Setting and Enrollment of Participants

Patients who presented at two child and adolescent general psychiatric outpatient clinics in the county of Västmanland, Sweden, between August 2011 and June 2013 were eligible for enrollment. These clinics provide services to children and adolescents 18 years old and younger who live within the catchment area (N = 37,494). There are no specialized anxiety clinics in the area. Referrals to the clinics are made by parents, primary care physicians, social services, school health services, and hospital departments. Data from the clinics’ patient administrative systems showed that, in 2010 (when procedures for diagnostic assessment did not include the systematic use of structured interviews), 9.7% of all admitted patients between the ages of 13 and 17 years (N = 1092) had received a diagnosis of anxiety. Patients were eligible if they met the age criteria (12 to 18 years old), regardless of their presenting symptoms. The exclusion criteria were inadequate Swedish-speaking skills and prior diagnosis of intellectual developmental disorder. During the predefined recruitment period, which totaled 63 weeks, 202 patients were found to be eligible for inclusion, whereas 77 were excluded: 28 declined to participate, 45 were missed as eligible and therefore not scheduled for the diagnostic interview, and 4 did not show up for the diagnostic interview. There were no significant differences in sex or age between the excluded and included patients. Clinical staff confirmed eligibility, checked for exclusion criteria, and obtained informed written consent to participate from both the adolescents and their parents. The project was approved by the Regional Ethical Review Board in Uppsala.

Schedule for Affective Disorders and Schizophrenia for School-Age Children – Present and Lifetime Version, 2009

The K-SADS (15) is a semi-structured diagnostic interview designed to guide clinicians in the collection of evidence for 33 DSM-IV (19) psychiatric disorders in children and adolescents between the ages of 6 and 17 years. It consists of a screening interview; symptom severity above the threshold determines which additional diagnostic supplements should be completed. The K-SADS is widely used as a diagnostic tool and reference standard in child and adolescent mental health research (13, 20, 21).

Spence Children’s Anxiety Scale

Anxiety symptoms were measured using the Spence Children’s Anxiety Scale (SCAS) (22), a 44-item (38 score-generating items and 6 positive filler items to reduce negative bias) Likert-type (0 = “never;” 3 = “always”) questionnaire designed to assess anxiety symptoms in children and adolescents. The SCAS provides a total score as well as scores on six subscales: panic attacks and agoraphobia, separation anxiety, physical injury fears, social phobia, OCD, and generalized anxiety. In the original study by Spence, the internal reliability coefficient for the total scale was 0.92, and it ranged from 0.60 to 0.82 for the subscales (22). In an evaluation of the psychometric properties of the Swedish translation of the SCAS, which included a subgroup of the current study sample, Swedish clinical cutoff scores were obtained. Results showed an internal reliability coefficient of 0.94 for the total score and demonstrated the scale’s ability to distinguish between adolescents with and without an anxiety disorder in a Swedish non-anxiety–specific clinical setting (23).

Functional impairment/negative life events

Self-reports of functional impairment and negative life events were collected from adolescents during their initial visits through a routine intake questionnaire. Items were related to problems within the family (any occurrence of child physical abuse in the family, violence between parents, or alcohol or drug problems in the family); difficulties in school (current need for extra help or current failed courses); interpersonal difficulties (school-related bullying or difficulty making new friends); somatic concerns (frequent headache or stomachache or pain in the neck, shoulders, back, or legs during the last three months); and sleep concerns (during the last three months). On the basis of the intake questionnaire, nine dichotomous variables were computed and organized into the five functional domains of family, school, interpersonal, somatic, and sleep.

Procedure

Clinical staff registered each patient’s demographic data, referral cause, and source and administered the SCAS as well as an intake questionnaire as part of a regular intake routine during each participant’s first or second visit. A time was then scheduled for each patient to complete the K-SADS interview (median time between intake visit and interview, seven days; interquartile range, 15 days).

Diagnostic Procedure

Five experienced clinicians (two psychiatrists, two psychologists, and one counselor) received extensive interview training by a K-SADS teacher before data collection. A free-marginal multirater kappa (24, 25) was chosen to calculate inter-rater reliability. This kappa statistic differs from Cohen’s kappa in that it allows for multiple raters and is not dependent on marginal distributions. Calibration meetings were held monthly throughout the study, and these involved video recordings of randomly selected interviews. The average inter-rater reliability and percent overall agreement (Po) during data collection was as follows: all diagnoses, 0.94 (P_o = 0.97); any anxiety, 0.92 (P_o = 0.96); separation anxiety, 0.83 (P_o = 0.92); social phobia, 0.92 (P_o = 0.96); specific phobia, 0.94 (P_o = 0.97); OCD, 1.00 (P_o = 1.00); generalized anxiety disorder, 1.00 (P_o = 1.00); and panic/agoraphobia, 0.94 (P_o = 0.97).

Adolescents and their parents were interviewed together. Interviewers determined the current presence of diagnoses on the basis of full DSM-IV criteria and information collected via the K-SADS only. Diagnoses were not classified as primary or secondary. Anxiety disorders included in the analyses of anxiety prevalence and homotypic comorbidity were social phobia (social anxiety), generalized anxiety disorder, specific phobia, separation anxiety disorder, panic disorder, agoraphobia, and anxiety not otherwise specified (unspecified anxiety disorder). To harmonize with the chapter on anxiety disorders in the DSM-5 (26), the diagnoses of OCD and post-traumatic stress disorder were excluded from the analysis of anxiety prevalence and homotypic comorbidity and included in the category of heterotypic comorbidity. If criteria for the DSM-IV diagnosis of panic disorder with agoraphobia (code 300.21) were fulfilled, both anxiety categories of Panic disorder and Agoraphobia were coded as “1” (i.e., “present”) in the analysis of prevalence and homotypic comorbidity. All non-anxiety disorders were clustered into categories in accordance with the organizational structure of the DSM-5. An overview of the categories and organization of the K-SADS–derived DSM-IV diagnoses is presented in supplementary Table S1.

Data Analysis

Statistical analyses were conducted with the use of SPSS 22 software running on the Windows 7 operating system. Between-group differences were analyzed with independent t-tests and chi-squared tests. Binary logistic regression analyses were conducted to examine the variables associated with anxiety and comorbidity. In multivariate models, separate for each covariate, we adjusted for sex and age.

Description of Sample

The study group consisted of 125 consecutively referred adolescents (72 girls; 57.6%) between the ages of 12 and 18 years (mean, 15.7; standard deviation, 1.5). The majority of the patients’ parents were separated (60.8%). There were multiple causes for referral, as described by the referral source; the most common were symptoms of attention-deficit/hyperactivity disorder (31.2%) and depression (29.6%). Problematic symptoms of anxiety were reported in 18.4% of patients at the time of referral. Among those referred for more than one cause (38 patients; 30.4%), the most frequent help-seeking causes were symptoms of anxiety (52.6%) and depression (52.6%).

The total number of diagnoses per participant ranged from 0 to 8 (median, 2; interquartile range, 1 to 3). The number of diagnoses was related to sex, with girls having more diagnoses than boys (girls: mean, 2.8; standard deviation, 1.7; boys: mean, 1.8, standard deviation, 1.3; t(123) = 3.7; p < .001). Eleven participants (8.8%) had no diagnoses.

Prevalence of Anxiety Disorders

Fifty-eight adolescents (46.4%; 95% confidence interval [CI], 37.9 to 55.1; 62.5% of all girls and 24.5% of all boys) were diagnosed with at least one anxiety disorder, with a female-to-male ratio of 3.5:1. The prevalences of the individual anxiety disorders were as follows: social anxiety, 27.2%; generalized anxiety, 17.6%; specific phobia, 16.8%; panic disorder, 9.6%; agoraphobia, 9.6%; separation anxiety disorder, 4.0%; and unspecified anxiety disorder, 1.6%. Significant associations with sex—with more girls meeting the criteria for diagnosis—were found for all anxiety disorders, except for generalized anxiety and panic disorder. The prevalence of anxiety disorders is presented in Table 1.

Clinical Presentation

There were no differences between K-SADS–defined anxious and non-anxious adolescents in terms of age, parents’ marital status, or ethnicity. However, there was a significant sex difference, with a higher proportion of girls found among anxious as compared with non-anxious adolescents (χ²(1) = 17.7; p < .001). Significant differences in clinical presentation were found, with higher proportions of prior admission to the clinic (χ²(1) = 7.2; p = .007) and symptoms of depression as the referral cause (χ²(1) = 4.4; p = .036) among anxious adolescents. Symptoms of attention-deficit/hyperactivity disorder as the referral cause were more common among non-anxious adolescents (χ²(1) = 7.5; p = .006). Anxious adolescents rated themselves as having more anxiety symptoms (t(119) = 6.7; p < .001) and more somatic and interpersonal problems, with higher proportions of frequent symptoms of pain (χ²(1) = 22.3; p < 0.001) and difficulties making friends (χ²(1) = 11.9; p = .001) as compared with non-anxious adolescents. There were no significant differences with regard to referral source, family problems, school functioning, or sleep concerns. There was a tendency for a higher proportion of anxious than non-anxious adolescents to have experienced school-related bullying (χ²(1) = 3.6; p = .058). Table 2 shows the demographic and clinical presentation characteristics of anxious and non-anxious adolescents.

Associations between clinical presentation and anxiety disorders

For each covariate, separate binary logistic regression analyses, which were adjusted for age and sex, were performed for the total sample to examine associations between clinical presentation and anxiety disorders. A referral cause of anxiety symptoms was not related to any anxiety disorder. Only 21% of K-SADS-defined anxious adolescents were referred for symptoms of anxiety. The category of “any anxiety” was associated with a referral that requested a neurodevelopmental assessment (odds ratio [OR], 4.2; 95% CI, 1.2 to 14.6; p = .023), self-reports of frequent body ache (OR, 4.3; 95% CI, 1.7 to 10.4; p = .001); and difficulties making new friends (OR, 4.5; 95% CI, 1.6 to 12.3; p = .004). Significant associations between generalized anxiety and clinical presentation were more frequently found than they were for the other anxiety disorders, with generalized anxiety being related to three variables: age (OR, 1.4; 95% CI, 1.0 to 2.1; p = .047); referral source (i.e., referral from a primary care physician; OR, 3.2; 95% CI, 1.2 to 8.7; p = .024); and self-reports of frequent body ache (OR, 5.6; 95% CI, 1.4 to 23.1; p = .016). Panic disorder was related to a referral cause of depressive symptoms (OR, 4.9; 95% CI, 1.3 to 18.7; p = .021), as was agoraphobia (OR, 6.9; 95% CI, 1.6 to 29.1; p = .009). Social anxiety and specific phobia were associated with difficulties making new friends (OR, 4.3; 95% CI, 1.7 to 11.1; p = .003 and OR, 3.8; 95% CI, 1.3 to 10.7; p = .012, respectively). Analyses of associations between self-rated anxiety symptoms showed that all anxiety categories, except separation anxiety and unspecified anxiety disorder, were related to the total SCAS scores (all ORs within the 95% CI of 1.0 to 1.1; all p values of <.02).

Patterns of Comorbidity

Only the 58 participants who had been diagnosed with anxiety disorders were included in the analyses of comorbidity, unless otherwise specified. Three (5.2%) adolescents had no comorbidity (i.e., only one anxiety disorder).

Homotypic comorbidity

Homotypic comorbidity was found in 25 adolescents (43.1%; 95% CI, 31.2 to 55.9; see Table 1). Homotypic comorbidity without concurrent heterotypic comorbidity was noted in two anxiety-disordered adolescents (3.4%). Age and sex were not significant predictors of total homotypic comorbidity among anxious adolescents, but sex was a significant predictor in the total sample, with more girls having additional anxiety disorders as compared with boys (OR, 7.3; 95% CI, 2.1 to 26.1; p = .002). The most frequent overlaps between anxiety disorders were social anxiety with concurrent generalized anxiety (20.7%) and social anxiety with concurrent specific phobia (20.7%). Six adolescents (10.3%) had both: social anxiety with concurrent generalized anxiety and concurrent specific phobia. In the total sample, the diagnoses of social anxiety, generalized anxiety, specific phobia, panic disorder, and separation anxiety disorder were significantly associated with the presence of additional anxiety. When analyses of homotypic comorbidity were limited to anxious adolescents, significant associations remained for the diagnoses of generalized anxiety, specific phobia, and panic disorder. Self-rated symptoms of anxiety were also predictive of homotypic comorbidity, which was congruent with associations between K-SADS–derived anxiety disorders and homotypic comorbidity. In the total sample and among the anxiety-disordered adolescents, a total SCAS score at or above a diagnostic cutoff was a significant predictor of homotypic comorbidity, as were continuous measures of total anxiety symptoms and disorder-specific symptoms. Among referral causes, the only significant predictor of homotypic comorbidity was found in the total sample with the referral cause of autistic symptoms (OR, 9.6; 95% CI, 1.1 to 80.6; p = .037). Adjusted ORs for the association between anxiety disorders, self-rated anxiety symptoms, and homotypic comorbidity are presented separately for the total and anxious samples in Table 3.

Heterotypic comorbidity

Categories of K-SADS–derived heterotypic comorbid diagnoses are presented as supplementary material (see Table S1). Heterotypic comorbidity was found in 53 anxiety-disordered adolescents (91.4%; 95% CI, 81.4 to 96.3). Heterotypic comorbidity without concurrent homotypic comorbidity was found in 30 adolescents (51.7%). Logistic regression analyses showed that no specific anxiety disorder was significantly associated with increased odds of overall heterotypic comorbidity. In separate multivariate models, adjusted for age and sex, single anxiety disorders were used as independent variables, in logistic regression analyses of associations between the four most frequent heterotypic categories and individual anxiety disorders. Significant relationships between three heterotypic categories and four anxiety disorders were observed: trauma disorders and panic disorder (OR, 8.6; 95% CI, 1.6 to 46.1; p = .012); trauma disorders and social anxiety (OR, 0.1; 95% CI, 0.1 to 0.6; p = .011); depressive disorders and generalized anxiety (OR, 4.1; 95% CI, 1.2 to 13.9; p = .024); and conduct disorders and agoraphobia (OR, 5.7; 95% CI, 1.1 to 29.9; p = .040). Overall, significantly higher rates of trauma-related disorders were observed among adolescents with anxiety than in those without (15.5% vs. 4.5%; p = .037). There were no other significant differences in heterotypic comorbid categories between anxious and non-anxious adolescents. Heterotypic comorbidity in the total sample and separately for anxious and non-anxious adolescents is presented in Table 1.

Strengths

There are several indicators of the validity and generalizability of the current investigation. First, participants were consecutively included irrespective of referral cause or presenting symptoms. Second, the diagnostic assessments were based on a standardized interview, thereby increasing both diagnostic precision and detection. Third, data collection was conducted in a non-anxiety–specialized setting and is thus likely to be generalizable to standard clinical care settings. Fourth, the proportion of girls observed in the current sample (58%) was similar to the proportion observed in referrals to the clinics during the 6-month period that followed the end of the study (63%).

Limitations

Several limitations of the current study should be acknowledged. Only 13 boys met the diagnostic criteria for an anxiety disorder, thus precluding analyses of sex differences beyond the prevalence of any anxiety disorder. A majority of parents did not report their educational level or socioeconomic status, which precluded the analysis of these demographics in relation to anxiety. Whereas the administration of the K-SADS typically involves separate interviews, adolescents and parents were interviewed together in the current study, which may have had an effect on the inclination of both informants to describe psychiatric symptoms. Moreover, diagnoses were based only on information collected during the K-SADS interview; interviewers were unaware of the results of questionnaires completed at intake or other additional information sources. This diagnostic procedure requires patients to be able to adequately recognize and verbalize their feelings and behaviors, but these skills can be impaired in psychiatrically referred youths for various reasons.