How accurate is the mini-cog test when used to assess dementia in general practice?

Background and rationale for review

In most parts of the world there are increasing numbers of older adults, and memory complaints and conditions such as Alzheimer's disease and other forms of dementia are becoming increasingly common as a result. Most individuals with memory difficulties will first seek out care or be identified in the healthcare system through their primary care health care providers, which may include family physicians or nurses. Therefore, there is a need for tools that could identify individuals who may have dementia or significant memory problems. These tools should also be able to rule out dementia in those individuals with memory complaints who do not have dementia or significant memory problems. Such tools in primary care must be relatively easy to use, quick to administer, and accurate so as to be feasible to use in primary care while at the same time not overdiagnose or underdiagnose dementia. The Mini-Cog, a brief cognitive screening tool, has been suggested as a possible screening test for dementia in primary care as it has been reported to be accurate and relatively easy to administer in primary care settings. The Mini-Cog consists of a memory task that involves recall of three words and an evaluation of a clock drawing task.

Study characteristics

We searched electronic databases for articles evaluating the Mini-Cog and this evidence is current as of January 2017. The purpose of our review was to compare the accuracy of the Mini-Cog for detecting dementia of any type in primary care settings when compared to in-depth evaluation conducted by dementia specialists. We included studies that evaluated individuals with any potential severity of dementia and regardless of whether previous cognitive testing had been completed prior to the Mini-Cog. Overall, our review identified four studies conducted in primary care settings that compared the accuracy of the Mini-Cog to detailed assessment of dementia by dementia specialists.

Quality of the evidence

Of the four studies included in the review, all except one study had limitations in how the Mini-Cog was evaluated, which may have led to an overestimation of the accuracy of the Mini-Cog in the remaining studies. Notably, the most problematic issue in study quality related to how participants were selected to participate in research studies, which may have further contributed to an overestimation of the accuracy of the Mini-Cog in most of the studies included in our review.

Key findings

The results of the highest-quality study Holsinger 2012 found that the Mini-Cog had a sensitivity of 76%, indicating that the Mini-Cog failed to detect up to 24% of individuals who have dementia (e.g. false negatives). In this same study, the specificity of the Mini-Cog was 73% indicating that up to 27% of individuals may be incorrectly identified as having dementia on the Mini-Cog when these individuals do not actually have an underlying dementia (e.g. false positives). We conclude that at the present time there is not enough evidence to support the routine use of the Mini-Cog as a screening test for dementia in primary care and additional studies are required before concluding that the Mini-Cog is useful in this setting.

Authors' conclusions: 

There is a limited number of studies evaluating the accuracy of the Mini-Cog for the diagnosis of dementia in primary care settings. Given the small number of studies, the wide range in estimates of the accuracy of the Mini-Cog, and methodological limitations identified in most of the studies, at the present time there is insufficient evidence to recommend that the Mini-Cog be used as a screening test for dementia in primary care. Further studies are required to determine the accuracy of Mini-Cog in primary care and whether this tool has sufficient diagnostic test accuracy to be useful as a screening test in this setting.

Read the full abstract...
Background: 

Alzheimer's disease and other forms of dementia are becoming increasingly common with the aging of most populations. The majority of individuals with dementia will first present for care and assessment in primary care settings. There is a need for brief dementia screening instruments that can accurately detect dementia in primary care settings. The Mini-Cog is a brief, cognitive screening test that is frequently used to evaluate cognition in older adults in various settings.

Objectives: 

To determine the accuracy of the Mini-Cog for detecting dementia in a primary care setting.

Search strategy: 

We searched the Cochrane Dementia and Cognitive Improvement Register of Diagnostic Test Accuracy Studies, MEDLINE, Embase and four other databases, initially to September 2012. Since then, four updates to the search were performed using the same search methods, and the most recent was January 2017. We used citation tracking (using the databases' ‘related articles’ feature, where available) as an additional search method and contacted authors of eligible studies for unpublished data.

Selection criteria: 

We only included studies that evaluated the Mini-Cog as an index test for the diagnosis of Alzheimer's disease dementia or related forms of dementia when compared to a reference standard using validated criteria for dementia. We only included studies that were conducted in primary care populations.

Data collection and analysis: 

We extracted and described information on the characteristics of the study participants and study setting. Using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) criteria we evaluated the quality of studies, and we assessed risk of bias and applicability of each study for each domain in QUADAS-2. Two review authors independently extracted information on the true positives, true negatives, false positives, and false negatives and entered the data into Review Manager 5 (RevMan 5). We then used RevMan 5 to determine the sensitivity, specificity, and 95% confidence intervals. We summarized the sensitivity and specificity of the Mini-Cog in the individual studies in forest plots and also plotted them in a receiver operating characteristic plot. We also created a 'Risk of bias' and applicability concerns graph to summarize information related to the quality of included studies.

Main results: 

There were a total of four studies that met our inclusion criteria, including a total of 1517 total participants. The sensitivity of the Mini-Cog varied between 0.76 to 1.00 in studies while the specificity varied between 0.27 to 0.85. The included studies displayed significant heterogeneity in both methodologies and clinical populations, which did not allow for a meta-analysis to be completed. Only one study (Holsinger 2012) was found to be at low risk of bias on all methodological domains. The results of this study reported that the sensitivity of the Mini-Cog was 0.76 and the specificity was 0.73. We found the quality of all other included studies to be low due to a high risk of bias with methodological limitations primarily in their selection of participants.