We conducted a cross-sectional observational study. The recruitment period went from February through October 2016. The study was approved by the Clinical Research Ethics Committee of Puerta de Hierro Hospital in Majadahonda, Madrid, Spain (03/16). All patients and the parents or legal guardians of those that were minors received information about the study and agreed to participate by signing the informed consent form. The study conformed to the ethical principles of the latest update of the Declaration of Helsinki.

Specialists from 46 child and adolescent psychiatry or paediatric neurology clinics in 30 provinces of 14 autonomous communities in Spain participated in the study.

The sample was obtained by recruiting the first 5 patients that visited each clinic that met the inclusion criteria. We obtained the data from the health records of the patient and from the information collected during the visit in which the patient was included in the study.

Patients of any race and gender aged 12–18 years with a confirmed diagnosis of attention deficit hyperactivity disorder (ADHD) based on the DSM-5 criteria were included in the study.12

We collected data on the year of birth, sex, weight, height, socioeconomic level (low, annual income <15000 euro; middle, annual income of 15000–45000 euro; high, annual income >45000 euro), alcohol use (>40g/day in male patients and >20g/day in female patients), tobacco use (daily consumption) and drug use (regular or occasional use). We also assessed the history of psychiatric disease or ADHD in first-degree relatives.

In regard to ADHD, we recorded the date of diagnosis and the type of ADHD diagnosed in the patient: predominantly inattentive (PI), predominantly hyperactive-impulsive (PH), or combined type (CT). We reviewed the health records of the patients following standard clinical procedure to collect information on the presence of psychiatric comorbidities (PSCs) based on the DSM-5 classification, the presence of non-psychiatric comorbidities based on the ICD-10 classification, and whether there was a past history of suicide attempts.

Adolescents were asked to complete the Social Anxiety Scale for Adolescents (SAS-A). This is a tool specifically designed for the evaluation of social anxiety.10,11

The SAS-A comprises 22 items, 18 of which are statements and 4 control questions, and the final score ranges between 18 and 90 points. Each item is answered on a 5-point Likert scale to indicate the extent to which the given statement “is true” for the respondent, with answer options ranging from 1 (“It is never true”) to 5 (“It is always true”). The SAS-A contains three subscales: fear of negative evaluation (FNE), with 8 items; social avoidance and distress in new situations (SAD-N), with 6 items; and social avoidance and distress experienced in general (SAD-G), with 4 items. Higher scores indicate greater severity.

Patients did not receive treatment as part of the study, but received medical treatment and/or care for their disorder based on the clinical judgement of the specialist in charge. We collected data on the history of pharmacological and nonpharmacological treatments received prior to the study as well as current treatment.

The primary endpoint of the study was the mean score on the SAS-A at the time of the visit, which were compared to reference scores obtained in the general healthy adolescent population. We estimated the mean reference value for the total score in the Spanish population at 43.17 points, with a standard deviation of 12.95. We calculated the sample size required to get a power of 100% in detecting significant differences greater than one standard deviation relative to reference values with an alpha of 0.05 (two-sided)10,11 using SPSS Sample Power, which was 234 patients.

We performed a descriptive analysis, calculating frequencies and percentages for qualitative variables and the mean, 95% confidence interval, standard deviation, median, minimum and maximum for quantitative variables. We compared qualitative variables by means of the Fisher exact test or the chi square test. We compared quantitative variables by the Student t test for independent samples as applicable. We performed factor analysis to assess differences in the SAS-A scores based on different factors, applying the Bonferroni or Games Howell correction based on the homogeneity of variance to control for errors in multiple comparisons. We performed exploratory multivariate analysis to assess the association between the total and subscale scores in the SAS-A and different control variables recorded in the study. We defined statistical significance as a P-value of less than .05. The analysis was performed with SPSS version 24.0.

The study included a total of 234 adolescents. Of all patients, 68.8% (159) were male and 31.2% (72) female, while the sex was not recorded in 3. The mean age was 14.9 years (95% CI, 14.6–15.1), with no differences based on sex (P=.798). We found no differences between the sexes in body mass index (P=.705), with a mean of 21.2kg/m2 (95% CI, 20.7–21.7). Five patients (2.2%) had obesity (body mass index ≥30kg/m2). Table 1 presents the distribution of socioeconomic levels and substance use, with no differences based on sex.

There was a history of ADHD in first-degree relatives in 32.1% of the patients (75). Three patients (1.3%) had a history of suicide attempts, all of them female, two with PI type and one with combined ADHD.

Of all patients, 51.7% (121) were classified as having combined ADHD, 37.2% (87) as having PI ADHD, and 9% (21) as having PH ADHD, while the type of ADHD had not been determined at the time of inclusion in the study in 2.1% (5). A higher proportion of female patients were in the PI group (48.6%, n=35) compared to male patients (32.1%; n=51) (P<.05).

We found no differences between male and female patients in the age of diagnosis of ADHD (P=.455). The mean age at diagnosis was 10.6 years (95% CI, 10.1–11), with a minimum of 3.1 and a maximum of 18 years. We found that ADHD was diagnosed significantly later in patients with the PI type compared to those with combined type (P=.022), with a mean difference of 1.4 years (95% CI 0.1–2.6), with no differences in relation to the other types of ADHD. The mean duration of ADHD was 4.2 years (95% CI, 3.8–4.7).

We found non-psychiatric comorbidity in 24.8% patients (58).

Of all patients, 50.4% (118) had psychiatric comorbidities (PSCs). We found no significant differences in the prevalence of PSCs based on sex or ADHD type. Of all patients in the sample, 26.5% (62) had 2 or more PSCs. The mean number of PSCs was 1.9 (95% CI, 1.7–2.1), with a range of 1–6 PSCs.

We present the frequency of each type of PSC by ADHD type in Table 2, where we used the DSM-IV classification to facilitate comparisons with previous studies. We found statistically significant differences in eating disorders (P=.017), which were more frequent in female adolescents. Anxiety disorders (P=.034) were more frequent in patients with PI type compared to those with CT. There was a higher proportion of behavioural disorders in patients with CT compared to patients with PI type (P<.001). Personality disorders were more frequent in patients with PH type compared to CT (P<.05).

Table 3 details the PCSs observed in our study and compares their prevalence with the prevalence found in other studies.

Sixty-five percent of the patients (152) had received behavioural therapy, psychological support, or both, and these treatments were more frequent in patients with PSCs (78.8%, n=93) compared to patients without (50.9%, n=59) (P<.0001). In addition, 97.9% of patients (229) had received some type of pharmacological treatment prior to inclusion in the study.

The most frequent ongoing treatment for ADHD during the study was methylphenidate (184 patients, 78.6%), followed by lisdexamfetamine (35 patients, 15%) and atomoxetine (10 patients, 4.3%). A current treatment was not received in 5 patients.

Forty-two patients were receiving pharmacological treatment for PSCs, 36 with one drug, 4 with two drugs and 2 with three drugs. The drugs were risperidone (14), sertraline (9), fluoxetine (6), aripiprazole (5), paroxetine (5), topiramate (3), olanzapine (2), quetiapine (2), citalopram (1), escitalopram (1), lorazepam (1) and venlafaxine (1).

We compared SAS-A scores by sex and ADHD type and to reference values in the healthy adolescent population (Table 4). We found statistically significant differences between the sexes, with higher scores (greater severity) in female compared to male adolescents in the FNE subscale (P=.003), SAD-N subscale (P=.016) and the total score (P=.01).

We found no differences in the FNE subscale scores based on the type of ADHD.

Scores were significantly higher on the SAD-N subscale in the PI group compared to the PH group (P=.003) and the CT group (P=.049).

Scores were significantly higher on the SAD-G subscale in the PI group compared to the PH group (P=.009) and the CT group (P=.045).

The total scores in the SAS-A were significantly higher in the PI group compared to the PH group (P=.003).

We assessed for differences in the total and subscale scores based on the treatment patients were receiving at the time of the study, and did not find any.

Table 5 presents the results of the exploratory multivariate analysis of the association between SAS-A total and subscale scores and the control variables, which included data for 204 patients. The equation explains only 15% of the variation (r2=0.151), so other factors that were not studied may have affected the results. In this analysis, we excluded obsessive-compulsive disorder from the group of anxiety PSCs, as this disorder is no longer included in the anxiety disorders in the DSM-5 classification. When it came to PSCs, we found the least favourable total scale and subscale scores in the SAS-A in adolescents with ADHD and a comorbid anxiety disorder compared those without PSCs (P<.01) independently of other variables. As for ADHD type, the group that scored least favourably in the total scale (P=.023) and SAD-N (P=.014) and SAD-G (P=.018) subscales was the PI group compared to the PH group. We also found that patients in the PI group scored higher in the SAD-N (P=.037) and SAD-G (P=.021) subscales compared to patients with the combined type, independently of other variables.

We compared the total scores in the SAS-A and the presence of psychiatric comorbidity by sex (Fig. 1) and by type of ADHD (Fig. 2) to reference values in the healthy adolescent population in Spain.10 On average, patients in our sample scored significantly higher compared to healthy Spanish adolescents. These differences were greater in patients with PSCs and in female patients. We found the greatest differences in the SAS-A scores relative to the reference population in patients with PI type and PSCs (Fig. 2).

Figure 1.

Total score in the SAS-A in adolescents with ADHD, by sex and by presence or absence of psychiatric comorbidity, compared to the reference value in the healthy adolescent population in Spain.

ADHD, attention deficit hyperactivity disorder; SAS-A, Social Anxiety Scale for Adolescents.

The presence of psychiatric comorbidity is represented in grey, and its absence in black.

Reference values in the healthy adolescent population are shown as a continuous horizontal line for male adolescents and a dotted horizontal line for female adolescents.10

(0.06MB).

Figure 2.

Total score in the SAS-A in adolescents with ADHD, by ADHD type and by presence or absence of psychiatric comorbidity, compared to the reference value in the healthy adolescent population in Spain (black horizontal line).10

ADHD, attention deficit hyperactivity disorder; SAS-A, Social Anxiety Scale for Adolescents.

Presence of psychiatric comorbidity: white circles; absence of psychiatric comorbidity: blue circles.

(0.07MB).