Homotypic Versus Heterotypic Continuity of Anxiety Symptoms in Young Adolescents: Evidence for Distinctions Between DSM-IV Subtypes

For SAD, homotypic and heterotypic continuity were roughly equal. Previous studies indicated considerable comorbidity between SAD and the other anxiety problems, especially with SoPh (Compton et al., 2000). In the present study comorbidity rates at wave 1 were high as well, not only with SoPh, but also with other types of anxiety. This suggests that SAD and the other anxiety problems may represent two sides of the same coin. However, homotypic continuity was as strong as heterotypic continuity, which supports the usefulness of SAD as a separate diagnostic construct. The distinction between SAD and other types of anxiety was further supported by another study, in which we conducted latent class analysis to assess the boundaries between SAD and SoPh in referred 8- to 11-year-olds (Ferdinand et al., 2006). Four different classes of individuals were detected; those with (1) low SAD and SoPh item scores on a self-report questionnaire, (2) high SAD and high SoPh item scores, (3) high SAD and low SoPh scores, and (4) low SAD and high SoPh scores. This also supported the idea that, despite high comorbidity rates, SAD may at least partially represent a separate phenomenon, that, for instance, may be subject to specific etiological influences that differ from influences that affect the course of other anxiety problems.

For GAD, homotypic continuity was almost as strong as heterotypic continuity. This is—at least partly—consistent with a previous study (Pine et al., 2001) in adolescents from the general population indicating that separation anxiety disorder and social phobia in adolescents did not predict future generalized anxiety disorder. This suggested rather strong homotypic continuity of GAD. However, continuity of GAD was more heterotypic than in the Pine et al. study, because in our study, heterotypic continuity was also considerable. This contrast may be due to methodological issues such as differences in sample characteristics (Pine et al. investigated older adolescents), assessment procedures (Pine et al. applied standardized interviews instead of self-report questionnaires), or statistical approach (Pine et al. used categorical diagnostic samples whereas the present study used dimensional scale scores). Further, Pine et al. did not use a block design for their regression analyses, but included all predictors in a forward stepwise logistic regression analysis. So, in essence, they did not test if one predictor predicted future GAD, over and above the effect of other predictors.

Compared to heterotypic continuity, homotypic continuity of SoPh symptoms was relatively strong. Pine et al. (2001) found that SoPh, but also GAD, in adolescence predicted future SoPh, independently of other types of anxiety, whereas SAD did not. Remarkably, in their study, GAD was a better predictor of future SoPh than SoPh itself. The aforementioned methodological differences between the Pine et al. study versus the present study may explain differences between findings.

Homotypic and heterotypic continuity of PD did not differ much. Analyses for boys and girls separately (Table 4) indicated that homotypic continuity was higher in girls than in boys. Previous studies already indicated that panic disorder tends to have a chronic course, in children as well as in adults (Biederman et al., 1997; Bruce et al., 2005). The prevalence of full blown panic disorder, and even of panic attacks, in adolescents is very low (Essau et al., 1999). The present study nevertheless suggests that in girls, the disorder may already begin with a chronic homotypic course at a very young age which is in accordance with the higher prevalence in females versus males (Goodwin et al., 2005), and with studies that retrospectively investigated age at onset, and that often point to childhood or adolescence as a starting point (Sheehan et al., 1981).

Comorbidity rates of OCD with other anxiety disorders are generally high (Heyman et al., 2001; Tukel et al., 2002), which was supported by the correlations among the wave 1 RCADS scale scores. The present study showed that heterotypic continuity of OCD symptoms was higher than homotypic continuity. To our knwoledge, previous studies that assessed homotypic continuity of OCD symptoms versus heterotypic continuity with other types of anxiety in young adolescents are not available, so we cannot compare our findings with previous work. Homotypic continuity was somewhat stronger in girls than in boys. This seems to suggest that, longitudinally, OCD symptoms in boys correlate differently with comorbid conditions than OCD symptoms in girls. Again, we were not able to find previous work on this topic. Future studies are needed to investigate if the differences between boys and girls we found can also be found in other samples, countries, and cultures.

In many studies, anxiety disorders are treated as one group of disorders (Barrett et al., 2001; Lipman et al., 2001; MacMillan et al., 2001; Roza et al., 2003; Shortt et al., 2001), and, even, some widely used assessment instruments do not contain scales that tap different anxiety dimensions (Achenbach, 1991a, b). Previous studies found considerable associations between different types of anxiety symptoms, which suggested the presence of one higher order factor (Nauta et al., 2004). Several studies with adults also found evidence for a higher order factor that explained the presence of different types of anxiety (Hettema et al., 2005; Krueger, 1999; Vollebergh et al., 2001). Given the magnitude of heterotypic continuity in the present study, a higher order factor is likely to be present. However, the present study also showed that considerable homotypic continuity is present as well, occurring separately from a general propensity for high anxiety levels. This indicates that each type of anxiety problem may, at least partly, represent a distinct taxonomic construct. Homotypic continuity was found specifically for SAD, GAD, and SoPh, and for PD in girls. This may indicate that SAD, GAD, SoPh, and PD represent diagnostic constructs that are at least partially distinct. Hence, in clinical practice, instruments are needed that measure different anxiety dimensions separately. Intruments that just assess on single anxiety dimension may not be sufficient. Further, the distinctions between different anxiety constructs indicate that, despite the evidence that similar treatment methods are generally efficacious for different types of anxiety problems, each type of anxiety might require a slightly different treatment approach, and development of specific treatment modules.

The sample consisted of young adolescents only. For older adolescents, different homotypic and heterotypic continuities could apply. Furthermore, questionnaires were used instead of clinical interviews; information about the presence or absence of DSM-IV (American Psychiatric Association, 1994) clinical diagnoses was not obtained. Even though RCADS symptom dimensions have proved to reflect corresponding DSM-IV anxiety disorders (Nauta et al., 2004), still, it may be the case that different results would have been obtained if DSM-IV diagnoses, that take account of the level of functional impairment, would have been used instead of RCADS scale scores.

Since different informants may provide different information, this study would have been more valuable if information regarding symptoms of different types of anxiety would also have been gathered from parents or teachers. Unfortunately, such information was not available.