1. Introduction

Dementia is an umbrella neurological syndrome resulting from more than 100 brain disorders, most common of which include Alzheimer’s disease, vascular diseases, Lewy bodies, frontotemporal disease etc. [1, 2]. Worldwide, it is estimated that 5–8% of general population aged 60 and over have been diagnosed with dementia in past years [2]. It has become one of the most common causes of dependency and disability among the elderly [1, 2].

Dementia produces an increasing societal burden resulting in a total economic costs of US$ 818 billion (1.1% of the world’s gross domestic product) in 2016 [2], which has been considered as the major challenging health issue in 21^st century [3, 4]. Many health professionals do not follow the regulations or facility policies about human rights and freedom, and still consider physical or chemical restraints as the inevitable approach to control patients’ behavioural symptoms and prevent the disruption of life-sustaining therapies [2]. Furthermore, dementia patients and their families are subject to prejudice, and their life quality is affected [1].

Since 1906 when Alzheimer’s disease has been associated with dementia, enormous investment has been allocated for studying dementia prevention and treatment [5, 6]. Ageing and family history are associated with genetic background which are easily recognized as the irreversible risk factors for dementia [2, 7, 8]. The environmental and behavioural risk factors, such as sedentary lifestyle, lower socioeconomic status (SES) and unhealthy diet have been circumstantially postulated as the risk factors for dementia [7, 9–12]. Studies have consistently revealed that negative psychological functioning, such as depressive symptoms [13] and neuroticism [14] are the risk factors for dementia development. However, to date, the aetiology and pathology of dementia are still not well understood and there is still no treatment currently available to cure dementia or to reverse its progressive course [2]. A statistical study of the prevalence of dementia, or its ultimate result–mortality, globally and in countries grouped by their socioeconomic characteristics may shed light on the role environmental factors play in aetiology of this condition.

Human species had lived in small hunter-gatherer groups for millions of years before they started to live in large scale societies some ten thousand years ago [15]. During the hunting and gathering period, humans have well adapted to cooperative breeding [16, 17], and then evolved alloparental care [18]. Therefore, human’s millions of years of adaptation suggest that biological foundations of human love have genetically shaped humans for flourishing in small communities [19, 20]. However, in the last few hundreds of years, human societies were industrialized quickly. The rapid industrialization has made most people grow up in just core families with few siblings, which is different from how humans had adapted for flourishing. Such discrepancies or mismatches have been associated with mental health in human population [21, 22].

It is well researched that positive psychological well-being has been implicated in health across adulthood [23]. Household creates a social environment which is salient to maintain health for the co-residential members. On a daily basis, the individual members encounter this environment, play their social role and enjoy the social relations [24]. Moreover, studies also showed that large household offers the residents the subjective happiness [25] leading to low risk for residents to develop various cancers [26], for instance female breast cancer [27, 28] and ovarian cancer [29]. Subjective happiness was associated with mental health significantly stronger than with physical health in people with disabilities [30] and hospital patients [31]. A recent study revealed that greater household size has the protecting role against children developing mental health disorders [32].

The above considerations directed us to try to identify whether smaller household could serve as a risk factor for people to develop and die of dementia. Therefore, in this study, we assessed, from a global perspective, whether large household has the inhibitory role in lowering the risk for the residents to die of dementia using empirical population level data obtained from international organizations. This approach has been already used by our team in a number of studies on various other aspects of human health [26, 28, 29, 33–40].

2. Materials and methods

3. Results

Fig 1 shows the relationship between household size and DMR. The relationship is negative along a power curve with moderately strong negative correlation (r = -0.6678, p < 0.001). Subsequent parametric analyses of log-transformed data and nonparametric analyses confirmed the relationship between DMR variables and the household size.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The relationship between household size and dementia mortality rate.

https://doi.org/10.1371/journal.pone.0263309.g001

Table 1 presents relationships between all the variables (dependent and independent) in Pearson r (above the diagonal) and nonparametric (below the diagonal) analyses. Worldwide (n = 169), Spearman’s rank correlation showed that household size was in significant negative correlation to DMR (r  =  -0.579, p < 0.001). This strength and direction of relationship were similar and observed in Pearson’s r household size and SMR variables (r = - 0.524, p<0.001). Worldwide, non-parametric analysis showed that DMR was associated with ageing (r = 0.533, p<0.001), GDP PPP (r = 0.497, p<0.001) and urbanization (r = 0.436, p<0.001). These strengths and directions of the relationship were observed in the Pearson analysis as well (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Data descriptives and bivariate correlations (Pearson, above diagonal & non-parametric below diagonal) between all variables.

https://doi.org/10.1371/journal.pone.0263309.t001

Table 2 shows the relationship between DMR and household size, ageing, GDP PPP and urbanization examined by controlling for the other three variables in a partial correlation analysis. Household size was the only independent variable to have significant correlation (r = −0.332, p < 0.001) with DMR independent of the other three variables (Table 2). Neither GDP nor urbanization showed a correlation with DMR independent of the other three variables despite the fact that each of them (GDP and urbanization) had a significant correlation to DMR in simple bivariate analyses. This suggests that household size was the independent risk factor for DMR. This suggestion was proved true in the subsequent Stepwise linear regression analyses (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Comparison of partial correlation coefficients between dementia mortality rate and each variable when the other three variables are controlled for.

https://doi.org/10.1371/journal.pone.0263309.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Multiple linear regression showing predicting effects of independent variables and identify the significant predictors of dementia mortality.

https://doi.org/10.1371/journal.pone.0263309.t003

Table 3 shows if and how much household size explained the correlations of the other three variables to DMR respectively. In the enter model, when household size was not considered as one of the independent variables, ageing was the only significant predictor of DMR (β = 0.357, p < 0.001). However, when household size was incorporated as an independent variable, it ranked as the strongest predictor of DMR (β = - 0.362, p < 0.001). Ageing was still a significant predictor of DMR (β = 0.226, p < 0.05), but the correlation strength has significantly decreased (z = 1.67, p<0.05). Both GDP and urbanization barely showed correlations with DMR regardless of household size inclusion.

Table 3 also shows that when household size was not included as one of the independent variables, ageing (R² = 0.203) was the only variable included as the significant predictor of DMR. However, when household size was included as an independent variable, household was selected as the variable having the greatest influence on DMR with R² = 0.263, while ageing was placed second increasing R² to 0.289. The other variables (GDP PPP and urbanization) were removed by the analysis as having no statistically significant influence on DMR.

Table 4 presents that, in general, household size is negatively associated with DMR in different country groupings. The highlight of these relationships was that household size was constantly in negative correlation to DMR. Small sample sizes in some categories of countries make differences between correlation coefficients found for them statistically insignificant and thus it is difficult to discern differences among those categories in the strength of correlations. However, in all categories, correlations are negative and often significantly different from zero. In economically developed country groupings, such as in the World Bank High income economics, WHO European Region and OECD all correlations are significant and moderately strong. Lower values of correlation coefficients in some groupings, especially those of lower income categories in the World Bank income classifications, may be a result of less variability in variables correlated when total ranges of their variation are artificially curtailed by categorisation. Averages of household size and dementia mortality in variously categorised countries show a tendency of larger household size to be associated with smaller dementia mortality.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Correlation of dementia mortality to household size in different country groupings.

Partial correlations are calculated when GDP, Urbanisation and Ageing are kept statistically constant.

https://doi.org/10.1371/journal.pone.0263309.t004

4. Discussion

The worldwide trend of increased DMR may have multiple aetiologies, which may act through multiple mechanisms. This study not only suggested that household size may be a major factor for dementia mortality at the population level, but also showed that household size was a determining risk factor overriding risk factors such as ageing, SES and urbanization. This study also revealed that the predicting effect of household size on dementia mortality was independent of the effects of other common risk factors, such as ageing, socio-economic status and urbanization.

Household relations are usually invested with the powerful norms, histories, and emotions that characterize family relationships [24]. From a life course perspective, they are related to the dementia risks in the essential pathways which are grounded on biological, psychological and social contributions [65]. In the industrialized societies, especially in the developed countries, most people grew up in core families with very limited number of playmates and little interaction with neighbours. This is different from how humans had adapted for flourishing through early cooperative breeding [16, 17], and then evolved alloparental care [18], both of which laid the biological foundations of human love which may be heritable generation by generation [19]. The mismatch between the way we live now and how our ancestors did has been postulated as the risk factor for mental disorders in young generation [21, 22].

The lack of cure for dementia, and of accurate diagnoses of specific causal factors, has made it difficult to target preventative interventions [66]. It is well established that appropriate psychological, social support and physical care have been the key strategies for the healthcare for dementia patients [67]. Studies have revealed that, psychologically, family life increases levels of purpose in life [68, 69], which may reduce by 30% risk of dementia [70, 71]. Large household may offer the residents more social engagement in life which is protective against dementia [72, 73].

Compiled 2,200 years ago, Huangdi Neijing has been the fundamental doctrinal source of Asian medicine. It illustrates how emotions are associated with the visceral organs which are in charge of five qi’s (translation: gas; meaning: emotions): happiness, anger, sadness, worry and fear. Among these five qi (emotions), only happiness makes the gas smooth [74], which keeps people healthy. In Western medicine, hundreds of years’ exploration of the manifestation of emotions through physiological responses (mind—body interaction) [75–77] has suggested that the formation of mental experiences (emotions) is closely associated with bodily responses [78, 79].

Alzheimer’s disease is the most common cause of dementia. The most prevalent hypothesis about this pathology is that it occurs when the beta-amyloid (toxic protein) begins to clump around neurons in the brain. The neurons degenerate, which leads to a decrease in synaptic plasticity and ultimately to cognitive decline. Oxytocin has been considered chemical responsible for targeting the removal or reduction of beta-amyloid or improvement of the cognitive ability of patients. An animal model study showed that toxic beta-amyloid damaged synaptic plasticity in mice’s brains, but this beta-amyloid-induced impairment of hippocampal synaptic plasticity in mice was reversed by oxytocin in mice [80]. Interestingly, this study also revealed that oxytocin did not improve brain’s synaptic plasticity in mice’s brains if oxytocin was the only treatment [80]. This may imply that oxytocin was produced as an auto-immune response to the neuron blockage by toxic beta-amyloid in mice’s brains. Similarly, another animal model study found that oxytocin strengthened social memory [81] and improved spatial memory [82] when mice were in their motherhood. In human studies, it has been reported that oxytocin selectively strengthened participants’ memory for social stimuli depending on the participants’ social contexts and individual attachment styles [82, 83].

Oxytocin has been implicated in many aspects of social functioning. The therapeutic effects of oxytocin have not only been explored in dementia prevention and treatment, but also have targeted the treatment of diseases for other aberrant social behaviour related disorders [84, 85], such as autism spectrum disorder [86–89], posttraumatic stress disorder [90, 91], schizophrenia [92, 93], and anxiety disorders [94, 95].

The nature of vascular dementia (VaD) is strongly associated with stroke [96, 97]. To date, there are few therapeutic options to protect cognitive decline arising from cerebrovascular diseases [98, 99]. Worldwide, prevention of strokes and management of post-stroke symptoms have been considered approaches to reduce the vascular dementia initiation [100]. Based on a number of studies in human and animal models, a systematic review conducted by Gutkowska and colleagues concluded that oxytocin has multiple roles in protecting cardiovascular system [101], which prevents VaD onset and might be the candidate treatment for VaD [99].

Frontotemporal dementia (FTD) is an umbrella term for a group of uncommon neurological diseases due to progressive damage to the frontal and/or temporal lobes of the brain. Empathy loss is one of hallmark symptoms of FTD [85, 88]. Oxytocin is an important mediator of social behaviour, potentially enhancing empathy and prosocial behaviours [89]. Finger and colleagues reported that intranasal oxytocin improved the behavioural symptoms in FTD, including levels of apathy and expressions of empathy [89]. And the benefit or efficacy was associated with dosage and time related [89, 91]. This beneficial effect of oxytocin also improved emotional expression processing [93, 95], empathy [102], anxiety [103] and cooperative behaviour [104] in healthy adults and autism patients. Therefore, the mechanism of upregulated oxytocin mediation of empathy and behavioural deficits have been postulated as a potential treatment approach in FTD [89].

Large household promotes more interpersonal interactions between household members that may offer mind-body interaction which offers biological protection against dementia through the therapeutic effects of oxytocin [105–109]. At the same time, positive psychosocial well-being produced by large household may exert a beneficial slow-down on dementia development [70]. Family members who receive more family support may feel comprehensive positive psychological well-being [70, 110–112]. However, a greater meaning (purpose) of life may be the most important psychological resource to lower dementia risk [70, 110]. Sutin and colleagues explored the protective effects of psychological functioning (life satisfaction, optimism, mastery, purpose in life, positive affect) on preventing high risk population from developing dementia [70]. The results revealed that people from large household showed more psychological well-being leading to lower risk of dementia onset [70]. Interestingly, purpose in life explained 30% dementia risk and this protecting role was independent of the competing effects of other multiple risk factors, such as chronic disease and low physical activities, genetic risk, psychological distress and socioeconomic status [70].

Household residents may interact with each other more often to create life satisfaction [25, 113]. They may also share healthcare knowledge, encourage each other to establish healthy lifestyle and utilize health care services in an effective way [114–116]. People with positive psychological well-being tend to practise healthy lifestyle, have more knowledge of health risk factors and attend regular physical examination [27]. The protecting role of such positive psychological well-being has been postulated to decrease the risk in the development of breast cancer and general cancers [27, 36, 117, 118].

Large household is even more important for pre-dementia patient, especially the young onset dementia. Generally, young onset dementia present non-specific signs and symptoms at the early onset in young patient, but they are progressing and irreversible [119]. With household residents’ observations and encouragement, the atypical dementia symptoms can be noticed by co-residential members, and accordingly they can have the examination in time, and proper treatment subsequently.

Additionally, from the perspective of evolution, a population with large household offers more chance to survive the natural selection. This produces the opportunity to have portion of a population with less fitness, for instance, dementia, removed through greater mortality rate without disturbing population’s essential activities [33, 36, 37, 120–123]. In other words, the genetic background of dementia in such population may be more often eliminated from the population with large household without affecting population as a whole. Therefore, population with large household may have less genes/mutations of dementia to incur high mortality rate of dementia. Furthermore, a population with large household means less birth control and high total fertility rate which allows more biological variation in fertility [124]. A portion of this additional variation, however small, provides the opportunity for the natural selection [124].

Interestingly, large household has shown its consistent and significant, but inverse correlations to dementia mortality rates in the developed world. Several unique phenomena in the developed world can explain this interesting relationship: 1) The fertility rate keeps falling down leading to small household/family size. 2) The cultural doctrines, especially individualism and independence, have reduced the interactions between people living in the developed world. 3) Importantly, dementia care delivery in the developed world is primarily through individual home support service, because dementia patients live with small family/household, or through nursing homes [125]. While in the developing world people live with big family/household for most of their life and they would receive the “informal” healthcare from family members and/or household residents [126], instead of the “formal” care through nursing homes. The fact that people live in bigger households in developing countries is a combined result of the economic situation and social customs. With lower incomes and less availability of housing people are more likely to share accommodation while social customs demand respect and care for elderly people, who, in the situation of low incomes cannot afford to have living arrangements independent of members of younger generations. In developed countries higher incomes, greater availability of accommodation, higher mobility and individualistic lifestyles of adults, together with well-resourced retirement situations of older persons allow individuals or couples to have separate households. In this situation, social interactions with other individuals become sporadic personal choices rather than a daily necessity.

5. Strengths and limitations in this study

Due to the inherent uncertainties of the worldwide estimates provided by the WHO and the World Bank, our results must be seen as showing relations between published estimates, not necessarily precisely measured actual phenomena. The relations, however, are the closest available approximation of actual situation.

Little work has been done on the dementia epidemiology studies, which may be due to extremely low onset rate making data collection difficult. For example, worldwide, mortality rate of dementia is only 26.61 per 100,000, and this presence is not noticeable. Therefore, the low prevalence rate of dementia would require unaffordable sample size for identifying small household as the potential risk factor for dementia in individual based epidemiological or laboratory approaches. Ecological studies are based on aggregated quantitative data zooming in the rare presence of dementia 100,000 times, which makes dementia presence noticeable for analysing the potential effects of dementia risk-modifying factors at population level. This also suggests the necessity of engaging ecological study into the epidemiology studies of rare chronic diseases such as dementia and cancer [28, 29, 36] and Type 1 diabetes [33].

Due to the nature of the cross-sectional data, a couple of intrinsic limitations should be mentioned. Firstly, the results in this study only showed the relationship between household size and dementia mortality rate as correlational, instead of causal. Secondly, the results based on the ecological approach in this work are subject to the “ecological fallacy”. Therefore, the protective role of large household size may not always hold true for each individual to predict their dementia specific death risk. Thirdly, dementia mortality has been associated with multiple risk factors which could have confounded the relationship between large household/family size and dementia mortality rate in this study. These factors include, but are not limited to, hypertension, obesity, head injury and hearing loss in middle life, and smoking, depression, diabetes and social isolation in later life. However, these factors have not been well established for explaining dementia mortality. Also considering unavailability of the data on these factors at population level, we cannot include them as the potential confounders for ruling out their competition with the detrimental effect of small household/family size on dementia mortality in this study. Finally, the data employed in this study might be crude. The WHO, United Nations and World Bank may have made some random errors arising from the methodologies used for collecting and aggregating the data.

Regardless of the strength and limitations of the data quality, we have showed that countries with large household size had lower dementia mortality rate in different data analysis models. The findings in this study may have shed light for further research into the subject with exposure based longitudinal cohort studies at population level. Accordingly, this may lead to far reaching public health implications in dementia and its prevention.

6. Conclusions

Independent of ageing, urban lifestyle and low socio-economic status, large household has been shown to have significant protective role against dementia mortality in this study. This forward-thinking study approach will help shed light on the further study into the protective role of positive psychological well-being against dementia. As part of dementia prevention, when large household/family size is impossible to achieve, healthcare practitioners should encourage people to increase their positive interactions with people from their neighbourhood or other fields.

Supporting information

Acknowledgments

The authors express appreciation to Dr Cherian Varghese from the Management of on communicable Diseases, Disability, Violence and Injury Prevention Department (NVI) of World Health Organization for the assistance in locating the data.

The authors also appreciate the great assistance from the reviewers in improving this study.