Investigating age-related differences in responses to the screening items for internalising disorders in three national surveys

Abstract

Background: Epidemiological studies typically report lower prevalence of mental disorders among older adults relative to middle-aged and young adults. A possible explanation is that age-related bias in the screening items of diagnostic instruments leads to older adults being differentially screened out of the full assessment. This study investigated potential age-related bias in screening items for internalising disorders in three epidemiological surveys. Method: Measurement invariance was estimated for the internalising disorder screening items in the 2007 and 1997 Australian National Survey of Mental Health and Wellbeing, and the 2001-2002 National Epidemiologic Survey on Alcohol and Related Conditions. These surveys assessed mental disorders using the Composite International Diagnostic Interview (CIDI) and the Alcohol Use Disorder and Associated Disabilities Interview Schedule–DSM-IV Version (AUDADIS-IV). A series of multi-group confirmatory factor analyses (CFA) were performed for each survey across older (65-85 years), middle (35-64 years) and young (16-34 years) adults. Results: Differences between successive CFA models for each survey were negligible, indicating measurement invariance across age groups for the CIDI and AUDADIS-IV screening items. Limitations: The number of items and symptoms representing internalising pathology differed between surveys. The samples excluded people in aged-care institutions. Conclusions: While findings do not rule out that other measurement errors may be present (e.g., age-related bias in the remaining items), these results support the validity of the screening items in the CIDI and AUDADIS-IV. Low prevalence estimates of internalising disorders in older adults are unlikely attributable to differences in response patterns to screening items.