Bayesian Estimation On Diagnostic Performance Of Dementia Screening Tests In The Absence Of A Gold Standard

Objectives: This study was conducted to estimate diagnostic performance of dementia screening tests including Mini-Mental State Examination (MMSE), Mini-Cog, Clock Drawing Test (CDT) and Ascertain Dementia 8 questionnaire (AD8) without a gold standard by Bayesian method. Methods: 2015 participants aged 65 years or more in eastern China were enrolled. 4 screening tests were administered and scored by specifically trained psychiatrists. The prior information of sensitivity, specificity of each screening tests are updated via Bayes’ theorem to a posterior distribution, using the survey data and the multinomial likelihood function. Results: The diagnostic characteristics of Mini-Cog, including sensitivity, specificity, PPV, NPV, especially the Youden index, performed well, even better than the combinations of several screening tests. Meanwhile, the Mini-Cog was quick and easy to administer, which needed only 2–4 minutes to complete compared with more than 10 minutes of MMSE, while having psychometric properties similar to the MMSE. Conclusion: Our study showed that Bayesian method was a suitable tool for evaluating dementia screening tests when a gold standard was not available. The Mini-Cog with excellent screening characteristics, could be considered to be used as a screening test to help screening patients for cognitive impairment and diagnose dementia early.