Development of a dementia screening tool for Thai elderly: 5-item dementia screening tool (the revised version)

Abstract

This study aims to develop a dementia screening tool for Thai elderly and evaluate measurement properties of the 5-item dementia screening tool (The revised version), compared with the original 14-item dementia screening tool (Original version) and the diagnostic criteria of the physicians. The study sample consisted of 765 elderly aged 60 or over, who resided in Chachoengsao and Nakhon Ratchasima provinces. The research instruments included the original 14-item dementia screening tool, the revised 5-item version, the Thai version of the Mini-Mental State Examination (MMSE-Thai 2002), the Thai-translated version of the Montreal Cognitive Assessment (MoCA) test, and medical diagnoses based on DSM-5 criteria. Data were analyzed using descriptive statistics, the Kolmogorov-Smirnov test for data distribution, and criterion validity analysis.

The result showed that participants had an average age of 70.20 years (SD=6.83). It was found that 15.24% of the sample exhibited cognitive impairment, with 6.78% of these individuals being diagnosed with dementia and 87.29% showing signs of mild cognitive impairment. The measurement properties of the revised 5-item memory performance test at a cut-off score of 8 scores were evaluated, with a reliability coefficient of 0.838. The test's performance varied across different cognitive conditions. (1) Cognitive Impairment Group: The test demonstrated high sensitivity (92.5) but low specificity (8.83) and accuracy (27.79). The positive predictive value (PPV) was low (23.18), while the negative predictive value (NPV) was high (78.79). The area under the curve (AUC-ROC) was 0.502. (2) Mild Cognitive Impairment Group: The test also showed high sensitivity (82.95), with low specificity (13.24) and accuracy (29.28). The PPV was low (22.22), and the NPV was high (77.22). The AUC-ROC was 0.481. (3) Dementia Group: The test had very low sensitivity (9.09), but high specificity (95.59) and accuracy (75.69). The PPV was low (39.10), and the NPV was high (77.87). The AUC-ROC was 0.523.

In summary, the evaluation of the measurement properties of the revised 5-item dementia screening tool suggested that this tool may serve as an alternative screening method for dementia, particularly for individuals with cognitive impairment and those with mild cognitive deficits. The relatively high sensitivity supports its potential in this role. However, due to the limitations of other measurement properties, which have relatively low values, it is essential to carefully consider the appropriateness of the tool before its use. Further studies and refinements are recommended to improve the tool’s effectiveness in identifying individuals at risk for, or exhibiting, cognitive impairments.