A Hierarchical Bayesian Latent Class Model for the Diagnostic Performance of Mini-Mental State Examination and Montreal Cognitive Assessment in Screening Mild Cognitive Impairment Due to Alzheimer's Disease.

BACKGROUND

The Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) are low costing and noninvasive neuropsychological tests in screening Mild Cognitive Impairment (MCI) due to Alzheimer's disease (AD). There is no consensus on which test performs better in detecting MCI due to AD based on the different imperfect reference standards. Therefore, we conducted a meta-analysis to assess the diagnostic performance of MMSE and MoCA for screening MCI due to AD in the absence of a gold standard.

METHODS

Six electronic databases were searched for relevant studies until April, 2022. A hierarchical Bayesian latent class model was used to estimate the pooled sensitivity and specificity of MoCA and MMSE in the absence of a gold standard.

RESULTS

90 eligible studies covering 21273 individuals for MMSE, 26631 individuals for MoCA were included in this meta-analysis. The pooled sensitivity was 0.71(95%CI: 0.67-0.74) for MMSE and 0.85(95%CI: 0.83-0.88) for MoCA, while the pooled specificity was 0.71(95%CI: 0.68-0.74) for MMSE and 0.79(95%CI: 0.76-0.81) for MoCA. MoCA was useful to "rule in" and "rule out" the diagnosis of MCI due to AD with higher positive likelihood ratio (4.07; 95%CI: 3.60-4.62) and lower negative likelihood ratio (0.18; 95%CI: 0.16-0.22). Moreover, the diagnostic odds ratio of MoCA was 22.08(95%CI: 17.24-28.29), which showed significantly favorable diagnostic performance.

CONCLUSIONS

It suggests that MoCA has greater diagnostic performance than MMSE for differentiating MCI due to AD when the gold standard is absent. However, these results should be taken with caution given the heterogeneity observed.