DOI: 10.11648/j.ajim.20261405.11 ISSN: 2330-4324

Diagnostic Accuracy and Feasibility of Brief Cognitive Screening in Adults with Type 2 Diabetes: A Systematic Review

Sharmila Bishowkarma, Gagan Karki, Ravi Rokaya, Bibek Shrestha
Show PDF Cite
Background Cognitive impairment and dementia are more common in people with type 2 diabetes yet often go undetected. Brief cognitive screening tests could enable earlier identification in routine care, but their diagnostic accuracy and practicality in this population remain uncertain. We systematically reviewed the diagnostic accuracy and feasibility of brief cognitive screening in this population. Methods The review was registered (PROSPERO CRD420261441730) and reported following the PRISMA-DTA guideline. We searched MEDLINE (via PubMed), Embase, CENTRAL and PsycINFO from January 2000 to April 2026, restricted to English-language reports, for studies evaluating a brief cognitive screen against an acceptable reference standard clinical diagnostic criteria or a comprehensive neuropsychological battery or reporting feasibility. Two reviewers independently screened, extracted and appraised studies (QUADAS-2 for accuracy; the Mixed Methods Appraisal Tool for feasibility). Sensitivity and specificity with Wilson 95% confidence intervals were synthesised structurally using paired forest and summary receiver-operating-characteristic plots, and the certainty of evidence was assessed with GRADE. Results Six of 557 records were eligible: four diagnostic-accuracy studies (802 adults) and two feasibility studies. All accuracy studies evaluated the Montreal Cognitive Assessment (MoCA) and one also evaluated the Mini-Mental State Examination (MMSE). At the conventional cut-off (<26) the MoCA was highly sensitive but poorly specific (sensitivity 88-94%, specificity 29-51%); lower, diabetes-specific thresholds improved specificity at the expense of sensitivity (for example 49%/91% at 20/21, and 81%/62% at an optimal 24/25). The single MMSE study showed the reverse profile (66%/96%). Risk of bias was appreciable in every study (incorporation bias, spectrum effects, within-sample thresholds), and a meta-analysis was not appropriate. Two studies supported nurse-delivered screening in primary care (MoCA; and the Mini-Cog, mean 118 seconds, 100% completion). Overall certainty of evidence was low to very low. Conclusions The MoCA is the only brief cognitive screen with replicated accuracy evidence in type 2 diabetes: sensitive but over-referring at the standard cut-off, with a lower threshold better balancing accuracy. Non-specialist screening appears feasible, but the accuracy evidence is limited and at material risk of bias. Diabetes-specific validation against clinical or neuropsychological reference standards is the priority for future research.

More from our Archive