5 - (IOP 19) Evaluating Measurement Differences Across Cultural Contexts in the English and Spanish-language CASI ADHD Subscales: Beyond Translation

Given the high prevalence of Attention-Deficit/Hyperactivity Disorder (ADHD) worldwide (Cortese et al., 2023), adapting and translating existing validated measures is essential for equitable measurement of clinical and research outcomes. Despite the need for culturally adapted assessments, few studies have tested cross-cultural measurement invariance of ADHD rating scales using modern psychometric methods. This study examines differential item functioning (DIF) in the ADHD subscale of the widely used parent-report Child and Adolescent Symptom Inventory (CASI) to assess measurement differences across English- and Spanish-speaking respondents in the United States (U.S.) and Mexico with implications for clinical and research uses of this measure.
Data were collected from 331 participants (6-12 years of age) from the U.S. and Mexico in school-based studies recruiting children with 6 or more ADHD symptoms within the Inattention (IA) and Hyperactivity/Impulsivity (H/I) domains (Pfiffner et al., 2016). We selected anchor items via ordinal logistic regression and inspecting item wording complemented this approach (e.g., choosing items with more direct translations). CFA confirmed the two-factor model of ADHD (IA and H/I) in the combined sample; fit statistics indicated a good fit (M2 = 241.08, CFI = 0.944, TLI = 0.920, RMSEA = 0.076, SRMR = 0.049). We assessed DIF iteratively using multigroup multidimensional IRT, comparing item functioning across language, on both polytomous (severity) and dichotomous (symptom count) scaling of the CASI.
Findings in the polytomous data indicated significant unsigned DIF at p < .05 with all seven non-anchor H/I items compared to two of the seven non-anchor Inattention items. The direction of effects suggest underreporting in H/I within the Spanish language sample. When CASI items were dichotomized, fewer were identified with significant DIF and the overall unsigned DIF effect size decreased from 1.270 (p = .018) to .613 (p = .106).
CASI scores are likely to underestimate true ADHD symptoms in Spanish (vs. English-speaking families, especially for H/I symptoms, which prior work finds less culturally equivalent than IA symptoms in Spanish vs English speaking samples (Gerdes et al., 2013). This effect attenuated when items were dichotomized as Symptom Present vs. Absent. However, this approach may not capture symptom severity in Spanish-speaking youth. This work highlights the importance of conducting cross-group comparisons with translated measures, and underscores the need for future work in a larger sample to examine whether and how ADHD H/I severity can be measured appropriately in Spanish speaking populations outside the U.S. The oral paper presentation will include related impairment data, as prior work indicates this may be a more appropriate assessment for Latine families.