Independent report

Tackling loneliness evidence review: executive summary

Updated 17 March 2023

What we need to know next about loneliness

We have updated what we know about loneliness since the 2018 Loneliness Strategy and related evidence review. We have done this through a voluntary network of experts, not a formal review. We did this to work out what we still need to know and how to start. This is important because the evidence base is growing as we hoped. A review of loneliness statistics by the Office for Statistics Regulation (OSR) in 2021 found that data on loneliness was increasingly important to a range of users including UK government and devolved administrations, local authorities, academics, charities and community groups. The review noted that while there are a range of official statistics on loneliness produced by the government, little local level demographic data exists, meaning some users, including charitable organisations, are producing their own statistics to fill data gaps.^{[footnote 1]} But this growth of evidence makes it hard to find it all, to make sense of it, and know what you can rely on for decision making, especially for those of us who are not specialist loneliness researchers.

As well as continuing to improve measurement, we have identified eight priority areas outlined below:

Measurement, prevalence, drivers and consequences

Loneliness affects most people, but chronic loneliness has been linked to poor physical health, mental health and poor personal wellbeing. There is now more data on loneliness for the UK population, including from during the pandemic using consistent measures for adults.

Reported loneliness is higher for those who are:

• 16-24 years old
• female
• single or widowed
• living with a limiting mental health condition
• renters
• lower neighbourhood belonging
• lower local social trust^{[footnote 2]}

There is also some indication loneliness is also linked to poorer academic attainment, lower school trust, and lower liking of school.^{[footnote 3]}

Prevalence across groups most affected

We now have much better representative data, complementing pre-2018 evidence, and we are getting closer to understanding how loneliness affects different age groups, to summarise:

• the overall rate of loneliness did not change for teenagers from 2006 to 2014^{[footnote 4]}
• however, loneliness did increase over teenage years (i.e. older teenagers were more lonely than younger ones)
• for teenagers, loneliness is related to negative social experiences including:
  • bullying from peers and siblings
  • arguments with parents
  • cluster of bad experiences including victimisation and family conflict^{[footnote 5]}

Consequences of, or associations with, loneliness

• For older adults, physical health has been extensively researched
• For adolescents, research is limited, but growing and includes a forthcoming synthesis on health and loneliness for non-clinical groups and three ESRC funded analysis projects with What Works Wellbeing^{[footnote 6]}
• Role of loneliness in predicting poor health and wellbeing including for teens^{[footnote 7]} and evidence of impact on education^{[footnote 8]} and employment is growing^{[footnote 9]}
• Experience of loneliness in teenagers is linked with self-harm behaviour
• Something called HPA-axis functioning seems to be a likely explanation for the link between loneliness and health outcomes. Higher levels of cortisol (a stress hormone) on waking, and a blunted cortisol awakening response seem to be associated with loneliness in adults and teenagers. This pattern increases risk for mental and physical health issues for all ages.^{[footnote 10]}

What we need to know next

We need to measure and monitor loneliness for subgroups of population other than by regions e.g. for different disabilities or illnesses - and we should not assume the drivers and consequences are the same for everyone. It will be helpful to look at what driving factors can be changed e.g. structural, physical spaces, policies etc, that offer maximum change. For example, forthcoming research shows that identifying as a sexual minority matters more for loneliness in some geographic regions than others.^{[footnote 11]}

• Establishing ‘norms’: We need ‘normative values’ for the measures of loneliness, so high levels and changes can be benchmarked.
• Age invariance: Are loneliness questions understood the same way by people of different ages and when we are at different points in the life course? Are there true differences in loneliness or are differences due to how groups understand and respond to the questions? This can help with comparisons across age groups.
• Suitability: Are the recommended measures suitable across vulnerable groups (e.g., people with distinct disabilities)?
• Intensity, frequency and duration: All aspects are found to be important, and can inform decisions about interventions.
  • What are the most valid indicators of loneliness severity?
  • Clarifying the difference between transient and chronic loneliness
• Other useful information: Some people will not be aware of feeling lonely, and may think of loneliness differently, or call it something else.
  • We need to test this and if so, suggest other useful objective measures to take alongside the measures of loneliness
  • Determine whether others — friends, partners, strangers — can meaningfully estimate the loneliness of those they know

How do passing feelings of loneliness become chronic?

We still don’t know the answer to this and we do not know when loneliness becomes a problem for individuals? Are the predictors of transient and chronic loneliness distinct?

What we need to know next

• Population surveys that have robust loneliness measures collected from the same people over time mean researchers can now investigate how types of loneliness, circumstances, health, work and education outcomes all interact (see longitudinal quantitative data table). Separate measurement of frequency, intensity and duration can facilitate that understanding.
• A systematic review of theoretical work looking at the time frame over which loneliness moves from passing to long term.

1. Life course approach

Studies to date have looked at specific age groups. Looking at how loneliness changes over people’s lives is now possible because of longitudinal studies that follow people over time (see section 5.1, table 1). Exploring differences between age groups is also possible in experimental work. For example, looking at the difference in impact of life events at different periods of life e.g., being widowed at different ages, how cumulative experiences of loneliness or certain life events impact loneliness over time, and how we continue to be lonely or exit from it.

What we can see so far is that loneliness in later life is higher for:

• women with experience of economic hardship and conflictual paternal relationship^{[footnote 12]}
• men with experience of prolonged bullying or parental substance abuse^{[footnote 13]}
• poverty in childhood^{[footnote 14]}

It also looks like loneliness can affect specific age groups in some activities e.g. the need for social connection at age 25-34 affects performance on some social tasks, but that does not appear to be the case for other age groups.^{[footnote 15]}

What we need to know next

There is a particular gap in studies looking at people in mid-life 25-60 and in understanding how prior experiences can impact later loneliness. Work on loneliness could explore what’s common or unique about loneliness for people of different ages, what helps at different ages, life stages and life transitions.

2. Social stigma and loneliness

We know there is a link between social stigma and loneliness:

Experience of discrimination, either directly or indirectly, as an extended family member or carer, are strong predictors of loneliness, reducing social networks and can cause relationship strain for several groups with findings around ethnicity, race, immigration, mental health, sexual orientation, homelessness, intellectual disabilities, and autism.^{[footnote 16]}

These experiences can also affect loneliness indirectly by affecting how people feel about themselves, reducing trust in others and/or increasing sensitivity to rejections which can lead to shying away from social interactions.^{[footnote 17]}

Loneliness itself is also socially stigmatized and those reporting loneliness are often perceived as socially inept, poorly adjusted, and generally incompetent, and that is confirmed in research where people are more unwilling to befriend someone described as lonely. Young men particularly feel more shame about feeling lonely than older people or women.^{[footnote 18]}

What we need to know next

• How do groups currently not researched who experience stigma experience loneliness? Includes people with disabilities, some sexualities/ages.
• How do we improve social skills and social skills-based interventions - which skills, whose skills, and how?
• What is the impact of stigma on family members and informal caregivers’ social needs? It’s well understood that the stigma exists, but less is known about the consequences for family members
• What and how do structural factors e.g., policies, legislation, service provision impact loneliness and / or relationships between loneliness and health?
• How do we avoid increasing stigma with interventions?

3. Societal culture

Culture, in this context, is the most intangible aspects of our way of life, including social behaviour, values, ideologies, beliefs, customs, habits, expectations, rules, and norms in communities. A common area of research on loneliness is whether societies are individualistic or collectivist in culture, but findings about how they affect loneliness are inconsistent. See also stigma and measurement.

What we need to know next

• What is the impact of structural, as well as individual, level factors on loneliness?
• What is the experience of loneliness in different cultures and across different ages?
• How does culture impact loneliness?
• What is the role of cultural immigrant groups juggling multiple cultures?
• How generalisable are findings and interventions between cultural contexts?
• Does culture affect the link between loneliness and health using a variety of health indicators?
• Look at a broader range of aspects of culture and at different units of analysis other than country level e.g., regions, neighbourhoods etc to understand how loneliness emerges and can be addressed.
• How does culture interact with different social demographics to affect loneliness?

4. Mental health

The Loneliness and Social Isolation in Mental Health Network is funding preliminary work in this area, including looking at groups with mental illnesses. There is a range of quality in evidence, but it has found association between loneliness and:

• a range of mental illnesses including dementia, paranoia, psychosis, anxiety, depression and becoming depressed (people reporting loneliness are more at risk of becoming depressed and depressed people are more at risk of becoming lonely)^{[footnote 19]}
• suicidal thoughts, behaviours, and attempts^{[footnote 20]}
• all health outcomes^{[footnote 21]}

Loneliness looks likely to be a significant predictor of both suicidal ideation and behaviour, with depression potentially a link between them. There is strong evidence that loneliness and poor social support predict worse outcomes for people with depression.^{[footnote 22]} The impact of social isolation and loneliness on mental health of healthy children and teens suggest that it increases risk of later depression and maybe anxiety between 4 months and up to 9 years later with some suggestion that the length of time feeling lonely impacting more on mental health than how intensely lonely they felt.^{[footnote 23]}

Types of interventions to tackle loneliness for general population and for people with mental health conditions

There is some preliminary evidence on interventions that may be acceptable and potentially helpful among people with mental health problems, both in psychological interventions aimed at changing the way people think about other people and about their relationships, and interventions focused more on practical support in extending social interactions^{[footnote 24]}. However, in most cases what we have are preliminary studies that show approaches are feasible and acceptable, but do not provide definitive evidence.

What we need to know next

There is a gap in the evidence around the mechanisms that link loneliness and mental health for:

• middle-aged adults - where suicide rates are high
• those from minority ethnic backgrounds
• men
• the impact of COVID on social isolation and loneliness for healthy children and its impact on later depression and anxiety risk

Prioritise research for groups with pre-existing mental health conditions as they are high risk for loneliness and social isolation and routes out are harder with potential to greatly improve both quality of life and clinical outcomes including:

• understanding the relationship and mechanisms between depression and loneliness and interventions
• longitudinal studies measuring loneliness and other mental health diagnoses such as Obsessive Compulsive Disorder, other specific anxiety disorders, personality disorder, psychotic illness, eating disorders and alcohol and substance use disorders (including bidirectional risk and multi-level risk factors individual, family, community, society)
• qualitative studies exploring mechanisms linking the experiences of loneliness, mental illness, and stigma for those living with long term mental health conditions
• specific populations that should be prioritised because of the potential for chronic loneliness to lead to mental health problems are:
  • children in care
  • care leavers
  • young adults especially those Not in Education Employment or Training (NEETS)
  • middle-aged adults
  • people who are LGBT
  • specific ethnic groups
  • refugees and asylum seekers
  • parents including perinatal loneliness
  • retired
  • people with disabilities and their carers, especially young carers

5. Place and context

Geographic region accounts for 5-8% variation in loneliness and the effect of gender, sexual orientation and ethnicity on loneliness differs by geographical region^{[footnote 25]}. Analysis from the pandemic finds loneliness higher in areas with higher numbers of people aged 16-24 years or unemployed during the pandemic.^{[footnote 26]}

Other evidence has focused on the place- based factors that affect loneliness and social isolation and suggest:

• living in remote areas is associated with poor transport, reduced local activity choices, social isolation of minorities, poor digital connectivity, and lack of opportunities to socialise outside of school, which increases loneliness^{[footnote 27]}
• young people also report feeling particularly lonely in densely populated urban areas, including at university, that are rich in social opportunities^{[footnote 28]}
• living in greener, more walkable, and less populated areas is linked with lower loneliness^{[footnote 29]}
• individuals with a higher sense of belonging to their neighbourhoods and higher trust in the inhabitants of their neighbourhoods feel less lonely and higher neighbourhood social cohesion has been associated with better mental- and physical health^{[footnote 30]}

We know more now about the place-based factors that affect loneliness and social isolation, but less about how they impact loneliness and related mental and physical health^{[footnote 31]}. This includes regional differences, deprivation, built and natural environment, land use patterns, urban design, street network connectivity, travel and digital infrastructure, area wealth, social cohesion, neighbourhood perception, and ethnic homogeneity.

What we need to know next

• What are the characteristics of places that facilitate social connection?
• What do people need/want from their communities?
• What living and workspaces should encourage belonging?
• How do the place and context impact the relationships to loneliness we find in the data?
• What is the role of various aspects of place?
• What constitutes place-based factors? And which aspects affect which groups?
• How do these place-based factors relate to loneliness?
  • Natural environmental factors - air quality, light pollution, noise, green and blue space
  • Connectivity - transport, digital etc
  • Built environment and public realm - homes, buildings, zoning, street space
  • Public structures - community assets like libraries, leisure facilities, cafes
  • Neighbourhood factors - social cohesion, deprivation, neighbourhood perception, group density, crime

How do these place-based factors affect:

• the risk factors for loneliness
• the transition from transient to chronic loneliness
• at different life stages and sensitive periods for placed based risk factors
• how they might be modified to reduce loneliness
• the interaction with individual differences
• the interaction with effectiveness and acceptability of interventions for loneliness

6. Workplace

This is a fast-growing area of interest and work is what we spend most of adult life doing. We can see that being in work is linked with a lower risk of loneliness. In work, we see loneliness linked to either personality factors or to the workplace (e.g. nature of the job or the work culture).

Work is also changing: we change jobs more than we used to, and digital and remote working is increasing; temporary rather than permanent contracts are associated with loneliness.^{[footnote 32]} Relationships at work are a top driver of job satisfaction and can affect wellbeing and performance. A sense of community and belonging in the workplace, as in any community, for people who’ve moved from their usual social networks^{[footnote 33]}. A lack of social connection at work can lead to lower commitment and and productivity, higher absenteeism and staff turnover, and those who report higher loneliness appear less approachable to colleagues.^{[footnote 34]}

What we need to know next

Because this is an area with an early-stage evidence base, mixed methods research and basic data are needed. Routinely collecting data on loneliness (e.g., in staff surveys and surveys of people in working age with different working patterns would help). Research questions include:

• What is the impact of workplace loneliness on work performance, teams and collaborations?
• How do relational aspects of work e.g. help giving/receiving, (peer) mentoring, care/support networks affect loneliness?
• How do working patterns impact loneliness?
• How does remote working impact loneliness?
• How should awareness of loneliness be embedded at an organisational level? Does this reduce stigma? (it should) and how can we ensure it does so?
• How do organisational culture and values impact the experience of loneliness?
• How do managers impact loneliness?
• What can managers do? How can work be structured to decrease loneliness?
• What do managers need?
• How do workplace transitions affect loneliness? E.g. starting new job, being promoted, retirement, life events such as caring or health conditions, new parenthood, bereavement
• How can shared activities at work help with loneliness as they do for wellbeing?
• How can organisations help to reduce loneliness in their local communities or customers? E.g. with staff volunteering

7. The economic case for tackling loneliness

Understanding the cost of loneliness and the return on investment from interventions is currently limited and is a significant gap. We need both robust evaluations and economic analysis embedded into the intervention trials, so that there is information on:

• relative programme costs based on costs of service development and delivery
• direct impacts on loneliness
• impacts on other areas such as use of health and social care services, work and education participation, informal care and willingness to volunteer

Much of the existing evaluation literature has focused on small scale, short -term, evaluation of interventions, but longer-term economic evaluation is needed. Different approaches can be used to address these gaps in understanding and could include:

• collecting data over longer time periods as part of evaluation
• pooling evidence on costs and effects from multiple small-scale studies
• linking evaluations to routinely collected datasets, for instance on health service utilisation

What we need to know next

Methodological research is needed in these areas. For example, the economic evaluation outcome measure used in health is the Quality Adjusted Life Year (QALY) and more recently the Wellbeing Adjusted Life Year (WELBY) so it is important to measure quality of life with e.g. EuroQoL EQ-5D instrument for adults, not just loneliness. There are known problems with typical health measures in this area where arguably they do not account well for psychological and social aspects. Wellbeing and other measures may also be used to capture impacts on loneliness in assessment of quality of life.

• How to develop quality of life instruments to assess the impact of loneliness on people’s lives?
• How well do quality-of-life instruments (Quality Adjusted Life Years - QALY) and wellbeing instruments (Wellbeing Year - WELBY) work in assessing the impact of loneliness in the lives of individuals?
• How well do the loneliness measures map to comparable utility values associated with different health states used to estimate QALYs?
• How might you ‘bolt on’ additional domains to existing quality of life measures to better capture loneliness?
• What is the relationship between loneliness and wellbeing measures used in evaluation and appraisal?
• How to identify and capture all key relevant economic aspects of loneliness?
• What are the key areas affected by loneliness?
• How to gather meaningful data for economic analysis.
• How to measure and research impacts that happen over many years?

8. Effectiveness of interventions

The number of studies examining interventions for loneliness has increased significantly in recent years and is diverse in terms of participant groups and intervention types, although most are for older people and western countries for groups or individuals. There are some school and community-based interventions with broad target groups. Social support and social network intervention strategies are most used because they are group face to face interventions, although digital is increasing.

A large meta-analysis in 2021 showed that interventions with the primary aim of reducing loneliness were effective despite a large variation between types of intervention and age groups. In 1 in 3 multiple strategies are used and examples of community programmes consisting of different interventions.^{[footnote 35]} Combining and targeting intervention strategies is recommended and broadening out the range of drivers of loneliness covered by interventions. The quality of evaluations now needs to be improved.

What we need to know next

• Systematic, robust, high quality evaluations of loneliness interventions in terms of methodology and analysis
• Analysis of duration of effects
• Effectiveness of interventions for different groups of people especially outside older age and which work better for different groups.
• Develop interventions that address other drivers and to understand what the key ingredients are for interventions across different situations.

1. Please see (OSR, 2021) ↩

2. Please see the following - (ONS, 2018; BBC Loneliness Study, Barreto et al., 2021a) ↩

3. Please see - (HBSC English data; Qualter, Hennessey, et al., Forthcoming) ↩

4. Please see (Qualter, Hennessey, et al., 2021; Qualter, Hennessey, et al., 2021; Yang, Petersen, & Qualter, 2020), complementing existing evidence ↩

5. Please see (Matthews et al., 2018) ↩

6. Please see (Eccles, Maes, et al., 2021) ↩

7. Ibid ↩

8. Please see (Qualter, Hennessey, et al., 2021) ↩

9. Ibid ↩

10. Please see (Adam et al., 2006; Jopling et al., 2021; Steptoe et al., 2004); ↩

11. Please see (Doyle and Molix, 201 5) ↩

12. Please see (Nicolaisen and Thorsen, 2014) ↩

13. Please see (Kamitya et al., 2014) ↩

14. Ibid ↩

15. Please see (Pearce et al., 2021) ↩

16. Please see (Gray, 2002; Vasilieou et al., 2017) ↩

17. Please see (Doyle & Molix, 2014; Smart Richman et al., 2016; Zhang et al., 2020). ↩

18. Please see (Barreto et al., 2021b). ↩

19. Please see (Solmi et al., 2020) ↩

20. Please see (McLelland et al., 2020) ↩

21. Please see (Park et al., 2020) ↩

22. Please see (Wang et al., 2018) ↩

23. Please see (Loades et al., 2020) ↩

24. Please see (Ma et al., 2020) ↩

25. Please see (Marquez, Long, et al., 2022) ↩

26. Please see (ONS, 2021a) ↩

27. Please see (Alwood, 2020) ↩

28. Please see (ONS, 2018) ↩

29. Please see (Domènech-Abella et al, 2020, Foster et al., 2015; Maas et al., 2009; Scharf et al., 2008) ↩

30. Please see (Kearns et al., 2015; Kress et al. 2020) ↩

31. Please see (Cacioppo and Cacioppo, 2018) ↩

32. Please see (Moens et al., 2021) ↩

33. Ibid ↩

34. Please see (Bycio, Hackett, & Allen, 1995; Ozcelik & Barsade, 2018). ↩

35. Please see (Christensen et al., 2021) ↩