Comparability of an ADHD Latent Trait Between Groups: Disentangling True Between-Group Differences From Measurement Problems

Measurement invariance of Attention Deficit/Hyperactivity Disorder symptom criteria as rated by parents and teachers in children and adolescents: A systematic review

This systematic review aimed to establish the extent to which each Attention Deficit/Hyperactivity Disorder (ADHD) symptom criterion is being assessed without being influenced (biased) by factors such as informant, sex/gender, and age. Measurement invariance (MI) testing using confirmatory factor analysis (CFA) is the prime statistical method to ascertain how these factors may affect the measurement and colour the perception or interpretation of symptom criteria. Such effects (non-invariance) can be operationalised in the form of altered association of a symptom criterion with the measured trait (expressed via variations in CFA loadings which represent the weight of each symptom criterion) due to the factor(s) and/or artificially alter the probability of endorsement of a particular symptom criterion (expressed via variations in the CFA threshold(s) representing how mild or severe a given symptom is). Based on a pre-registered protocol (CRD42022276105), we searched PubMed, Global Health, Embase and PsycInfo up to 21-02-23 for studies that included MI assessments on specific ADHD symptom criteria in individuals aged 0-18 years old, using parental and/or teacher report. Self-reports were excluded, given the poor reliability of self-report in ADHD. All included studies met specific COnsensus-based Standards for the selection of health Measurement Instruments (COSMIN) criteria. Results were synthesised in tabular form, grouping results by factors (e.g. informant) from 44 studies retained. Most comparisons indicated both metric (same loadings) and scalar invariance (same thresholds) with regard to informant, gender, age, temporal (repeated assessments) and co-morbidity. Therefore, the available evidence supports the current diagnostic criteria. However, findings could have been improved by systematic reporting of the direction of bias and its effect size. There appears to be a bias towards reporting MI instead of non-invariance. More studies in the literature are needed where the amalgamation of information provided by different informs and the association of specific symptoms with comorbidity are analysed.

Utility for Clinical Practice of a Bifactor ADHD Model in an Elementary School Population in Mexico

We assessed the appropriateness of the bifactor model for a Mexican ADHD symptom questionnaire (BMQ-ADHD) applied to parents and teachers of elementary school children. With a sample of 765 reports of children's behavior (48.7% with ADHD A1 criteria, 42.6% girls, mean age 8.5 years [± 1.6 sd]), we examined construct validity, measurement invariance, differences for gender and school level, and the appropriateness of using summed scores. The BMQ-ADHD questionnaire was characterized by good construct validity for the bifactor model for parents' and teachers' reports. For both corpora, we detected invariance for gender and school level. There were differences in ADHD symptoms by gender, but not by school year. The summed scores may represent the factors accurately for females but may present difficulty for males in the parents' questionnaires. The present study revealed good BMQ-ADHD psychometric properties for a unidimensional-hierarchical ADHD scale segregated by gender for parents' and teachers' reports.

Is the endorsement of the Attention Deficit Hyperactivity Disorder symptom criteria ratings influenced by informant assessment, gender, age, and co-occurring disorders? A measurement invariance study

OBJECTIVES

This study aims to ascertain whether the differences of prevalence and severity of attention deficit hyperactivity disorder (ADHD) are true or whether children are perceived and rated differently by parent and teacher informant assessments (INFAs) according to gender, age, and co-occurring disorders, even at equal levels of latent ADHD traits.

METHODS

Use of latent trait models (for binary responses) to evaluate measurement invariance in children with ADHD and their siblings from the International Multicenter ADHD Gene data.

RESULTS

Substantial measurement noninvariance between parent and teacher INFAs was detected for seven out of nine inattention (IA) and six out of nine hyperactivity/impulsivity (HI) items; the correlations between parent and teacher INFAs for six IA and four HI items were not significantly different from zero, which suggests that parent and teacher INFAs are essentially rating different kinds of behaviours expressed in different settings, instead of measurement bias. However, age and gender did not affect substantially the endorsement probability of either IA or HI symptom criteria, regardless of INFA. For co-occurring disorders, teacher INFA ratings were largely unaffected by co-morbidity; conversely, parental endorsement of HI symptoms is substantially influenced by co-occurring oppositional defiant disorder.

CONCLUSIONS

Our findings suggest general robustness of Diagnostic and Statistical Manual of Mental Disorders ADHD diagnostic items in relation to age and gender. Further research on classroom presentations is needed.