Screening for dementia is usually considered important, but only if accuracy of detection is sufficient and treatments are available and effective. In the National Dementia Strategy for England, one of the three main areas promoted was early diagnosis with acknowledgement that much of this role falls to primary care. The majority of dementia and pre-dementia cases in the community and in primary care remain undetected. One in three of those diagnosed remains unaware of their diagnosis. GPs in the UK are encouraged actively to look for people with dementia through annual screening as well an opportunistic testing of older people attending primary care with any significant health concern. The UK government has also encouraged case finding for dementia on acute admission to secondary care services using a dementia CQUIN (Commissioning for Quality and Innovation) which meant that overall, between 80% and 90% of patients aged 75 years and over were screened and assessed.
Screening for dementia is usually considered important, but only if accuracy of detection is sufficient and treatments are available and effective. In the National Dementia Strategy for England, one of the three main areas promoted was early diagnosis with acknowledgement that much of this role falls to primary care.1 The majority of dementia and pre-dementia cases in the community and in primary care remain undetected.2 One in three of those diagnosed remains unaware of their diagnosis.3 GPs in the UK are encouraged actively to look for people with dementia through annual screening as well as opportunistic testing of older people attending primary care with any significant health concern.4 The UK government has also encouraged case finding for dementia on acute admission to secondary care services using a dementia CQUIN (Commissioning for Quality and Innovation), which meant that overall, between 80% and 90% of patients aged 75 years and over were screened and assessed.5 Similarly, the Alzheimer’s Foundation of America has run a National Memory Screening Program, and cognitive assessments have formed part of the Medicare Annual Wellness Visit in the United States.6, 7
Without tools, clinical recognition of dementia is often low especially in cases of those who have mild dementia or mild cognitive impairment (MCI).8 Rarely do clinicians use standardized criteria, such as those provided by DSM-5, let alone advanced cognitive tests.9 Most GPs rely on their clinical judgement, occasionally enriched with a basic cognitive screening tool such as the Mini-Mental State Examination (MMSE).10 It is thought that systematic, routine use of a simple tool in primary care would prove cost-effective.11 Beyond the detection of early dementia, some hope to identify pre-dementia in the form of MCI, which affects 15% of adults over 75 and progresses to dementia at a rate of about 5% per year, depending on risk factors.12 That said, there is also a case against routine screening. In particular, inaccurate screening would be problematic and false positives (without a second stage screener) could create anxiety, excessive medical tests, and inappropriate treatment. Additionally, false negatives would give patients and families false reassurance and potentially delay essential treatment or support.
Mini-Mental State Examination
The MMSE was published in 1975 as a relatively simple practical method of grading cognitive impairment.13 Since then it has become the most commonly used cognitive screener in clinical practice, both for dementia and MCI.14 While the MMSE may never have been intended as a diagnostic (case-finding) tool, it has been extensively investigated as a diagnostic test of dementia and, to a lesser extent, examined as a diagnostic screen for MCI. Many are attracted by the brevity of the instrument, which typically takes a little over 5 minutes in healthy individuals, and the fact it used to be royalty free (although in 2001 copyright was acquired by Psychological Assessment Resources [www.minimental.com/] so that there is now a fee attached to its use). In clinical practice, the main applications of the MMSE are to help clinicians grade the severity of cognitive change and to help with cognitive screening, either by ruling out those without a cognitive disorder with as few false negatives as possible, or perhaps by pointing towards cases with suspected but unconfirmed dementia.
The MMSE has an internal structure of 20 individual tests covering 11 domains, including: orientation, registration, attention or calculation (serial sevens or spelling), recall, naming, repetition, comprehension (verbal and written), writing, and construction. However, it only tests for recall of three items whereas more modern tools use five or more item recall. Internal consistency appears to be moderate with Cronbach alpha scores reported between 0.6 and 0.9.15 Test-retest reliability has been examined in several studies, and in those where re-examination took place within 24 hours, reliability by Pearson correlation was usually above 0.85. Using RASCH analysis (named after Georg Rasch), it is possible to grade the completion difficulty of each item on the MMSE. Relatively difficult items are: the recall of three words, citing the correct date, copying the pentagon design, spelling ‘world’ backwards, and completing serial sevens. Conversely, relatively simple items are: naming the correct country, registering three words, following the command, and naming an object. Acceptability is generally high, but it falls in those with definite or suspected impairment, who may be reluctant to expose perceived deficits. All questions are designed to be asked in the order listed, with omissions scored as errors, giving a maximum score of 30. However, there is some ambiguity in several items, leading to the structured MMSE.16
Approximately 250 validation studies have been published using the MMSE as the principal tool, or as a comparator tool, but many are underpowered and/or lack an adequate criterion standard, and hence can give a misleading impression of accuracy. Nevertheless, this extensive evidence base means scores are fairly well understood by health professionals and can be adjusted on the basis of normative population data. For example, Crum et al. tested an extensive group of 18,056 participants in the US Epidemiologic Catchment Area study and presented distributions by age and educational levels.17 Some groups have provided norms for each item on the MMSE by age group, yet there remains uncertainty regarding optimal cut-off threshold for each condition under study. A cut-off of <24 in persons with at least 8 years of education was recommended as significant by Folstein and colleagues.8 13 Some individuals with MCI or early dementia and a background of extensive education may experience a ceiling effect with the MMSE. In other words, the MMSE may lack subtle tests necessary to detect early cognitive changes, particularly regarding recall.
The MMSE has been extensively investigated as a diagnostic test for current dementia, either on its own or against comparison scales. O’Connor et al. conducted one of the first adequately powered studies of the MMSE, using a cut-off <24 in 586 who received a Cambridge Examination for Mental Disorders of the Elderly (CAMDEX)/Cambridge Cognition Examination (CAMCOG) interview as a gold standard.18 They found that the sensitivity of the MMSE was 86% and specificity 92%. In 2013, Linn et al. found a pooled sensitivity estimate of 88.3% (95% confidence interval [CI] 81.3–92.9) and a pooled specificity estimate of 86.2% (95%CI 81.8–89.7) but only in a small sample of studies.19 In 2009, Mitchell undertook the first meta-analysis of the MMSE in dementia; however, sample size was limited and this meta-analysis was updated and revised in 2013 with 45 studies.20, 21 This included community studies, primary care studies, and, most commonly, studies in specialist settings where the prevalence of dementia is relatively high. The prevalence of each condition in each setting strongly influences the performance of a test. High-prevalence settings favour case-finding with few false positives, but at the expense of false negatives. Low-prevalence settings favour screening with few false negatives but at the expense of frequent false positives. The most recent meta-analysis was published in 2015 and included 108 MMSE studies involving 36,080 subjects (of whom 10,263 had dementia).22 The most common cut-off values to define participants with dementia were <23 and <24, and the prevalence was 28%. Taking all studies to date (see Table 11.1), the best estimate (using a meta-analytic bivariate random-effects model) is that the MMSE has a sensitivity of 81.3% (CI 80.6–82.1%) and specificity of 89.1% (CI 88.7–89.5%). Positive predictive value (PPV) was calculated as 74.8% (CI 74.0–75.6%) and the negative predictive value (NPV) was 92.3% (CI 92.0–92.6%). The positive clinical utility index was 0.608 (95% CI 0.598–0.618), suggesting that this can be categorized as ‘fair’ for case-finding; negative clinical utility index was 0.822 (CI 0.819–0.825), suggesting that this can be categorized as ‘excellent’ for screening. However, performance deteriorates when clinicians use the MMSE to look for dementia in a combined group of healthy people mixed with those with MCI.
|Purpose of test||Sensitivity||Specificity||PPV||NPV||Overall correct||LR+||LR–||CUI+||CUI–|
|Dementia vs healthy controls|
|Detection of dementia||81.3% (80.6–82.1%)||89.1% (88.7–89.5%)||74.8% (74.0–75.6%)||92.3% (92.0–92.6%)||86.9% (86.5–87.2%)||7.45 (7.19–7.73)||0.21 (0.20–0.22)|
|Dementia vs mixed MCI/healthy|
|Detection of dementia||71.6% (69.8–73.4%)||93.5% (92.8–94.2%)||85.1% (83.5–86.7%)||86.4% (85.4–87.3%)||86.0 (85.2–86.8)||11.01 (9.863–12,33)||0.30 (0.28–0.32)|
|Mild cognitive impairment vs healthy controls|
|Detection of MCI vs HC|
HC = healthy controls; MCI = mild cognitive impairment; PPV = positive predictive value; NPV = negative predictive value; LR+ (likelihood ratio+) = sensitivity/(1 – specificity); LR– (likelihood ratio–) = (1 – sensitivity)/specificity; CUI+ (Clinical Utility Index +) = sensitivity × PPV; Clinical Utility Index – (Clinical Utility Index –) = specificity × NPV.
A Cochrane review specifically examined the merits of the MMSE in people aged 65 years and over in community and primary care settings.23 The authors conducted a meta-analysis of 28 studies in the community (44 articles) and six studies in primary care (eight articles). In the community, the pooled accuracy at a cut point of 24 (15 studies) showed a pooled sensitivity of 0.85 (95% CI 0.74–0.92) and a specificity of 0.90 (95% CI 0.82–0.95); at a cut point of 25 (10 studies) its sensitivity was 0.87 (95% CI 0.78–0.93), whereas specificity was 0.82 (95% CI 0.65–0.92). The authors estimated that based on these results, one would expect 85% of people with dementia to be correctly identified with the MMSE, while 15% would be wrongly classified as not having dementia; 90% of those tested would be correctly identified as not having dementia while 10% would be false positives and might be referred for further testing. In other words, although the MMSE can be used as part of a diagnostic evaluation for dementia, it should not be used in isolation to confirm or exclude disease.
One critical question is whether the MMSE retains sufficient accuracy when screening for early stages of dementia. People with early dementia are particularly at risk of being overlooked.24 Provisional evidence from three studies suggests a small reduction in accuracy when attempting to detect those with mild dementia. For example, in a specialist hospital or memory clinic, Heinik et al. found that the area under the receiver operating characteristic (ROC) curve was 0.96 for all dementias, but 0.89 for very mild dementia;25 similarly, Meulen and colleagues found that the area under the ROC curve was 0.95 for all dementias, but only 0.87 for mild dementia.26 However, it should be noted that a cut-off threshold higher than ≤23 is recommended when looking for mild dementia. Yoshida et al. found a 95% sensitivity and an 83% specificity looking for mild dementia in a Japanese memory clinic at a threshold of ≤28, which would give ‘good’ clinical utility for screening (clinical utility index + = 0.789) and case-finding (clinical utility index – = 0.786).27 At a lower threshold of ≤25, sensitivity fell to 76% but specificity increased to 97%, which would also have ‘good’ clinical utility for screening (clinical utility index + = 0.800) and case-finding (clinical utility index – = 0.727). In a sub-analysis of 88 people with mild Alzheimer’s scoring >20 on the MMSE, Kalbe and colleagues found that the MMSE had a sensitivity of 92% and a specificity of 86% (PPV 85.2%, NPV 92.2%), which again would imply ‘good’ clinical utility for screening (clinical utility index + = 0.781) and case-finding (clinical utility index – = 0.796).28 Regarding diagnosis of mild dementia in primary care, Kilada and colleagues found adjustment of the MMSE cut-off to ≤27 was required. Grober et al. examined the value of MMSE in 317 primary care attendees with mild dementia (Clinical Dementia Rating [CDR] of 1.0 and 0.5 but without MCI). In this study, at a cut-off of ≤23 sensitivity was 53% and specificity 90% (PPV 52.7%, NPV 90.1), but at a cut-off of ≤26 sensitivity was 73% and specificity 73% (PPV 36.0%, NPV 92.7%), suggesting only ‘fair’ clinical utility.29
MCI is not the only pre-dementia condition, but it is the most studied. In 2015, a meta-analysis found 21 qualifying studies with a sensitivity estimate of 0.62 (95% Cl, 0.52–0.71) and specificity of 0.87 (95% Cl, 0.80–0.92).15 Mitchell re-examined this data in 2016 and found 40 relevant studies.30 In 2018, the most recent meta-analysis expanded this to 46 qualifying studies.31 Most used less robust cross-sectional definitions of MCI, rather than longitudinal. The majority used the Mayo clinic criteria suggested by Petersen and colleagues,32 and recruited from memory clinics or secondary care, with only a handful recruited directly from the community. Samples were not matched but recruited from convenience samples. Thus, the mean age of those with MCI was 73.2 years and healthy controls 71.0 years. The proportion of females in MCI studies was 44% and in controls 46.9%. Regarding education, the mean number of years of education in people with MCI was 9.79 versus 9.64 in controls. Perhaps the major question regards cut-off thresholds. In terms of cut-off scores, 12 studies examined <29, 9 studies <28, 17 studies <27, and 9 studies <26. After weighting, Mitchell found that the meta-analytic sensitivity was 59.7% (58.6–60.7%) and specificity was 80.2% (79.4–81.0%). PPV was 72.1% (71.1–73.2%) and NPV 69.9% (69.0–70.7%).30 This data is summarized in Table 11.1. Breton et al. found slightly lower accuracy rates with a sensitivity of 66.4% (95% CI 60.5–71.8%) and specificity of 73.5% (95% CI 68.6–77.8%).31 The positive clinical utility index was 0.431%, suggesting that this can be categorized as ‘poor’ for case-finding MCI (95% CI 0.418–0.444) and negative clinical utility index was 0.561 (95% CI 0.553–0.568), suggesting that this can be categorized as ‘fair’ for screening (ruling out) those without MCI.
Given the limitations of the MMSE diagnostically, it would be surprising but not impossible that the MMSE might be a valuable risk prediction tool. A Cochrane review from Arevalo-Rodriguez and colleagues examined whether the MMSE can be used to predict (rather than diagnose) dementia.33 They included 11 studies comprising 1,569 people with MCI who were followed for conversion to dementia (n = 4), AD (n = 8), or VaD (n = 1). Arevalo-Rodriguez established the diagnosis of MCI not just using conventional Petersen and revised Petersen criteria, but also using Matthews 2008 criteria, and using the CDR scale score of 0.5, criteria which are fairly broad. It should be noted that using the QUANDAS2 appraisal tool,34 all 11 studies had a high risk of bias in at least one domain. Looking in more detail, most studies came from samples of older people in memory clinics. Various thresholds were used to define a positive MMSE (≤21, ≤26, ≤28, ≤29), and follow-up times ranged from 15 months to seven years; over this time 36% on average developed dementia.
Results suggested that the accuracy of a baseline MMSE showed a wide range of sensitivities (23–76%) and specificities (40–94%). Overall, the authors found that at the median specificity of 88%, the sensitivity was only 40% for prediction of dementia and 54% for prediction of AD.33 They calculated that in a hypothetical cohort of 100 MCI patients with a 36% incidence of dementia, the number of missed cases (false negatives) would be 18 patients, while eight MCI patients would be overdiagnosed (false positives). Further modelling of the accuracy can be generated from a pre-test/post-test Baysian graph, calculated in a commentary by Mitchell.35 This showed that the predictive accuracy of using the MMSE to spot later dementia (or AD) at a prevalence of 36%, would be 45% for dementia and 52% for AD when the test is positive (PPV); and 86% for dementia and 89% for AD when the test is negative (NPV).