Study design and variables

A two-center cross-sectional survey-based study was conducted to investigate the moderating role of gender on the relationship between ADHD symptoms and functional impairment among university students. The independent variable was ADHD symptoms, while the dependent variables included total functional impairment and its different subdomains (i.e., family, work, school, life skills, self-concept, social, and risk). The Wender–Utah Rating Scale-25 (WURS) was utilized as a screening tool to assess ADHD symptoms across a broad developmental period and symptom range [19]. Throughout the article, ADHD symptoms are referred to as childhood attention deficit and hyperactivity symptoms (CAS). Somatization, depression, anxiety, hostility, and self-concept were screened and included as control variables, making it possible to control for the effects of these psychiatric symptoms in students without an autism spectrum disorder (ASD), bipolar disorder (BD), or any psychotic disorder (PD) diagnosis.

Setting

This research study was carried out during the 2021–2022 academic year and involved 1087 undergraduate students from the Istanbul University-Cerrahpasa, Florence Nightingale Faculty of Nursing and 2051 students from the Istanbul University-Cerrahpasa, Cerrahpasa Faculty of Medicine. Both nursing and medical students are considered undergraduate students in Turkey. The study was conducted in the students’ classrooms after obtaining the necessary permissions from the faculty boards. The purpose and methodology of the study were clearly explained to the students, and they were invited to participate. Email addresses were collected from participants for communication and for referral to the psychiatry outpatient clinic if their CAS scores exceeded the clinical threshold on the WURS. The scales were administered by four of the researchers, who were present in the classroom to provide any necessary assistance. The data were collected with an online survey tool. All surveys were consolidated into a single link for access, which was shared with students through tools such as the smartboard in the classroom. The link was sent via email or text message to those who preferred this method, allowing them to access it through their devices (e.g., tablets or mobile phones).

Participants

Participation in the study was voluntary basis, and participants did not receive any compensation for their participation. Exclusion criteria included an antipsychotic treatment history and a diagnosis of ASD, BD, or PD due to the possibility of disruption in reality testing. Students with subclinical symptoms were not excluded, as their symptoms were screened and controlled. Additionally, students currently undergoing treatment for ADHD were excluded because such treatment can be related to functional outcomes.

A total of 703 students participated in the study. Four were excluded because they met criteria for psychosis, and two were excluded because of a diagnosis of BD. An additional four students were excluded because they were taking antipsychotic medications. Therefore, the study group consisted of 693 students. (Fig 1) The flowchart illustrates the process of selecting the student sample, inclusion and exclusion criteria, and the characteristics of the final study group.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of participant selection.

https://doi.org/10.1371/journal.pone.0321767.g001

Measurement tools

In this sample, ADHD symptoms were assessed using WURS [17], and functional impairment was assessed through the Weiss Functional Impairment Rating Scale: Self-report (WFIRS-S) [20]. Psychiatric symptoms were screened using the Brief Symptom Inventory (BSI) [21].

Ethical standards

The procedures used in this study adhered to the tenets of the Declaration of Helsinki. The Istanbul University-Cerrahpasa, Cerrahpasa Medical School Ethics Committee approved all study materials (06.04.2021, E-67665420-604.02.01-69040). Informed consent was obtained from the participants who agreed to participate in the study.

Statistical analysis

The initial dataset was screened for variable ranges, response patterns, outliers, and missing values. Four participants were excluded for careless or constant response patterns, six participants as univariate outliers, and three participants as multivariate outliers. Ultimately, 680 individuals were included in the analysis set. The scores for seven participants on the WFIRS-S work dimension and three on the WFIRS-S risk dimension could not be computed because participants selected the N/A option for all questions. As they constituted less than 5% of the cases, the mean substitution technique was used [26], wherein the means of the series were calculated and used to fill in the missing values for the relevant variables.

To evaluate our hypotheses, we performed a moderation analysis, which relies on normality in the distribution of variables due to its regression-based framework. To assess normality, we examined the skewness and kurtosis of the study variables. The values fell within the acceptable range of +/–2, suggesting univariate normality [27], except for the WFIRS-S risk scores. Nevertheless, we did not apply any transformation to this variable and used it in its original form, as deviations from normality tend to have a minimal impact on results in large samples [26]. We checked the dataset for multicollinearity by examining VIF and tolerance values, as strong correlations between predictor variables can weaken their individual contribution to predicting the dependent variable [26]. The results showed that all VIF values were under 5 [28], and tolerance values exceeded .10 [26], suggesting that multicollinearity was not a concern.

Prior to the main analysis, we computed descriptive statistics (including means, standard deviations, and ranges of the variables) and Pearson’s product-moment correlations to examine the bivariate relationships among the study variables. The PROCESS Macro Version 4.2 [29] is a widely used tool for conducting moderation analyses and offers pre-designed model templates. In alignment with our research objectives, we selected Model 1 from these templates to examine both the direct relationship between CAS and functional impairment (H1) and the moderating effect of gender on this relationship (H2). Both the independent variable (X) and the moderator variable (W), along with their interaction term (X*W), are included as predictor variables in this model. It also allows for the incorporation of control variables into the regression function. This model enables researchers to address multiple questions simultaneously. First, it facilitates the examination of whether there is a significant relationship between the independent and dependent variables while controlling for the effects of the moderator, the interaction term, and the control variables. Second, it allows researchers to evaluate whether the addition of the interaction term leads to a statistically significant increase in the explained variance of the dependent variable. If the interaction term is found to be statistically significant, the strength and significance of the relationship between the independent and dependent variables are re-evaluated across different levels of the moderator variable. This analysis helps to clarify how the moderator affects this association.

In our study, CAS was the independent variable, and gender functioned as the moderator. Impairments in seven distinct domains as well as overall functioning were addressed as dependent variables. Additionally, we incorporated participants’ comorbid symptoms—anxiety, depression, negative self-concept, somatization, and hostility—into the model as control variables. Given that there were eight dependent variables, we conducted eight moderation models for each functional impairment score. In each model, we first examined the direct relationship between CAS and functional impairment as well as the moderating effect of gender on this relationship. If the analyses showed a statistically significant difference in R-square (ΔR-square) for the interaction term, we subsequently analyzed the association between CAS and functional impairment separately for women and men to assess how this relationship differs based on participants’ gender. SPSS Version 23.0 was used for all analyses.

CAS and functional impairment

ADHD is a disorder that can cause significant functional impairment [30]. The findings of this study support research showing that ADHD plays a significant role in overall functional impairment [4,5,30]. Longitudinal studies have reported different rates of ADHD diagnosis persistence, ranging from 4% to 76% during the transition to adulthood [31]. Additionally, a study found that a childhood ADHD diagnosis is associated with an 80% rate of functional impairment in adulthood, regardless of whether ADHD symptoms persist at the clinical level [2]. Our study supports these findings and shows that CAS contributes to the general functional impairment observed in university students. Therefore, we suggest that clinical assessments should include an evaluation of both the history of CAS and current psychiatric symptoms to better understand functional impairment, even in university students who have not been diagnosed with ADHD. In addition, the results of our study suggest that well-functioning individuals may still experience impairments due to CAS. Given the moderate positive correlation between CAS and functional impairment scores, we suggest that students, particularly men, with a higher symptom burden are at greater risk for functional impairment and should be evaluated. However, we are unable to comment on the role of treatment, as cases receiving ADHD treatment were not included in the study. Nevertheless, we can conclude that students with higher CAS who were not treated may have higher impairments.

The moderating effect of gender on overall functional impairment

Several studies have reported that gender can have an effect on ADHD and its outcomes [9,32,33]. While most of these studies have found that men are more strongly affected by ADHD symptoms [9,32], Fedele et al. [33] conducted a large clinical study consisting primarily of females (78%) and found that women are more affected. However, the gender distribution of the sample in that study [33] was inconsistent with previous reports obtained from clinical and community populations [8–10], suggesting that it may not represent the general ADHD profile. Conversely, other studies have emphasized the similarity between men and women in terms of ADHD outcomes [34], although comorbid psychiatric symptoms were not evaluated [33,35] or controlled for [32,34]. Thus, the effect of concomitant psychiatric symptoms on functional impairment could not be eliminated in these studies questioning the association between ADHD and functional impairment.

In our study, a broad range of psychiatric symptoms were screened, allowing for more specific CAS results. Consistent with a large body of evidence [9,32], the results showed that CAS impairs both men and women in university students, but men are more likely to be affected negatively. This highlights the importance of evaluating and treating CAS in clinical practice for children and university students. In addition, clinicians should consider that men are more vulnerable to CAS than women. However, although we found men to be more vulnerable to CAS, there was a significant positive relationship between CAS and functional impairment in our women participants as well. Given studies reporting difficulties in recognizing ADHD in females [8–10], we emphasize the need to assess ADHD symptoms in women during childhood and college years.

One notable aspect of our findings was that, while the moderating effect of gender was statistically significant, the R-square change values remained relatively small. This is consistent with prior research indicating that moderation effects in psychological models often produce small effect sizes [30]. A possible explanation for this is the inclusion of multiple control variables, particularly comorbid psychiatric symptoms, which are known to contribute substantially to functional impairment [17,18]. Given that five of the six control variables in our models represented different psychiatric symptoms, it is likely that their collective influence accounted for a significant portion of the explained variance, thereby reducing the prominence of the interaction term. Despite this, prior literature suggests that even small but significant moderation effects can yield meaningful insights into complex psychological relationships, particularly in applied research settings [30]. Thus, while the effect sizes were modest, they still provide valuable information about the moderating role of gender in CAS-related functional impairment.

The moderating effect of gender on subdomains of functional impairment

In this study, we observed different results in subdomains of functional impairment in terms of the moderating effect of gender. The associations between CAS and functional impairments in the social and risk domains were stronger in men than in women. Studies have reported varying results regarding the effect of ADHD symptoms on women compared to men in terms of psychosocial functioning. Some have suggested that women are more affected [9], while others have indicated that both genders are similarly affected [34]. In a study that examined psychosocial functioning in adolescents with ADHD, it was found that women were more vulnerable to psychosocial problems than men, particularly internalization symptoms [36]. Conversely, in our sample, the relationship between CAS and social impairment in men was stronger when psychiatric symptoms were controlled, including internalization symptoms. Based on this result, we suggest that men may be more prone to social impairments due to CAS. However, it is important to note that our sample consisted of high-functioning individuals.

Gender was also found to be a moderator in the relationship between CAS and risk-taking behaviors. Previous studies have reported that adults with ADHD are more prone to risk-taking behaviors, such as gambling, substance use, unprotected sex, and careless driving [37–39]. Shoham et al. investigated factors associated with risky behaviors in adults with ADHD and found that externalizing symptoms, which were reported to be more frequent in men [8–10], were significantly and positively associated with risky behaviors that may lead to physical or non-physical negative consequences [40]. This is consistent with our finding that men exhibited higher levels of risky behaviors associated with CAS.

The effects of gender have been investigated for some sub-dimensions of functional impairment (e.g., psychosocial and cognitive) [9,11,12,34,36], and the results of the few studies investigating gender differences in risk-taking behaviors are consistent with our findings [37,38]. Adolescent men with ADHD have been shown to be more likely to engage in risky behaviors [41,42]. Additionally, in an adult ADHD sample, men were found to be more vulnerable to risky behaviors than women. Notably, that study predominantly included participants with low education levels (below secondary school) [43]. In our study, we showed that this tendency among men is also present at higher education levels. Consistent with the literature [37–39], in our sample, both men and women with high CAS had a greater likelihood of engaging in risky behaviors, but the association between CAS and risky behaviors was stronger among men. Therefore, it can be concluded that men are more prone to ADHD symptom-related risky behaviors, even if they have a high educational level.

Strengths and limitations

The study has several strengths. First, it screened a wide range of current psychiatric symptoms and controlled for them statistically, leading to specific findings related to ADHD symptoms. Second, it involved university students, a high-functioning sample, showing that some impairments persist even in such populations. Third, functional impairment was evaluated using a comprehensive scale. For instance, when assessing the school sub-dimension, various factors such as relationships with authority figures, scheduling difficulties, note-taking, and homework completion were considered, rather than solely focusing on exam grades. Functional impairment was also assessed in multiple subdomains.

However, the study has some limitations as well. First, the data were obtained from self-report scales and were not validated through other sources. Second, the cross-sectional design limits inferences about causality. Third, retrospective assessment of CAS has the potential for bias. Fourth, the sample was predominantly women, likely due to the gender composition of the faculties where the study was conducted. Therefore, the results must be confirmed in more balanced samples in terms of gender distribution. Fifth, gender was assessed through a dichotomous question offering only the options “man” and “woman,” meaning individuals identifying outside of these categories may not have been represented in the study. Finally, as this was a two-center study, the results cannot be generalized to all university students from different universities or departments.