The relationship between loneliness and healthy ageing indicators in Brazil (ELSI-Brazil) and England (ELSA): sex differences

Abstract

Objectives:

To estimate five harmonized healthy ageing indicators covering functional ability and intrinsic capacity among older women and men from Brazil and England and evaluate their association with loneliness.

Study Design:

Cross-sectional study.

Methods:

We used two nationally representative samples of men and women aged 60 years and over from the Brazilian Longitudinal Study of Aging (ELSI-Brazil) wave 2 (2019–2021) (n=6,929) and the English Longitudinal Study of Ageing (ELSA) wave 9 (2018–2019) (n=5,902). Healthy ageing included five separate indicators (getting dressed, taking medication, managing money, cognitive function, and handgrip strength). Loneliness was measured by the 3-item UCLA Loneliness Scale. Logistic regression models stratified by sex and country were performed.

Results:

Overall, age-adjusted healthy ageing indicators were worse in Brazil compared to England for both men and women. Considering functional ability, loneliness was negatively associated with all indicators (ranging from OR=0.26 (95%CI 0.13–0.52) in English men regarding the ability to take medication to OR=0.49 (95%CI 0.27–0.89) in Brazilian women regarding the ability to manage money). Considering intrinsic capacity, loneliness was negatively associated with a higher cognitive function (OR=0.72; 95%CI 0.55–0.95 in English women) and a higher handgrip strength (OR=0.61; 95%CI 0.45–0.83 in Brazilian women). Lonely women demonstrated lower odds of a higher number of healthy ageing indicators than men in both countries.

Conclusions:

Country-specific social environments should be targeted by public policies to decrease loneliness and promote healthy ageing later in life.

Introduction

In 2015, the World Health Organization defined healthy ageing as developing and maintaining the functional ability that enables well-being later in life, comprising functional ability, intrinsic capacity, and environment components ¹. Since then, few proposed healthy ageing scores emerged, using different indicators of functional ability and intrinsic capacity ^2,3, but not including data from Brazil. More recently, the World Health Organization proposed five harmonized indicators based on nationally representative data from 42 countries worldwide, including Brazil, and covering the same components ⁴.

Common approaches to measuring healthy ageing are essential in cross-country comparisons aiming at establishing the distinct impact of a particular environment on ageing, especially between upper-middle- and high-income countries. Due to a faster demographic change, upper-middle-income countries had to adapt more quickly to population ageing than high-income countries despite having poorer health and social welfare infrastructures. Brazil and England, respectively, are good examples of those countries. They show a distinct ageing process but with similar universal public primary-care-oriented health systems with public participation, funded through general taxation and controlled by the Secretary of State for Health with decentralization at the local level ⁵. In 2020, the Brazilian population aged 65 years and over was 10% of a 211,8-million population ⁶. In England, this age group represented 18.5% of a 56,5-million population ⁷.

The environment also encompasses the support and relationships, related to the quality of support provided by people or animals at home, work, and in other aspects of daily activities ⁸. When there is a lack of support and relationships, loneliness, the subjective feeling of social isolation and not belonging ⁹ can emerge.

Loneliness has been considered a global epidemic among the world’s older adult population ¹⁰ and is associated with various adverse health outcomes, such as physical and mental health problems ¹¹ and mortality ¹². About 11–18% of depressive symptoms in England can be attributable to loneliness ¹³. The mechanisms underlying poor health outcomes rely on a diminished capacity of self-regulation related to lifestyle and sleep quality ¹¹, neuroendocrine stress ¹⁴, inflammatory responses, and immune deregulation ¹¹. However, few studies evaluated the impact of loneliness on positive outcomes, such as healthy ageing. One study in Amsterdam did not find any association between loneliness and longevity (≥90 years) among men and women ¹⁵, although longevity was not necessarily healthy. Other studies found that loneliness decreases the odds of “ageing well” in older German individuals (≥40 years) ¹⁶ and in adults (≥18 years) from Finland, Poland, and Spain ¹⁷. However, studies comparing high-income with upper-middle-income countries are lacking. Therefore, this study aimed at estimating and comparing healthy ageing indicators among older women and men from Brazil and England and evaluating the burden of loneliness on healthy ageing indicators.

Methods

Data source

This is a cross-sectional analysis using data from wave 2 (2019–2021) of the Brazilian Longitudinal Study of Ageing (ELSI-Brazil) and wave 9 (2018–2019) of the English Longitudinal Study of Ageing (ELSA). Both studies are nationally representative of the community-dwelling population aged 50 and over ^18,19. ELSI-Brazil sampling procedures included a probabilistic sample design, combining geographical stratification in clustering stages, covering 70 municipalities from the five great Brazilian geographic regions ¹⁸. The final wave 2 sample comprised 9,949 previous and refreshed participants. ²⁰. Of those, 6,929 had 60 years and over and were eligible for the current analysis. ELSA initial sample was drawn from the Health Survey for England in 1998/1999/2001, an annual cross-sectional survey ¹⁹ using a complex design with clustering effects. ELSA sample has periodically been refreshed to maintain the representation of younger older adults. The final wave 9 sample comprised 7,289 core participants eligible for the nurse visit (i.e., an additional collection of biological samples and handgrip strength measures). Of those, 5,902 were 60 years and over and were included in this analysis.

ELSI-Brazil has been approved by the Research Ethics Committee of the Oswaldo Cruz Foundation, Minas Gerais (protocol 34649814.3.0000.5091). The ELSA has been approved by National Research Ethics Service (London Multicentre Research Ethics Committee - MREC/01/2/91).

Healthy ageing indicators

We included the five individual healthy ageing indicators proposed by the World Health Organization:

• Functional ability: it included three basic daily activities. Participants were asked about having difficulties in (1) getting dressed, (2) taking medication, and (3) managing money. Participants were considered “able of” when they did not report difficulties in any of the activities.

• Intrinsic capacity: it included cognitive and vitality sub-domains. (4) Cognitive function was evaluated by the delayed word list learning test, where the interviewer read ten words, and participants were asked to repeat as many words as they could after five minutes. Cut-offs considering the lowest 20^th percentile were created for each country, stratified by sex, 5-year age group up to 85 years, and years of schooling (lowest or highest), to classify the absence of healthy ageing. The lowest schooling level included incomplete formal education in each country (incomplete first-level “≤7y” in Brazil and 0-level or equivalent “≤11y” in England). The vitality sub-domain was evaluated by (5) handgrip strength. It was objectively assessed through the best of three attempts in the dominant upper limb by a handgrip dynamometer. Cut-offs considering the lowest 20^th percentile were created for each country, stratified by sex, and 5-year age group, to classify the absence of healthy ageing. All values used as cut-offs can be found in Supplemental Table 1. In England, handgrip strength was assessed during the nurse visit. The waves 8 (2016–2017) and 9 (2018–2019) nurse samples were designed to be analyzed as a whole i.e. 2,837 from wave 8 and 2,061 from wave 9.

Loneliness

Loneliness was measured by the 3-item University of California (UCLA) Loneliness Scale ⁹. A valid scale in each country, containing three questions about how often the participant feels lack of companionship, left out, and isolated from others. Each question had three possible answers (i.e., hardly ever or never, some of the time, and often), generating a score ranging from 3 to 9. According to the final score, the participants were classified into “no loneliness” (scores of 3– 5) and “loneliness” (scores of 6–9)²¹.

Covariates

Covariates included sociodemographic characteristics [age groups (60–69, 70–79, and ≥ 80), wealth (quintiles)], living alone (yes or no), having a partner (yes or no), social network contact (infrequent or frequent), multimorbidity (yes or no), and depressive symptoms (yes or no). Wealth included the total non-pension household wealth and was measured differently in each country, considering its cultural peculiarities. In Brazil, wealth included any home or other property (less mortgage) and vehicle assets at the current value. In England, it included financial wealth (savings and investments), any home and other property (less mortgage), any business assets, and physical wealth such as artwork and jewels owned by the household minus any debt, comprising 22 components. A detailed description of the wealth variable can be found at http://elsi.cpqrr.fiocruz.br/en/guidelines-to-use/ for Brazil and http://bit.ly/1yrRgHd and http://bit.ly/1awp6iZ for England. In both countries, the components were either observed or imputed. Social network contact included the frequency of face-to-face meetings with children, family, or friends who do not live with the participant. Each category (i.e., children, family, and friends) used a 6-point Likert scale ranging from less than once a year/never to three or more times a week. The final score combined the categories ranging from 3 to 18, which higher scores indicating greater face-toface contact frequency ¹³. Infrequent social network contact included scores lower than 7, which means answering “less than once a year/never” or “once or twice a year” in at least two categories. Multimorbidity included a self-reported history of at least two medical diagnoses of cardiovascular disease (hypertension, stroke, heart attack, angina or heart failure), high cholesterol, neurologic disease (Parkinson’s or Alzheimer’s disease), chronic obstructive pulmonary disease, diabetes, arthritis, asthma or cancer. Depressive symptoms were measured by the validated 8-item Center for Epidemiologic Studies – Depression (CES-D) Scale ^22,23, using the cut-off point of four or more as depressive symptoms ²⁴. The scale contains eight questions about depressive symptoms experienced during the week before interview.

Statistical analysis

We estimated and plotted in charts the prevalence of each functional ability indicator (i.e., ability to get dressed, ability to take medicine, and ability to manage money) and median of each intrinsic capacity indicator (i.e., cognitive function and handgrip strength) by country and sex at age groups. Covariates and loneliness age-adjusted prevalence by country and sex, allowing comparisons between Brazil and England. To examine the association between loneliness and each healthy ageing indicator, we adjusted separate logistic regression models considering each healthy ageing indicator as the dependent variable, including loneliness as the independent variable, and adjusting for other covariates. Considering the complex data design of ELSI-Brazil and ELSA, we incorporated the study design (i.e., cluster and stratification for ELSI-Brazil and cluster for ELSA) and weights using the Survey package (v4.0). All analyses were done separately by sex in the softwares R, RStudio, and Stata 17.0.

Results

In both countries, the included participants aged 60 and over (6,929 from ELSI-Brazil and 5,902 from ELSA) had different missing data on each healthy ageing indicator and, therefore, each model had different samples. Participants were younger in Brazil: 55% of the Brazilian participants were 60–69 years, while only 45% were the same age in England, justifying the age standardization approach used to compare prevalence. With regards to sex, in both countries, most participants were women (54.4% in Brazil and 53.9% in England).

Figure 1 shows the prevalence of functional ability indicators [(A) ability to get dressed, (B) take medication, and (C) manage money] according to age and country. Of the above-mentioned indicators, healthy ageing was similar among women and men within countries. Overall, when comparing countries we observed that the prevalence of functional ability indicators was higher in England than in Brazil, mainly at 70–79 and ≥80 years, except for the prevalence of the ability to get dressed, which was higher in Brazilian women (98.1% vs. 88.6% among Brazilian and English aged 60–69 years) and men (98% vs. 89.5% among Brazilian and English aged 60–69 years).

Figure 1 –

Prevalence of functional ability indicators [(A) ability to get dressed, (B) take medication, and (C) manage money] according to age and country – ELSI-Brazil (2019–21) and ELSA (2018–19).

Figure 2 shows the median of intrinsic capacity indicators [(A) cognitive function and (B) handgrip strength] according to age and country. Median cognitive function and handgrip strength were higher among English than Brazilians for both men and women in all age groups, demonstrating higher intrinsic capacity indicators in England.

Figure 2 –

Median of intrinsic capacity indicators [(A) cognitive function and (B) handgrip strength] according to age and country – ELSI-Brazil (2019–21) and ELSA (2018–19).

The age-adjusted prevalence of loneliness was similar across countries and sex: among women, the age-adjusted prevalence was 18.8% in Brazilians and 22.5% in English. Among men, it was 14.7% in Brazilians and 16.2% in English. Overall, covariates were worse among women than men in both countries. However, they generally differed by country. Age-adjusted prevalence of older women who have a partner was lower among Brazilians than English (39.7% vs. 56.3%, respectively); both women and men showed higher social network contact in Brazil than in England (94.7% vs. 81.8% among women, respectively, and 93.8% vs. 79.7% among men, respectively). The age-adjusted prevalence of men with multimorbidity was lower among Brazilians than English (34.9% vs. 43.4%, respectively).

Table 2 presents the fully adjusted models of the association between loneliness and healthy ageing indicators in older women and men from Brazil and England. Loneliness was negatively associated with a higher number of healthy ageing indicators in women than men, irrespectively of nationality. Regarding functional ability, loneliness was negatively associated with all indicators among Brazilian women (get dressed: OR:0.37; 95%CI: 0.22–0.64; take medication: OR:0.37; 95%CI: 0.22–0.63; and manage money: OR:0.49; 95%CI: 0.27–0.89), whereas in England it was associated with the ability to manage money in women (OR:0.38; 95%CI: 0.200.72) and take medication in men (OR:0.26; 95%CI: 0.13–0.52).

Table 2 -.

Fully adjusted models of the association between loneliness and healthy ageing indicators in older women and men from Brazil and England – ELSI-Brazil (2019–21) and ELSA (2018–19).

	Women				Men
	Brazil		England		Brazil		England
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Ability to get dressed
No loneliness	1.00	Ref.	1.00	Ref.	1.00	Ref.	1.00	Ref.
Loneliness	0.37	0.22–0.64	0.80	0.57–1.31	0.67	0.23–1.95	1.23	0.80–1.90
Ability to medication take
No loneliness	1.00	Ref.	1.00	Ref.	1.00	Ref.	1.00	Ref.
Loneliness	0.37	0.22–0.63	0.91	0.37–2.21	0.76	0.35–1.67	0.26	0.13–0.52
Ability to money manage
No loneliness	1.00	Ref.	1.00	Ref.	1.00	Ref.	1.00	Ref.
Loneliness	0.49	0.27–0.89	0.38	0.20–0.72	0.96	0.39–2.34	0.78	0.31–1.95
Higher function
No loneliness	1.00	Ref.	1.00	Ref.	1.00	Ref.	1.00	Ref.
Loneliness	0.86	0.63–1.15	0.73	0.55–0.95	1.06	0.71–1.57	0.77	0.54–1.10
Higher handgrip strength
No loneliness	1.00	Ref.	1.00	Ref.	1.00	Ref.	1.00	Ref.
Loneliness	0.61	0.45–0.83	1.12	0.80–1.56	0.60	0.34–1.05	1.06	0.70–1.61

OR: odds ratio; CI: confidence interval. Ref.: Reference. Models were made separately by each healthy ageing indicator (i.e., getting dressed, taking medication, managing money, cognitive function, and handgrip strength) and adjusted for age group, wealth, living alone, having a partner, social network contact, multimorbidity, and depressive symptoms.

Concerning intrinsic capacity, lonely individuals were less likely than non-lonely to have a higher cognitive function among English women (OR:0.73; 95%CI: 0.55–0.95) and a higher handgrip strength among Brazilian women (OR:0.61; 95%CI: 0.45–0.83).

Since depressive symptoms could also be a mediating path between loneliness and healthy ageing, additionally we tested the interaction effect between loneliness and depressive symptoms (Supplemental Table 2). Most of the models was not significant, demonstrating that loneliness might exert an independently effect on healthy ageing, irrespectively of depressive symptoms.

Discussion

Our findings showed disparities in healthy ageing indicators across countries, demonstrating better health among English women and men, except for the ability to get dressed, which was higher among Brazilians. The age-adjusted loneliness prevalence was similar across countries and sex. Nevertheless, consistently in both countries, loneliness is inversely associated with a higher number of healthy ageing indicators among older women than men, demonstrating that the burden of loneliness decreases health more among women.

Comparing the prevalence of each healthy ageing indicator is difficult because most of the previously published studies compared the prevalence of functional abilities according to categories of activities of daily living, such as basic or instrumental. In general, Brazil has worse functional ability based on basic activities of daily living ²⁵ and instrumental activities of daily living (including taking medication and managing money) than England ²⁶. However, in 2015, a study found comparable age- and sex-adjusted prevalence in six basic activities of daily living (including getting dressed) (82.1% in Brazil vs. 83.7% in England) ²⁷. In the present study, the higher ability to get dressed in Brazil could be attributable to warmer weather, where comfortable, soft, and fewer clothes are needed.

Although intrinsic capacity was worse in Brazil, the same pattern did not occur for all functional ability indicators, reinforcing that functional ability derives from the interaction between intrinsic capacity and environment. Therefore, healthy ageing also depends on the social, economic, political, and built environments where older adults live and should be targeted to build age-friendly communities. Studies comparing countries with different contexts are needed ²⁸, mainly because the social environment, often measured by the satisfaction of personal relationships and frequency of social network contact, seems to be more essential in achieving well-being in later life among women and men than the built environment ²⁹. Positive aspects of personal relationships have been reported to decrease the risk of poor functional ability among Brazilians ³⁰ and English ³¹.

In our study, loneliness was associated with all healthy ageing indicators, independently of depressive symptoms. As mentioned above, comparisons between studies that investigated the association between loneliness and functional ability indicators should be made with caution since these indicators have been indirectly measured in the literature, differently from this study. Longitudinal studies which included “difficulties in getting dressed” ^16,32 or “poor handgrip strength” ³³ as one item of the outcome consistently found a positive association with loneliness. These results suggest that inflammatory responses ¹¹ caused by loneliness, may generate sarcopenia, which is a major contributor to the risk of functional ability decline and physical frailty ³⁴. Furthermore, cross-sectional and longitudinal evidence indicated that loneliness was negatively associated with cognitive function ³⁵. The linked mechanisms were unclear and warranted further investigation ³⁵. Nevertheless, there is some evidence that this relationship could be associated with too much or too little activity of certain neurotransmitters and hormones ¹⁴, probably leading to brain damage.

Some implications emerge from our findings for public policies: irrespectively of the social context, promoting healthy ageing encompasses organizations that enable people’s relationships and an infrastructure that empowers social connections ³⁶. In the health sector, we encourage multi-professional teams to track loneliness with simple questions and to provide group activities to support and increase the sense of belonging. Recently in the United Kingdom, general practitioners have been offering a social prescribing that links those in need to a range of services that support social, emotional, and practical needs ³⁷. Intersectionally, age-friendly environments that enable tackling loneliness include having access to community spaces to get together and accessibility when going outside, such as street infrastructure and public transport network ³⁶. Further research can evaluate the effectiveness of these actions on healthy ageing.

Although longitudinal studies reported being a woman as a determinant of loneliness ³⁸, we did not find that women suffer more from loneliness than men. However, our results demonstrated that the burden of loneliness unequally decreases healthy ageing among older women than men in Brazil and England, corroborating previous results in England ³⁹. Distinctively from our results, longitudinal ³² association between loneliness and cognitive impairment ³⁵ or items of difficulties in getting dressed/cognitive impairment ³² was increased among older men but not women in China. Some possible explanations could be speculated. Firstly, differences might be attributable to a lack of uniformity in measuring loneliness. Secondly, different loneliness patterns occur in each country even when showing similar overall loneliness scores using the 3-item UCLA Loneliness Scale. English more often answer that they hardly ever feel a lack of companionship than Americans ⁴⁰. In the current study, older Brazilian women were more likely not to have a partner than English, which can be compensated by the higher social network contact. These are social environment variables previously described to be associated with loneliness. Thirdly, worse social determinants in women, such as living alone and not having a partner, may have a higher buffering effect on healthy ageing.

This study has some strengths and limitations that must be considered when interpreting our results. We highlight the inclusion of recent data from two nationally representative samples of older adults as a strength that ensures external and internal validity, along with the methodological studies’ rigor and comparative procedures. As a potential limitation, analyzing healthy ageing indicators separately limited interpretations. However, healthy ageing scores are a new approach, and their wide use depends on further investigations. Another potential limitation relates to the wealth measurement consistency, a significant covariate in countries’ comparisons, that was limited due to countries’ particularities. Finally, the cross-sectional design limits establishing the associations’ direction, not ruling out reverse causality bias.

In conclusion, despite having different contexts, consistently in Brazil and England, lonely women were less likely to age healthy than men. The environment in which older adults live should be targeted by public policies to empower social connections, decrease loneliness feelings and promote healthy ageing in later life, mainly among women.

Table 1 –

Age-adjusted prevalence of healthy ageing indicators and other participant’s characteristics among women and men from Brazil and England – ELSI-Brazil (201921) and ELSA (2018–19).

Variable (%)	Age-adjusted prevalence
	Women				Men
	Brazil		England		Brazil		England
	%	95% CI	%	95% CI	%	95% CI	%	95% CI
Other participant’s characteristics
Loneliness	18.8	16.2–22.0	22.5	20.7–24.0	14.7	11.4–19.0	16.2	14.5–18.0
Living alone	28.6	24.1–34.0	32.2	30.4–34.0	20.3	17.4–24.0	19.4	17.6–21.0
Having a partner	39.7	37.0–43.0	56.3	54.4–58.0	71.7	69.2–74.0	72.2	70.1–74.0
Frequent social network contact	94.7	93.5–96.0	81.8	80.0–83.0	93.8	92.2–95.0	79.7	77.7–82.0
Multimorbidity	50.8	46.6–55.0	44.6	42.7–47.0	34.9	32.1–38.0	43.4	41.2–46.0
Depressive symptoms	33.3	30.1–37.0	32.7	30.8–35.0	20.9	17.7–24.0	22.1	20.3–24.0

CI: confidence interval.