Clock Drawing Test Performance in Rural Vietnam: Reliability and Concurrent Validity Relative to the Mini‐Mental State Examination in a Community‐Based Sample
Aki Yazawa, Thuy Thi Phuong Pham, Chau Que Nguyen, Aoi Ito, Danh Cong Phan, Masahiko Hachiya, Thanh Tien Hoang, Hung Thai Do, Tetsuya Mizoue, Yosuke InoueABSTRACT
Aim
Some cognitive screening tools can be difficult to administer in population‐based epidemiological studies because they are time‐consuming and rely on familiarity with formal testing. The Clock Drawing Test (CDT) is a brief, pragmatic tool requiring minimal verbal instruction, but its performance in low‐ and middle‐income countries remains unclear. We examined the reliability and performance characteristics of the CDT and its concurrent validity relative to the Mini‐Mental State Examination (MMSE) performance in rural Vietnam.
Methods
Data came from 1073 adults aged 55–65 years participating in the second wave (2023) of the Khánh Hòa Cardiovascular Study. The CDT was scored using the 6‐point Shulman method (lower scores indicate better performance). Inter‐rater reliability was assessed using intraclass correlation coefficients (ICC). Construct validity was examined via Spearman's correlations with MMSE scores. Known‐groups validity was tested across education levels. Concurrent validity relative to MMSE performance was evaluated by examining the discriminatory performance of the CDT using ROC analysis, sensitivity, specificity, predictive values, and agreement between CDT‐ and MMSE‐based classifications using Cohen's κ .
Results
Inter‐rater reliability between two raters was good (single‐measure ICC = 0.79), and CDT scores correlated negatively with MMSE ( ρ = −0.30, p < 0.001). CDT performance varied by education ( p < 0.001). ROC analysis showed modest discriminatory accuracy (AUC = 0.68); a cutoff ≥ 3 yielded sensitivity 67% and specificity 62%, with limited agreement between CDT and MMSE classifications ( κ = 0.13).
Conclusions
In rural Vietnam, the CDT demonstrated acceptable reliability and modest concurrent validity relative to MMSE performance. Agreement between CDT‐ and MMSE‐based classifications was limited, suggesting that the two measures should not be considered interchangeable. Although a high negative predictive value was observed in this community‐based sample, its implications for cognitive screening remain uncertain. Further validation against clinical diagnostic outcomes and in higher‐risk populations is warranted.