Introduction

According to the WHO, around 50 million people worldwide are affected by dementia [1]. The neurodegenerative disease causes progressive damage related to memory and recognition, spatial and temporal orientation and communication. It induces behavioral changes such as aggressive behaviors and results in increased needs for care. Dementia has a strong impact on the lives of those affected and their caring relatives [2,3,4], and results in a global burden on societies of an estimated 818 billion USD as of 2015 [5]. While Alzheimer’s Disease (AD) is seen as the most common form of dementia, most people in old age suffer from mixed forms of dementia [6].

While dementia was previously understood as a disease of the elderly, a life span perspective now increasingly prevails. Today it is well known that AD starts decades before diagnosis [7]. The fact that there is currently no cure for dementia and only limited progress in new treatments has paved way for a new focus on the preservation of cognition and the prevention of dementia. Based on more recent findings of large cohort studies in the Western world on dementia and cognitive disorders which report higher prevalence and incidence rates for earlier cohorts [8,9,10,11], the focus on influential lifestyle factors impacting brain health at the population level has gained momentum. This has facilitated research on modifiable risk factors over the life span. The current paper provides an overview on well-established modifiable risk factors for dementia and discusses the implications for a public brain health agenda.

Methods

We conducted a systematic literature search on dementia risk factors in Medline via the PubMed interface following PRISMA guidelines [12]. The search included English systematic reviews and meta-analyses published in the 5 years before 11 May 2018. Risk factors were searched separately using the following search strings: (review OR meta-analysis) AND (dementia OR alzheimer OR cognitive decline OR cognitive impairment) AND (sport OR exercise OR physical activity OR physical inactivity)/(social activities OR social network OR social support OR social interaction OR social participation OR social isolation OR loneliness)/(education OR mental demand OR mental stimulation OR cognitive demand OR cognitive stimulation OR cognitively AND active)/(nutrition OR diet OR nutrient OR dietary supplement OR mediterranean OR food)/(alcohol)/(smoking OR smoke OR tobacco)/(blood pressure OR hypertension)/(diabetes)/(adiposity OR obesity OR overweight)/(depression OR depressive)/(hearing loss OR hearing impairment OR hearing)/(craniocerebral injury OR cci OR concussion OR brain injury OR brain trauma)/(anticholinergic medication OR anticholinergic OR anticholinergics).

We assessed literature in two steps. In the first step, titles and abstracts of all identified articles were screened, and in the second step the remaining full texts were assessed. The basis for assessment was our predefined list of criteria which included all systematic reviews and meta-analyses with populations above 18 years and all forms of dementia. We excluded articles that solely focused on cognitive impairments or included other disorders like Parkinson`s Disease.

Results

Physical activity

There is a strong evidence base suggesting that physical activity has a protective effect against dementia [13,14,15] and against AD in particular [15,16,17]. Moreover, Guure et al. [18] found protective effects of physical activity for all-cause dementia (ACD) and AD, but not for vascular dementia (VaD), and Stephen et al. [17] found a protective effect of leisure time physical activity against AD. Results strongly support the idea that regular physical activity throughout the life course is protective against dementia while physical inactivity poses a risk factor. While it is generally assumed that more physical activity is connected to stronger effects, it remains unclear which activities undertaken at which intensity, duration and frequency are most protective against dementia [5, 19]. There is evidence that even minor physical activity, compared to none, is connected to lower dementia risk [20], suggesting that every step counts.

Social activities

Humans are social beings and social activities contribute to protecting cognitive functioning and reducing dementia risk. One recent meta-analysis addressed the role of social interactions as a protective factor against dementia [21]. Kuiper et al. [21] show that low social participation as well as less-frequent contact and social isolation in later life increase dementia risk. The roles of network size and satisfaction with one’s social network are less clear.

Education, mental demands at the workplace and a cognitively active lifestyle

A large body of literature showed that low education is linked to increased dementia risk. Two recent meta-analyses demonstrate the protective effect of education against AD [16] and dementia [22]. Focusing on a potential dose–response relationship, Xu et al. [22] showed a 7% reduction in dementia risk for every additional year of education (RR 0.93, 95% CI 0.92–0.94; p = 0.000). Results implicate that cognitive activity, starting at a very early age, is a protective factor. This is consistent with the notion that once established, a cognitive reserve may then act as a buffer against age-related impairments [23].

The majority of the population spends a great amount of their time at the workplace. Therefore, work has a strong influence on the quantity and quality of daily cognitive activities. However, this field remains under-researched to date. One review reported evidence for a protective effect of high job control and work complexity, relating to people and data, against cognitive decline and dementia [24]. Further studies point to the relevance of mentally stimulating workplaces for brain health [25].

While education mostly takes place in childhood and young adolescence, a cognitively active lifestyle (e.g. with regard to leisure activities) is also important in midlife and beyond. Yates et al. [26] conducted a series of meta-analyses on this topic. Four out of five meta-analyses showed significant associations between participation in cognitively demanding leisure activities and reduced risk of cognitive impairment and dementia. The studies refer to a diverse set of activities with typical elements being reading, playing games (e.g. chess or card games), crossword puzzles and sometimes theater or artistic activities.

Diet

What we eat and drink has an impact on our health and wellbeing. With regard to dementia prevention, the so-called Mediterranean Diet (MeDi), a specific dietary pattern, is at the center of research interest. The MeDi is a balanced diet rich in fibers and fresh vegetables with unsaturated fats (olive oil), fish and little red meat. Recent meta-analyses show a consistent pattern of protective associations between a MeDi and AD [27] or dementia [28,29,30].

Furthermore, different dietary components were investigated in the identified meta-analyses. Results suggest a protective effect of unsaturated fats [14, 27], antioxidants and vitamin B [27], vitamins E and C, flavonoids, niacin and folate [28] as well as positive effects of coffee and tea [31]. Detrimental effects were reported for saturated fats [32] and low vitamin D. More research in this field is needed.

Alcohol and smoking

Excessive drinking increases dementia risk [33]. Nevertheless, there is evidence that mild consumption of alcohol may be protective against dementia [28, 33]. More research is needed to understand the non-linear, potentially J-shaped relationship between alcohol intake and dementia risk. Also, it is clear that excessive alcohol consumption is linked to multiple negative outcomes including addiction as well as specific forms of alcohol-related dementias [34].

Smoking bears a variety of serious health risks. A large evidence base suggests that smoking is also risk factor for dementia [14, 28] as well as for ACD, AD and VaD specifically [16, 27, 35]. Zhong et al. [35] suggest that dementia risk for former smokers reverts to approximately the same as for non-smokers. Although not yet firmly established, there may also be a link between passive smoking and risk for cognitive impairment or dementia [36].

(Treatable) chronic conditions as risk constellations for dementia

There are a number of, mainly chronic, disorders related to dementia risk, and their combination may even intensify the detrimental effects. Hypertension is a risk factor for dementia, especially when experienced during midlife [14], and antihypertensive treatment can reduce the risk for ACD [37] as well as AD and VaD [38]. In addition, diuretics were associated with reduced risk for dementia and AD [39]. Evidence suggests that diabetes, especially type 2 diabetes T2D, is a strong risk factor for dementia [14, 40], AD [41], VaD and non-VaD [40], and there is evidence that insulin sensitizers are connected to reduced dementia risk [42]. Obesity is common, and all reviewed meta-analyses [43, 44] and reviews [14, 45] show a clear connection between midlife obesity and dementia. More recently, the role of peripheral hearing loss as a common risk factor for dementia and AD was highlighted [6], but associations are inconsistent [46]. Whether this effect is mediated via impairment-related withdrawal from social interactions or could rather be conceptualized as a prodromal symptom remains unclear. Furthermore, depression is a well-known risk factor for dementia [14], ACD, AD and VaD [47]. As is the case with hearing impairment, depression, particularly later in life, could also be a prodrome of dementia [48].

Further risk factors and emerging fields for prevention

Traumatic brain injury (TBI) is primarily a consequence of traffic-related accidents and falls; however, sports injuries and indirect forces such as shock waves from battlefield explosions may also cause it. TBI is related to AD-risk [49, 50].

Although we did not identify any reviews or meta-analyses, there are reports on anticholinergic drug use and increased dementia risk in elderly individuals [51, 52]. This is important because a substantial proportion of the elderly uses anticholinergic drugs which are prescribed for various indications related to neurology, psychiatry and internal medicine. Table 1 gives an overview on systematic reviews and meta-analyses of dementia risk and protective factors.

Table 1 Systematic reviews and meta-analyses of dementia risk and protective factors
Full size table

Discussion

Compelling evidence from research synthesis of longitudinal observational studies

This meta-review is based on systematic reviews and meta-analyses of a large body of longitudinal observational studies. Results provide evidence for the protective effect of typical lifestyle factors such as physical activity, social interactions and a cognitively active lifestyle, as well as the Mediterranean Diet and non-smoking. There is also emerging evidence that mental demands at work matter. Heavy drinking is clearly risky, especially for brain health. We highlighted a number of potentially treatable chronic conditions which are clearly associated with increased dementia risk such as hypertension, type 2 diabetes, obesity and depression. Although the evidence base regarding the negative effects of hearing loss is still small, it clearly shows preventive potential. In addition, TBI is a risk factor for developing dementia. More research is needed on anticholinergic drug use.

For most of the mentioned associations, strength, consistency, temporality, plausibility and a dose–response relationship are established [53]. Information on plausibility comes from research examining mechanisms underlying each of the associations. Three broad mechanisms are discussed: (1) the increase of cognitive reserve (e.g. via education or mental demands at work); (2) the reduction of brain inflammation (e.g. via MeDi); and (3) the reduction of brain damage from vascular, neurotoxic or oxidative stress (e.g. by quitting smoking or treating diabetes). Risk/protective factors can have an impact via one main mechanism (e.g. education increases cognitive reserve), or via more than one mechanism (e.g. physical activity) [6].

Findings from longitudinal observational population-based studies often investigating the lifestyle of individuals over decades in real life scenarios provide a solid foundation for proposing prevention strategies. That many of the risk factors are interrelated and potential effects need to be separated to avoid confounding is often cited as limitation for such studies. Confounding is usually dealt with by statistical adjustment; however, this approach could be incomplete and may become more difficult as the magnitude of risk factors becomes smaller. In addition, more research is needed to gain insight into the effects of interactions between factors that often co-occur (ExE), and between modifiable and non-modifiable risk factors (ExG). This is important because non-modifiable factors may to some extent have an impact on the effects of lifestyle factors. For example, the presence of apolipoprotein E ε4 may moderate the relationship between smoking and AD [35]. In a large German cohort study, Luck et al. [54] showed that physical activity even in late life reduced conversion to dementia and AD or in delays in the onset of clinical manifestations. Indices of interaction indicated no significant interaction between low physical activity and the APOE ε4 allele for general dementia risk, but a possible additive interaction for AD-risk. Although highly valuable, to mention each interaction is beyond the scope of this paper.

Limited results from randomized controlled prevention trials (RCTs)

Following the principles of evidence-based medicine [55, 56], it would be ideal to report evidence from randomized controlled trials (RCTs) for each risk or protective factor mentioned above. Evidence-based medicine classifies studies according to grades of evidence on the basis of the research design, using internal validity. The highest grade is reserved for RCTs and the lowest grade is applied to case series and expert opinion. Observational studies fall at intermediate levels [57]. Andrieu et al. [58] reviewed single-domain lifestyle interventions (RCTs) addressing individuals with or without risk factors for AD over the past three decades. In general, evidence was inconclusive, although several trials on physical activity, cognitive training, or antihypertensive interventions showed some evidence of efficacy. Small and selected samples, inappropriate timing of the intervention, short duration of intervention and follow-up were the most important shortcomings of these trials [59, 60]. It is important to keep in mind that in the last decade our perspective on timing in dementia has substantially changed. While dementia was previously viewed as a disease of old age, there has been a paradigm shift towards recognizing that the disease starts much earlier and has a long preclinical phase. This shift is also reflected in newer diagnostic criteria including milder forms of cognitive impairment, an additional focus on non-memory impairments and criteria related to biomarkers [7, 61,62,63]. Shifting diagnostic criteria towards earlier manifestations has an impact on how we think about prevention in terms of primary and secondary prevention [64]; it also clearly suggests a life span perspective. Furthermore, successfully tackling lifestyle changes in short-duration single-domain interventions is challenging, as lifestyle is usually rooted in a broader context and shows a certain degree of stability over longer time periods [65]. That said, the results from these short-duration single-domain interventions with limited follow-up periods are not surprising.

Based on these observations, the fact that multiple modifiable risk factors operate jointly and that there are age-dependent and combined effects of the risk factors (GxE, ExE), researchers have progressed to multi-domain interventions which address multiple risk factors at once in large groups of individuals with reasonable follow-up periods. The prototype for such studies is the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability FINGER [66, 67]. This study investigated four intensive lifestyle-based strategies—diet, exercise, cognitive training, and vascular risk management—over 2 years in more than 1200 individuals aged 60–77 years who are at risk of dementia according to the CAIDE risk score. The intervention showed small but positive effects on cognition in high-risk individuals. However, it must be noted that small individual effects may have a substantial impact on brain health on the population level [68]. Similar interventions which were not initially targeted to high-risk groups have failed to show positive results [58, 69]. Multi-domain intervention trials making use of new technology such as the “Healthy Aging through Internet Counseling in Elderly” (HATICE) program are under way [70]. AgeWell.de, a recently initiated German multi-domain intervention study in high-risk individuals, is focused on the primary care setting. Although multi-domain interventions seem promising for selected prevention in high-risk individuals, questions remain regarding the dose needed to change behavior, the optimal timing of the intervention during the life course, target groups, modes of delivery, and implementation settings.

The classical research chain of identifying a single risk factor to proof from concept studies, and small and large RCTs to broad implementation studies might not be the ideal framework for investigating lifestyle factors for brain health and the prevention of dementia. Immediately investigating multi-domain interventions in large RCTs seems promising in high-risk individuals, paving the way for selected primary intervention strategies. However, this line of research is still in its early phases.

Is it too early for a public brain health agenda?

The histories of epidemiology and public health have many examples of when pure observation informed successful preventive efforts. It started with John Snow, considered as one of the fathers of modern epidemiology. He identified a public water pump on Broad Street in London as the source of the cholera outbreak in 1854 and persuaded the local council to disable the pump by removing its handle. This has been credited as ending the outbreak. Many other examples followed. Similar to infectious diseases in former centuries, non-communicable diseases such as dementia are currently a huge challenge for health and social systems. Despite the limited evidence from prevention trials so far, we argue that it is time to discuss a public brain health agenda consisting of risk reduction strategies. Although these suggestions are mainly based on dementia risk factor knowledge from a large body of well-conducted longitudinal observational studies, this available knowledge can be used. Some protective factors such as education and lifelong learning cannot be investigated in trials, while other common constellations such as obesity reduction in midlife and late life dementia would require follow-up time-frames exceeding a researcher’s life. The recently observed reduced incidence rates in later-born cohorts in some Western countries suggest that changes in risk factors drive dementia incidence and that risk reduction may have a substantial public health impact [8,9,10,11]. The changes in risk factors responsible for currently declining age-specific incidence rates in the western world may include improvements in living conditions, education and better healthcare, particularly an improved management of cardiovascular risk factors [71]. Furthermore, lifestyle recommendations such as promoting a physically active lifestyle, social integration and a healthy diet are safe and confer many additional health benefits. Considered together, the time seems right for the establishment of a public brain health agenda.

Multiple projections addressed the potential for the prevention of risk and estimated the brain health impact [72]. Norton et al. 2014 estimated the total impact of a set of risk factors—diabetes, midlife hypertension, midlife obesity, physical inactivity, depression, smoking, and low educational attainment—on the reduction of worldwide AD prevalence to be 28.2%, or 9.6 million attributable cases [72]. According to population projections, 30.5% of current AD cases in Germany (305,000 cases) could be attributable to the risk factors considered. Therefore, a 10–50% reduction of all seven risk factors could have potentially prevented 23,000–130,000 of the current AD cases in Germany [73].

Based on the results regarding risk and protective factors for dementia, we suggest corresponding key actions. Table 2 shows ten suggested key actions and recommendations for a brain health agenda: (1) increase physical activity, (2) foster social integration, (3) improve education and foster lifelong learning, (4) provide mentally stimulating workplaces, (5) foster a cognitively active lifestyle, (6) propose a healthy Mediterranean-like diet, (7) reduce alcohol consumption, (8) stop smoking, (9) prevent, diagnose and treat chronic conditions, (10) reduce anticholinergic medication in the elderly. In a further step, these key actions should be translated into more specific public health initiatives. For example, to promote physical activity, activities may range from fostering individual exercise programs to increasing school-based physical education and walkability promoting land-use policies [74]. Different initiatives could be systematically developed from social ecological models (SEMs) that describe the interactive characteristics of individuals and environments underlying health outcomes. SEMs recognize individuals as embedded within larger social systems and have long been recommended to guide public health practice. Based on earlier work, McLeroy et al. [75] offered five SEM levels specifically related to health behaviors: intrapersonal factors, interpersonal processes and primary groups, institutional factors, community factors and public policy [75,76,77]. Therefore, key actions should be further elaborated and broken down to initiatives related to each of the SEM levels. This will also shift attention from the individual to the importance of social and political environments to prevent dementia and improve brain health. Although the majority of key actions may read as individual actions associated with behavior change, one can think of various environmental changes to support such a public health agenda. For instance, increased walkability of cities and improved infrastructure for biking may contribute to more physical activity. Also, neighbourhood centres and multigenerational housing projects could support social integration. Additionally, measures to increase school education and free, online university courses can support lifelong learning, and subsidized access to cultural events may help facilitate a cognitively active lifestyle within vulnerable populations. Food labelling could help consumers to make healthier food choices. Furthermore, environmental changes could address the creation of cognitively stimulating workplaces and support mobility in old age. Returning to George Snow, he could have educated London residents not to drink the water from the pump on Broad Street to change individuals’ behavior. We know, however, that he did something more powerful: he tackled the environment and persuaded the local council to disable the pump.

Table 2 Public brain health agenda: ten key actions and recommendations
Full size table