Abstract
Objective: To estimate the prevalence and examine the course of mild cognitive impairment (MCI), amnestic type, using current criteria, within a representative community sample.
Methods: Retroactive application of MCI criteria to data collected during a prospective epidemiologic study was performed. The subjects were drawn from voter registration lists, composing a cohort of 1,248 individuals with mean age of 74.6 (5.3) years, who were nondemented at entry and who were assessed biennially over 10 years of follow-up. The Petersen amnestic MCI criteria were operationalized as 1) impaired memory: Word List Delayed Recall score of <1 SD below mean; 2) normal mental status: Mini-Mental State Examination score of 25+; 3) normal daily functioning: no instrumental impairments; 4) memory complaint: subjective response to standardized question; 5) not demented: Clinical Dementia Rating Scale score of <1.
Results: At the five assessments, amnestic MCI criteria were met by 2.9 to 4.0% of the cohort. Of 40 persons with MCI at the first assessment, 11 (27%) developed dementia over the next 10 years. Over each 2-year interval, MCI persons showed increased risk of dementing (odds ratio = 3.9, 95% CI = 2.1 to 7.2); 11.1 to 16.7% progressed to Alzheimer disease and 0 to 5.0% progressed to other dementias. Over the same intervals, 11.1 to 21.2% of those with MCI remained MCI; of 33.3 to 55.6% who no longer had MCI, half had reverted to normal.
Conclusions: In this community-based sample, 3 to 4% of nondemented persons met MCI operational criteria; despite increased risk of progressing to dementia, a substantial proportion also remained stable or reverted to normal during follow-up. Amnestic MCI as currently defined is a high-risk but unstable and heterogeneous group.
Mild cognitive impairment (MCI) is one of several terms describing “a cognitive state intermediate between normal aging and dementia,”1 clinically suggesting a risk or prodromal state for Alzheimer disease (AD) and perhaps other dementias. Although conceptually the term might include a broad range of cognitive impairments short of dementia, Petersen et al.2 defined MCI more specifically as a condition characterized by the presence of isolated memory impairment beyond that expected for age and education, in the absence of dementia. Persons meeting this amnestic definition of MCI are then presumed to be at increased risk of progressing (or “converting”) to dementia.3 This concept has been challenged both by others whose studies suggest that “MCI” is, in fact, early or incipient AD4 and by those who find that MCI is an unstable condition with poor predictive validity for AD.5 Some argue that MCI cannot be a diagnostic entity and that it appears to increase risk not because it creates a predisposition for AD, but because 20% of those with MCI already have AD.6 Nevertheless, a growing number of research groups, using varying definitions and modifications of the original criteria, are already undertaking intervention trials in persons with MCI. However, a sound epidemiologic basis for these definitions and trials is notable by its absence.
Clinical criteria for amnestic MCI as proposed by the Mayo Clinic group include 1) memory complaint, 2) abnormal scores on memory tests, 3) normal general mental status, 4) normal daily functioning, and 5) absence of dementia.2. We report here the results of the post-hoc application of these published criteria to data collected prospectively over 10 years in a well-characterized community-based cohort participating in an epidemiologic study of dementia. Our goal was to estimate the prevalence and incidence of MCI and examine its course and outcomes, exploring the effects of varying case definitions of MCI.
Methods.
Study site and population.
As described previously,7 the Monongahela Valley Independent Elders Survey (MoVIES) was conducted in the largely rural mid-Monongahela Valley of southwestern Pennsylvania. A 1:13 random sample was drawn from the age strata 65 to 74 and 75+ years from the voter registration lists for the selected communities. Entry criteria were age 65+ years, living in the community at the time of recruitment, fluent in English, with at least sixth grade education. The random sample yielded 1,422 subjects with a response rate of 59%. A further 259 volunteers who met the same entry criteria were also enrolled, for a total original cohort of 1,681 participants aged 65+ years recruited during 1987 to 1989.8 At study entry (wave 1), the MoVIES cohort had a mean (SD) age of 72.9 (5.9) and was 57.8% female and 97.1% white. Their median educational level was high school graduate, and 56.6% had high school (graduate) or higher education. Survivors were followed through successive biennial data collection waves until 2001. Most of the attrition between successive waves was due to death (average 9.9%) and less for other reasons such as dropout and relocation (average 2.8%).
Biennial screening and risk factor assessment.
At study entry and approximately every 2 years thereafter, all participants underwent a standardized assessment described previously.7,9⇓ Included were a cognitive test battery incorporating the Mini-Mental State Examination (MMSE)10 (used as the basis for the MCI criterion of “normal general mental status” in the analyses reported here) and the neuropsychological panel of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD).11 The Word List Delayed Recall component of the CERAD panel served as the basis for the MCI criterion of “abnormal memory test scores.” The assessment also included standardized questions about subjectively experienced memory functioning (the basis of “memory complaint” for MCI) and the Older Americans Resources and Services (OARS) Scale12 for ability to perform instrumental activities of daily living (IADLs) (the basis for “normal daily functioning”).
Additional assessments.
Depression was assessed using a modified version of the Center for Epidemiologic Studies–Depression (CES-D) (mCES-D) Scale.13,14⇓ In this interviewer-administered version of the scale, the range of scores is 0 to 20, with higher scores indicating greater depression. We used a threshold score of ≥5 on the mCES-D to reflect significant depressive symptoms, as reported previously.14–16⇓⇓ Number of regularly taken prescription drugs was used as a measure of overall morbidity.15,16⇓ The APOE-4 genotype was assessed using PCR; the presence of at least one E4 allele was treated as APOE-4 positive.17,18⇓
Selection for diagnostic assessment.
Based on screening cognitive test scores at each wave, the following subgroups of participants were selected for a detailed standardized clinical evaluation: 1) those meeting criteria for “cognitive impairment,” that is, those with test scores at or below the 10th percentile of the cohort on the MMSE or on at least one memory test and one other test8; 2) those meeting criteria for “cognitive decline,” that is, those whose test scores had declined from a previous wave by amounts greater than or equal to the 95th percentile of decline experienced by the cohort; and 3) a group of “cognitively unimpaired” control subjects with test scores above the cognitive impairment threshold at study entry, matched on age and sex to persons with dementia at study entry. Those selected for diagnostic assessment at any wave in any of the above subgroups were automatically selected for reassessment at all subsequent waves.7
Diagnostic assessment.
As described in detail earlier,7 the clinical evaluation for diagnostic assessment was derived from the assessment protocols of CERAD11 and the University of Pittsburgh Alzheimer’s Disease Research Center19 and were conducted blind to the screening results. Subsequently, all available information was reviewed for a consensus-based final diagnosis, including the presence or absence of dementia according to the Diagnostic and Statistical Manual of Mental Disorders, 3rd ed. rev. (DSM-III-R),20 the stage of dementia according to the Clinical Dementia Rating (CDR) Scale,21 and the presence of AD according to the clinical criteria proposed by the Work Group of the National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association.22 We have previously reported8 the sensitivity, specificity, and predictive value of our screening criteria for dementia according to the CDR; based on the low false-negative rate at baseline, those not selected for clinical assessment are assumed to have CDR = 0. Based on all available information about the emergence of cognitive and functional decline, a date of onset was estimated for all cases of dementia, as reported previously.7 As noted, participants were assessed in a series of biennial data collection waves starting in 1987. Data from waves 2 (1991 to 1993) to wave 6 (1998 to 2001) were used for these analyses as several variables relevant to the current analyses were not collected at study entry (wave 1).
Post-hoc application of criteria for MCI.
The original criteria for amnestic MCI proposed by the Mayo group2 were operationalized as follows for application to MoVIES data: 1) Memory complaint: a report by the participant of memory being fair or poor (compared with good or excellent) in response to a standard question asking for a self-rating of memory. Note that this response was not a spontaneous complaint and was based only on subjective report, as not all participants in this community-based survey had informants. 2) Abnormal scores on memory tests: scores worse than 1 SD below the mean for the MoVIES cohort (excluding those with CDR of ≥0.5) at baseline on Delayed Recall of the 10-item CERAD Word List. CERAD did not include a test of nonverbal memory. For comparison, we also calculated impaired memory and the prevalence of MCI using the threshold of 1.5 SD below the mean, as proposed by the Mayo group.2 3) Normal general mental status: scores of >24/30 on the MMSE.10 In addition, we also examined a more stringent definition requiring all nonmemory tests to be scored better than 1 SD below the mean. 4) Normal daily functioning: self-report of independent functioning on all instrumental activities of daily living (IADLs) on the OARS Scale.12 5) Absence of dementia: CDR score of <1.
Statistical methods.
From data collected at each wave (2 to 6), we calculated the frequencies and proportions of participants who met each of the operational criteria for MCI as listed above and the prevalence of MCI using these criteria. We restricted each analysis to those subjects with complete data on all variables related to that analysis. The percentage of subjects with missing data on different variables was very low, ranging from 0.2 to 3.0%. The single outlier was APOE genotype, which was missing in 33.9% of the wave 2 (1989 to 1991) cohort; genotyping was carried out between 1994 and 1996, after several members of the original (wave 1) cohort had already been lost to follow-up (primarily from mortality). In preliminary analysis, the missingness of the APOE variable did not depend on MCI status conditional on other covariates, indicating that it was a covariate-dependent missing process.23 Therefore, valid inferences can be drawn from generalized estimating equations (GEEs)24 based on cases with complete data.25
At each wave 2 through 6, we then calculated the frequencies and proportions of participants with MCI, compared with unimpaired participants, whose status at the subsequent wave was one of the following: AD, other dementia, still MCI, no longer MCI (three subgroups: reverted to normal, worsened but not demented, and mixed), and deceased. A subject with MCI could worsen, for example, by dropping his or her MMSE score to below 25 or reporting at least one IADL impairment, without meeting independent criteria for dementia (CDR = 1); a subject could also have a mixed outcome by improving on one criterion and worsening on another, without meeting criteria for dementia.
Combining data from all waves, GEE analyses24 were implemented using PROC GENMOD in SAS version 8 (Cary, NC).26 We used unstructured correlation patterns to specify correlations within subjects, with logistic link for binary outcomes. Significance was defined as a p value of <0.05. The following associations were examined:
1) The association between subjective memory loss (memory complaint), as reported by the subject, and objectively measured decline in memory (Delayed Recall) between that wave and the preceding wave, adjusting for age, sex, education, and score at the previous wave (to control for floor effects). 2) The cross-sectional associations of MCI with age, sex, education, number of prescription drugs, depression, and APOE-4 genotype. 3) The risk of mortality during the follow-up period among those with MCI compared with the unimpaired, adjusting for age, sex, and education. 4) The risk for persons with MCI, compared with normal (unimpaired) participants, of progressing to AD over 2 years (by the next wave), adjusting for age, sex, and education. 5) The factors associated with progression to AD (vs all other outcomes) using all available 2-year transitions among participants with MCI. Age, sex, education, number of prescription drugs, depression (mCES-D score 5+), and APOE-4 genotype (at least one E4 allele) were examined.
All models were fit again including a variable to control for baseline recruitment status (random vs volunteer sample), with no change in results. Therefore, we report results without inclusion of this variable.
Results.
Table 1 shows the sample size and relevant demographic characteristics of the cohort at each wave 2 through 6.
Table 1 MoVIES: nondemented cohort size, demographic characteristics, and prevalence of MCI at five consecutive data collection waves
Table 2 shows the prevalence of the individual “criteria” for MCI items 1 through 4 among those who met criterion 5, that is, with CDR score of <1. With use of the example of wave 2, 14.7% had abnormal memory (scores of ≥1 SD below the mean on Word List Delayed Recall), 89.2% had normal mental status (MMSE score of >24), 72.8% had normal daily functioning (able to perform all IADLs on the OARS Scale), and 36.7% had a subjective complaint of memory loss. Upon combination of all four of the above in an objective approximation of Petersen’s criteria, 3.2% of the cohort at wave 2 met these criteria for MCI.
Table 2 Individual and combined criteria for MCI and related conditions (MoVIES waves 2–6)
Subjective memory complaint.
Eliminating the requirement of memory complaint as a criterion changed the prevalence of MCI from 3.2 to 6.3% at Wave 2 and comparably in other waves (see table 2). We examined the association of subjective memory complaint (in response to a standard question) to observed decline in memory (Word List Delayed Recall) scores since the previous data collection wave 2 years earlier, using GEE adjusting for age, sex, education, and previous score. Decline in scores was associated with memory complaint: (coefficient 0.31, p < 0.001); that is, those with memory complaints had a decline 0.31 larger than those without these complaints. However, across waves, 59.9 to 69.4% of subjects with decline in memory scores between successive waves did not have a memory complaint, whereas 19.8 to 34.6% of subjects with memory complaints did not show a decline in scores from the previous wave.
Cross-sectional associations with MCI.
With use of GEE, the only factors associated with MCI were age (odds ratio [OR] = 1.04, 95% CI = 1.01 to 1.08, p = 0.02), male sex (OR = 1.9, 95% CI = 1.3 to 2.8, p = 0.001), and low education (OR = 1.8, 95% CI = 1.2 to 2.7, p = 0.002) (not shown in tables).
Mortality.
There was no difference in 2-year mortality risk between MCI and unimpaired participants (OR = 1.4, 95% CI = 0.8 to 2.5, p = 0.2).
MCI transitions.
Table 3 shows the transition to and from MCI status between biennial data collection waves, under post-hoc application of our operational version of the Mayo amnestic MCI criteria.
Table 3 MCI transitions between biennial data collection waves (MoVIES waves 2–6)*
Interpretation of table 3 is as follows, with use of the example of transitions between wave 2 and wave 3, shown in the second column (in the third, fourth, and fifth columns, the corresponding figures can be examined for the wave 3 to 4, wave 4 to 5, and wave 5 to 6 transitions, bearing in mind the aging and increasing mortality of the cohort): In column 2, the first row shows that 10% of MCIs at wave 2 vs 2.1% of subjects at wave 2 had transitioned to AD by wave 3. The second row shows the same for transition to any dementia. The third row shows that 10.0% of MCI subjects at wave 2 had died (without becoming demented) by wave 3. The fourth row shows that 17.5% of MCI subjects at wave 2 still met criteria for MCI at wave 3. The fifth row shows that 55.0% of MCI subjects had reverted to non-MCI by wave 3. Of these, approximately half did so by “improving”; that is, their memory scores improved to <1 SD below the mean, or they no longer reported a memory complaint, or both. Approximately one-fourth of them became non-MCI by worsening (although not to the level of dementia); that is, MMSE score dropped to below 25, or at least one IADL impairment was now reported, or both. In about one-fourth of them, the transition to non-MCI showed a mixed picture, that is, improved on one criterion and worsened on another. The bottom row shows that 1.9% of normal individuals at wave 2 had become MCI; this might be considered the “incidence” of MCI if all underlying assumptions were true.
Risk of progression to dementia.
Based on GEE, combining data from all waves, the OR for MCI progressing specifically to AD was 4.2 (95% CI = 2.2 to 8.0) after adjustment for age, sex, and education; the OR for progression to any dementia was 3.9 (95% CI = 2.1 to 7.2).
Factors associated with progression to AD/dementia.
With use of GEE, based on all available 2-year transitions among participants with MCI at all waves combined (93 observations based on 76 distinct subjects), only greater age predicted progression to AD (for every year of age, OR = 1.1, 95% CI = 1.02 to 1.3, p = 0.02) compared with all other outcomes over each 2-year follow-up interval. We may have had insufficient power in this sample to detect additional associations. Even if our 93 observations of MCI over the course of the study had represented 93 distinct subjects, if half of them had a given risk factor, the power to detect an OR of 2.5 with a significance level of 0.05 would have been only 46%.
Other outcomes.
In addition to short-term (2-year) outcomes, we also examined outcomes over the entire follow-up period. Of 40 individuals who met criteria for MCI at wave 2 and were followed for 10 years till wave 6, 11 (27.5%) developed dementia, 9 (22.5%) of these 11 developing AD, over the course of follow-up. The interval from wave 2 assessment to the estimated date of onset of dementia ranged from approximately 1 year (0.7 year) to 10 years (10.3 years). This finding should be interpreted with the recognition that at least some who died in the interim, without being diagnosed as having AD, might have developed signs of dementia had they lived. However, as noted, there was no significant difference in mortality between MCI and unimpaired participants over 2-year intervals.
Discussion.
The term “mild cognitive impairment” was first defined operationally by the New York University group27 based on the Global Deterioration Scale28 and redefined shortly thereafter at the Max Planck Institute in Germany,29 based on DSM-III-R20 and International Classification of Disease-1030 criteria. Its currently most widespread use is that proposed by the Mayo Clinic group, which regards amnestic MCI as a state with high probability of “conversion” to AD.2,3⇓
In our community sample, across five biennial data collection waves over 10 years, between 10 and 17% of those with operationally defined amnestic MCI progressed to AD every 2 years, with a fourfold elevation in risk compared with those who were unimpaired according to these criteria. The “incidence” of MCI was 1.9 to 2.8% over 2-year periods; that is, 2 to 3% of previously normal individuals developed “MCI” over that period. Between 11 and 21% of those with MCI remained MCI, and 33 to 56% became non-MCI, half of whom reverted to normal or unimpaired. Among those who survived 10 years of follow-up after meeting criteria for MCI, 27% developed dementia (23% developed AD). These findings, in the context of previous literature, have implications for the definition of MCI, bearing in mind the nature of the samples and the specific criteria employed in different studies.
An acknowledged limitation is that our study retroactively fitted the Mayo MCI criteria to data collected from the late 1980s onward, before Mayo criteria had been published. Further, our epidemiologic design required us to operationalize the Mayo criteria to an extent beyond that recommended by their originators. We used actual test scores rather than age- or education-adjusted scores so as to be able to examine the effects of age and education on MCI. Yet, our data confirm the basic premise of “MCI” as defined by the Mayo group, namely, that persons with isolated memory impairment are at greater risk of progressing to AD and other dementias than those without this deficit. On the other hand, our data also suggest that when the criteria are applied objectively in the nonclinical community setting, they define an unstable category; further, that a change in even one criterion can change an individual’s classification and therefore the estimated prevalence of MCI. This finding is not inconsistent with the caveat laid down by the Mayo group2 that their criteria are intended to serve as guidelines to be used by expert clinicians interpreting the data on a case-by-case basis and exercising clinical judgment. However, as emphasized in a recent comprehensive review of the MCI literature, clear definitions and good normative data will be critical in identifying the subgroups with incipient AD vs not only other dementing disorders but also normal aging and conditions causing reversible cognitive impairment.1
The prevalence of MCI cannot be estimated from the original Mayo study, which was based on 76 consecutively diagnosed patients from a “general community clinic” population in Rochester, MN.2 Several European groups have applied their interpretations of the Mayo criteria to community-based studies, reporting prevalence ranging from 2.8 to 6.1%.5,31-34⇓⇓⇓⇓ Following a similar approach, we found prevalence to range from 3 to 4% in our rural Pennsylvanian cohort, as reported here. Another term, “cognitive impairment, no dementia” (CIND) is recognized by the presence of cognitive impairment in the absence of dementia, based on clinical examination and neuropsychological testing, without excluding other neurologic, psychiatric, or medical causes of cognitive impairment. The term CIND was first used by the Canadian Study of Health and Aging (CSHA).35 In that large, nationally representative epidemiologic study, the prevalence of CIND was about 16%, twice that of dementia. A category with “circumscribed memory loss” or isolated memory deficit, similar to amnestic MCI, had a prevalence of 5.3% and accounted for the single largest subgroup within CIND in the Canadian study.35 The Religious Orders Study (ROS) used the term MCI in a definition that appears identical to CIND and found a frequency of 26.4%36; this figure is remarkably similar to the 23.4% prevalence of CIND as defined in Indianapolis Study of Health and Aging,37 despite ROS being a volunteer rather than community-based sample. In the Cardiovascular Health Study (CHS) Cognition Study, amnestic MCI was present in 6% of the sample, whereas MCI of the multiple-cognitive deficits type had a 16% prevalence.38 In the Leipzig Longitudinal Study of the Aged, the prevalence of MCI varied from 3 to 20% depending on the concepts applied.34 Clearly, estimated prevalence is largely a function of definition and operational criteria.
The course and outcome of MCI have also varied across studies. In the Rochester (Mayo) sample, 12% of MCI patients “converted” or progressed annually to AD compared witho 1 to 2% of control subjects.2 In a sample of elderly individuals recruited by advertisement in Boston, MA, a CDR score of 0.5 identified 123 subjects with “questionable AD”; they had a variety of trajectories over 3 years of follow-up, during which 18.7% of them were diagnosed as “probable AD.”39 In St. Louis, MO, 100% of referred subjects rated as CDR = 0.5 progressed to dementia over 9.5 years of follow-up, suggesting that MCI, as defined by the Washington University group, essentially represents early AD.4 Rates of progression or conversion to dementia are predictably lower in the community-based samples than in the referral-based clinic samples. In our study, between 10 and 17% of those with MCI progressed to dementia over 2-year intervals. In the general practice-based Eugeria study in Montpellier, France,5 MCI subjects had a 3-year conversion rate of 11.1%. The community-based Personnes Agées QUID (PAQUID) Study in Bordeaux, France, found an annual conversion rate of 8.3% with good specificity but less than ideal sensitivity; and 40% of those with MCI had reverted to normal after 2 years.33 Thus, the apparent homogeneity of MCI and its seemingly inexorable progression to AD in clinic samples may be strongly related to referral patterns and selection factors. The dichotomous distinction of MCI from dementia allows the progression from MCI to dementia to be considered in clinically convenient terms but fails to capture the heterogeneity of observed outcome1 or even the heterogeneity of MCI itself.
If MCI is essentially incipient AD,4 then factors associated with, and predictive of, AD should bear a similar relationship to MCI. If MCI is, on the other hand, an unstable and heterogeneous category, few, if any, factors are likely to be found consistently associated with it. Greater age and lesser education were associated with MCI in our study, as in several others.31,40⇓ We also had the unusual finding that MCI was associated with male sex, consistent with our previous findings that men in our rural older Pennsylvanian cohort obtained scores equal to or worse than those of women on all tests41 and had higher incidence of questionable dementia (CDR = 0.5).7 In Finland, a long-term prospective study found MCI was associated not with contemporaneously measured blood pressure or cholesterol, but with midlife hypercholesterolemia and systolic blood pressure elevation.31 In the CHS Cognition Study, MCI was associated with African American race, depression, APOE-4 genotype, cortical atrophy, and MRI-identified infarcts.40 However, the Eugeria study5 found no differences among MCI and normal groups with respect to age, sex, education, cognitive scores, family history, medication use, exposure to anesthesia, illness over the previous year, depressive illness, APOE genotype, or hypoperfusion on SPECT neuroimaging.
Progression from MCI to AD in our study was associated only with age; lack of power is a potential explanation, given the small numbers of persons who met criteria for MCI. Here, the literature is even more sparse. In Rochester, the APOE-4 genotype strongly predicted progression of memory-impaired persons to AD.42 In memory-impaired referral samples in Toronto43 and New York,27 scores on specific memory or language tests or both predicted further cognitive decline and/or progression to AD. In the volunteer sample in Boston, the overall level of functional impairment as well as difficulties with specific daily functions involving, for example, judgment and problem solving, discriminated well between those who would and would not progress to AD within 3 years.39
Thus, there is empirical support for the general thesis that a condition of isolated memory deficit, short of dementia, can be defined in the population. Further, persons with this condition are significantly more likely to develop frank dementia over time than those without this deficit. However, the rates at which this progression occurs and predictors of progression are widely discrepant across studies and populations. These discrepancies are partly related to the nature of the population (clinical/referral vs community based) and length of follow-up. They also seem due in large part to the definitions that were used in different studies,34,44⇓ although one study found that variations in case definition affected prevalence but not outcomes.45
As revealed by the studies reviewed above, experts are divided over how the memory deficit criterion should be defined: subjective report, objective testing, or both. If a memory complaint is required, it seems essential that the patient/study participant have insight, or a reliable informant, or both; in the Canadian study, memory complaints appeared unnecessary for case definition.45 If objective assessment is required, there is no consensus on which and how many memory tests are needed (e.g., whether both visual and verbal memory should be impaired), how many standard deviations below the mean should be considered impaired, or on which norms the means and standard deviations should be based. The Mayo criteria for memory impairment are based on age and education norms that happen to be available for Rochester, but appropriate norms are not available for all populations. The requirement for functional impairment to be absent raises issues of perceptiveness on the part of the patient and informant, the degree of cognitive challenge in the patient’s everyday activities, and the presence of competing morbidities, such as sensory or motor impairment, which can also compromise daily functioning. The CSHA study found intact ADL to be unnecessary for case definition45; the usefulness and validity of the ADL criterion were also challenged in the Eugeria study.5 The criterion of normal mental status is sometimes, as in our study, taken to imply scores in the normal range on a screening test of cognitive functioning; however, in the Mayo study, this criterion referred to normal scores on adult intelligence tests.2 It is also sometimes interpreted as the presence of normal scores in all nonmemory domains of cognitive functioning, that is, ensuring that the memory deficit is isolated. A natural extension of this argument, not discussed further here, is that any isolated cognitive deficit, whether in memory or other domain, could be included under the broad rubric of MCI. This is, in fact, the case for the category “age-associated cognitive decline,”46 which in the Eugeria study had better predictive value for AD than MCI.5 The requirement that dementia be absent for the diagnosis of MCI is understood to mean not that underlying brain disease is absent, but merely that the individual does not meet diagnostic criteria for dementia at the time of assessment.
Conceptually, it may seem counterintuitive that individuals could have cognitive impairment, isolated memory impairment, and no dementia and yet not meet criteria for amnestic MCI. In the MoVIES Study, individuals having an MMSE score below the threshold of 25 were operationally designated as cognitively impaired and referred for diagnostic evaluation. In our retrospective application of the Mayo MCI criteria, “normal mental status” criterion was operationalized as scores of ≥25. Thus, by definition, these two groups became mutually exclusive. Approximately 10% of the cohort at each wave had cognitive impairment (MMSE ≤ 24) but not dementia (CDR < 1). Among them, the proportion of subjects with isolated memory impairment (Word List Recall < 1 SD below the mean) ranged from 38 to 54% across waves 2 through 6. These individuals may not have met criteria for dementia because they had no functional impairment or because they had cognitive impairment only in a single domain. At the same time, they may have failed to meet the Mayo criteria for MCI because they did not have subjective memory complaints.
As the term MCI is increasingly used to identify those likely to progress to dementia, we recommend that predictive validity should be the basis for its definition. The next generation of criteria should be based on the characteristics most predictive of progression to dementia in prospective studies of large, representative, community-based samples.
Acknowledgments
Supported in part by grants AG07562, AG05133 and K24 AG02035 from the National Institute on Aging, NIH.
The authors thank the MoVIES research staff for project coordination and data collection, management, and analysis and the MoVIES participants for their cooperation over the 15 years of the study.
- Received December 15, 2003.
- Accepted March 8, 2004.
References
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.