Psychometric validation of the Vietnamese Hospital Anxiety and Depression Scale (HADS) in adults with type 2 diabetes mellitus: a multi-site cross-sectional study

Anxiety and depression are common in type 2 diabetes mellitus (T2DM) and are associated with poorer self-management and clinical outcomes. This study evaluated the measurement properties of the Vietnamese Hospital Anxiety and Depression Scale (HADS) in adults with T2DM. In this multi-site cross-sectional study, 465 adults with T2DM from three clinics in Ho Chi Minh City completed the HADS and Patient Health Questionnaire-9 (PHQ-9). Internal consistency was estimated using ordinal reliability coefficients. Structural validity was evaluated using exploratory and confirmatory factor analyses, followed by bifactor modelling. Construct validity was examined through convergent and discriminant validity analyses, and clinic-level heterogeneity was evaluated using Kruskal-Wallis tests for scale-level comparisons and eta-squared η² estimates for item-level variance. Ordinal McDonald’s ωt was 0.765 (HADS-A) and 0.846 (HADS-D). Although parallel analysis suggested three factors, a two-factor solution was retained because the additional factor fragmented the depression items without representing a distinct construct. This structure was confirmed by confirmatory factor analysis only after incorporating two theory-driven residual correlations (RMSEA = 0.084, 90% CI [0.075, 0.094]; CFI = 0.922; TLI = 0.904; SRMR = 0.076). Bifactor analysis demonstrated a substantial general distress factor (ωh = 0.839; ECV = 0.608) while indicating meaningful residual variance within the anxiety and depression subscales. Discriminant validity was supported for HADS-D but not HADS-A (√AVE-A = 0.552 < r = 0.652). Both subscales correlated moderately with the PHQ-9. In an exploratory analysis of cross-site score comparability, three anhedonia-related depression items demonstrated comparatively larger differences across the three clinics (η² = 0.18–0.21), indicating caution when comparing or pooling scores across these settings. The Vietnamese HADS demonstrates acceptable structural validity and partial construct validity in adults with T2DM. The bifactor structure supports the total score as a broad, cross-sectional index of general psychological distress. However, diagnostic accuracy and criterion validity remain to be established. Therefore, the total score is currently supported strictly as a continuous indicator of emotional burden in research and routine care, rather than a formal screening tool. Type 2 diabetes affects nearly 600 million people worldwide and carries a substantial psychological and functional burden. Anxiety and depression are particularly common among people with diabetes, and both conditions make it harder for people to follow their treatment and keep their blood sugar levels in check. The Hospital Anxiety and Depression Scale (HADS) is one of the most widely used such tools in chronic disease settings, designed to avoid physical symptoms that might reflect illness rather than psychological distress. However, its measurement properties had not been formally evaluated in Vietnamese adults with type 2 diabetes, limiting its use in clinical practice and research. We tested the Vietnamese HADS in 465 adults attending three different diabetes outpatient clinics in Ho Chi Minh City. The questionnaire proved reliable, and its original two-part structure (one section for anxiety, one section for depression) was confirmed after minor statistical adjustments to account for wording overlap in the Vietnamese translation. The total score best captures overall emotional distress and should be the primary measure reported, while the individual subscale scores offer useful but more limited additional information. Responses to some depression-related questions varied notably across the three clinic sites, which suggests caution is needed when comparing scores across settings in multi-site studies. This study provides initial psychometric evidence for interpreting the Vietnamese HADS in adults with type 2 diabetes in Vietnam, while highlighting important limitations that require further evaluation.