ReviewLiving alone and risk of dementia: A systematic review and meta-analysis
Introduction
The number of people living with dementia globally is increasing. In 2015 an estimated 47 million people were affected and this is forecast to rise to 131 million by 2050 (Prince et al., 2015). To date there are no effective disease modifying pharmacological interventions, so research has focused on understanding the impact of modifiable lifestyle factors (Livingston et al., 2017). These modifiable lifestyle factors or putative risk factors for dementia are understood to be present throughout the life course (Livingston et al., 2017). Risk factors can be predictive or explanatory (Schooling and Jones, 2018), and the nature of risk is yet to be established in the case of lifestyle factors and dementia. However, modification of lifestyle factors can slow the rate of cognitive decline even after disease onset (Deschaintre et al., 2009).
One way of understanding the importance of a modifiable lifestyle factor is by calculating the Population Attributable Fraction (PAF) as used in the Lancet Commission for dementia (Livingston et al., 2017). The PAF takes into account both the relative risk (RR) and prevalence of the risk factor in the population and is defined as the proportion of incident cases that are attributable to the risk factor. Assuming a causal relationship, if the risk factor in question were to be eliminated, the PAF indicates the proportion of incident cases that would be reduced as a result (Mansournia and Altman, 2018).
Using pooled effect sizes for RRs, Livingston and colleagues (Livingston et al., 2017) reported PAFs for well-established dementia risk factors which included smoking (13.9 %), depression (10.1 %), physical inactivity (6.5 %), hypertension (5.1 %), diabetes (3.2 %) and obesity (2.0 %). In addition to these factors, Livingston et al. (2017), the UK National Institute for Clinical Excellence (NICE, 2015) and the US National Institute for Health (Daviglus et al., 2010) have identified social isolation as a potential modifiable risk factor for dementia.
Livingston and colleagues identified a systematic review and meta-analysis of social relationship risk factors which reported the relative risk of developing incident dementia as 1.57 (95 % CI: 1.32–1.85) associated with low social contact and 1.41 (95 % CI: 1.13–1.75) associated with low social participation (Kuiper et al., 2015). Using the RR for low social contact in combination with an estimate for prevalence from a study on loneliness in the elderly by the Economic and Social Research Council (ESRC, 2003), Livingston et al. (2017) reported the PAF for low social contact as 5.9 %, similar to hypertension and physical inactivity. However, the authors noted that this figure was likely to be a conservative estimate in the absence of more accurate prevalence data (Livingston et al., 2017). Additionally, the use of social contact as a measure of social isolation is limited by the lack of consensus on the definition. Each of the eight studies included in Kuiper and colleagues’ meta-analysis used different definitions of social contact. Four studies captured a frequency component in their definition such as ‘visiting children or other relatives (never vs. at least weekly)’ (Crooks et al., 2008) and four studies used an absolute definition e.g. ‘visiting friends (No vs. Yes)’ (He et al., 2000). In addition to the heterogeneity conferred by the frequency aspect of these categories, not all social contact is equitable. Social contact involving friends, but not increased contact with family, is found to be associated with a reduced risk of dementia (Sommerlad et al., 2019). Adding further to the overall heterogeneity in the reported PAF, the prevalence data relied on data from the ESRC study (2003), which defined low social contact as that of ‘less than monthly’. Thus, the use of low social contact and its associated PAF as reported by the Lancet Commission (2017) is likely to be a crude measure of the impact of social isolation on incident dementia in the population.
Arguably a more objective measure of social isolation is that is that of living alone. The metric of living alone lends itself more favourably to a precise binary definition and, in countries where census data are collected, prevalence data are accurate and readily available.
Thus, the aims of this review were to further investigate social isolation as a risk factor for dementia by conducting a systematic review of the literature on living alone and incident dementia, pooling the effect sizes and calculating the associated PAF.
Section snippets
Systematic search and study selection
A systematic literature search was conducted using search strings from the previously published systematic review on social relationships and the risk of incident dementia (Kuiper et al., 2015). Three databases (Embase, Medline and PsycInfo) were searched from inception to 8th August 2019. Duplicates were removed, a title and abstract screen conducted and then the full texts of the remaining studies were assessed.
Studies were included if they were: longitudinal cohort studies; comprised of
Study selection
A total number of 36,402 articles were identified from the initial database search. After removal of duplicates, articles were screened on title and abstract leaving a total of 227 articles, which were subjected to full text screen. This resulted in 12 studies (Akbaraly et al., 2009; Arai et al., 2004; Bickel and Cooper, 1994; Chen et al., 2011; Fratiglioni et al., 2000; He et al., 2000; Helmer et al., 1999; Holwerda et al., 2014; Paillard-Borg et al., 2009; Rawtaer et al., 2017; Rodriguez et
Discussion
This is the first review to systematically draw together the literature on living alone and risk of incident dementia. The meta-analysis demonstrated an elevated risk of incident dementia in people who live alone. The PAF calculation indicates 8.9 % of the cases of incident dementia in those aged 65 and over are attributable to living alone suggesting that social isolation is more important than previously thought in understanding risk of dementia.
The findings are particularly relevant in
Declaration of Competing Interest
None.
References (41)
Social regulation of leukocyte homeostasis: the role of glucocorticoid sensitivity
Brain Behav. Immun.
(2008)- et al.
Modulation of hippocampal cell proliferation, memory, and amyloid plaque deposition in APPsw (Tg2576) mutant mice by isolation stress
Neuroscience
(2004) - et al.
Influence of social network on occurrence of dementia: a community-based longitudinal study
Lancet
(2000) - et al.
Social relationships and risk of dementia: a systematic review and meta-analysis of longitudinal cohort studies
Ageing Res. Rev.
(2015) - et al.
High cortisol levels are associated with cognitive impairment no-dementia (CIND) and dementia
Clin. Chim. Acta
(2013) - et al.
Does loneliness contribute to mild cognitive impairment and dementia? A systematic review and meta-analysis of longitudinal studies
Ageing Research Reviews
(2019) - et al.
The lancet international commission on dementia prevention and care
Lancet
(2017) - et al.
Leisure activities and the risk of dementia in the elderly: results from the Three-City Study
Neurology
(2009) - et al.
Sociodemographic factors associated with incidence of dementia among senior citizens of a small town in Japan
Care Manag. J.
(2004) A review of adjusted estimators of attributable risk
Stat. Methods Med. Res.
(2001)