https://orcid.org/0000-0002-9115-304X
Figures
Abstract
The aim of this paper is to study the association between childhood circumstances and loneliness in older adults in Europe. Based on rich information collected by the Survey on Health, Ageing, and Retirement in Europe (SHARE) on childhood characteristics and individual characteristics at age 50+, the study is able to control for personality traits, socioeconomic and demographic factors, social support and health in later life, and country-specific characteristics. The analyses show strong correlations between life circumstances in childhood and feeling lonely in older age; these correlations remain significant after adjusting for covariates. While ill health is the main factor correlated with loneliness at 50+, as expected, the analysis of the relative importance of the determinants reveals that personality traits account for more than 10% of the explained variance and that life circumstances during childhood account for 7%. Social support at older ages is the second highest category of factors, accounting for 27%—with, interestingly, support at home and social network characteristics contributing about 10% each, engaging in activities and computer skills accounting for 7% of the explained variance. Demographic and socioeconomic factors account for 6% and country-level characteristics contribute 5%. This paper points out the relevance of early life interventions to tackling loneliness in older age, and it shows that early interventions and interventions aiming at increasing social support in later life need to be adapted to all personality types. Thus, the role of childhood circumstances and the mechanisms explaining the association between loneliness in childhood and loneliness in later life deserve more attention in future research.
Citation: Guthmuller S (2022) Loneliness among older adults in Europe: The relative importance of early and later life conditions. PLoS ONE 17(5): e0267562. https://doi.org/10.1371/journal.pone.0267562
Editor: Enrico Mossello, University of Florence, ITALY
Received: August 13, 2021; Accepted: April 12, 2022; Published: May 18, 2022
Copyright: © 2022 Sophie Guthmuller. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: This work uses data from SHARE Waves 6 and 7 (DOIs: 10.6103/SHARE.w6.800, 10.6103/SHARE.w7.800), see Börsch-Supan, Brandt et al. (2013) for methodological details. The SHARE data collection has been funded by the European Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, DASISH: GA N°283646) and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, SHARE-COVID19: GA N°101015924) and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, and VS 2020/0313. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C, RAG052527A) and from various national funding sources is gratefully acknowledged (see [http://www.share-project.org)]www.share-project.org). Börsch-Supan, A., M. Brandt, C. Hunkler, T. Kneip, J. Korbmacher, F. Malter, B. Schaan, S. Stuck, S. Zuber and S. C. C. T. on behalf of the (2013). "Data Resource Profile: The Survey of Health, Ageing and Retirement in Europe (SHARE)." International Journal of Epidemiology 42(4): 992-1001. Other researchers can access these data in the same manner.
Funding: This work is partly funded by the Leibniz Science Campus Ruhr to which the author is affiliated. https://lscr.rwi-essen.de/en/loneliness-among-ageing-european-populations The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The author has declared that no competing interests exist.
Introduction
Loneliness has been a growing topic of interest in Europe over the last decade [1–4], as it has been shown to be linked with ill health and to increase with age. Loneliness is correlated with a higher risk of developing mental conditions (e.g., depression, dementia) and a deterioration in physical health (e.g., less active lifestyles, diabetes, stroke, coronary heart disease), as shown for instance in [5–11]. Loneliness is also linked with all causes of mortality [12–15] and has an impact on health care utilization [16]. In 2016, 6% of the European population declared that they felt lonely most of the time, and this proportion reached 9% for the population aged 65+ years [17].
In addition to ill health, which is one of the major predictors of loneliness among older adults, another set of individual characteristics in later life is known to be significantly correlated with loneliness. Among these characteristics are demographic and socioeconomic factors, household characteristics (e.g., size of household, marital status), and the amount of social support (participation in social activities, frequency of contact with family members and friends, size of social network, closeness to social network). For an overview, see, for instance [2, 18]. Recent studies have also shown that certain personality traits—in particular, neuroticism and extraversion–are significantly associated with loneliness [19–21].
However, there is less empirical evidence on the effect of life events on loneliness, especially of early life circumstances. The link between early life conditions and loneliness at age 50+ can be explained through several underlying mechanisms. The exposure to stressful events in childhood (e.g. financial distress, ill health, physical harm) or the child social environment (e.g. the parent–child relationship, friendships) might influence its social development and self-esteem, which could have direct and indirect long-lasting effects on the child’s risk of suffering from loneliness later in life. Numerous studies have found significant effects of early life conditions on later life outcomes, including health-related variables, educational attainment, and employment [22–30].
The majority of the studies that examined the determinants of loneliness among older adults are cross-sectional [31–34] with few exceptions, to the best of our knowledge. For instance, the paper by Aartsen and Jylhä analyzed the onset of loneliness based on a prospective study in Tampere, Finland; they found that life events such as losing a partner have a significant effect on loneliness [35]. Using national survey data in Sweden, Dahlberg et al. studied the link between social engagements in early life and social engagement in later life and how it is correlated with feelings of loneliness [36]. The only study that focused on the link between early life circumstances and loneliness in later life is the one by Kamiya et al. [37]. The study showed that poor childhood socioeconomic status and parental substance abuse have direct effects on loneliness at older ages in Ireland, after accounting for demographic, socioeconomic, and health factors.
This paper aims to contribute to the existing literature by answering the following questions. First, are life circumstances in childhood significantly associated with loneliness in later life in Europe? If the answer is yes, does the significant association remain after controlling for personality traits, demographic and socioeconomic factors, social support and health in later life, and country-specific fixed effects? Second, what is the relative importance of the link between childhood circumstances and loneliness in later life, compared to later life conditions, including country-specific characteristics?
This study addresses these questions using rich individual-level data from 17 countries in Europe allowing accounting for a larger set of individual characteristics in early life and later life, and in particular, country specific fixed effects. Loneliness is measured with the R-UCLA Loneliness Scale. It uses indirect questions that do not mention the word “loneliness” in order to account for under-reporting related to the social stigma of suffering from loneliness [38–40].
The remainder of the paper is structured as follows. Section “Background” presents the evidence from the empirical literature and the underlying mechanisms that explain the link between early life circumstances and loneliness at older ages. Section “Subjects and methods” describes the data and the estimation strategy. The results are reported in Section “Results”. Section “Discussion” discusses the results and limitations of the study.
Background
Loneliness is defined as “a situation experienced by the individual as one where there is an unpleasant or inadmissible lack of (quality of) certain relationships. This includes situations in which the number of existing relationships is smaller than is considered desirable or admissible, as well as situations where the intimacy one wishes for has not been realized” [18]. Therefore, loneliness is a multidimensional phenomenon that may include (subjective) social isolation—the (perceived) “lack of, or deficit in, the quantity of a social network” [41], and emotional isolation—“the lack of person(s) to whom one feels attached” [41].
While social isolation and loneliness are strongly associated, lonely persons are not necessarily socially isolated, and persons with a small social network do not necessarily feel lonely. Hence, loneliness is a subjective and complex feeling moderated by different individual factors that can be grouped as follows: childhood circumstances, personality traits, demographic and socioeconomic factors, social support and health, and macro-level characteristics. The remainder of this section “Background” presents the evidence from the literature and describes the underlying mechanism through which these factors could predict later life loneliness.
Childhood circumstances
The mechanisms through which early social relationships can be associated with loneliness in later life are mainly grounded in social developmental theories [42]. Based on Bowlby and Ainsworth’s theory of attachment [43–45], it has been shown that the parent–child relationship—and the mother–child relationship in particular—plays a crucial role in developing secure attachment skills, leading to better social, emotional, and cognitive skills later in life [46, 47]. A lack of (close) friendships is similarly detrimental to the development of social skills, self-esteem, and self-perception, and it leads to increased long-lasting feelings of rejection [48, 49]. The role of friendship has been significantly linked to later life outcomes related to social development and cognitive functioning, as well as to mental health and well-being [50–53]. More generally, a number of empirical studies have shown the importance of the link between early social relationships and outcomes in later life. For instance, Ejlskov et al. showed that relationship adversities throughout life, including in childhood, increase the risk of feeling lonely in later life [54]. Previous research also revealed that socially isolated children tend to have lower subsequent educational attainment and be part of a less advantaged social class in adulthood, and that they are more likely to be psychologically distressed in adulthood [55].
The link between early life conditions and loneliness at age 50+ can be explained by exposure to stressful events in childhood, such as financial distress, ill health, and physical harm, which have been shown to have long-lasting effects on a child’s well-being and various later life outcomes; health-related variables, educational attainment, and employment [22–30]. Early life conditions may have a direct link to loneliness in later life through personal constraints (deficit in social skills, low self-esteem, powerlessness, expectations about self-efficacy, self-perceived lack of disclosure to others) [18, 56] that remain over the course of life. They may also have an indirect link—for instance, through educational attainment, employment status, or relationship status—that in turn affect loneliness at age 50+. A couple of studies examined the correlation between early life conditions and risk of loneliness in later life [37, 57]. Kamiya et al. found that parental substance abuse and financial distress during childhood are associated with an increased level of loneliness in later life [37]. Childhood trauma (and adulthood trauma) were independently related to the most distressed loneliness classes [58].
The importance of religion in the family is generally linked with a set of norms and cultural constraints that favor collectivistic values and moderate self-definition and role conception of individuals within the group. In this view, growing up in a family with strong collective values influences self-perception on being alone, feeling lonely, and social connectedness [59]. It has been shown that, in collective societies, individuals with fewer social connections tend to feel lonelier, as they do not comply with a cultural value in which the focus is the group. They therefore tend to feel more responsible for their lack of social network and their feeling of loneliness [60, 61]. In more individualistic societies, individuals tend to be more independent, oriented to themselves, and thus might be less impacted by fewer contacts or social isolation.
Personality traits
Recent research shows that the propensity of feeling lonely is also linked to personality. Neurotic and anxious personalities are expected to be positively linked to loneliness, whereas extraversion, agreeableness, and conscientiousness tend to be protective personality traits [62]. Using the five big personality traits, Wang et al. found that neuroticism was significantly correlated with a higher level of loneliness in older adults [20]. In a longitudinal study following subjects from adolescence to midlife, neuroticism was found to predict levels of subjective health and loneliness later in life [63]. Extraverted older adults, on the other hand, were found to suffer less from loneliness [21, 64]. using multivariate models controlling for individual confounding factors in adulthood, Buecker et al. found that all personality traits were significantly associated with loneliness, except openness [65]. Neuroticism was found to predict the development of subjective health and loneliness later in life.
Demographic and socioeconomic factors
The literature has shown that loneliness varies across age groups and life stages. Among older adults, one would expect that loneliness increases with age and the likelihood of being alone. With a higher life expectancy, women are also more exposed to loneliness than men. However, the evidence from the literature regarding these two factors of age and gender are rather mixed, indicating that other associated factors, such as health and social support, might have a larger impact on loneliness [66–69].
Another factor associated with age and gender relates to work situation and wealth. Having employment is expected to be linked with more frequent social interactions and higher socioeconomic status, which in turn leads to a lower risk of loneliness. Empirical results indeed found that the retired and unemployed populations have a larger likelihood of suffering from loneliness. The higher risk of loneliness among those with low levels of income can be partly explained by their more frequent experience of stressful situations in order to be able to make ends meet, and less frequent participation in unaffordable social activities. See, for instance, [70, 71].
Social support in later life
Social support is one of main moderators of loneliness in later life. Comparing the onset of loneliness in different countries, Sundström et al. found that living alone was the most consistent factor related to a higher level of loneliness [33]. Adverse family life events among older adults have also been found to be largely associated with loneliness. Vozikaki et al. showed that a child’s departure from home or the death of a partner are major predictors of steady feelings of loneliness [31]. Beyond family characteristics, another part of the literature focuses on the link between social connection and loneliness in older adults. Several papers in Europe and in the US showed that having a network of confidants has a greater effect in preventing loneliness than living alone [10, 34, 72–74]. In addition to regular contact with family and friends, taking part in social activities is also a protective factor of loneliness. Niedzwiedz et al. even found that social participation reduces the socioeconomic differences in loneliness among older adults in Europe [70]. The environment, such as living in a rural area or in a large city, is associated with higher levels of loneliness, although living in a deprived region matters more [75–77].
Access to technology and the internet, in particular social media, has also been linked to mental health issues and loneliness, especially among young adults and adolescents [78]. Among older adults, internet communication and social media tools are used primarily to stay in contact with children, other family members, and friends. As such, they are associated with a lower level of loneliness [79].
Health conditions in later life
Ill health is among the main predictors of social isolation and is therefore significantly correlated with loneliness. Older adults with ill health are more likely to suffer from loneliness—see, for instance, Barlow et al. and Meltzer et al. who studied levels of loneliness in chronically ill persons or adults with mental disorders [80, 81]. Health conditions—such as chronic diseases, IALD limitations, depressive symptoms, and subjective ill health status—are found to be strong risk factors for social isolation and loneliness [31, 33, 82].
Country-specific characteristics
In addition to these individual characteristics, country-specific characteristics also explain higher risks of loneliness. Macro-level demographic and economic factors, inequalities, cultural norms and values, levels of safety, and the existence and extent of public and social policies are among the country-specific characteristics that were found to be associated with social isolation and loneliness [32, 33, 83–85].
Subjects and methods
Survey on Health, Ageing, and Retirement in Europe (SHARE)
The data from the Survey on Health, Ageing, and Retirement in Europe (SHARE) provide the information for individuals aged 50+ on health, socioeconomic status, and social and family networks. The cross-national panel database covers 27 European countries and Israel. The regular questionnaire (SHARE) has been carried out every two years since 2004. In addition to the regular questionnaire, a SHARELIFE questionnaire was carried out in 2009 (wave 3) and in 2017 (wave 7). The SHARELIFE questionnaire is a retrospective survey collecting information about past and early life experiences related to employment, health, family, and housing situation. The questionnaire followed a so-called life history calendar (LHC), helping respondents remember the chronology of past events. The individuals included in the study are those aged 50+ who participated in wave 6 [86] and replied to the SHARELIFE history questionnaire in wave 7 [87, 88]. The individuals live in Austria, Germany, Sweden, Spain, Italy, France, Denmark, Greece, Switzerland, Belgium, Israel, the Czech Republic, Luxembourg, Portugal, Slovenia, Estonia, and Croatia.
Loneliness at 50+.
The data to characterize individuals at 50+ come from wave 6 (2015). The short version of the Revised UCLA Loneliness Scale [38–40, 89] is used as the outcome of interest. This scale is based on the answers to three items: How much time do you feel…: (i) you lack companionship; (ii) left out; (iii) isolated from others? Participants responded to these three items on a three-point Likert scale: “often” (3), “some of the time” (2), “hardly ever or never” (1). The addition of points from the three items determines the R-UCLA Loneliness, which ranges from 3 (not lonely) to 9 (very lonely). This paper studies the determinants of moderate to severe levels of loneliness, taking into account the differences in self-assessed loneliness across countries. Therefore, respondents are defined with a level of loneliness scale, in which the fourth country-specific quartile was defined as “feeling lonely” and those in the first, second, and third quartiles were defined as “not lonely.” Focusing on the highest quartile of the lonely population is most relevant for policy implications. From a statistical point of view, this definition is also used to account for the non-normal distribution of the loneliness scale among respondents [70, 90].
Childhood circumstances.
Information on childhood (before age 17) is retrieved from the wave 7 SHARELIFE questionnaire. Participants were asked to rank how they felt during childhood about a number of items. For childhood circumstances, the following items were considered: having a group of friends they felt comfortable spending time with; their relationship with their mother/father; whether their family was pretty well off financially, about average, or poor; their subjective health status; and the importance of religion at home. A dummy variable was also added to indicate whether the child was the only child in the household. Finally, the variable “never being physical harmed” was recoded as (never vs. often, sometimes, rarely) from the question “How often did your mother / father / or another person that raised you, push, grab, shove, throw something at you, slap or hit you?”
Personality traits.
The 10-item Big Five Inventory (BFI-10) is used to study the association between personality and feelings of loneliness, introduced for the first time in wave 7. It is an established personality inventory measuring the “Big Five” personality dimensions with two items each. Introduced by Rammstedt and John in 2007 [91, 92] the BFI-10 is an ultra-short measure of personality, which is especially suitable for multi-theme surveys in which assessment time and questionnaire space are limited. As such, the BFI-10 measures the following five personality traits: openness vs. closedness to experience; conscientiousness vs. lack of direction; extraversion vs. introversion; agreeableness vs. antagonism; and neuroticism vs. emotional stability. Each personality trait has a score ranging from low to high, as 1(0.5)5 [91, 92].
Demographic and socioeconomic characteristics.
As previously shown in the literature, differences in gender, age, and highest education level (divided into low, or medium and high), and employment status are controlled for as drivers of hampering factors of loneliness. Wealth, as an economic resource variable equalized by the OECD equivalence scale, is used and defined as country-specific quintiles to account for the difference in employment status. Wealth includes household financial (income, money in bank accounts, etc.) and real assets (value of own residence or vehicle).
Social support at 50+.
To measure social support at home, the marital status and size of the household are included. Marital status is defined as being married or in a civil partnership, compared to divorced/separated, never married, or widowed. The size of the household considers the number of individuals (one, two, three, or more). The set of variables available in the module dedicated to social network in wave 6 is used to measure the quality of the social support received inside and outside of the family. The social network is defined as persons with whom one most often talks about important things—the person can be a family member, friend, neighbor, acquaintance, etc.). The size of the network (the number of persons in the social network), the frequency of contact with the social network (daily, several times a week, about once a week, less than once a week to never), the mean closeness of the social network (defined as a categorical variable with three levels: close, very close, extremely close), and the geographical proximity to the social network (same household, less than 1 kilometer away, between 1 and 25 kilometers away, more than 25 kilometers away) are included. The geographical environment is identified by proxy through information on the area where the household is located (a rural area or village, a big city, the suburbs or outskirts of a big city, a large town, a small town).
To measure social participation, information on participation in activities and self-assessed computer skills are used. The variables of number of activities the last twelve months and level of satisfaction with activities were combined to create a categorical variable that takes into account the level of satisfaction of individuals reporting participation in any activity. Two subgroups are defined of those who replied that they have participated in any activity: those who are satisfied with no activity (rated 5 or more on a scale of 1 to 10) and those who are not satisfied with no activity (rated 4 or less). The number of activities is then composed of the following items: none & unsatisfied; none & satisfied; one, two, three, four & more. The variable of computer skills consists of the following categories: excellent or very good, good, fair, poor, I never used a computer.
Health status at 50+.
To control for ill health status as a major loneliness driver, the following are used: information on the number of chronic diseases (none, one, two or more); the EURO-D scale that measures depression; and the Global Activity Limitation Index (GALI), which defines ill health as “being limited” for six months or more in activities people usually do [93].
Sample.
The working sample has the following main characteristics. It includes 27,623 observations for which there were no missing values in the variables used in the analysis. The initial sample includes 33,523 observations (See S1 Appendix for more detailed information on the exclusion criteria). The analysis of the missing observations by variable shows that the proportion of missing observations ranges from 0.04% to 6.06%. The analyses were run including a missing category when applicable, and the estimates and the findings do not vary with and without the inclusion of the missing observations. In addition, a very small proportion of some variables (maximum 1%) are imputed values—except for income, for which the proportion of imputed values is around 30%. Multiple imputation techniques (hot-deck method and fully conditional specification method) have been used by the SHARE project data team. (See detailed description of the methods and the procedures in [94] and [87]. A sensitivity analysis was performed to assess whether the results are affected by the imputed values, and it concludes that the estimates are robust. The results are available upon request.
The working sample is 57.10% female, with an average age of 67 years old; 57.94% are retired, and 71.53% are married. 37.31% have a low educational level. The most frequent household size is two individuals (57.45%), and 19.56% are single households. 35% live in a rural area. Regarding loneliness, the average loneliness scale is 3.85 (min = 3, max = 9), and 17.11% of individuals have a loneliness score within the fourth country-specific quartile. The figures are reported in Tables 1–4.
Statistical analysis
Descriptive analysis.
A descriptive analysis is performed to compare the individual characteristics of lonely vs. not lonely individuals. Tests of difference in mean between the two groups are used for age and the UCLA loneliness scale. Tests of difference in proportion between the two groups are used for binary variables and for each item of the categorical variables. Standard errors are adjusted for clustering at the country level. The Pearson Chi2 test of independence is also reported for the categorical variables.
Multivariate equation model.
The equation defining the relationship between loneliness at age 50+ and its determinants can be written as follows:
(1)
Li is a dummy variable indicating whether an individual i has a level of loneliness at its country-specific fourth quartile of the R-UCLA scale. Ci is a vector of childhood circumstances (wealth; health; quality of relationship with mother, father, and friends; importance of religion). Pi is a vector of the five scores of the personality traits (extraversion, agreeableness, conscientiousness, neuroticism, and openness). The vector Di includes a set of individual socioeconomic and demographic variables at age 50+. Si is composed of social support, activities, and network variables at age 50+. Hi is a vector of individual health information. Ic is a country-specific fixed effect to account for differences in loneliness level between countries. ei is the error term accounting for unobserved characteristics. Eq 1 is estimated with a logit model and clustered standard errors at the country level. Five specifications of the model were estimated, including each category of factors consecutively. Specification 1 includes the vector Ci. Specification 2 includes the vectors Ci and Pi. Specification 3 includes Ci, Pi, and Di. Specification 4 includes Ci, Pi, Di, and Si. Specification 5 is the full specification and includes Ci, Pi, Di, Si, and Hi. All five specifications are estimated with country fixed effects (Ic).
The variables included in the model were selected following a stepwise (forward hierarchical selection by category of factors, and backward selection within category) procedure and the Akaike Information Criteria (AIC). The pseudo R2 and the proportion of correctly specified outcomes are reported as measures of model fit for each of the specifications. Potential multicollinearity issues were ruled out using the correlation matrix and Variance Inflation Factor (VIF). Results available upon request.
Decomposition of the effects.
Once the influence of each characteristic on the probability of feeling lonely is known, an analysis of the importance of the association between different group of factors (e.g., life circumstances in childhood, social support in later life) and loneliness in later life can be conducted. To measure the relative importance of each category of factors, the relative contribution of each these categories within the explained variance of Eq 1 is computed [95]. The computation follows McKelvey and Zavoina in 1975 [23, 96], breaking down the pseudo R2 to measure the share of variance explained by each category of variables Xc associated with a coefficient βc, using linear predictions of loneliness. The ratio (Rc) of the contribution of each category of variable c is computed as follows:
(2)
where
is the linear prediction of Eq 1. Standard errors of the ratios are bootstrapped iterations.
Overall, the following p-value thresholds are reported in the statistical analysis; * p < 0.05, ** p < 0.01, *** p < 0.001. Given the sample size, results with p-values below 0.01 are commented as statistically significant in the text.
Results
Descriptive analysis
Among the population reporting feeling lonely at age 50+, the proportion of individuals reporting often having friends with whom they felt comfortable spending time with during childhood is significantly lower compared to the population without feelings of loneliness at age 50+. Similarly, the frequency of growing up with no other child in the household is 5.7 percentage points higher among individuals feeling lonely at age 50+. The quality of the relationship with parents is also negatively correlated with loneliness in older age. The proportions of individuals with excellent or very good relationships with their mother or father are significantly lower, while the proportions of those reporting a fair or poor relationship are higher among the subsample of lonely individuals aged 50+. This pattern is also found when looking at health and wealth indicators; the proportions of having experienced physical harm, having bad health, and growing up in a poor household are all significantly higher among the lonely subgroup at age 50+. Finally, individuals growing up in a household in which religion was very important are also more frequent among the lonely population in older age (see Table 1).
The two populations also differ by their personality traits. The score for agreeableness, conscientiousness, and openness are significantly lower within the lonely population, while the score for neuroticism is significantly higher (see Table 1).
Regarding the demographic and socioeconomic factors, the proportion of women is 9.7 percentage points higher among those feeling lonely compared to those without feelings of loneliness. Lonely individuals at age 50+ are around two years older. The proportion of employed individuals is lower among lonely individuals, while the proportion of “other types of employment” status is higher among the lonely population. This category includes the unemployed, homemakers, self-employed, and permanently sick individuals. The proportions of individuals with low educational level and lower income (1st and 2nd quintiles) are also higher among individuals feeling lonely. Conversely, the proportion of individuals with higher income (4th and 5th quintiles) is higher among the population without feelings of loneliness (see Table 2).
In terms of health, individuals with more than one chronic disease, limitations on activities, and symptoms of depression are significantly more common among the population feeling lonely at age 50+ (see Table 2).
Table 3 reports the quantity and quality of social connection by loneliness status at age 50+. As expected, the two subgroups differ in their social connections. First, social connections within the household are found to be protective factors for loneliness. The proportion of married individuals is 18.0 percentage points lower among the lonely population. The proportions of widowed, never married, and divorced individuals are higher within the lonely population. Similarly, one-person households are more frequent among the lonely group.
Second, the size and the tightness of the social network also seem to play an important role in impeding feelings of loneliness. The average size of the network—the number of persons with whom one can discuss important things is slightly lower among lonely individuals (around 2.5 vs. 2.7). The level of closeness with the network, frequency of contact, and proximity to the social network, are significantly different between the populations with and without feeling of loneliness. The proportion of the sample with daily contact with the network is on average 6.3 percentage points lower within the group of lonely individuals. The proportion of the sample reporting feeling extremely close to their network is 10.5 percentage points lower among the lonely population. The average geographical proximity to the persons within the network is also higher for individuals feeling lonely. The area of living, however, does not seem to differ between the two groups.
Third, the loneliness status differs with the engagement in activities and the level satisfaction. The proportion of individuals reporting participating in no activities and being unsatisfied is 7.5 percentage points higher in the lonely group. The proportion of those engaging in more than two social activities are also significantly lower in the population feeling lonely at age 50+. Interestingly, the proportion of those with at least a good level of computer skills is significantly lower in the lonely population; while the proportion of individuals reporting having never used a computer is 12.8 percentage points higher within the lonely group (see Table 4).
Multivariate analysis
Childhood circumstances.
First, the association between childhood circumstances and loneliness at age 50+ can be examined. Fig 1 displays the odds ratios of the variables describing childhood circumstances for the five different specifications. This allows for the comparison of the size of the association between childhood circumstances and loneliness at age 50+, after controlling additionally for personality traits, socioeconomic and demographic factors, social support, and health. Having friends with whom they could comfortably spend time as a child is negatively correlated with loneliness in later life. The size of the association decreases gradually with the number of factors included in the model, but it remains significant in all specifications, including in the full model that takes into account the influence of personality traits, health, social support, and socioeconomic and demographic variables. In the full specification (5), the odds of loneliness at age 50+ is 1.15 higher for individuals who sometimes had comfortable friends, and 1.24 significantly higher for those who rarely or never had comfortable friends, compared to those who replied that they often had comfortable friends to with whom to spend time during childhood. In the remainder of this section “Multivariate analysis”, the odds ratios presented are those of the full specification (5).
Note: The figure displays odds ratios of the variables describing childhood circumstances for the five different specifications. Specification (1) includes C, a vector of childhood circumstances (wealth; health; quality of the relationship with mother, father, and friends; importance of religion), and I, country-specific fixed effects. Specification (2) includes (1) and P, a vector of the five scores of personality traits (extraversion, agreeableness, conscientiousness, neuroticism, and openness). Specification (3) includes (2) and D, a set of individual socioeconomic and demographic variables at age 50+. Specification (4) includes (3) and S, social support, activities, and network variables at age 50+. Specification (5) includes (4) and H, a vector of individual health information at age 50+. Reading example: in the full specification, reporting having had rarely or never friends with whom spending comfortable time in childhood increases the risk of feeling lonely at age 50+ by 1.24. The pseudo R2 and the proportion of correctly specified outcomes are respectively for each specification: (1): 0.0364; 82.89%, (2): 0.0600; 82.99%, (3): 0.0864; 82.96%; (4): 0.1181; 83.39%, (5): 0.1699; 84.09%. * p < 0.05, ** p < 0.01, *** p < 0.001.
The quality of the parent–child relationship during childhood is also a significant factor impeding loneliness in later life, especially the mother–child relationship. Those having had a fair or poor relationship with their mother as a child have 1.34 higher odds of feeling lonely at age 50+ compared to those with an excellent relationship. In contrast, the father–child relationship does not have a significant effect. Similarly, having been the only child in the household during childhood is not a significant factor of loneliness once the later life factors are included (D, S, and H). However, having experienced physical harm as a child increases the likelihood of loneliness at age 50+ by 1.11. The odds of loneliness at age 50+ is 1.09 higher when one had bad health during childhood and 1.21 significantly higher when one grew up in a household with poor wealth. The importance of religion in the household as a child is significantly correlated with feelings of loneliness in later life.
Personality traits.
The correlation between personality and loneliness at age 50+ is analyzed by adding the scores of the Big Five personality traits in the model from the second specification. Fig 2 reports the odds ratios of each of the Big Five personality traits for specifications 2, 3, 4, and 5. The size of the association between personality and loneliness at age 50+ remains stable, whatever the specification. An increase of 0.5 in the score for extraversion decreases the odds of feeling lonely by 0.25. The score for conscientiousness is also negatively associated with loneliness, but not significantly in the full specification. Agreeableness and openness seem to be protective personality traits as well, but their effects are not as strong as those of extraversion and conscientiousness. Individuals with a neurotic personality have a higher likelihood of feeling lonely (odds ratio = 1.20).
Note: The figure displays odds ratios of the five scores of personality traits for four different specifications. Specification (2) includes P, a vector of the five scores of personality traits (extraversion, agreeableness, conscientiousness, neuroticism, openness); C, a vector of childhood circumstances (wealth; health; quality of the relationship with mother, father, and friends; importance of religion); and I, country-specific fixed effects. Specification (3) includes (2) and D, a set of individual socioeconomic and demographic variables at age 50+. Specification (4) includes (3) and S, social support, activities, and network variables at age 50+. Specification (5) includes (4) and H, a vector of individual health information at age 50+. Reading example: In the full specification, an increase of 0.5 in the score of neuroticism increases the odds of loneliness by 1.20. The pseudo R2 and the proportion of correctly specified outcomes are respectively for each specification: (1): 0.0364; 82.89%, (2): 0.0600; 82.99%, (3): 0.0864; 82.96%; (4): 0.1181; 83.39%, (5): 0.1699; 84.09%. * p < 0.05, ** p < 0.01, *** p < 0.001.
Demographic and socioeconomic factors.
Fig 3 displays the odds ratios of demographic and socioeconomic variables added in specifications 3, 4, and 5. As expected, the odds of loneliness at age 50+ decreases with income and education. In the full specification, only individuals in the fifth quintiles of income are more likely to feel lonely, and individuals with low education are 1.10 more likely to feel lonely. Being employed is a significant protective factor against feeling lonely in older age, compared to being retired or being a homemaker. When controlling for variable categories C, P, S, socioeconomic factors, and gender, loneliness at age 50+ decreases with age (for individuals aged 50+). Once health factors are included in the model, the significant gender difference in the likelihood of loneliness disappears.
Note: The figure displays odds ratios of individual socioeconomic and demographic variables at age 50+ for three different specifications. Specification (3) includes D, a set of individual socioeconomic and demographic variables at age 50+; P, a vector of the five scores of personality traits (extraversion, agreeableness, conscientiousness, neuroticism, openness); C, a vector of childhood circumstances (wealth; health; quality of the relationship with mother, father, and friends; importance of religion), and I, country-specific fixed effects. Specification (4) includes (3) and S, social support, activities, and network variables at age 50+. Specification (5) includes (4) and H, a vector of individual health information at age 50+. Reading example: in the full specification, being employed decreases the odds of loneliness by 0.23. The pseudo R2 and the proportion of correctly specified outcomes are respectively for each specification: (1): 0.0364; 82.89%, (2): 0.0600; 82.99%, (3): 0.0864; 82.96%; (4): 0.1181; 83.39%, (5): 0.1699; 84.09%. * p < 0.05, ** p < 0.01, *** p < 0.001.
Social support.
Fig 4 reports the odds ratios of the variables related to the social environment of individuals at age 50+, variables added in specifications 4 and 5 of the model. Having company at home seems to be a protective factor against loneliness. Being widowed, divorced, or never married increases the risk of loneliness compared to being married. Similarly, the risk of loneliness at age 50+ decreases with the size of the household.
Note: The figure displays odds ratios of social support, activities, and network variables at age 50+ for two different specifications. Specification (4) includes S, social support, activities, and network variables at age 50+; D, a set of individual socioeconomic and demographic variables at age 50+; P, a vector of the five scores of personality traits (extraversion, agreeableness, conscientiousness, neuroticism, openness); C, a vector of childhood circumstances (wealth; health; quality of the relationship with mother, father, and friends; importance of religion); and I, country-specific fixed effects. Specification (5) includes (4) and H, a vector of individual health information at age 50+. Reading example: in the full specification, being divorced increases the risk of loneliness by 1.24. The pseudo R2 and the proportion of correctly specified outcomes are respectively for each specification: (1): 0.0364; 82.89%, (2): 0.0600; 82.99%, (3): 0.0864; 82.96%; (4): 0.1181; 83.39%, (5): 0.1699; 84.09%. * p < 0.05, ** p < 0.01, *** p < 0.001.
Having a network of persons with whom to talk about important things is also an important impeding factor. The risk of loneliness decreases significantly with the size of the network. The reduction in the risk of loneliness also depends on the frequency of contact with the network and the closeness of the network. For instance, the odds of feeling lonely increases by 1.59 when the frequency of contact is less than once a week to never, compared to those with daily contacts. The area of living does not seem to have a significant impact on loneliness at age 50+; however, the geographical proximity of the network is key to reducing the risk of feeling lonely.
Taking part in activities is protective against loneliness for those wishing to participate. Individuals who declared participating in any activity and being unsatisfied with it have an increased odds of loneliness of 1.76. In contrast, taking part in four or more activities significantly reduces the risk of loneliness. Lastly, having sufficient computer skills impedes the odds of loneliness at age 50+. Indeed, individuals who reported having poor computer skills or having never used a computer have increased odds of loneliness of 1.3.
Health at age 50+.
Fig 5 reports the odds ratio of the health variables included in specification 5 of Eq 1. Loneliness at age 50+ significantly decreases with health. The odds are 1.14 higher with more than one chronic disease, 3.41 higher with symptoms of depression, 1.40 higher with limitations with activities.
Note: The figure displays odds ratios of individual health information at age 50+ for the full specification. Specification (5) includes H, a vector of individual health information at age 50+; S, social support, activities, and network variables at age 50+; D, a set of individual socioeconomic and demographic variables at age 50+; P, a vector of the five scores of personality traits (extraversion, agreeableness, conscientiousness, neuroticism, openness); C, a vector of childhood circumstances (wealth; health; quality of the relationship with mother, father, and friends; importance of religion); and I, country-specific fixed effects. Reading example: in the full specification, having limitation with daily activities increases the odds of loneliness by 1.40. The pseudo R2 and the proportion of correctly specified outcomes are respectively for each specification: (1): 0.0364; 82.89%, (2): 0.0600; 82.99%, (3): 0.0864; 82.96%; (4): 0.1181; 83.39%, (5): 0.1699; 84.09%. * p < 0.05, ** p < 0.01, *** p < 0.001.
The relative importance of the determinants of loneliness at age 50+
To address the final research question of this study, the relative contribution of each category of factors is computed in the variance explained by the model. Table 5 reports the proportions of the explained variance by category of factors with bootstrapped confidence intervals. As expected, in the full specification, ill health status at age 50+ is the main risk factor for loneliness. It contributes to 43.32% of the explained variance, with mental health being the major contributor among the health measures.
Social support in later life is the second highest category of factors and accounts for 27.05% of the explained variance. Social connection at home explains 10.26%. The social network overall explains 9.87% of the model, while engaging in activities and having computer skills in later life account for 6.92% of the explained variance. Life conditions and circumstances during childhood contribute 7.50% and personality traits contribute 10.42%. Demographic and socioeconomic factors account for 6.50% of the explained variance. Country-level characteristics contribute 5.20%.
Discussion
This paper aimed to study the determinants of loneliness in older adults in Europe; in particular, it investigated whether childhood circumstances are significantly associated with loneliness at older ages. Using data from the SHARE, examining individuals who replied to the SHARELIFE questionnaire in wave 7, an analysis was conducted on loneliness in 17 countries in Europe. The analysis showed significant correlations between life circumstances in childhood and feeling lonely in older age. These correlations remain significant when controlling for personality traits, demographic and socioeconomic characteristics, social support and ill health in later life, and country-specific characteristics. While ill health is the main factor correlated with loneliness at 50+, as expected, the analysis of the relative importance of the determinants reveals that personality traits account for 10.42% of the explained variance and that life circumstances during childhood account for 7.50%. Social support at older ages is the second highest category of factors, accounting for 27.05%—with, interestingly, support at home contributing 10.26%, social network characteristics contributing 9.87% and engaging in activities and computer skills accounting for 6.92% of the explained variance. Demographic and socioeconomic factors account for 6.50% and country-level characteristics contribute 5.20%.
The following limitations must be discussed. One might question how well survey participants remembered their childhood. The SHARELIFE questionnaire uses different techniques to facilitate the recall mechanisms—in particular, an event history calendar with multiple dimensions and visual sequential recollection [88, 97]. These tools have been proven to limit errors in reproducing past life events [98, 99]. For individuals with very high levels of loneliness at 50+, there is a risk that part of the significant association between adverse events in childhood and loneliness in later life might be due to emotional bias. For instance, individuals suffering from depression have been found to recall more accurately and value more intensely adverse past events than healthy individuals [28, 100, 101]. However, given the size of the effect, the association found in this study cannot be explained by this recall bias alone [102]. This study reveals specific types of loneliness, such as situational loneliness, which is due to a specific crisis or loss or to chronic loneliness. For instance, it would be interesting to measure the specific effects of life events at older ages, such as the recent death of a spouse or a divorce, but the number of individuals experiencing these types of events was too low to be introduced in the model. The effects of these events are measured by the marital status factor in the model.
Based on the results and the limitations of this study, the following policy implications can be drawn. First, the findings of this paper confirm the importance of social networks and support in older age. Public interventions aiming at increasing activities and contact among older adults should take into account the different personality traits in adapting the types of activities and interaction possibilities—in particular, for those more prone to suffering from loneliness. This paper also points out the relevance of early life interventions to tackle loneliness in older age. This reaffirms findings from other studies that the quality of relationships at school and at home, as well as adverse life events during childhood, are significant predictors of education, employment, and health in later life [24, 25, 37, 103, 104], though these previous studies did not take into account as many confounding factors in the analysis of the correlation between loneliness and personality as this study did. The importance of personality traits is also confirmed here. The association between personality and loneliness can be explained by a fixed or constant component of personality that remains or changes only marginally over the course of the lifetime, and partly by smaller effects of periods of intense loneliness on personality development [62, 105]. In this view, early interventions and interventions aimed at increasing social support in later life need to be adapted to all personality types [106].
The role of childhood circumstances and the mechanisms explaining the association between loneliness in childhood and loneliness in later life deserve more attention in future research. In light of the trend of increasing childhood loneliness in the past decade [107], as well as the impact of the COVID-19 pandemic on children’s level of loneliness and their expected well-being later in life, this research is now more important than ever.
Acknowledgments
The author would like to thank the editor and the two anonymous reviewers for the feedback and the numerous useful comments received.
The statistical analysis was carried out with wu.cloud a cloud computing facility of the Institute for Statistics and Mathematics, Vienna University of Economics and Business, Austria.
References
- 1. Heinrich LM, Gullone E. The clinical significance of loneliness: A literature review. Clinical Psychology Review. 2006;26(6):695–718. pmid:16952717
- 2. Hawkley LC, Cacioppo JT. Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms. Annals of Behavioral Medicine. 2010;40(2):218–27. pmid:20652462
- 3. Valtorta N, Hanratty B. Loneliness, isolation and the health of older adults: do we need a new research agenda? Journal of the Royal Society of Medicine. 2013;105(12):518–22.
- 4.
Yang K. Loneliness. A Social Problem: Routledge Advances in Sociology; 2019. 240 Pages p.
- 5. Fees BS, Martin P, Poon LW. A Model of Loneliness in Older Adults. The Journals of Gerontology: Series B. 1999;54B(4):P231–P9.
- 6. Hawkley LC, Burleson MH, Berntson GG, Cacioppo JT. Loneliness in everyday life: Cardiovascular activity, psychosocial context, and health behaviors. J Pers Soc Psychol. 2003;85(1):105–20. pmid:12872887
- 7. Cantarero-Prieto D, Pascual-Sáez M, Blázquez-Fernández C. Social isolation and multiple chronic diseases after age 50: A European macro-regional analysis. PLOS ONE. 2018;13(10):e0205062. pmid:30356322
- 8. Luanaigh CÓ, Lawlor BA. Loneliness and the health of older people. International Journal of Geriatric Psychiatry. 2008;23(12):1213–21. pmid:18537197
- 9. Christiansen J, Larsen FB, Lasgaard M. Do stress, health behavior, and sleep mediate the association between loneliness and adverse health conditions among older people? Social Science & Medicine. 2016;152:80–6.
- 10. Ge L, Yap CW, Ong R, Heng BH. Social isolation, loneliness and their relationships with depressive symptoms: A population-based study. PLOS ONE. 2017;12(8):e0182145. pmid:28832594
- 11. Smith RW, Barnes I, Green J, Reeves GK, Beral V, Floud S. Social isolation and risk of heart disease and stroke: analysis of two large UK prospective studies. The Lancet Public Health. 2021;6(4):e232–e9. pmid:33662329
- 12. Luo Y, Hawkley LC, Waite LJ, Cacioppo JT. Loneliness, health, and mortality in old age: A national longitudinal study. Social Science & Medicine. 2012;74(6):907–14.
- 13. Steptoe A, Shankar A, Demakakos P, Wardle J. Social isolation, loneliness, and all-cause mortality in older men and women. Proceedings of the National Academy of Sciences. 2013;110(15):5797.
- 14. Holt-Lunstad J, Smith TB, Baker M, Harris T, Stephenson D. Loneliness and Social Isolation as Risk Factors for Mortality. Perspectives on Psychological Science. 2015;10(2):227–37. pmid:25910392
- 15. Rico-Uribe LA, Caballero FF, Martín-María N, Cabello M, Ayuso-Mateos JL, Miret M. Association of loneliness with all-cause mortality: A meta-analysis. PLOS ONE. 2018;13(1):e0190033. pmid:29300743
- 16. Gerst-Emerson K, Jayawardhana J. Loneliness as a Public Health Issue: The Impact of Loneliness on Health Care Utilization Among Older Adults. American Journal of Public Health. 2015;105(5):1013–9. pmid:25790413
- 17.
European Quality of Life Survey. In: Eurofound, editor. https://www.eurofound.europa.eu/data/european-quality-of-life-survey2016.
- 18. Gierveld JdJ. A review of loneliness: concept and definitions, determinants and consequences. Reviews in Clinical Gerontology. 1998;8(1):73–80.
- 19. Abdellaoui A, Chen HY, Willemsen G, Ehli EA, Davies GE, Verweij KJH, et al. Associations between loneliness and personality are mostly driven by a genetic association with Neuroticism. Journal of Personality. 2018;87(2):386–97. pmid:29752830
- 20. Wang B, Dong X. The Association Between Personality and Loneliness: Findings From a Community-Dwelling Chinese Aging Population. Gerontology and Geriatric Medicine. 2018;4. pmid:30035191
- 21. Long MV, Martin P. Personality, Relationship Closeness, and Loneliness of Oldest Old Adults and Their Children. The Journals of Gerontology: Series B. 2000;55(5):P311–P9.
- 22. Ma J, Yang Y, Wan Y, Shen C, Qiu P. The influence of childhood adversities on mid to late cognitive function: From the perspective of life course. PLOS ONE. 2021;16(8):e0256297. pmid:34398901
- 23. Tubeuf S, Jusot F, Bricard D. Mediating role of education and lifestyles in the relationship between early-life conditions and health: evidence from the 1958 British cohort. Health Econ. 2012;21 Suppl 1:129–50. pmid:22556004
- 24. Healthy Currie J., Wealthy, and Wise: Socioeconomic Status, Poor Health in Childhood, and Human Capital Development. Journal of Economic Literature. 2009;47(1):87–122.
- 25. Pakpahan E, Hoffmann R, Kröger H. The long arm of childhood circumstances on health in old age: Evidence from SHARELIFE. Advances in Life Course Research. 2017;31:1–10.
- 26. Flores M, Kalwij A. The associations between early life circumstances and later life health and employment in Europe. Empirical Economics. 2014;47(4):1251–82.
- 27. van den Berg GJ, Deeg DJH, Lindeboom M, Portrait F. The Role of Early‐Life Conditions in the Cognitive Decline due to Adverse Events Later in Life. The Economic Journal. 2010;120(548):F411–F28.
- 28. Liu RT. Childhood Adversities and Depression in Adulthood: Current Findings and Future Directions. Clin Psychol (New York). 2017;24(2):140–53. pmid:28924333
- 29. van den Berg GJ, Doblhammer G, Christensen K. Exogenous determinants of early-life conditions, and mortality later in life. Social Science & Medicine. 2009;68(9):1591–8.
- 30. Ferraro KF, Schafer MH, Wilkinson LR. Childhood Disadvantage and Health Problems in Middle and Later Life. American Sociological Review. 2015;81(1):107–33. pmid:27445413
- 31. Vozikaki M, Papadaki A, Linardakis M, Philalithis A. Loneliness among older European adults: results from the survey of health, aging and retirement in Europe. Journal of Public Health. 2018;26(6):613–24.
- 32. Fokkema T, De Jong Gierveld J, Dykstra PA. Cross-National Differences in Older Adult Loneliness. The Journal of Psychology. 2012;146(1–2):201–28. pmid:22303621
- 33. Sundström G, Fransson E, Malmberg B, Davey A. Loneliness among older Europeans. European Journal of Ageing. 2009;6(4):267–75. pmid:28798610
- 34. Hawkley LC, Hughes ME, Waite LJ, Masi CM, Thisted RA, Cacioppo JT. From social structural factors to perceptions of relationship quality and loneliness: the Chicago health, aging, and social relations study. J Gerontol B Psychol Sci Soc Sci. 2008;63(6):S375–S84. pmid:19092047
- 35. Aartsen M, Jylhä M. Onset of loneliness in older adults: results of a 28 year prospective study. European Journal of Ageing. 2011;8(1):31–8. pmid:21475393
- 36. Dahlberg L, Andersson L, Lennartsson C. Long-term predictors of loneliness in old age: results of a 20-year national study. Aging & Mental Health. 2016;22(2):190–6. pmid:27802772
- 37. Kamiya Y, Doyle M, Henretta JC, Timonen V. Early-Life Circumstances and Later-Life Loneliness in Ireland. The Gerontologist. 2013;54(5):773–83. pmid:24009171
- 38. Russell D, Peplau LA, Ferguson ML. Developing a Measure of Loneliness. Journal of Personality Assessment. 1978;42(3):290–4. pmid:660402
- 39. Russell D, Peplau LA, Cutrona CE. The revised UCLA Loneliness Scale: concurrent and discriminant validity evidence. J Pers Soc Psychol. 1980;39(3):472–80. pmid:7431205
- 40. Russell DW. UCLA Loneliness Scale (Version 3): Reliability, Validity, and Factor Structure. Journal of Personality Assessment. 1996;66(1):20–40. pmid:8576833
- 41.
Mullins LC. Loneliness. In: Birren JE, editor. Encyclopedia of Gerontology (Second Edition). New York: Elsevier; 2007. p. 93–8.
- 42. Marangoni C, Ickes W. Loneliness: A Theoretical Review with Implications for Measurement. Journal of Social and Personal Relationships. 1989;6(1):93–128.
- 43.
Bowlby J. A secure base: parent-child attachment and healthy human development. New York: Basic Books; 1988.
- 44.
Ainsworth MDS. Infancy in Uganda: infant care and the growth of love. Baltimore: Johns Hopkins Press; 1967.
- 45. Chris Fraley R. Attachment Stability From Infancy to Adulthood: Meta-Analysis and Dynamic Modeling of Developmental Mechanisms. Personality and Social Psychology Review. 2002;6(2):123–51.
- 46. Posada G, Trumbell JM. Attachment in Infancy and Early Childhood. Encyclopedia of Infant and Early Childhood Development2020. p. 82–91.
- 47. Chen E, Brody GH, Miller GE. Childhood close family relationships and health. Am Psychol. 2017;72(6):555–66. pmid:28880102
- 48. Boivin M, Bégin G. Peer status and self-perception among early elementary school children: The case of the rejected children. Child Development. 1989;60(3):591–6. pmid:2737009
- 49. Bishop JA, Inderbitzen HM. Peer Acceptance and Friendship: An Investigation of their Relation to Self-Esteem. The Journal of Early Adolescence. 1995;15(4):476–89.
- 50.
Sakyi KS, Surkan Pj Fau—Fombonne E, Fombonne E Fau—Chollet A, Chollet A Fau—Melchior M, Melchior M. Childhood friendships and psychological difficulties in young adulthood: an 18-year follow-up study. (1435-165X (Electronic)).
- 51. van Harmelen AL, Kievit RA, Ioannidis K, Neufeld S, Jones PB, Bullmore E, et al. Adolescent friendships predict later resilient functioning across psychosocial domains in a healthy community cohort. Psychological Medicine. 2017;47(13):2312–22. pmid:28397612
- 52. Hartup WW, Stevens N. Friendships and adaptation in the life course. Psychological Bulletin. 1997;121(3):355–70.
- 53. Burr JA, Han SH, Peng C. Childhood Friendship Experiences and Cognitive Functioning in Later Life: The Mediating Roles of Adult Social Disconnectedness and Adult Loneliness. The Gerontologist. 2020;60(8):1456–65. pmid:32573696
- 54. Ejlskov L, Bøggild H, Kuh D, Stafford M. Social relationship adversities throughout the lifecourse and risk of loneliness in later life. Ageing and Society. 2020;40(8):1718–34.
- 55. Lacey RE, Kumari M, Bartley M. Social isolation in childhood and adult inflammation: Evidence from the National Child Development Study. Psychoneuroendocrinology. 2014;50:85–94. pmid:25197797
- 56. Fry PS, Debats DL. Self-efficacy beliefs as predictors of loneliness and psychological distress in older adults. Int J Aging Hum Dev. 2002;55(3):233–69. pmid:12693547
- 57. Thoresen S, Aakvaag HF, Strøm IF, Wentzel-Larsen T, Birkeland MS. Loneliness as a mediator of the relationship between shame and health problems in young people exposed to childhood violence. Social Science & Medicine. 2018;211:183–9. pmid:29958130
- 58. Hyland P, Shevlin M, Cloitre M, Karatzias T, Vallières F, McGinty G, et al. Quality not quantity: loneliness subtypes, psychological trauma, and mental health in the US adult population. Social Psychiatry and Psychiatric Epidemiology. 2019;54(9):1089–99. pmid:30293176
- 59. Rokach A. The Effect of Gender and Culture on Loneliness: A Mini Review. Emerging Science Journal. 2018;2(2).
- 60. Heu LC, van Zomeren M, Hansen N. Lonely Alone or Lonely Together? A Cultural-Psychological Examination of Individualism–Collectivism and Loneliness in Five European Countries. Personality and Social Psychology Bulletin. 2018;45(5):780–93. pmid:30264659
- 61.
Rokach A. 4—Loneliness, gender, and culture. In: Rokach A, editor. The Psychological Journey To and From Loneliness: Academic Press; 2019. p. 85–96.
- 62. Mund M, Freuding MM, Möbius K, Horn N, Neyer FJ. The Stability and Change of Loneliness Across the Life Span: A Meta-Analysis of Longitudinal Studies. Personality and Social Psychology Review. 2019;24(1):24–52. pmid:31179872
- 63. Mund M, Neyer FJ. The Winding Paths of the Lonesome Cowboy: Evidence for Mutual Influences Between Personality, Subjective Health, and Loneliness. Journal of Personality. 2016;84(5):646–57. pmid:26112403
- 64. Kamath M, Kanekar S. Loneliness, Shyness, Self-Esteem, and Extraversion. The Journal of Social Psychology. 1993;133(6):855–7.
- 65. Buecker S, Maes M, Denissen JJA, Luhmann M. Loneliness and the Big Five Personality Traits: A Meta–Analysis. European Journal of Personality. 2020;34(1):8–28.
- 66. Yang K, Victor C. Age and loneliness in 25 European nations. Ageing & Society. 2011;31(8):1368–88.
- 67. Dykstra PA. Older adult loneliness: myths and realities. European Journal of Ageing. 2009;6(2):91. pmid:19517025
- 68.
Rokach A. 7—Loneliness and life’s ages and stages. In: Rokach A, editor. The Psychological Journey To and From Loneliness: Academic Press; 2019. p. 143–71.
- 69. Madsen KR, Holstein BE, Damsgaard MT, Rayce SB, Jespersen LN, Due P. Trends in social inequality in loneliness among adolescents 1991–2014. Journal of Public Health. 2019;41(2):e133–e40. pmid:30053062
- 70. Niedzwiedz CL, Richardson EA, Tunstall H, Shortt NK, Mitchell RJ, Pearce JR. The relationship between wealth and loneliness among older people across Europe: Is social participation protective? Preventive Medicine. 2016;91:24–31. pmid:27471027
- 71. Hansen T, Slagsvold B. Late-Life Loneliness in 11 European Countries: Results from the Generations and Gender Survey. Social Indicators Research. 2016;129(1):445–64.
- 72. Litwin H, Shiovitz-Ezra S. Social network type and subjective well-being in a national sample of older Americans. The Gerontologist. 2011;51(3):379–88. pmid:21097553
- 73. Litwin H, Stoeckel KJ. Confidant Network Types and Well-Being Among Older Europeans. The Gerontologist. 2013;54(5):762–72. pmid:23749390
- 74.
Shiovitz-Ezra S. 30 Confidant networks and loneliness Active ageing and solidarity between generations in Europe. In: Börsch-Supan A, Brandt M, Litwin H, Weber G, editors.: De Gruyter; 2013. p. 349–58.
- 75. Finlay JM, Kobayashi LC. Social isolation and loneliness in later life: A parallel convergent mixed-methods case study of older adults and their residential contexts in the Minneapolis metropolitan area, USA. Social Science & Medicine. 2018;208:25–33. pmid:29758475
- 76. Beere P, Keeling S, Jamieson H. Ageing, loneliness, and the geographic distribution of New Zealand’s interRAI-HC cohort. Social Science & Medicine. 2019;227:84–92.
- 77. Victor CR, Pikhartova J. Lonely places or lonely people? Investigating the relationship between loneliness and place of residence. BMC Public Health. 2020;20(1):778. pmid:32456626
- 78. Hunt MG, Marx R, Lipson C, Young J. No More FOMO: Limiting Social Media Decreases Loneliness and Depression. Journal of Social and Clinical Psychology. 2018;37(10):751–68.
- 79. Zhang K, Kim K, Silverstein NM, Song Q, Burr JA. Social Media Communication and Loneliness Among Older Adults: The Mediating Roles of Social Support and Social Contact. The Gerontologist. 2020.
- 80. Barlow MA, Liu SY, Wrosch C. Chronic illness and loneliness in older adulthood: The role of self-protective control strategies. Health Psychology. 2015;34(8):870–9. pmid:25528177
- 81. Meltzer H, Bebbington P, Dennis MS, Jenkins R, McManus S, Brugha TS. Feelings of loneliness among adults with mental disorder. Social Psychiatry and Psychiatric Epidemiology. 2013;48(1):5–13. pmid:22570258
- 82. Tsur N, Stein JY, Levin Y, Siegel A, Solomon Z. Loneliness and subjective physical health among war veterans: Long term reciprocal effects. Social Science & Medicine. 2019;234. pmid:31254967
- 83. Barreto M, Victor C, Hammond C, Eccles A, Richins MT, Qualter P. Loneliness around the world: Age, gender, and cultural differences in loneliness. Personality and Individual Differences. 2021;169.
- 84. Aartsen M, Morgan D, Dahlberg L, Waldegrave C, Mikulionienė S, Rapolienė G, et al. Exclusion From Social Relations and Loneliness: Individual and Country-Level Changes. Innov Aging. 2020;4(Suppl 1):712–3.
- 85. Hansen T, Kafková MP, Katz R, Lowenstein A, Naim S, Pavlidis G, et al. Exclusion from Social Relations in Later Life: Micro- and Macro-Level Patterns and Correlations in a European Perspective. International Journal of Environmental Research and Public Health. 2021;18(23).
- 86.
Börsch-Supan A. Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 6. In: Release version: 8.0.0. SHARE-ERIC. Data set, editor. 10.6103/SHARE.w6.8002022.
- 87.
Börsch-Supan A. Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 7. In: Release version 8.0.0. SHARE-ERIC. Data set, editor. 10.6103/SHARE.w7.8002022.
- 88.
Bergmann M, Scherpenzeel A, Börsch-Supan A. SHARE Wave 7 Methodology: Panel Innovations and Life Histories. Munich; MEA Max Planck Institute for Social Law and Social Policy. 2019.
- 89. Hughes ME, Waite LJ, Hawkley LC, Cacioppo JT. A Short Scale for Measuring Loneliness in Large Surveys: Results From Two Population-Based Studies. Research on Aging. 2004;26(6):655–72. pmid:18504506
- 90. Pikhartova J, Bowling A, Victor C. Does owning a pet protect older people against loneliness? BMC Geriatrics. 2014;14(1):106. pmid:25240250
- 91. The Rammstedt B. 10-Item Big Five Inventory. European Journal of Psychological Assessment. 2007;23(3):193–201.
- 92. Rammstedt B, John OP. Measuring personality in one minute or less: A 10-item short version of the Big Five Inventory in English and German. Journal of Research in Personality. 2007;41(1):203–12.
- 93. van Oyen H, Van der Heyden J, Perenboom R, Jagger C. Monitoring population disability: evaluation of a new Global Activity Limitation Indicator (GALI). Soz Praventivmed. 2006;51(3):153–61. pmid:17191540
- 94.
De Luca G, Celidoni M, Trevisan E. Item nonresponse and imputation strategies in SHARE Wave 5. Munich:: Munich Center for the Economics of Aging (MEA) at the Max Planck Institute for Social Law and Social Policy (MPISOC). 2015.
- 95. Shorrocks AF. Inequality Decomposition by Factor Components. Econometrica. 1982;50(1):193–211.
- 96. McKelvey RD, Zavoina W. A statistical model for the analysis of ordinal level dependent variables. The Journal of Mathematical Sociology. 1975;4(1):103–20.
- 97. Börsch-Supan A, Brandt M, Schröder M. SHARELIFE—One century of life histories in Europe. Advances in Life Course Research. 2013;18(1):1–4. pmid:24797462
- 98. Belli RF. The structure of autobiographical memory and the event history calendar: potential improvements in the quality of retrospective reports in surveys. Memory. 1998;6(4):383–406. pmid:9829098
- 99. Havari E, Mazzonna F. Can We Trust Older People’s Statements on Their Childhood Circumstances? Evidence from SHARELIFE. European Journal of Population / Revue Européenne de Démographie. 2015;31(3):233–57.
- 100. Hardt J, Rutter M. Validity of Adult Retrospective Reports of Adverse Childhood Experiences: Review of the Evidence. Journal of child psychology and psychiatry, and allied disciplines. 2004;45:260–73. pmid:14982240
- 101. Urban EJ, Charles ST, Levine LJ, Almeida DM. Depression history and memory bias for specific daily emotions. PloS one. 2018;13(9):e0203574–e. pmid:30192853
- 102. Patten SB, Wilkes TCR, Williams JVA, Lavorato DH, el-Guebaly N, Schopflocher D, et al. Retrospective and prospectively assessed childhood adversity in association with major depression, alcohol consumption and painful conditions. Epidemiology and Psychiatric Sciences. 2015;24(2):158–65. pmid:24480045
- 103. Arpino B, Gumà J, Julià A. Early-life conditions and health at older ages: The mediating role of educational attainment, family and employment trajectories. PLOS ONE. 2018;13(4):e0195320. pmid:29621290
- 104. Almond D, Currie J, Duque V. Childhood Circumstances and Adult Outcomes: Act II. Journal of Economic Literature. 2018;56(4):1360–446.
- 105. Mund M, Neyer FJ. Loneliness effects on personality. International Journal of Behavioral Development. 2018;43(2):136–46.
- 106. Layden EA, Cacioppo JT, Cacioppo S. Loneliness predicts a preference for larger interpersonal distance within intimate space. PLOS ONE. 2018;13(9):e0203491. pmid:30188950
- 107. Lempinen L, Junttila N, Sourander A. Loneliness and friendships among eight-year-old children: time-trends over a 24-year period. Journal of Child Psychology and Psychiatry. 2018;59(2):171–9. pmid:28892142