Mild cognitive impairment in a population-based epidemiological study



This article is corrected by:

  1. Errata: ERRATUM Volume 8, Issue 3, 153, Article first published online: 4 September 2008

Dr Manabu Ikeda MD, PhD, Department of Psychiatry and Neuropathobiology Faculty of Medical and Pharmaceutical Sciences, Kumamoto University 1-1-1, Honjo, Kumamoto 860-8556, Japan. Email:


Background:  It would be of considerable importance to be able to estimate the rate at which subjects with mild cognitive impairment (MCI) progress to dementia in a cohort of a population-based epidemiological study and to establish simple diagnostic methods for the identification of people at high-risk of dementia. Subjects in a community based elderly cohort of MCI were followed longitudinally.

Methods:  Subjects were selected from participants in the first epidemiological study conducted on all persons aged 65 years and older residing at home in Nakayama, Japan, using the Mini-Mental State Examination (MMSE). Mild cognitive impairment was defined as memory deficit with objective memory assessment, without dementia, impairment of general cognitive function or disability in activities of daily living. The conversion rate was calculated using the person–year method.

Results:  At baseline, the sample consisted of 104 subjects selected from 1162 community dwellers aged over 65 years of age. During the 5-year follow up, 11 subjects (10.6%) were diagnosed with Alzheimer's disease (AD), five (4.8%) were diagnosed with vascular dementia (VaD) and six (5.8%) were diagnosed with dementia of other etiology. In this survey, the annual conversion rate of MCI to AD was calculated as 8.5% per 100 person–years and shifted to dementia at a rate of 16.1% per 100 person–years. The proportion of subjects with MCI who progressed to develop dementia was the same as in previous reports. However, nearly 40% of subjects returned to normal.

Conclusion:  The MCI concept, as currently defined, is an unstable and heterogeneous group.


Recently, the term ‘mild cognitive impairment’ (MCI) was proposed to describe the transitional state between normal cognition and Alzheimer's disease (AD).1,2 Mild cognitive impairment is increasingly recognized as an important public health issue because it is common and is associated with significant morbidity, especially the development of clinically diagnosed AD.3–5 As high-risk individuals for dementia, subjects with MCI could be a particularly suitable population for preventive intervention. Longitudinal studies of case series have revealed an increased risk of AD in MCI subjects, with a conversion rate ranging from 7 to 20% per year.3,6–8

Most studies investigating the natural history of MCI have been conducted on samples of subjects recruited in specialized outpatient clinics, such as memory clinics for AD. Such samples are highly selective and it is essential to identify subjects at high risk of dementia from community based surveys in order to carry out early intervention. To our knowledge, one community based prospective cohort study reported that the annual conversion rate to AD was 8.3% over 5 years.9 The incidence andoutcomes of MCI in the general population are still largely unknown.

Standardized memory scales, such as the Wechsler Memory Scale Revised (WMS-R), can be used to identify subjects satisfying a strict definition of MCI.1,2 However, it may be very difficult to undertake such time-consuming examinations in community based epidemiological surveys.

Nakayama study for MCI10

Nakayama is a Japanese rural community adjacent to Matsuyama City, a metropolis on Shikoku Island. We extracted a group of subjects with MCI using the Mini-Mental State Examination (MMSE),11,12 based on the results of the first Nakayama study,13 and tried to estimate the rate at which subjects with MCI shifted to dementia.


The first Nakayama study included all residents over 65 years of age living in the rural community between January 1997 and March 1998 by means of a door-to-door survey with a three-phase design. Of 1438 inhabitants, 1162 (81%) completed the protocol. A more detailed description of the methods has been published elsewhere.13,14

In the present study, we selected subjects who were participants of the first Nakayama study and who satisfied the following criteria: (i) normal general cognitive function, with MMSE ≥ 24; (ii) objective memory impairment, assessed by three-word recall in MMSE (delayed recall 0/3 or 1/3); (iii) neuropsychiatric examination findings of an absence of dementia or depression, diagnosed by geriatric neuropsychiatrists according to DSM-IIIR15 criteria; and (iv) no impairment of activities of daily living (ADL) evaluated using the physical self-maintenance scale (PSMS) and instrumental activities of daily living scale (IADL).


A 5 year follow up was conducted on all subjects between April and December 2003. A senior neuropsychiatrist administered the MMSE to subjects, whereas a public health nurse interviewed one family member of each subject using the PSMS and IADL.16 Subjects who were hospitalized or otherwise institutionalized were included in the study. Cranial computed tomography (CT) was conducted on all subjects whose MMSE score declined by more than 2 points from baseline.17

The diagnosis of dementia was established according to DSM-IIIR criteria. Subsequently, demented subjects were classified into subgroups based on the cause of dementia. Alzheimer's disease was defined according to the National Institute of Neurological and Communicative Disorders and Strokes-Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) criteria for probable AD18 and vascular dementia (VaD) was also defined according to NINDS-AIREN criteria.19

The conversion rate was calculated using the person–year method.20


The sample consisted of 104 subjects at baseline; 59 women and 45 men. The mean age of female and male subjects was 75.5 ± 6.7 years (range 65.1–90.2) and 73.6 ± 6.8 years (range 65.1–101.4), respectively.

Five years after the first Nakayama study, 14 subjects had died, 13 had moved to other communities (mainly due to their institutions) and six refused to participate in the follow-up investigation. Eleven (10.6%) subjects were diagnosed with AD (five men, six women), five (4.8%) were diagnosed with VaD (three men, two women) and six (5.8%) were diagnosed with dementia of other etiology. There were nine (8.7%) subjects who remained MCI (three men, six women). Furthermore, 40 (38.5%) subjects showed a restored MMSE score (Table 1). In our survey, the conversion rate from MCI to AD was 8.5% per 100 person–years and from MCI to dementia it was 16.1% per 100 person–years for 5 years.

Table 1.  Outcomes for 104 subjects with mild cognitive impairment and 74 control subjects at the 5 year follow up
 MCI subjects (n = 104)Control (n = 74)
  1. MCI, mild cognitive impairment.

Died (n)149
Moved to another community (n)132
Refused to participate in follow up (n)60
Alzheimer's disease (n)110
Vascular dementia (n)52
Other type of dementia (n)61
MCI (n)91
Free of dementia and MCI (n)4059

The control group consisted of 74 participants (41 men, 33 women) at baseline. The mean age of female and male subjects was 75.4 ± 7.2 years (range 65.1–89.2 years) and 73.2 ± 6.7 years (range 65.1–92.4 years), respectively. There were no significant differences in age or in the gender ratio between the MCI and control groups. Of the 74 participants without dementia or MCI at baseline, nine subjects had died, two had moved to other communities, two were diagnosed with VaD (one man, one woman) and one was diagnosed with dementia of other etiology, one with MCI, and none developed AD (Table 1).


This is the first report of 5 year outcomes of MCI in a population-based study of dementia in Asia. Our survey differs from previous investigations in the following aspects. First, even in the screening interview, subjects were examined directly by a neuropsychiatrist and a cranial CT was performed on all subjects with any sign of dementia. Second, we have continued follow up over 5 years in the Nakayama community after the first Nakayama investigation with a definite examination at 5 years. Therefore, we were able to analyze the conversion rate using the person–year method.

Several studies have been undertaken to determine the natural course of MCI in attempt to estimate the ‘conversion’ rate to AD in this group.21 As expected, most longitudinal studies of case series revealed a much increased risk of AD in MCI subjects.1,22,23 Subjects with MCI may constitute a particularly suitable population for preventive interventions.

In previous clinic-based reports, MCI has been found to progress to AD at rates of 7–20% per year.1,23,24 Standardized episodic memory examinations (e.g. WMS-R) with comprehensive neuropsychological tests have been suggested as a definitive means of identifying MCI. Therefore, measuring cognitive functions using MMSE, WMS-R, the Wechsler Adult Intelligence Scale Revised (WAIS-R) and the auditory verbal learning test were adopted in those studies. The differences in these rates are probably related to the different instruments and cut-off limits chosen to define MCI across the studies.

To our knowledge, there have been a number of community based prospective cohort studies following community dwelling MCI elderly people for several years.9,25–27 In one of these studies,9 comprehensive test batteries for the evaluation of global mental status (MMSE), visual memory (Benton's Visual Retention Test (BVRT)), verbal fluency (Isaacs Set Test), visuospatial attention (Zazzo's Cancellation Test) and simple logical reasoning and attention (Wechsler's Digit Symbol Test) were used. However, it may be difficult to administer comprehensive tasks in an ordinal epidemiological survey. Not only is it time consuming and expensive, but it is also a demanding task to ask of large numbers of participants, particularly the very old. Therefore, it is an unsuitable method for identifying those at high risk of AD among the general public.

There is an increasing need for fast and efficient cognitive screening instruments suitable for detecting MCI from normal aging individuals.28,29 Such screening tests would lighten the burden on patients and physicians, economize medical resources and provide opportunities for dementia prevention and treatment when there is evidence that effective interventions exist.30 In the present study, we used MMSE to select subjects who exhibit MCI. The total MMSE score was used to screen subjects and select them for neuropsychological evaluation/diagnosis; then, a subset of the MMSE (three-word recall) was used to further classify subjects as MCI. The MMSE is widely used as a screening instrument for cognitive decline or cognitive impairment in population-based studies, as well as in clinical practice, and is a well-validated instrument for assessing global cognitive function.31 The MMSE is administered easily and quickly; thus, it is suitable for use in the community. The conversion rate of MCI in the present study was almost the same as in a previous community based MCI study with strict memory examinations.9

In the present study, we did not check for subjective memory complaints corroborated by the informant, which are considered characteristic of MCI based on strict criteria.2 Observations by knowledgeable informants regarding an individual's cognitive abilities in everyday functioning have been shown to be sensitive and reliable for the detection of MCI.5,32 However, it is difficult to use informant-based scales as screening tools because, in modern Japanese society and in many Western societies, considerable numbers of people live alone.

Conventionally defined MCI has reasonable predictive value and specificity for AD. Our finding supports the hypothesis that MCI is an intermediate stage before dementia or AD. However, the lack of stability was even greater in MCI, because approximately 40% of our prevalent cases returned to normal. Mild cognitive impairment was very unstable across time in previous studies (Table 2).9,12,26 Therefore, people displaying MCI, as currently defined, are an unstable and heterogeneous group. There is epidemiological evidence that many subjects labeled as having MCI do not worsen over time and may revert to normal cognitive abilities. A diagnosis of MCI as a predementia stage of AD in such individuals would be inaccurate and carry a heavy personal and societal burden. The heterogeneity within MCI has been noted and a new classification has been proposed, based predominantly on neuropsychological profiles, and includes amnestic or single-memory MCI, multiple-domain MCI and single non-memory MCI.33 Conversely, concepts such as ‘age-associated memory impairment’ (AAMI) and ‘age-associated cognitive decline’ (AACD) consider such mild cognitive deficits to fall within the limits of normal aging in Europe.34,35 Conceptually, these terms may include a broad range of cognitive impairments. Enlarging the definition of subjects with acquired cognitive impairment not qualifying as dementia would allow the screening of more subjects at risk of dementia and possibly yield more sensitivity in the prediction of the etiological subtypes.

Table 2.  Instability of subjects with mild cognitive impairment
ReferenceFollow-up interval (years)MCI still MCI, Stable (%)MCI converted non-MCI, Improved, reverted to normal (%)
  1. MCI, mild cognitive impairment.

Larrieu et al.92 (total 5)5.4–6.941.4–43.2
Ganguli et al.262 (total 10)11.1–21.235.7–55.0
Ishikawa et al.1258.738.5

Although MCI does not constitute a homogeneous clinical syndrome, the recognition that MCI is a risk state towards further cognitive decline is clinically relevant and the control of risk factors, such as systolic hypertension, hypercholesterolemia, diabetes mellitus and stroke, may delay progression to dementia. Therefore, MCI may be a promising therapeutic target and an important target for screening and possible early intervention.


The authors thank the officials of the Nakayama Home Health Care Support Center, especially Ms Kaori Iimori and Ms Michiko Nishimura. This work was supported, in part, by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (to MI). The authors are especially grateful to Professor Naoji Amano (Department of Psychiatry, Shinshu University School of Medicine, Matsumoto, Japan) for inviting them to the 19th Japanese Psychogeriatric Society meeting, which took place in Matsumoto, Japan, in June 2004.