Theoretical Perspectives on Comorbidity of Learning Disorders

• In Event: 2-103 - Comorbidity in Specific Developmental Disorders: Examining Prevalence Rates, Cognitive Causes, and Implications for Treatment

Integrative Statement

Prevalence studies as well as clinical manuals (e.g. DSM-5) report a substantial comorbidity between different learning disorders, i.e. “dyslexia” and “dyscalculia”, which poses a challenge for diagnostics and therapy.
Interestingly, learning disorders also show an increased comorbidity in a broader sense, with other neurodevelopmental disorders like ADHD. However, the causes for the clinical co-occurrence beyond chance are not yet well understood. Is there a cognitive overlap (common aetiology), are symptoms of one disorder just a by-product of another (symptom phenocopy), or are comorbidities the sum of independent causes (additivity)?
In the recent state of knowledge, comorbid learning (and neurodevelopmental) disorders seem to be additive in their underlying cognitive profiles. This means that the presentation of comorbid clinical symptoms is associated with the sum of specific cognitive deficits of each isolated disorder. The respective “additivity pattern” needs to be discussed from both a theoretical and a methodological perspective.
Theoretically, additivity findings can be (a) directly or (b) indirectly helpful to unravel the reasons for comorbidity. When (a) both isolated disorders share similarly strong cognitive deficits in some areas which co-appear in the comorbid profile, a common aetiology is plausible and informative for diagnostics and intervention. However, if (b) the comorbid impairments equal the sum of two fully dissociated isolated profiles (i.e. distinct deficits of disorder A + distinct deficits of disorder B), no direct explanation for shared roots lies at hand. In this case, further research with other candidate causes should follow.
Future research should also pay more attention to the “mental hierarchy” of cognitive markers under study: Domain-general abilities like intelligence, processing speed or working memory likely influence development across content domains. Accordingly, they are plausible as the common – if any – origin of the comorbidity that we observe. By contrast, domain-specific measures like facets of reading, spelling or mathematical competence, are helpful on a phenomenological level – to describe isolated and comorbid profiles. However, the closer these measures are related to the diagnostic criteria of one or the other disorder (e.g. basic calculation deficits in isolated and comorbid dyscalculia), the less can they be regarded as latent cognitive causes.
Irrespective of the theoretical account, additivity findings depend on the way of statistical testing. Additivity, i.e. the absence of interaction effects within a double-dissociation design (2x2 ANOVA), equals the statistical null hypothesis (“The two disorders’ profiles are independent.”). This means that keeping type I error low would increase the possibility to interpret comorbidity (falsely positive) as the sum of independent causes. There are several strategies to cope with this problem, e.g. a more liberal α-level or equivalence testing, defining a region around zero which is practically equivalent to no effect. Such a priori decisions on meaningful effect sizes are highly recommendable for future (replication) research.
Taken together, disentangling different levels of the evidence in comorbidity research is a job for empirical, but also review studies. If and how such findings can be transferred to the design of interventions needs to be discussed.

Authors

• Christin Schwenk, TU Dortmund, Faculty of Rehabilitation Sciences
  Presenting Author

• Jörg-Tobias Kuhn, TU Dortmund, Faculty of Rehabilitation Sciences
  Non-Presenting Author