Can Late Life Social or Leisure Activities Delay the Onset of Dementia?

One of the questions asked most often by older individuals is whether there are activities they might engage in to prevent or postpone the onset of dementia. This question is of consequence not only to older individuals but to society. The prevalence of dementia rises exponentially, doubling with every 5 years of chronological age between ages 65 and 85.1 A 5-year delay in onset of symptoms would halve the number of dementia patients. In the United States, where the economic costs of caring for demented older persons are over 100 billion dollars per year, the discovery of activities that delay the onset of dementia would likely have the same consequences as the discovery that diet and exercise delay the onset of cardiovascular disease has had on improving the health of older individuals and extending life span. In an article in this issue of the Journal of American Geriatrics Society, Fabrigoule et al.2 report that traveling, odd jobs or knitting, and gardening are associated with a lower risk of developing dementia in a cohort of 2040 randomly selected home-dwelling residents of Gironde, France, aged 65 and older, who were nondemented at their baseline screening and who were then followed for at least 3 years. This study highlights two significant methodological issues inherent in identifying late life risk or protective factors for dementia. The first issue is dissociating protective or risk factors from the effects of early prediagnosis symptomatology. A major problem in determining whether successful engagement in an activity delays the onset of cognitive impairment or whether the lack of successful engagement is an early symptom was well stated by Hultsch et al.,3 who investigated the association of cognitively demanding activities and cognitive performance. These investigators stated that “given the present data set… we are unable to demonstrate whether activity styles act to maintain cognitive performance or whether healthy cognitively able individuals choose, or are more able to participate, in an active life style …” Hultsch et al. suggested that only through a longitudinal study could this be determined. Fabrigoule et al.2 have carried out a longitudinal study to circumvent this difficulty. Yet the problem persists. Dementia was diagnosed in their cohort using DSM-III-R criteria. DSM-III and DSM-III-R criteria have been adopted worldwide because they are highly accurate; when a subject is diagnosed as demented using these criteria, there is more than a 95% likelihood that the diagnosis is correct.4, 5 But this accuracy is obtained by the conservative nature of these criteria, which require the presence of both functional and cognitive impairment, with cognitive impairment involving at least two different areas of cognition. In Alzheimer's disease, which accounts for about two-thirds of demented subjects older than age 65, one of the major defining characteristics is an insidious onset. Symptoms commonly begin months to years before diagnosis. The very earliest symptoms of AD are usually forgetfulness, but in a significant number of patients include loss of initiative as well as psychiatric symptoms such as anxiety or depression.6 Any of the above may lead to a reduction in complex activities by individuals in the incipient but undiagnosed phase of dementia. And cognitive impairment as measured on one psychometric test has been reported as early as 6 years before diagnosis of Alzheimer's disease (C. Kawas, personal communication). In the Gironde study, social and leisure personal activities were self-reported at baseline. Incident cases of dementia were diagnosed 1 to 3 years after baseline screening. From an unadjusted univariate analysis, it appeared that the incidence of dementia was reduced in individuals who described themselves as being involved without difficulty in certain activities-sports, traveling, visits, associations, child care, reading, watching television, odd jobs or knitting, playing parlor games, gardening. When the analysis was carried out using a proportional hazard model and included both age and cognitive performance as covariates, only three types of activities remained significantly protective against the onset of dementia-traveling, odd jobs or knitting, and gardening-and then only if individuals engaged in two of these three specific activities. This certainly suggests that many of the activities that appeared to be protective in the univariate analysis but dropped out in the multivariate analysis were engaged in by individuals who were ‘cognitively able’ at baseline and not by individuals who may have been in the earliest stages of their dementing illness. Fabrigoule et al.2 point out that a history of discontinuation of the specific social and leisure activities because of difficulties with these activities may represent an important contribution in identifying symptoms of early dementia and, thus, amplify current instrumental activities of daily living schedules. In this regard it is important to note that participation or lack of participation in a ‘golden age’ club was not a significant protective activity or risk factor in their analysis. In regard to the three remaining types of activities, it is surprising that these investigators did not determine whether the introduction of educational status into their proportional hazard analysis would have affected the significance of these activities. Traveling, in particular, is an activity more likely to be engaged in by those of higher socioeconomic status. The authors did show that including prior occupation in the analysis did not eliminate these remaining variables, but they apparently did not take into account years of education. The Gironde survey of dementia, together with a survey carried out in Dordogne, both areas around Bordeaux in France (together termed the Paquid study), represents one of the major epidemiological studies of prevalence, incidence, and risk factors of dementia.7-9 In the prevalence phase of this study, Dartigues et al.7 reported that the relative risk (adjusted for age and gender) of dementia was 1.94 among illiterates compared to those with secondary education; this is similar to the relative risk of 2.04 observed in similarly stratified groups observed by Zhang et al.10 in Shanghai, China, who compared those with 6 or more years of education with illiterates. These results have been confirmed in a number of other prevalence studies (summarized by Katzman11). Although there was no significant effect of education on dementia incidence in the Framingham study,12 Stern et al.,13 in a 4-year follow-up of the North Manhattan study, found a relative risk of dementia of 2.02 (95%CI, 1.3–3.06) in those with less than 8 years of schooling and if low education were combined with low occupation (eg, unskilled worker), the relative risk was 2.87 (95%CI, 1.32–3.84). Similar findings from the incidence phase of the Shanghai survey of dementia show that individuals with no education are 2.37 (95% CI, 1.59–3.59) times as likely to develop dementia (manuscript in preparation). It is very likely, although not proven, that education operates as a protective risk factor by delaying the onset of Alzheimer symptomatology without altering the time of onset of pathological changes. A relative risk of 2 in the illiterate compared with those with secondary education can be interpreted as about a 5-year delay in the appearance of symptoms in those with secondary education. Another example of the overall problem of separating risk factors from early symptomatology is the presence of a history of depression late in life. During the 1980s a number of case control studies were carried out with the goal of identifying risk factors for Alzheimer's disease. A collaborative group, terming themselves ‘EURODEM’ but including a number of US investigators, combined data from 11 case control studies for reanalysis, thereby greatly increasing the power of the analyses. Unexpectedly, a history of depression of late onset proved to be a significant risk factor (relative risk = 2.4, (95% CI 1.4–4.4).14 But depression frequently occurs in Alzheimer's disease.15, 16 Was depression a causal factor or simply an early symptom? We do not know the answer at the present time. Despite these methodological problems, one should not dismiss the importance of this report from the Gironde study. One can argue convincingly that activities protective against dementia should exist. Indeed, one might anticipate that cognitively demanding activities might increase or maintain brain reserve and delay the onset of dementia. As pointed out by the authors, a stimulating environment has been shown to increase brain weight and cognition in rodents. The protective effect of early education may well act in a similar manner. Katzman has speculated that the action might be to increase the number of interconnections or synapses between nerve cells.11 That the ability to form new synapses continues throughout life is suggested by studies of neuronal arborization of the human brain during aging, although this ability is much reduced in the elderly. Thus, if engagement in specific social or leisure activities in late life were found to reduce the incidence of or delay the onset of dementia, there would be a reasonable biological basis for its action. Ultimately, the validation of putative protective factors will require prospective experimental investigations, that is, longitudinal studies in which a suitably selected sample of older volunteers, who are taught and encouraged to participate in the identified activity, is compared with a placebo group. But until that can be carried out, randomized longitudinal studies such as the Gironde study are needed to identify which activities should be tested.