Which Cutoff Value of the Montreal Cognitive Assessment Should Be Used for Post-stroke Cognitive Impairment? A Systematic Review and Meta-Analysis on Diagnostic Test Accuracy

International Journal of Stroke

Wei, X., Ma, Y., et al. (2023).

International Journal of Stroke, 18(8), 908-916.

This systematic review and meta-analysis investigates the accuracy of the Montreal Cognitive Assessment (MoCA) in detecting post-stroke cognitive impairment.

National Nature Science Foundation of China; Natural Science Foundation of Gansu Province (China); Fundamental Research Funds for the Central Universities (China); Science and Technology Bureau Project of Lanzhou Chengguan District (China)

From database inception to March 23, 2023

Cohort, cross-sectional, or case-control studies

24

The optimal cutoff on the MoCA for detecting post-stroke cognitive impairment was 21/22. This cut-off value yielded the following psychometrics across all included studies:<br /> <ul> <li>pooled sensitivity = 0.78;</li> <li>pooled specificity = 0.86;</li> <li>pooled positive likelihood ratio (LR+) = 5.40;</li> <li>pooled negative likelihood ratio (LR-) = 0.25;</li> <li>pooled diagnostic odds ratio (OR) = 21.0; and</li> <li>pooled area under the curve (AUC) = 0.89.</li> </ul> Limitations to these findings included heterogeneity between studies and an overall lack of data regarding specific cutoff values.

The authors of this review conducted a sub-group analysis to determine the optional cut-off scores based on geographical region. <br /><br />Within Asian countries, the optimal cutoff was 22/23, which yielded the following psychometrics: <ul> <li>pooled sensitivity = 0.84;</li> <li>pooled specificity = 0.86;</li> <li>pooled LR+ = 5.8;</li> <li>pooled LR- = 0.19;</li> <li>pooled diagnostic OR = 31; and</li> <li>pooled AUC = 0.92.</li> </ul> In non-Asian countries, the optimal cutoff was 22/23, which yielded the following psychometrics:<br /> <ul> <li>pooled sensitivity = 0.65;</li> <li>pooled specificity = 0.85;</li> <li>pooled LR+ = 4.4;</li> <li>pooled LR- = 0.41;</li> <li>pooled diagnostic OR = 11; and</li> <li>pooled AUC = 0.80.</li> </ul> <span style="color: #333333;">Limitations of this review included a limited numbers of studies within each subgroup, heterogeneity between studies, and an overall lack of data regarding specific cutoff values.</span>

The authors of this review conducted a sub-group analysis to determine the optional cut-off scores based on stroke type. <br /><br />For ischemic stroke (including transient-ischemic attack), the optimal cutoff was 20/21, which yielded the following psychometrics:<br /> <ul> <li>pooled sensitivity = 0.73;</li> <li>pooled specificity = 0.93;</li> <li>pooled LR+ = 9.8;</li> <li>pooled LR- = 0.29;</li> <li>pooled diagnostic OR = 34; and</li> <li>pooled AOC = 0.91.</li> </ul> For ischemic and hemorrhagic stroke, the optimal cutoff was 22/23, which yielded the following psychometrics: <ul> <li>pooled sensitivity = 0.82;</li> <li>pooled specificity = 0.72;</li> <li>pooled LR+ = 3.0;</li> <li>pooled LR- = 0.25;</li> <li>pooled diagnostic OR = 12; and</li> <li>pooled AUC = 0.85.</li> </ul> <span style="color: #333333;">Limitations of this review included a limited numbers of studies within each subgroup, heterogeneity between studies, and an overall lack of data regarding specific cutoff values..</span>

The authors of this review conducted a sub-group analysis to determine the optional cut-off scores based on stroke chronicity. <br /><br />For the acute/subacute phase, the optimal cutoff was 21/22, which resulted in the following psychometrics:<br /> <ul> <li>pooled sensitivity = 0.83;</li> <li>pooled specificity = 0.78;</li> <li>pooled LR+ = 3.8;</li> <li>pooled NR- = 0.22;</li> <li>pooled diagnostic OR = 17; and</li> <li>pooled AUC = 0.87.</li> </ul> For the non-acute phase (i.e., any time greater than 4 weeks), the optimal cutoff was 22/23, which resulted in the following psychometrics: <br /> <ul> <li><span style="color: #333333;">pooled sensitivity = 0.76;</span></li> <li><span style="color: #333333;"> pooled specificity = 0.90;</span></li> <li><span style="color: #333333;"> pooled LR+ = 8.0;</span></li> <li><span style="color: #333333;"> pooled LR- = 0.27;</span></li> <li><span style="color: #333333;"> pooled diagnostic OR = 30; and</span></li> <li><span style="color: #333333;"> pooled AUC = 0.91.</span></li> </ul> Limitations of this review included a limited numbers of studies within each subgroup, heterogeneity between studies, and an overall lack of data regarding specific cutoff values..