Where does attention-deficit/hyperactivity disorder fit in the psychopathology hierarchy? A symptom-focused analysis

Abstract

Modern psychopathology classification systems position attention-deficit/hyperactivity disorder (ADHD) with different groups of conditions, either with externalizing or neurodevelopmental. As such, the optimal placement of ADHD in modern classification systems remains unclear. We advanced the literature by mapping ADHD symptoms onto three transdiagnostic psychopathology spectra-externalizing, neurodevelopmental, and internalizing-and their symptoms. ADHD symptoms had varied associations with different spectra, with subsets of symptoms relating most strongly to externalizing, others to neurodevelopmental, and still others to internalizing. Impulsivity, poor schoolwork, and low perseverance were most closely tied to externalizing, cognitive disengagement symptoms (e.g., confused, stared blankly, daydreamed) and immaturity were most closely tied to neurodevelopment, and cognitive disengagement symptoms were also tied to internalizing. Our findings advise against conceptualizing and treating ADHD as a unitary construct and against placing ADHD exclusively under either externalizing or neurodevelopmental spectra. Symptom-focused approaches will better inform modern psychopathology classification systems. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Figure 1. Availability of measures from three informants (i.e., youth, caregiver, and teacher) across three waves (i.e., baseline, one-year follow-up, and two-year follow-up) of the ABCD Study data

Note. Because the Attention Problems scale consists of different numbers of items across the three informants, we depicted the availability of each attention problem across informants and waves.

Figure 2. Utility of path, MIMIC, and network models in informing attention problems’ associations with psychopathology spectra

Note. MIMIC (Multiple Indicators Multiple Causes) models consider if individual attention problems are associated with psychopathology spectra above and beyond general attention problems, whereas network analysis informs attention problems’ associations with individual psychopathology symptoms. Because MIMIC models consider multiple levels of the psychopathology hierarchy at once (i.e., individual attention problems and general attention problems) and because network models consider symptom-by-symptom as opposed to symptom-to-spectrum associations (as tested in the path models), both methods are more specific than the path models

Figure 3. Illustrations of path, MIMIC, and network models

Note. Panel a) illustrates a path model where attention problems were regressed onto a psychopathology spectrum (i.e., externalizing), with age and sex as covariates. Panel b) illustrates a MIMIC (Multiple Indicators Multiple Causes) model, where attention problems were regressed onto 1) a psychopathology spectrum, 2) age and sex, and 3) a general attention problems factor while allowing for age and sex to covary with psychopathology. Panel c) is a sample psychometric network. Circles represent individual attention problems and psychopathology symptoms. Lines connecting these circles indicate symptom-by-symptom associations (or edges), with blue lines reflecting positive associations and orange lines reflecting negative associations. We primarily focused on associations across psychopathology (i.e., attention problems with externalizing, neurodevelopment, and internalizing symptoms).

Figure 4. Attention problems’ associations with externalizing, social responsiveness, and internalizing in path models

Note. The squares reflect the standardized regression coefficients, and the shaded area around them reflects the 83% confidence interval. To avoid overly stringent comparisons of estimates, we depicted 83% rather than 95% confidence intervals for both the path and MIMIC model results (Payton et al., 2003; Schenker & Gentleman, 2001). Eighty-three percent confidence intervals allow readers to interpret non-overlapping confidence intervals as significant.

Figure 5. Attention problems’ associations with externalizing and social responsiveness from the MIMIC models

Note. Each attention problem’s association with (1) externalizing is plotted along the x-axis and (2) social responsiveness along the y-axis. Eighty-three percent confidence intervals around the associations with externalizing are displayed as the length of the shaded boxes, and the width of the shaded boxes reflects the confidence intervals for social responsiveness. Rectangular boxes reflect the standard errors for attention problems’ associations with externalizing and social responsiveness. CON = concentration difficulties; DAY = daydreamed; FOG= confused; FIN = failed to finish tasks; IMA = immature; IMP = impulsive; INA = inattention; HYP = hyperactive; SBL = stared blankly; WRK = poor schoolwork.

Figure 6. Non-zero edges in the psychometric network based on the one-year follow-up caregiver-reported data

Note. Green indicates a positive edge, and orange indicates a negative edge. This figure only reports the non-zero edges (Table S15 contains all edge weights).