This study aimed to elucidate the gender differences in internalizing and externalizing problems in children and adolescents aged six to eighteen with ADHD using parent and teacher reports extracted from a clinical database. Consistent with studies in the West, females with ADHD presented with fewer externalizing problems than their male counterparts. In addition, females presented with more internalizing problems although it was not statistically significant.
Gender difference in externalizing problems
In contrast to previous clinical studies in Asia where no gender difference was found in externalizing problems, the gender difference noted in our study could be due to two reasons. Firstly, our sample had more females which have increased the statistical power. Secondly, Singaporean parents, similar to most Asian parents, are highly concerned with academic success . Singaporean parents may have a lower threshold to bring their daughters to the clinic when they are facing only academic difficulties or attention deficits, without displaying any behavioral issues.
Gender difference in internalizing problems
Females with ADHD were reported to have more internalizing problems but the difference did not reach statistical significance. This could be explained by several reasons. Firstly, only parent-reports were used in assessing the internalizing problems. It has been shown that parent-reports were less sensitive and they tended to underestimate internalizing problems compared to self-reports [34,35]. Secondly, combining different internalizing syndrome scales into one score may have masked the gender differences in individual syndrome scale. Females were found to be significantly more withdrawn and depressed while no significant gender differences were noted in somatic complaints or anxiety. In a study using self-report questionnaire, males with ADHD reported to experience higher level of anxiety than their female counterparts . The opposite directions of gender differences in individual components of internalizing problems may have resulted in the lack of gender difference in the overall internalizing score.
Relationship between other factors and internalizing and externalizing problems
In the multiple linear regressions, the regression models could only explain a small percentage of the variance in internalizing and externalizing problems respectively, suggesting that some important predictors were not included in the regression models. After controlling for potential confounders, there was no evidence of a statistically significant association between gender and internalizing problems and only a weak and marginally significant association between gender and externalizing problems. Therefore, the gender differences observed in the univariate analysis could be contributed by gender differences in the relationship between these confounding factors and internalizing or externalizing problems. However, the gender effects of these confounding factors in the ADHD population are unclear. Their potential effects were discussed below.
Effect of age
Adolescents were found to have more internalizing and externalizing problems than children. After controlling for other variables including gender, age remained a significant predictor of internalizing and externalizing problems. This could be due to the increase in academic demands and higher social expectation of impulse control with increasing age. Inattention, hyperactivity and impulsivity problems may become more impairing in secondary schools, leading to the development of more internalizing and externalizing problems. Previous literature has shown that an earlier age of onset is associated with more externalizing problems and less internalizing problems . In this study, the age of being diagnosed with ADHD was used instead of the age of onset which was not available. Since most symptoms of ADHD would manifest in early childhood, it is possible that those diagnosed at adolescence may have a longer duration of untreated ADHD leading to a greater severity of impairment. To test this hypothesis, accurate age of onset is needed. However, recall bias is a common obstacle in retrospective research, especially in ADHD studies where symptoms can manifest at a very young age.
Effect of socioeconomic status
In this study, children and adolescents with ADHD of lower socioeconomic status had more internalizing and externalizing problems. Among studies on socioeconomic status and ADHD prevalence, it remained inconclusive whether low socioeconomic status could be a risk factor for ADHD . Our results suggest a need for intervention targeting internalizing and externalizing problems in ADHD children from families of low socioeconomic status. Raising public awareness of ADHD may facilitate early help seeking from parents of lower education levels.
Effect of physical punishment and spouse aggression
Consistent with the literature, physical punishment such as caning and spanking was associated with more internalizing and externalizing problems . Based on social learning theory , physical punishment can model aggressive behavior of a child and the theory has been confirmed repeatedly in empirical research . Males were found to be more likely than females to receive physical punishment from parents and the effect of physical punishment also differed between genders . Furthermore, self-control was reported to be lower in males disciplined with spanking but no significant difference was noted in females disciplined with spanking . All these evidences suggest that males with ADHD are more likely to receive physical punishment and subsequently perceive poorer self-control and display more aggressive behaviors than females with ADHD. Similar to physical punishment, psychological aggression between parents could have modelled aggressive behaviors in children. While most studies on the effect of physical punishment were done in general population, this study only included participants with ADHD. Future studies should explore whether ADHD could be a moderator on the effect of physical punishment on internalizing and externalizing problems. A recent local study has reported that use of physical punishment was more common in mothers of ADHD children compared to typically developing children but the causal relationship between the use of physical punishment and the children behavioral problems remains unclear . As physical punishment is more widely accepted in the Asian culture, its effect on children development may be different from the West.
Gender differences in thought problems
Although males with ADHD had more thought problems than females with ADHD according to teacher-reports, no statistically significant gender difference was noted on individual items. Since these items described symptoms from a range of heterogeneous disorders including psychotic disorders and obsessive-compulsive disorder, more comprehensive assessment of these disorders is needed before we can establish their severity and prevalence for gender comparison. Furthermore, the low prevalence of thought problems in children and adolescents necessitates a larger sample for comorbidity study. The gender differences in thought problems observed by teachers in this study may reflect problems in inattention, impulsivity or hyperactivity e.g. “can’t keep his/her mind off certain thoughts”, “twitching”, “deliberately harm self”, “pick nose” and “strange behaviors”.
Implications and further studies
Unlike most clinical studies, participants in this study were all newly diagnosed with ADHD and were treatment naïve. Hence, the results were not confounded by interventions. The study has also included the largest number of females with ADHD among studies on gender differences in ADHD in Asia to date. With more female subjects, we were able to capture the gender differences that were not reported previously among young persons with ADHD in Asia. The findings suggest that in the clinical population, males and females with ADHD could present with different comorbidity profiles. Awareness of the gender-specific comorbidity profiles and management addressing the specific needs is warranted.
Although gender differences in ADHD symptoms and comorbidity have been reported more than two decades ago, the causes were not clearly understood. Structural and functional neuroimaging studies have shown numerous gender differences in neuroanatomy and neural network activities . For example, it has been suggested that overproduction of striatal dopamine receptors could be the cause of hyperactivity in male early developmental period . At the molecular level, gender differences in genetics and sex hormones could modulate neuronal development and led to gender differences in neurodevelopmental disorders such as ADHD . However, more research is needed to understand the biological basis of gender differences in ADHD symptoms. Other than biological mechanism, social factors such as socioeconomic status, use of physical punishment and psychological aggression between parents could have an effect on the gender differences in ADHD as suggested in our study. Each of these variables necessitates further research with more well established instruments in order to understand their effect on gender differences.
The lack of externalizing symptoms in females with ADHD may have delayed referral to the professional. A study on barriers to treatment in ADHD has suggested that parents of daughters with high risks of ADHD were less likely to seek help than parents of males with similar problems . Parental perception of ADHD treatment played an important role too as Bussing et al  have suggested a higher perceived stigma for ADHD treatment of daughters than of sons. More research on parental understanding on ADHD, threshold to treatment and stigma is needed in order to improve help seeking in females with ADHD. Furthermore, whether the behavior is labelled as problematic or not could be cultural dependent. Multiethnic study is needed to explore the different cultural thresholds and Singapore, with its multiethnic population, could be a potential study site.
In making the diagnosis, clinician should be mindful of the gender differences in the symptoms presented. Gender specific threshold in symptom lists or diagnostic instrument could be a possible solution. However, there is a risk of overdiagnosis in females and it was suggested that the impairment criterion should remain the same for both genders . At the moment, clinicians are still recommended to follow the same diagnostic criteria for both genders. Moreover, further research is needed to explore gender differences in neuropsychological tests in both the ADHD and general population in order to understand whether a gender specific cut off score is necessary. To prevent misdiagnosis, clinician should consider ADHD as a possible differential diagnosis apart from a mood disorder when they assess females with depressed mood and social withdrawal.
As most studies in female ADHD were based on clinical samples, characteristics of those that were not referred remained largely unknown. Future study using a community sample would help us understand the characteristics of this non-referred ADHD population. A community study would also promote understanding on the inherent gender difference on attention, hyperactivity and impulsivity in the general population. This information would be crucial in development of any gender-specific diagnostic tool or criteria. Apart from clinical criteria, future studies should explore gender differences in commonly used neuropsychological assessment in the ADHD population. This would allow more accurate interpretation of assessment.
There were several limitations in this study. Firstly, the participants were not recruited by nationwide population sampling although they were recruited from the largest child and adolescent psychiatric center in Singapore. Secondly, database was incomplete due to missing data. Thirdly, the database was collected more than ten years ago but it was the largest and most comprehensive of the ADHD clinical population available at the time of the study. Since gender differences in ADHD in externalizing and internalizing problems have been consistently reported over the years, we expect minimal effect of the time gap on the generalizability of our results.