CIND and MCI in the Italian elderly

The diagnosis of cognitive impairment, no dementia (CIND) was at first used in the Canadian Study of Health and Aging¹ for individuals older than 65 years whose problems with memory or other areas of cognition were not sufficiently severe to meet the criteria for dementia proposed in the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R),² but who were judged to be distinct from those cognitively normal of the same ages. Mild cognitive impairment (MCI) criteria proposed by Petersen et al. include memory symptoms, objective impaired memory function for age and education, preserved general cognitive function, and intact activities of daily living (ADL) in persons not fulfilling diagnostic criteria for dementia.³ MCI, in its more recent definitions, may involve only memory domain in subjects with otherwise unaffected cognitive functions, or another single nonmemory domain or even, in a broader definition, multiple cognitive domains.⁴

Precise operational criteria for the diagnosis of both conditions are substantially lacking. Likewise, data on the predictiveness of the different definitions in the individuation of high-risk subjects are scarce. In this study, we sought to evaluate in the Italian Longitudinal Study on Aging (ILSA), a large population-based study, the prevalence of cognitive impairment according to different definitions and subtypes, and to evaluate the progression to dementia in a 3-year follow-up period. Moreover, we studied the relationships of cognitive impairment with daily functioning, cardiovascular diseases, and vascular risk factors, to examine the influence of these variables in the transition to dementia.

METHODS

RESULTS

Of the 5,632 individuals of the original ILSA sample, 170 who at the prevalence day were dead or no longer resident at the address reported in the demographic lists were considered as noneligible. In the remaining 5,462 eligible individuals, the overall participation rate, similar for men and women, was 82.8%, which left 4,521 participants. Of these, the MMSE was administered to 3,425 (75.8%; mean age 74.4 ± 5.7 years, 52.4% men), the DCT was administered to 3,153 (69.7%; mean age 74.0 ± 5.7 years, 53.4% men), and the BSRT was administered to 2,967 (65.6%; mean age 73.9 ± 5.6 years, 53.2% men). Overall, the final study sample included the 2,830 participants (62.6%; 53.7% men) with all three tests performed. The mean age of individuals for whom all three tests were available was 73.7 ± 5.6 vs 75.7 ± 5.7 years in nonparticipants, and the mean education level 6.4 ± 4.6 vs 5.3 ± 4.4 years. Sixty-two subjects were diagnosed as having dementia at baseline and were excluded from further analyses, which left a sample of 2,768 subjects (mean age 73.6 ± 5.5 years, 53.6% men).

Overall, 264 participants were diagnosed with CIND and 445 with MCI. Subjects with CIND represented 9.5% (95% CI 8.4 to 10.6), and those with MCI represented 16.1% (95% CI 14.7 to 17.5) of the study population. Among different subcategories, the prevalence was 3.9% (95% CI 3.2 to 4.6; n = 107) for CIND with normal results on both the BSRT and the DCT, 1.8% (95% CI 1.3 to 2.3; n = 50) for amnestic CIND, 2.3% (95% CI 1.7 to 2.9; n = 65) for single nonmemory, and 1.5% (95% CI 1.0 to 1.9; n = 42) for multidomain CIND. The prevalence was 7.0% (95% CI 6.0 to 7.9; n = 195) for amnestic MCI, 7.8% (95% CI 6.8 to 8.8; n = 215) for single nonmemory, and 1.3% (95% CI 0.9 to 1.7; n = 35) for multidomain MCI.

Tables E-2 and E-3 report the distribution at baseline of demographics, vascular disease and risk factors, ADL and IADL in individuals with CIND or MCI, and relative subtypes compared with subjects cognitively normal at baseline. In subjects with CIND, considered as a whole, there was a significantly higher frequency of female sex, diabetes, myocardial infarction, stroke, and deficit in one or more ADL or IADL. Subjects with CIND showed a lower education level in all subtypes, and those with multidomain CIND showed a significantly higher frequency of hypertension and heart failure. MCI subjects, as a whole, were slightly younger than those with normal cognition. Deficit in one or more ADL or IADL was more common in these subjects, with the remarkable exception of amnestic MCI. Again, hypertension and heart failure were significantly more common in subjects with multidomain MCI. No differences were found in vascular diseases or risk factors and functional activities between subjects with amnestic MCI and those with normal cognition.

The cohort at risk for dementia consisted of 2,768 individuals. Of these, 316 (11.4%) died before they were contacted the second time in 1995. All 316 death certificates and 152 hospital clinical records were collected. Information from GPs, proxy respondents, medical records, and death certificates was considered reliable to define the occurrence or not of dementia for 210 of deceased individuals (66.5%). The follow-up procedures were completed by 1,992 (81.2%; mean age 72.9 ± 5.3 years, 52.8% men) of the 2,452 survivors. By entering demographic variables in a logistic regression model with being a nonparticipant as the dependent variable, we observed that nonparticipants were significantly older (RR for 1-year increment of age, 1.04; 95% CI 1.02 to 1.06), whereas sex (RR for men, 0.94; 95% CI 0.76 to 1.16) and education (RR for 1-year increment, 0.99; 95% CI 0.97 to 1.01) proved to be nonsignificant predictors.

The final data used were for 2,202 individuals (1,992 survivors and 210 deceased with reliable information; mean age 73.3 ± 5.4 years, 54% men) for a total of 8,520 person-years available for the analysis. The mean follow-up time was 3.9 ± 0.7 years. A total of 65 new dementia cases were identified, including 6 cases detected in individuals who died between the two surveys. Table E-4 indicates the number of cases and the sex-specific incidence rates (new cases per 1,000 person-years) of dementia at follow-up in the overall study population, in cognitively normal subjects, and in the different subtypes of CIND and MCI. Incidence of dementia (per 1,000 person-years) was 7.63 in the total sample, 5.02 in subjects with normal cognition at baseline, 21.37 in CIND, and 13.59 in MCI. In CIND, the lowest rates were found in subjects with deficits only on the MMSE and normal BSRT and DCT, and the highest in subjects with multidomain CIND. Among MCI subtypes, the lowest incidence was in amnestic and the highest was in multidomain MCI. Incidence of dementia in both CIND and MCI was higher in women, whereas incidence in subjects with normal cognition at baseline was higher in men.

To take into account the weight of comorbidity in the development of dementia, the predictive effect of cognitive decline was evaluated in a Cox regression analysis controlled for age, sex, and education level, and including all the other diseases and risk factors under study in the ILSA (hypertension, angina pectoris, myocardial infarction, heart failure, atrial fibrillation, stroke, diabetes, peripheral artery disease, smoking, wine consumption, cholesterol), together with ADL and IADL (table 1). In this multivariate analysis, cognitive decline at baseline significantly increased the risk of being demented at the follow-up examination. Considering individuals with normal cognition as the reference category, CIND and MCI both increased by almost three times the risk of dementia at follow-up. Among baseline variables, only a previous stroke and impairment in one or more IADL significantly and independently increased the risk of dementia at follow-up.

The predictive effect of cognitive decline in the transition to disability was also evaluated. At follow-up examination, information on ADL and IADL was obtained for 1,939 (97.3%) of the 1,992 participants still alive. Considering those totally independent at baseline, subjects with CIND or MCI were dependent at the follow-up examination in one or more ADL in 27.5% of cases vs 19.3% of individuals with normal cognition at baseline (p = 0.001), and dependent in one or more IADL in 42.6% vs 33.6% (p = 0.012). Table 2 reports the results of two models of logistic regression evaluating the net effect of cognitive decline on development of disability in ADL and IADL, including the other variables under study and controlling for disability level at baseline. Both CIND and MCI increased the risk of dependency at the follow-up examination. The observed effect was of the same magnitude for diseases such as stroke or heart failure.

Among subjects cognitively healthy at baseline, 33 became demented at follow-up. Table 3 shows differences in demographics, vascular diseases and risk factors, and ADL and IADL at baseline for cognitively normal subjects according to the presence of dementia at follow-up. Individuals who became demented at follow-up were significantly older and with a lower education level at baseline. They also had with higher frequency a previous stroke (12.1% vs 4.1%, p = 0.024) and impairment in one or more IADL (75.8% vs 43.1%; p < 0.001). To evaluate independent predictors of dementia at follow-up in these subjects, the same variables were included in a logistic regression analysis. Results showed that, among baseline variables, significant predictors of developing dementia at follow-up were age (RR for 1-year increment, 1.17; 95% CI 1.07 to 1.27), male sex (RR, 4.70; 95% CI 1.79 to 12.32), previous stroke (RR, 3.36; 95% CI 1.06 to 10.63), and impairment in one or more IADL (RR, 2.67; 95% CI 1.04 to 6.81).

DISCUSSION

We evaluated the contemporary presence of CIND and MCI in a population-based study. Both conditions were frequent in the Italian elderly population (approximately 1,095,000 prevalent cases expected for CIND and 1,860,000 for MCI) and significantly predicted progression to dementia in a 3-year period. Demographic variables, function at baseline, vascular disease, and vascular risk factors were differently associated with CIND, MCI, and the single subtypes under study and showed an effect on the progression to dementia at follow-up. Although incidence of dementia was significantly higher in individuals with a diagnosis of CIND or MCI, half of dementia cases in the study population developed in subjects judged as cognitively healthy at baseline. Our methodologic approach might raise the issue of whether changes in brain function that lead to MCI, CIND and dementia could be seen as a continuum, which may reflect a single pathologic process. Studies with repeated follow-up assessment are warranted to confirm or exclude the hypothesis of a linear progression from one condition to another. Identification of people at different stages across the process may also have implications for selective treatment options.

Differences in the study populations and in the diagnostic criteria used may reduce the validity of comparisons of frequency of CIND and MCI across studies. For instance, representativeness and generalizability may vary, considering that some studies were multicenter and population based,^1,21,22 some including also institutionalized persons,^1,22 whereas others were based on cohorts of nondemented elderly,^10,23 on subsamples of population-based cohorts,²⁴ or on GP research networks.²⁵ Also, in the Cardiovascular Health Study Cognition Study, the diagnosis of MCI required a documented decline from a previous level of functioning detected with an annual neuropsychological testing,²¹ whereas in other studies,^22,23,25 MCI diagnosis was based on the criteria proposed by Petersen et al.^3,26 The overall prevalence of MCI in our study was 16.1%, close to the 18.8% reported in the Cardiovascular Health Study Cognition Study.²¹ Our rate of 7.0% for amnestic MCI was comparable with the 5.0% reported in an urban community in northern Manhattan²³ and the 6.0% found in the Cardiovascular Health Study Cognition Study.²¹ In the Eugeria Longitudinal Study of Cognitive Aging, the prevalence of amnestic MCI was 3.24%,²⁵ close to the 3.02% reported in the Canadian Study of Health and Aging without the requirements of subjective memory symptoms and intact IADL.²² Our prevalence rates were 7.8% for single nonmemory and 1.3% for multidomain MCI, whereas the combined prevalence of MCI multiple cognitive deficit type (including single nonmemory MCI) was 15.7% in the Cardiovascular Health Study Cognition Study.²¹ In the northern Manhattan sample, the overall prevalence for MCI was 28.3%, the prevalence of single nonmemory MCI ranged from 2.1% to 5.2%, and the prevalence of multidomain MCI was 12.1%.²³ CIND diagnosis in the different surveys was based on Canadian Study of Health and Aging criteria,¹ on poor performance on one or more neuropsychological scales and informant questionnaires,²⁴ or on a MMSE score > 1 SD below the age- and education-specific mean.¹⁰ CIND prevalence was 9.5% in our sample and 16.8% in the Canadian Study of Health and Aging.¹ The prevalence of CIND was 11.5% in a sample from the Sacramento Area Latino Study on Aging²⁴ and 14.8% in the population aged 75 years or older of the Kungsholmen Project.¹⁰

Differences in study population, inclusion criteria and diagnostic definitions do also apply to progression rates of cognitive impairment to dementia. Depending on the definition of cognitive impairment and duration of follow-up, reported progression rates were 12% per year,³ 10% to 17% over 2-year intervals,²⁷ 35.1% and 44% over 3 years,^10,28 and 35.7% in a 5-year period.²⁹ Higher rates were reported in studies based on clinical series or elderly volunteer selection, rather than on population-based samples. In our study, the conversion rate to dementia in individuals with normal cognition at baseline during the follow-up period was 2%. The conversion rate of subjects with CIND was, overall, 8.2%, ranging from 1.2% of individuals with deficits only on the MMSE to 9.8% for amnestic CIND, 15.8% for single nonmemory CIND, and 19.0% for multidomain CIND. Conversion to dementia interested 5.2% of MCI subjects and ranged from 3.4% in individuals with amnestic MCI to 5.6% in single nonmemory MCI, and 15.0% in multidomain MCI. These data underscore that there are cases in which impairment of cognitive areas other than memory may be detected as first manifestation of dementia, and that the association of multiple deficits may increase the clinical predictiveness toward progression to dementia. Incidence of dementia per 1,000 person-years was 5.02 in cognitively healthy subjects, 13.59 in MCI, and 21.37 in CIND. The highest incidence rates were found in subjects with multidomain involvement, for both CIND (56.02 per 1,000 person-years) and MCI (40.60 per 1,000 person-years). In both CIND and MCI, incidence of dementia was higher in subjects with single nonmemory involvement than in amnestic types.

The involvement of ADL and IADL in the definition of and in subjects with CIND and MCI is controversial. Pure amnestic MCI should not include problems in ADL, whereas a minimal involvement of IADL could be present.⁴ We decided not to exclude MCI on the basis of performance on ADL. In the older population, ADL may be compromised as a consequence of comorbid conditions, the impact of which is difficult to disentangle from cognitive problems.³⁰ Our approach was rewarded by the fact that in amnestic MCI, the frequency of ADL and IADL involvement was not different from cognitively healthy subjects (22.1% vs 24.9%; p = 0.381 for ADL and 38.5% vs 44.1%; p = 0.128 for IADL). The complex relationship between functional activities and cognition is further confirmed by the fact that deficit in one or more IADL at baseline was significantly associated with an increased risk of dementia at follow-up, even when individuals with normal cognition at baseline were considered and, conversely, that the diagnosis of CIND or MCI at baseline significantly increased the probability of dependency at the follow-up examination.

Vascular disease and risk factors such as stroke, diabetes, myocardial infarction, heart failure, and hypertension were related to CIND, MCI, and their different subtypes. The effect of vascular risk factors on cognitive impairment and dementia has been previously reported.^31–34 In our population, hypertension and heart failure showed a consistent and significant association with multidomain deficits in both CIND and MCI. The occurrence of a previous stroke was a strong predictor of dementia at follow-up. It is of interest that this was also found in individuals who, according to our evaluation, were defined as without cognitive impairment at baseline. Approximately half of cases of dementia developed in cognitively healthy individuals. These findings may contribute to our knowledge of pathogenic aspects of dementia, suggesting that factors other than cognition may be identified as predictors of development of the disease in the older population. Stroke and vascular risk factors such as hypertension, in particular, may impact cognition at different time distances from the index event or first diagnosis.^34–36

This study has possible limitations, including the fact that results are referred only to participants in the ILSA who completed the overall tests battery and the limited neuropsychological evaluation encompassing only global cognition, memory, and attention. The problem of attrition may be relevant in studies evaluating cognitive impairment at a population level. In the hypothesis that missing examinations is not a random phenomenon, missing data caused by refusals in studies on cognitive performance in the elderly may lead to an underestimation of true frequency of cognitive impairment.³⁷ In our study, nonresponse increased with advanced age and lower education level, introducing a possible distortion in results. However, our data derive from a national representative population-based survey. The study sample was randomly selected and included urban, suburban, and rural population. Both free-dwelling and institutionalized individuals were considered. The use of incident cases arising within a general population allows a major generalizability of results and may provide a more complete picture of the condition under study. The mean sampling interval was of 3.9 ± 0.7 years. Studies with long intervals between the cross-sectional examinations are at risk of missing a relevant number of patients. We tried to reduce this effect also by using multiple sources of information for deceased individuals. A particular aspect of our survey was the opportunity to take into account the occurrence of major age-related diseases in the whole cohort. The assessment of the different conditions in the ILSA relied on standardized direct examination performed by trained neurologists or geriatricians, not merely on self-reported information. Finally, the number of subjects included allowed us to create subgroups with enough subjects to reliably control for the effect of age and education level on performances on neuropsychological tests.

ACKNOWLEDGMENT

The authors thank Ms. Maria Elena Della Santa for help with manuscript preparation.

Appendix 1