattentive type; HIT ¼ hyperactive-impulsive type; CT ¼ combined type;
PICGAS ¼ Parent Interviewer Children’s Global Assessment Scale; BIS ¼ Brief Impairment Scale; CI ¼ 95%Confidence Interval. Age
and number of non-ADHD diagnoses were used as covariates in all regressions where these factors were not the outcome of interest.
The OR with this footnote are estimated as the conservative lower bound. *p < .05; **p < 01.
a
Usual interaction could not be tested because one or more cells in the analysis had a zero and the estimate of the interaction was
undefined. Instead, Omnibus test is based on conservative analysis in which a single observation was created to obtain a lower
bound on the association, as described [in the text].

and a quarter of them had been suspended or ex- Gender ratio and treatment. The age-adjusted
pelled from school during the last year, none of the gender ratio obtained for ADHD (2.3) is consistent
girls with CT had gone through these demoralizing with findings from other community studies (Arnold,
experiences. 1996). However, our findings do not support previ-
The other interaction indicated that girls with IT ous reports that girls with ADHD are more likely to
were at a greater risk for any anxiety disorder than have the inattentive type (Biederman et al., 2002) or
boys with this type. Further analyses (data not that the proportion of girls with IT is higher than the
shown) suggest that separation anxiety was the only other types (Carlson et al., 1999). Additional re-
disorder within the anxiety umbrella, for which a search is needed to ascertain if these findings are
similar pattern was observed. Inattentive girls have a associated with cultural, sampling, or informant
much higher prevalence of SAD than boys (50.0% vs. source factors. Nevertheless, it is important to note
6.7%). Levy et al. (2005) also reported an IT/gender that a significantly higher female:male ratio with IT
interaction for this disorder in their community is not a finding consistently reported in the research
sample. literature.
The results that boys and girls with ADHD sub- Children with ADHD in the community were un-
types present different patterns of comorbidity are der-treated. The absence of an ADHD/gender inter-
especially interesting and suggest that contradict- action for services received argues against previous
ory comorbidity findings in the ADHD-gender observation that under-treatment of girls in clinics is
literature may be associated with differential refer- associated with preferential treatment settings, such
ral practices and/or unexplored interactions of as school (girls) vs. outpatient clinics (boys) (Arnold,
gender with ADHD types. In addition, analyses of 1996; Heptinsall & Taylor, 2002). This finding is also
comorbidity patterns by ADHD types and gender inconsistent with the view that girls with ADHD are
can result in a better understanding of childhood under-treated relative to boys. Nevertheless, a lower
psychopathology. rate of medication treatment was received by girls
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
838 José J. Bauermeister et al.

(1.8%) even though our data suggest that this are not impaired and do not need referral to treat-
treatment is not associated with a negative attitude ment.
towards medication. The general findings that boys and girls with
ADHD present similar risks and sequela argue
Limitations. Since teacher reports were not avail- against gender differences in prognosis, response to
able, our findings are limited to reports from parents treatment, and eventual outcome. However, this is
who may not be fully cognizant of their child’s be- not to say that gender could not mediate the conse-
havior at school. Also, with the exception of the quences of living with ADHD. Girls and boys face
PICGAS, no independent measures were obtained for different gender-related challenges in their life
the correlates examined. Consequently, the associ- courses that need to be taken into account in a
ations reported may have been affected by shared treatment program (Arnold, 1996). These and other
method variance. Second, we did not use gender- questions related to differential clinical course and
specific symptom thresholds to ascertain ADHD. outcome for boys and girls with ADHD need to be
Recent research suggests that there may be a small studied longitudinally and with a comprehensive
number of girls who have behaviors and impairment battery of gender sensitive measures.
that are consistent with ADHD but do not meet DSM
IV symptom criteria (Waschbusch & King, 2006).
Third, our ADHD types were defined in terms of
symptoms criteria without consideration of addi- Acknowledgements
tional cross-situational, age of onset, and impair- This research was supported by NIMH funded grants
ment criteria. This decision, modeled after other (MH54827 Canino (PI) and P01-MH 59876-02 Ale-
researchers (Graetz et al., 2005; Levy et al., 2005), grı́a (PI), and from P20 MD000537-01 Canino (PI),
resulted in an increased number of children ana- and 1R24ND000152-01 Cabiya (PI) from the Na-
lyzed but discrepant sample sizes for the ADHD and tional Center for Minority Health Disparities, and by
ADHD type analyses. It also may have resulted in McNeil Pediatrics Division of McNeil PPC, Inc., Fort
more heterogeneity across ADHD types. Finally, our Washington, PA.
general conclusion that there are no striking inter-
actions of correlates with gender does not imply that
subtle interactions might be found with larger sam- Correspondence to
ple sizes. Although there were no statistically signi-
ficant interactions in our sample of 1,886 children, José J. Bauermeister, 177 Las Caobas Street, San
the standard errors and confidence bounds reported Juan, Puerto Rico, 00927-4230, USA. Tel: (787) 763-
in Table 2 clearly show that our data are consistent 1946; Fax: (787) 758-4561; Email: jjbauer@prtc.net
with possible interactions.

References
American Psychiatric Association (1994). Diagnostic
Conclusion and implications
and statistical manual of mental disorders (4th edn).
In conjunction with an emerging cross-cultural Washington, DC: Author.
literature on children sampled from the community, Angold, A., Costello, E.J., & Erkanli, A. (1999). Comor-
our data suggest that differences between boys and bidity. Journal of Child Psychology and Psychiatry,
girls with ADHD were similar to those obtained for 40, 57–87.
Arcia, E., & Conners, C.K. (1998). Gender differences in
children without ADHD. Nonetheless, gender differ-
ADHD? Journal of Developmental and Behavioral
entially affects comorbidity of the ADHD subtypes
Pediatrics, 19, 77–83.
with mood and anxiety disorders. Arnold, L.E. (1996). Sex differences in ADHD: Confer-
Clinicians should be aware of the possibility that ence summary. Journal of Abnormal Child Psycho-
girls with the inattentive type are more at risk than logy, 24, 555–569.
boys for anxiety disorders, possibly SAD, and that Bauermeister, J.J., Matos, M., Reina, G., Salas, C.C.,
boys with the combined type are a greater risk for Martı́nez, J.V., Cumba, E., et al. (2005). Comparison
depression. Thereby reducing the risk of misdiag- of the DSM-IV combined and inattentive types of
nosis, particularly for girls with the inattentive type ADHD in a school-based sample of Latino/Hispanic
whose ADHD could be overlooked and diagnosed as children. Journal of Child Psychology and Psychiatry,
anxiety. Schools also need to develop programs to 46, 166–179.
Bauermeister, J.J., Shrout, P.E., Ramı́rez, R., Bravo,
help boys with ADHD effectively without resorting to
M., Alegrı́a, M., Martı́nez-Taboas, A., et al. (in press).
suspension or expulsion as the primary mean of
ADHD correlates comorbidity and impairment in
handling school problems. This practice can in- community and treated samples of children and
crease the risk of demoralization of boys with the adolescents. Journal of Abnormal Child Psychology.
disorder and school dropout. Also, the fact that girls Biederman, J., Kwon, A., Aleardi, M., Chouinard, V.A.,
are less subject to school suspension should not be Marino, T., Cole, H., et al. (2005). Absence of gender
misconstrued to mean that girls with the disorder effects on attention deficit hyperactivity disorder:
Ó 2007 The Authors
Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.
 ADHD and gender 839

Findings in nonreferred subjects. American Journal of National Institute of Mental Health Methods for the
Psychiatry, 162, 1083–1089. Epidemiology of Child and Adolescent Mental Disor-
Biederman, J., Mick, E., Faraone, S., Braaten, E., ders (MECA) Study. Social Psychiatry and Psychiatric
Doyle, A., Spencer, T., et al. (2002). Influence of Epidemiology, 33, 162–173.
gender on attention deficit hyperactivity disorder in Graetz, B.W., Sawyer, M.G., & Baghurst, P. (2005).
children referred to a psychiatric clinic. American Gender differences among children with DSM-IV
Journal of Psychiatry, 159, 36–42. ADHD in Australia. Journal of the American Academy
Bird, H.R., Andrew, H., Shwab-Stone, M., Goodman, of Child and Adolescent Psychiatry, 44, 159–168.
S.H., Dulcan, M., Richters, J., et al. (1996). Global Graetz, B.W., Sawyer, M.G., Hazell, P.L., Arney, F., &
measures of impairment for epidemiologic and clin- Baghurst, P. (2001). Validity of DSM-IV ADHD sub-
ical use with children and adolescents. International types in a nationally representative sample of Aus-
Journal of Methods in Psychiatric Research, 6, 295– tralian children and adolescents. Journal of the
307. American Academy of Child and Adolescent Psychi-
Bird, H.R., Canino, G.J., Davies, M., Ramı́rez, R., atry, 40, 1410–1417.
Chávez, L., & Duarte, C. (2005). The Brief Impairment Heptinsall, E., & Taylor, E. (2002). Sex differences and
Scale (BIS): A multidimensional scale of functional their significance. In S. Sandberg (Ed.), Hyperactivity
impairment for children and adolescents. Journal of and attention disorders of childhood (pp. 99–125).
the American Academy of Child and Adolescent Cambridge: Cambridge University Press.
Psychiatry, 44, 699–707. Hudson, W.W. (1982). Methodological observations on
Bravo, M., Ribera, J., Rubio-Stipec, M., Canino, G., applied behavioral science. A measurement package
Shrout, P., Ramirez, R., et al. (2001). Test–retest for clinical workers. Journal of Applied Behavioral
reliability of the Spanish version of the Diagnostic Science, 18, 229–238.
Interview Schedule for Children (DISC-IV). Journal of Jensen, P.S., Rubio-Stipec, M., Canino, G., Bird, H.,
Abnormal Child Psychology, 29, 433–444. Dulcan, M.K., Schwab-Stone, M.E., et al. (1999).
Canino, G., Shrout, P.E., Alegrı́a, M., Rubio-Stipec, M., Parent and child contributions to diagnosis of mental
Chávez, L.M., Ribera, J.C., et al. (2002). Methodo- disorder: Are both informants always necessary?
logical challenges in assessing children’s mental Journal of the American Academy of Child and Ado-
health services utilization: Cross cultural considera- lescent Psychiatry, 38, 1569–1579.
tions. Mental Health Service Research, 4, 97–107. Lahey, B.B., Pelham, W.E., Loney, J., Lee, S.S., &
Canino, G., Shrout, P.E., Rubio-Stipec, M., Bird, H.R., Willcutt, E. (2005). Instability of the DSM-IV Sub-
Bravo, M., Ramirez, R., et al. (2004). The DSM-IV types of ADHD from preschool through elementary
rates of child and adolescent disorders in Puerto Rico: school. Archives of General Psychiatry, 62, 896–902.
Prevalence, correlates, service use, and the effects of Levy, F., Hay, D., Bennett, K., & McStephen, M. (2005).
impairment. Archives of General Psychiatry, 61, 85– Gender differences in ADHD subtype comorbidity.
93. Journal of the American Academy of Child and
Carlson, C.L., Shin, M., & Booth, J. (1999). The case for Adolescent Psychiatry, 44, 368–376.
DSM-IV subtypes in ADHD. Mental Retardation and Messer, S.C., Angold, A., Costello, E.J., & Burns, B.J.
Developmental Disabilities Research Reviews, 5, 199– (1996). The Child and Adolescent Burden Assessment
206. (CABA): Measuring the family impact of emotional
Gaub, M., & Carlson, C. (1997). Gender differences in and behavioral problems. International Journal of
ADHD: A meta-analysis and critical review. Journal of Methods in Psychiatric Research, 6, 261–284.
the American Academy of Child and Adolescent Research Triangle Institute. (2001). SUDAAN User’s
Psychiatry, 36, 1036–1045. Manual, Release 8.0. Research Triangle Park, NC:
Gershon, J. (2002). A meta-analytic review of gender Research Triangle Institute.
differences in ADHD. Journal of Attention Disorders, Waschbusch, D.A., & King, S. (2006). Should sex-
5, 143–154. specific norms be used to assess attention-deficit/
Goodman, S.H., Hoven, C.W., Narrow, W.E., Cohen, P., hyperactivity disorder? Journal of Consulting and
Fielding, B., Alegrı́a, M., et al. (1998). Measurement of Clinical Psychology, 74, 179–185.
risk for mental disorders and competence in a
psychiatric epidemiologic community survey: The Manuscript accepted 22 January 2006

Ó 2007 The Authors

Journal compilation Ó 2007 Association for Child and Adolescent Mental Health.