How to Make Societies Thrive? Coordinating Approaches to Promote Well-being and Mental Health

Reader’s guide

Acronyms and abbreviations used in this report

Table 1. ISO codes for countries and world regions

AUS	Australia	FIN	Finland	LVA	Latvia
AUT	Austria	FRA	France	MEX	Mexico
BEL	Belgium	GBR	United Kingdom	NLD	Netherlands
CAN	Canada	GRC	Greece	NOR	Norway
CHE	Switzerland	HUN	Hungary	NZL	New Zealand
CHL	Chile	IRL	Ireland	OECD	OECD average
COL	Colombia	ISL	Iceland	POL	Poland
CRI	Costa Rica	ISR	Israel	PRT	Portugal
CZE	Czech Republic	ITA	Italy	SVK	Slovak Republic
DEU	Germany	JPN	Japan	SVN	Slovenia
DNK	Denmark	KOR	Korea	SWE	Sweden
ESP	Spain	LTU	Lithuania	TUR	Türkiye
EST	Estonia	LUX	Luxembourg	USA	United States

Table 2. List of acronyms and abbreviations used in the report

ABC	Act Belong Commit
B4IG	Business for Inclusive Growth
CBT	Cognitive behavioural therapy
CLPM	Cross-lagged panel model
COPD	Chronic obstructive pulmonary disorder
COVID-19	Coronavirus disease of 2019
EBSA	Emotionally-based school avoidance
EEG	Electroencephalogram
EHIS	European Health Interview Survey
EU-SILC	European Union Statistics on Income and Living Conditions survey
EQLS	European Quality of Life Survey
GDP	Gross domestic product
GHQ-12	General Health Questionnaire
GP	General practitioner
HiAP	Health in All Policies
IA	Impact assessment
IPS	Individual Placement and Support
IPV	Intimate partner violence
K-6	Kessler Scale
LBTQI	Lesbian, gay, bisexual, transgender and intersex
MHI-5	Mental Health Inventory
MWIA	Mental well-being impact assessment
NCD	Non-communicable disease
NEET	Young people not in education, employment or training
OECD	Organisation for Economic Cooperation and Development
PISA	Program for International Student Assessment
PHQ-8	Patient Health Questionnaire
PTSD	Post-traumatic stress disorder
RCT	Randomised control trial
SEL	Social and emotional learning
SF-12 / SF-36	Short Form Health Status
SUHI	Surface urban heat island
(S)WEMWBS	(Short) Warwick-Edinburgh Mental Wellbeing Scale
UBI	Universal Basic Income
UKHLS	Understanding Society – United Kingdom Household Longitudinal Study
WHO	World Health Organization
WHO-5	WHO-5 Wellbeing Index

Indicator definitions: Mental health outcomes

Mental health outcomes are defined differently throughout this report, depending on the specific mental health outcome, and specific tool, that is used. This section provides more detailed explanations of each mental health screening tool, divided into sections focused on mental ill-health (encompassing outcomes relating to mental distress and depression) and positive mental health. For more information about tools that can be used to measure population level mental health outcomes, refer to Measuring Population Mental Health (OECD, 2023[1]).

Cross-country data on population mental health outcomes remains patchy. Therefore throughout this report, data from existing internationally comparative surveys are used to illustrate the interlinkages between mental health and different well-being domains, details of which can be found in this section. The country coverage is primarily European, given data availability. Similarly, the mental health tools used in this report reflect current availability of internationally harmonised data on mental health, rather than an endorsement of any one mental health screening tool.

Tools to measure symptoms of mental ill-health

Mental Health Inventory (MHI-5) The Mental Health Inventory-5 (MHI-5) is a five-item scale to screen for symptoms of psychological distress. It is drawn from the 38-item Mental Health Inventory (MHI) and included in the 20-item and 36-item versions of the Short Form Health Survey (SF-20 and SF-36) (Berwick et al., 1991[2]; Kelly et al., 2008[3]). The questions tap into both negative and positive affect, with three items focusing on low/depressed mood and two on nervousness/anxiety (although the tool itself is not used to present these aspects separately). The MHI-5 has been found to be a reliable measure of mental health status, and has been validated against both depressive, and to a lesser degree also anxiety disorders (including generalised anxiety and panic disorder) in general population and patient samples in a range of countries (Yamazaki, Fukuhara and Green, 2005[4]; Hoeymans et al., 2004[5]; Elovanio et al., 2020[6]; Gill et al., 2007[7]; Rumpf et al., 2001[8]; Strand et al., 2003[9]; Thorsen et al., 2013[10]). There is some evidence that removing the two anxiety-related items does not reduce the effectiveness of the MHI in detecting depression, although this has not been examined in studies in which a formal diagnosis according to clinical criteria was used as a gold standard (Yamazaki, Fukuhara and Green, 2005[4]).

This report uses the MHI-5 screener tool as it is used in the 2013 and 2018 waves of the European Union Statistics on Income and Living Conditions (EU-SILC) surveys. For the figures in this report, those whose MHI-5 score is less than or equal to 52 are considered to be at risk for mental distress, and those whose score is greater than 52 are not considered to be at risk.

Table 3. MHI-5 Questionnaire with scoring breakdown

	All of the time	Most of the time	A good bit of the time	Some of the time	A little of the time	None of the time
During the past month, how much of the time:
1. Have you been a happy person? (reverse coded)	1	2	3	4	5	6
2. Have you felt calm and peaceful? (reverse coded)	1	2	3	4	5	6
3. Have you been a very nervous person?	1	2	3	4	5	6
4. Have you felt downhearted and blue?	1	2	3	4	5	6
5. Have you felt so down in the dumps that nothing could cheer you up?	1	2	3	4	5	6

Note: All items are added together to provide a total score from 5 to 30, which is then transformed into a variable ranging from 0-100 using a standard linear transformation. Higher values indicate better mental health, with the following cut-off points for various degrees of psychological distress: 68 or less mild, moderate or severe, 60 or less moderate or severe, 52 or less severe.

Source: Kelly, M.J. et al. (2008[3]), “Evaluating cutpoints for the MHI-5 and MCS using the GHQ-12: a comparison of five different methods”, BMC Psychiatry 8, 10, https://doi.org/10.1186/1471-244X-8-10.

General Health Questionnaire (GHQ-12) The 12-item General Health Questionnaire (GHQ-12) is a measure to detect psychological distress by focusing on affect (negative and positive), somatic symptoms and the functional impairment of respondents. The GHQ-12 has been translated into many languages and extensively validated in general and clinical populations worldwide (particularly against depression and anxiety disorders), including among adolescent samples (Hankins, 2008[11]; Gilbody, 2001[12]; Baksheev et al., 2011[13]). Originally intended as a unidimensional measure, there is some debate about the dimensionality of the GHQ-12, with many factor-analytical studies supporting a range of multidimensional structures (e.g. anxiety and depression, social dysfunction, loss of confidence) (Gao et al., 2004[14]). However, more recent evidence points to these results likely being an expression of method-specific variance caused by item wording, supporting the notion that treating the scale as a unitary construct would minimise bias (Hystad and Johnsen, 2020[15]). The GHQ-12 is subject to copyright restrictions and can thus not be republished in this report.

This report uses the GHQ-12 as it appears in the United Kingdom Household Longitudinal Study (UKHLS), for inclusion in cross-lagged panel models illustrating the bidirectional relationship between selected well-being outcomes and mental distress.

Patient Health Questionnaire (PHQ-9 / PHQ-8) The full Patient Health Questionnaire (PHQ) contains 59 questions, with modules focusing on mood, anxiety, alcohol, eating and somatoform disorders. PHQ-9 is a nine-question survey designed to detect the presence and severity of depressive symptoms, and it directly maps onto the DSM-IV and DSM-5 symptom criteria for major depressive disorder. The PHQ-8 questionnaire removes the final question regarding suicidal ideation. While a one-factor structure for both PHQ-8/9 has been identified, more recent studies support a two-factor model composed of affective and somatic factors (Sunderland et al., 2019[16]). Both instruments have shown acceptable diagnostic screening properties across various population and clinical settings, age groups, and cultures/ ethnicities, in addition to being also a reliable and valid measure of depression severity (Manea, Gilbody and McMillan, 2012[17]; Moriarty et al., 2015[18]; Kroenke et al., 2009[19]; Huang et al., 2006[20]; Kroenke, Spitzer and Williams, 2001[21]; Richardson et al., 2010[22]). The close alignment between the PHQ-8/9 and the DSM make it subject to the same criticism, including a potentially Western-focused construct of depression, relative to longer self-reported scales with less constrained symptom sets (Zimmerman et al., 2012[23]; Haroz et al., 2017[24]).

This report makes use of the PHQ-8, as it appears in the European Health Interview Survey (EHIS). For the figures in this report, those whose PHQ-8 score is greater than or equal to 10 are considered to be at risk for depression, and those whose score is less than 10 are not considered to be at risk. Other scoring conventions are possible (refer to the Notes section of Table 4 below).

Table 4. PHQ-9/8 questionnaire with scoring breakdown

	Not at all	Several days	More than half the days	Nearly every day
Over the last two weeks, how often have you been bothered by any of the following problems:
1. Little interest or pleasure in doing things	0	1	2	3
2. Feeling down, depressed or hopeless	0	1	2	3
3. Trouble falling or staying asleep, or sleeping too much	0	1	2	3
4. Feeling tired or having little energy	0	1	2	3
5. Poor appetite or overeating	0	1	2	3
6. Feeling bad about yourself – or that you are a failure or have let yourself or your family down	0	1	2	3
7. Trouble concentrating on things, such as reading the newspaper or watching television	0	1	2	3
8. Moving or speaking so slowly that other people could have noticed. Or the opposite – being so fidgety or restless that you have been moving around a lot more than usual	0	1	2	3
9. Thoughts that you would be better off dead or of hurting yourself in some way	0	1	2	3

Note: The last item in italics is the question on suicidal ideation that is added for the PHQ-9. Scoring can be done in two ways: (1) via an “algorithm diagnosis” of either major depression or other depression; or (2) via summing all items and applying different cut-off scores for depression severity. In the algorithm diagnosis that adheres to DSM definitions, the first or second item (depressed mood or anhedonia) have to present at least “more than half the days”, and combined with at least 5 of the total symptoms or 2 to 4 symptoms also present at this frequency constitutes major depression and other depression, respectively. In the second form of categorization, all items are added together to provide a total score of depression severity, with scores ranging from 0-24 for the PHQ-8 and 0-27 for the PHQ-9: 0-4 none, 5-9 mild depression, 10-14 moderate depression, 15-19 moderately severe depression, 20-24/27 severe depression. A score of ≥10 indeed typically represents clinically significant depression regardless of diagnostic status.

Source: Kroenke et al. (2009[19]), “The PHQ-8 as a measure of current depression in the general population”, Journal of Affective Disorders, Vol. 114(1-3), pp. 163-73, http://dx.doi.org/10.1016/j.jad.2008.06.026; Kroenke et al. (2001[21]), “The PHQ-9: Validity of a brief depression severity measure”, Journal of General Internal Medicine, Vol. 16/9, pp. 606-613, http://dx.doi.org/10.1046/j.1525-1497.2001.016009606.x.

Tools to measure positive mental health

WHO-5 Well-being index (WHO-5) The World Health Organization Well-Being Index (WHO-5) is a short questionnaire of 5 questions that focus on a respondent’s positive affect. The questionnaire, adapted from the longer WHO/ICD-10 Depression Diagnosis and DSM-IV Depression scale by selecting a subset of positively phrased items, has first been used in a project on well-being measures in primary health care by the WHO Regional Office in Europe in 1998 and since been translated into more than 30 languages (WHO, 1998[25]; Topp et al., 2015[26]). The WHO-5 has been applied as a generic scale for well-being across a wide range of study fields and countries, as a sensitive screening tool for depression as well as an outcome measure in clinical trials (Topp et al., 2015[26]). Studies of younger and elderly persons indicated a unidimensional structure for this scale (Topp et al., 2015[26]).

This report makes use of the WHO-5, as it appears in the 2016 wave of the European Quality of Life Survey (EQLS). For the figures in this report, those whose WHO-5 score is less than or equal to 52 are considered to have poor psychological well-being, and those whose score is greater than 52 are considered to have good psychological well-being. This scoring convention has been used by a number of publications, including Eurofound (Sándor et al., 2021[27]; WHO Collaborating Center for Mental Health, 1998[28]), however other scoring conventions are possible (see notes of Table 5 below).

Table 5. WHO-5 questionnaire with scoring breakdown

	All of the time	Most of the time	More than half the time	Less than half the time	Some of the time	At no time
Over the past two weeks…
1. I have felt cheerful and in good spirits	5	4	3	2	1	0
2. I have felt calm and relaxed	5	4	3	2	1	0
3. I have felt active and vigorous	5	4	3	2	1	0
4. I woke up feeling fresh and rested	5	4	3	2	1	0
5. My daily life has been filled with things that interest me	5	4	3	2	1	0

Note: All items are added together to provide a total score from 0 to 25, which is then multiplied by 4 to normalise to a 0 (worst possible well-being) to 100 (best possible well-being) score. A cut-off score of less than or equal to 50, or less than or equal to 52 (Sándor et al., 2021[27]), are often used as indicative of reduced well-being, validated in studies using the WHO-5 for the screening of depression and for the predicting patient mortality.

Source: Topp et al. (2015[26]), “The WHO-5 well-being index: A systematic review of the literature”, Psychotherapy and Psychosomatics, Vol. 84/3, pp. 167-176, http://dx.doi.org/10.1159/000376585.

The Warwick-Edinburgh Mental Well-Being Scale (WEMWBS) The 14-item WEMWBS scale was developed with funding from NHS Health Scotland in 2005 to measure mental well-being (conceived of as “both feeling good and functioning well”), taking the Affectometer 2 instrument as starting point (Warwick Medical School, 2021[29]). Some studies confirmed a unidimensional structure for WEMWBS, while others identified three residual factors relating to affective well-being, psychological functioning or eudaimonia, and social relationships (Shannon et al., 2020[30]; Koushede et al., 2019[31]). A shorter, 7-item version of the scale, the SWEMWBS is also available, focusing slightly less on affect (Stewart-Brown et al., 2009[32]). (S)WEMWBS has been validated in various populations and among different subgroups including adolescents, clinical samples and ethnic minority samples, has been translated into more than 25 languages and validated in Norwegian, Swedish, Italian, Dutch, Danish, German, French and Spanish. Both scales have been shown to be sensitive to changes that occur in mental wellbeing promotion and mental illness treatment and prevention projects (Koushede et al., 2019[31]). Both instruments can distinguish mental well-being between subgroups, but the SWEMBS has been found to be less sensitive to gender differences compared to the longer version (Koushede et al., 2019[31]; Ng Fat et al., 2017[33]).

The cross-lagged panel models included in this report use SWEMWBS as a measure of psychological flourishing (see section on the cross-lagged panel model below for further details).

Table 6. (S)WEMWBS questionnaire with scoring breakdown

	None of the time	Rarely	Some of the time	Often	All of the time
Over the last two weeks…
1. I’ve been feeling optimistic about the future	1	2	3	4	5
2. I’ve been feeling useful	1	2	3	4	5
3. I’ve been feeling relaxed	1	2	3	4	5
4. I’ve been feeling interested in other people	1	2	3	4	5
5. I’ve had energy to spare	1	2	3	4	5
6. I’ve been dealing with problems well	1	2	3	4	5
7. I’ve been thinking clearly	1	2	3	4	5
8. I’ve been feeling good about myself	1	2	3	4	5
9 I’ve been feeling close to other people	1	2	3	4	5
10. I’ve been feeling confident	1	2	3	4	5
11 I’ve been able to make up my own mind about things	1	2	3	4	5
12 I’ve been feeling loved	1	2	3	4	5
13 I’ve been interested in new things	1	2	3	4	5
14 I’ve been feeling cheerful	1	2	3	4	5

Note: Items in italics represent the 7-item shorter version of the scale (SWEMWBS). For the 14-item scale, all items are summed, yielding a total score ranging from 14-70. For the 7-item scale, raw scores are transformed into a 7-35 metric score (see conversion table here: https://warwick.ac.uk/fac/sci/med/research/platform/wemwbs/using/howto/swemwbs_raw_score_to_metric_score_conversion_table.pdf).For both scales, higher scores indicate greater levels of positive mental health. (S)WEMWBS scores approximate to a normal distribution, permitting parametric analysis. For categorical scoring, cut-off points for high, average and low mental well-being can be generated using two approaches: (1) a statistical approach putting the cut-off point at +/- one standard deviation, placing approximately 15% of the sample into high well-being and 15% into low well-being categories; or (2) a benchmarking approach against validated measures of depression, e.g. a score of 41-44 as indicative of possible/mild depression and a score of >41 as indicative of probable clinical depression, using the Center for Epidemiologic Studies Depression Scale (CES-D) as benchmark. WEMWBS is protected by copyright. Those wishing to use WEMWBS can obtain a licence to do so. Please go to https://warwick.ac.uk/wemwbs/using for information on the type of licence you will require and details on how to apply. A free of charge “non-commercial” licence is available to public sector organisations, charities, registered social enterprises and to researchers employed in Higher Education Institutions. Any further enquiries can be directed to [e-mail].

Source: Warwick-Edinburgh Mental Well-being Scale (WEMWBS) © NHS Health Scotland, University of

Warwick and University of Edinburgh, 2006, all rights reserved; Warwick Medical School (2021[29]), The Warwick-Edinburgh Mental Wellbeing Scales – WEMWBS, https://warwick.ac.uk/fac/sci/med/research/platform/wemwbs/.

Indicator definitions: Well-being outcomes

Figures throughout this report show the share of the population in OECD countries who experience some form of a well-being deprivation, spanning the full range of dimensions of the OECD Well-being Framework, disaggregated by mental health outcome: the overall population, those at risk for poor mental health, and those not at risk for poor mental health. The mental health outcome categories vary depending on the type of mental health tool used (refer to the above section). Well-being deprivations are binary categories, for example: those who are unemployed, who live below the poverty line, who are lonely, who do not feel safe walking alone at night, who feel excluded from society, etc. The following tables in this section provide more information on how each deprivation is constructed, along with a brief introduction to the data sources.

European Union Statistics on Income and Living Conditions (EU-SILC)

The European Union Statistics on Income and Living Conditions (EU-SILC) survey has been conducted annually since 2003 in affiliated European countries. The core module primarily focuses on income, employment and housing outcomes, however each year a rotating ad-hoc module is implemented to conduct a deep-dive on a specific topic. In 2013 and 2018, the ad-hoc module focused on well-being, and included the five questions comprising the Mental Health Inventory (MHI-5) screening tool for mental distress (see Table 3 above for more information).

The well-being deprivations (and in some cases, resilience factors, meaning above a threshold of well-being in a given dimension) described in Table 7 below all come from the 2018 EU-SILC survey, with the exception of those marked by an asterisk (*), which come from the 2013 survey wave. The ad hoc well-being modules in 2018 and 2013 have overlapping question sets, but they are not identical to one another, therefore some indicators included in 2013 were not repeated five years later. Not all of the indicators in the below table come from the ad-hoc module; many of those relating to material conditions are included in the core module and are thus collected annually.

There are 26 OECD member states with EU-SILC data in this report: Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom.

Table 7. Well-being deprivation and resilience variable definitions from the EU-SILC survey

Well-being dimension	Variable	Deprivation definition
Income and wealth	Receive any benefits	Received at least one of the following the year prior: unemployment benefits, disability benefits, survivors' benefits, sickness benefits, family and child allowances, housing allowances, education allowances, or other social exclusion allowances.
	Difficulty making ends meet Financially secure	Report ability to make ends meet with "with great difficulty," "with difficulty”, or "with some difficulty." Report ability to make ends meet “fairly easily”, “easily” or “very easily”.
	Bottom quintile household income	Household net disposable income is defined as the sum of all household members' gross personal income components, minus all taxes, social insurance contributions and inter-household transfers.
	At risk of severe material deprivation	Inability to afford / presence of arrears for any four of the following nine items: (1) has arrears on mortgage or rent payments, utility bills, hire purchase installments, or other loan payments, (2) afford paying for one week's annual holiday away from home, (3) afford a meal with meat, chicken, fish (or veg equivalent) every second day, (4) afford unexpected financial expenses, (5) afford a telephone (including a mobile phone), (6) afford a colour TV, (7) afford a washing machine, (8) afford a car, (9) able to continue keeping the home adequately warm.
	Receive disability benefits	Received a non-zero, positive sum of disability benefits the year prior. Disability can be either physical or mental impairments that inhibit one from employment. Disability benefits include: disability pensions (for those below the standard retirement age), early retirement because of inability to work, care allowances, economic integration of the handicapped. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
Work and job quality	Long hours worked	Work more than 49 hours per week in both a main job and additional jobs, combined. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Dissatisfied with job	Report a score of less than 5 on a scale from 0 (not at all satisfied with their job) to 10 (completely satisfied). For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Temporary contract	Have a work contract of limited duration. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Unemployed Employed	Spent over half of the previous year in unemployment, defined as not employed during the reference period, available and actively seeking work. Spent over half of the previous year in employment, including both full- and part-time employment. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Bottom decile gross wage	The bottom decile of gross (neither taxes nor social contributions have yet been extracted at source) employee cash or near cash income. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old), who are full-time workers.
	Unable to work	Spent over half of the previous year disabled and/or unfit to work. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Self-reported early retirement	Spent over half of the previous year retired, and are under the age of 65.
	Self-reported underemployment	Work fewer than 30 hours per week and give the reason for this lower number of hours worked as due to being “underemployed” (self-reported). For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
Housing	Rents home	Rent home at either a prevailing/market rate, or at a reduced rate.
	No Internet at home	Do not have an Internet connection for personal use at home, either because it is unaffordable, or for other reasons.
	Noise from neighbors or street	Report having problems with noise from neighbors or from outside / the street.
	Problems with accommodation	Have at least two of the following problems: (1) leaking roof, damp walls/floors/foundation, or rot in window frame or floor; (2) inability to keep home adequately warm; (3) lack a bath or shower in dwelling; (4) lack indoor flushing toilet for sole use of the household).
	Housing cost overburden No housing cost overburden	Spend more than 40% disposable household income on housing costs (either rent, or mortgage costs – including both principal repayment and mortgage interest). Households that are not overburdened by housing costs (refer to above definition).
	Overcrowded accommodation	This indicator follows the OECD Housing database definition, which is as follows. A household is considered as living in overcrowded conditions if less than one room is available in each household: for each couple in the household; for each single person aged 18 or more; for each pair of people of the same gender between 12 and 17; for each single person between 12 and 17 not included in the previous category; and for each pair of children under age 12. Rooms refer to bedrooms, living and dining rooms.
	Dissatisfied with accommodation*	Report a score less than 5 on a scale from 0 (not at all satisfied with accommodation) to 10 (completely satisfied).
Knowledge and skills and educational attainment	Low education (full pop) High education (full pop)	Highest level of education attained is less than primary, primary or lower secondary (ISCED levels 0-2), full population (all ages). Have obtained tertiary or above levels of education (ISCED levels over 2), full population (all ages).
	Low education (young people)	Highest level of education attained is less than primary, primary or lower secondary (ISCED levels 0-2), limited to those currently aged 25 to 34.
Environmental quality	Pollution, grime or other environmental factors Live in an area free of pollution, etc.	Report that pollution, grime or other environmental factors (e.g., smoke, dust, unpleasant smells, polluted water), in the area close to where the household lives, are a problem. Report that none of the above are problems in the area close to where the household lives.
	Dissatisfied with recreational and green areas*	Report a score less than 5 on a scale from 0 (not at all satisfied) to 10 (completely satisfied); respondents are asked to rate their satisfaction with recreational or green areas in the space where they live, which is defined as areas close to their residence.
Safety	Don't feel safe walking alone at night*	Report feeling “very unsafe” or “a bit unsafe” when walking alone at night.
	Crime, violence or vandalism in the area Live in an area free of crime	Report that “yes” the household has a problem with crime, violence or vandalism in the area in which they live. Report that “no” the household does not have a problem with crime, violence or vandalism in the area.
Work-life balance	Dissatisfied with commute time*	Report a score less than 5 on a scale from 0 (not at all satisfied with commuting time) to 10 (completely satisfied).
	Dissatisfied with time use Satisfied with time use	Report a score less than 5 on a scale from 0 (not at all satisfied with the amount of time they have to do things they like doing) to 10 (completely satisfied). Report a score greater than 5.
	No time or money for leisure activities	Do not regularly participate in a hobby or leisure activity either because of inability to afford to, or for other reasons.
Social connections and social capital	Dissatisfied with personal relationships	Report a score less than 5 on a scale from 0 (not at all satisfied with personal relationships) to 10 (completely satisfied).
	Could receive no non-material help from others	Report that “no” to the possibility (whether actual help is currently needed or not) of asking for and receiving moral help or services from relatives, friends, neighbours or other people they know.
	Feel lonely Rarely or never lonely	Felt lonely “all of the time” or “most of the time” over the past four weeks. Felt lonely “a little of the time” or “none of the time” over the past four weeks.
	Live alone	Live alone (respondent is sole household member).
	Could receive no material help from others	Report that “no” to the possibility (whether actual help is currently needed or not) of asking for and receiving material help (e.g., money, consumer goods, etc.) from relatives, friends, neighbours or other people they know.
	Do not see friends and family monthly	Do not get together with friends, family, or relatives for a drink or meal at least once a month, either because of inability to afford it, or for other reasons.
	No trust in others Have trust in others	Report a score less than 5 on a scale from 0 (do not trust other people at all) to 10 (completely trust). Report a score greater than 5.
Civic engagement	Feel excluded from society	Report a score greater than 5 on a scale from 0 (not at all feel excluded from society) to 10 (feel completely excluded).
	No trust in the police	Report a score less than 5 on a scale from 0 (do not trust the police at all) to 10 (completely trust).
	No trust in the legal system	Report a score less than 5 on a scale from 0 (do not trust the legal system at all) to 10 (completely trust).
	No trust in politics	Report a score less than 5 on a scale from 0 (do not trust the political system at all) to 10 (completely trust).

Source: OECD categorisations based on the 2013 and 2018 European Union Statistics on Income and Living Conditions (EU-SILC) (n.d.[34]), (database), https://ec.europa.eu/eurostat/web/microdata/european-union-statistics-on-income-and-living-conditions.

European Health Interview Survey

The European Health Interview Survey (EHIS) collects data on health status and determinants for citizens of EU countries. All of the figures in this report use data from the second wave of data collection in 2014.

EHIS includes the eight individual questions that make up the Patient Health Questionnaire (PHQ-8) to measure risk for major depressive disorder (refer to Table 4 for more information). Table 8 below lists the different well-being deprivations that are included in this report.

There are 22 OECD member states with EHIS data in this report: Austria, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Sweden and the United Kingdom.

Table 8. Well-being deprivation variable definitions from the EHIS survey

Well-being dimension	Variable	Deprivation definition
Health and healthy behaviours	Long-standing health problems	Report having a long-standing health problem (i.e., suffer from any illness or health problem of a duration of at least 6 months).
	Daily activities limited	State being “severely limited” or “limited but not severely” in daily activities because of a health problem.
	Never exercise	Report never engaging in physical activities (including doing sports, fitness or recreational (leisure) physical activities that cause at least a small increase in breathing or heart rate for at least 10 minutes continuously).
	Low back disorder	Report having suffered from a low back disorder or other chronic back defect in the past 12 months.
	Poor health	Report their health is “bad” or “very bad” in general.
	Severe bodily pain	Report having suffered from “very severe” or “severe” bodily pain over the past 4 weeks.
	High blood pressure	Report having suffered from high blood pressure in the past 12 months.
	Neck disorder	Report having suffered from a neck disorder or other chronic neck defect in the past 12 months.
	Arthrosis	Report having suffered from asthma (including allergic asthma) in the past 12 months.
	Allergy	Report having suffered from an allergy (such as rhinitis, eye inflammation, dermatitis, food allergy or other allergies, but excluding allergic asthma) in the past 12 months.
	Smoke every day	Report smoking tobacco daily.
	Urinary incontinence	Report having suffered from urinary incontinence in the past 12 months.
	Obese	Have a body mass index (BMI) >= 30.
	Rarely eat fruit	Report eating fruit “less than once a week” or “never”.
	Diabetes	Report having suffered from diabetes in the past 12 months.
	Asthma	Report having suffered from asthma (including allergic asthma) in the past 12 months.
	Coronary heart disease	Report having suffered from coronary heart disease or angina in the past 12 months.
	Bronchitis	Report having suffered from bronchitis in the past 12 months.
	Kidney problems	Report having suffered from kidney problems in the past 12 months.
	Rarely eat vegetables	Report having suffered from urinary incontinence in the past 12 months.
	Stroke	Report having suffered from a stroke in the past 12 months.
	Heart attack	Report having suffered from a myocardial infarction (heart attack) in the past 12 months.
	Cirrhosis	Report having suffered from cirrhosis of the liver in the past 12 months.

Source: OECD categorisations based on European Health Interview Survey (EHIS) wave 3 data (Eurostat, n.d.[35]) (database), https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:European_health_interview_survey_(EHIS).

European Quality of Life Survey

The European Quality of Life survey (EQLS) is, unlike EU-SILC and EHIS, not conducted by national statistical offices, but administered by Eurofound to monitor quality of life indicators in Europe. The data used in this report come from the 2016 round of data collection. The survey includes the World Health Organization (WHO-5) affect-based measure of positive mental health (see Table 5 for more information). Table 9 below lists the well-being deprivations that are included in this report.

There are 24 OECD member states with EQLS data in this report: Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Türkiye and the United Kingdom.

Table 9. Well-being deprivation variable definitions from the EQLS survey

Well-being dimension	Variable	Deprivation definition
Income and wealth	Receive any benefit	Anyone in the household received at least one of the following over the past 12 months: unemployment benefits, disability benefits, retirement pension, family and child allowances, housing allowances, or other social assistance.
	Difficulty making ends meet	Report ability to make ends meet with "with great difficulty," "with difficulty, or "with some difficulty."
	Severe material deprivation	A household is considered at risk for severe material deprivation if it cannot afford two of the following six items: (1) Keeping your home adequately warm (2) Paying for a week’s annual holiday away from home (3) Replacing any worn-out furniture (4) A meal with meat, chicken, fish every second day if you wanted to (5) Buying new, rather than second-hand, clothes (6) Having family or friends for a drink or meal at least once a month.
	Receive child or family benefits	Anyone in the household received child, family or care allowances over the past 12 months.
	Receive unemployment benefits	Anyone in the household received unemployment benefits over the past 12 months.
	Have arrears for paying bills	Household has been in arrears (unable to pay as scheduled) at any time in the past 12 months, for: utility bills, such as electricity, water or gas.
	Receive disability benefits	Anyone in the household received unemployment benefits or disability pension over the past 12 months.
	Have arrears for telephone expenses	Household has been in arrears (unable to pay as scheduled) at any time in the past 12 months, for: telephone, mobile or Internet connection bills.
	Have arrears for formal loans and credit	Household has been in arrears (unable to pay as scheduled) at any time in the past 12 months, for: payments related to consumer loans, including credit card overdrafts.
	Have arrears for rent	Household has been in arrears (unable to pay as scheduled) at any time in the past 12 months, for: rent or mortgage payments for accommodation.
	Have arrears for informal loans	Household has been in arrears (unable to pay as scheduled) at any time in the past 12 months, for: payments related to informal loans from friends or relatives not living in your household.
Work and job quality	Do not work preferred amount of hours	Those who work either more, or fewer, hours per week than their stated preference. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Work more than preferred hours	Those who work more hours per week than their preferred stated amount. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Perceived lack of job options	Those who state it would be “very unlikely” or “rather unlikely” for them to find a job of a similar salary, were they to lose or quit their job (asked only to those in paid work). For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Work long hours	Those who work more than 49 hours per week, in both a main job and additional jobs, combined. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Temporary contract	Have a temporary contract, which includes: a fixed term contract of less than 12 months, a fixed term contract of 12 months or more, a temporary employment agency contract, an apprenticeship or other training scheme, no written contract, or other. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Dissatisfied with job	Report a score of less than 5 on a scale from 1 (not at all satisfied with their job) to 10 (completely satisfied). For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Perceived sense of job insecurity	Those who think they are “very likely”” or “rather likely” to lose their job within the next 6 months. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Unemployed	Unemployed (constructed variable by Eurofound). For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Work fewer than preferred hours	Those who work fewer hours per week than their preferred stated amount. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	Unable to work	Unable to work due to long-term illness or disability (constructed variable by Eurofound). For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
Housing	Rents home	Rent home, either from social, municipal or non-profit housing provider, or from a private landlord or company.
	Perceived shortage of space in home	Report that “yes”, their accommodation has a shortage of space.
	Difficult to access recycling services	Report it is “very difficult” or “rather difficult” to access recycling services, including collection of recyclables (in terms of physical access, distance, opening hours, etc.).
	Difficult to access grocery stores	Report it is “very difficult” or “rather difficult” to access grocery shops or supermarkets (in terms of physical access, distance, opening hours, etc.).
	Dissatisfied with housing	Report a score of less than 5 on a scale from 1 (not at all satisfied with their accommodation) to 10 (completely satisfied).
	Problems with accommodation	Report having at least two of the following problems: (1) house has rot in windows, doors or floors; (2) house has damp or leaks in walls or roofs; (3) house lacks an indoor flushing toilet; (4) house lacks a bath or shower; (5) house lacks facilities (heating or cooling) to keep house a comfortable temperature.
	Feel likely will be necessary to move due to affordability	Those who have report that it is “very likely” or “likely” they will need to leave their accommodation within the next six months due to affordability.
Health and healthy behaviours	Used GP services	Used GP, family doctor or health centre services in the past 12 months.
	Chronic illness	Report “yes” to having any chronic (defined as lasting 6 months or more) physical or mental health problems, illnesses or disabilities.
	Used hospital services	Used hospital or medical specialist services in the past 12 months.
	Daily activities limited	Those who report their daily activities are “severely” or “to some extent” limited by their chronic (defined as lasting 6 months or more) physical or mental health problems, illnesses or disabilities.
	Dissatisfied with health services	Report a score of less than 5 on a scale from 1 (not at all satisfied with health services in the country) to 10 (completely satisfied).
	Poor general health	Report their health is “bad” or “very bad” in general.
	Used emergency services	Used emergency healthcare services in the past 12 months.
	Used online medical consultation services	Used medical consultations online or by telephone in the past 12 months.
Knowledge and skills and educational attainment	Low education (full pop)	Highest level of education attained is less than primary, primary or lower secondary (ISCED levels 0-2), full population (all ages).
	Low education (young people)	Highest level of education attained is less than primary, primary or lower secondary (ISCED levels 0-2), limited to those currently aged 25 to 34.
Environmental quality	Pollution in area	Those who report “major” or “moderate” problems with either air quality or litter/rubbish on the street in the area (defined as the immediate area of the respondent’s home).
	Difficult to access green space	Report “very difficult” or “rather difficult” to access recreational or green areas.
Safety	Feel unsafe walking alone in neighbourhood after dark	Those who “strongly disagree” or “disagree” with the statement, “I feel safe when I walk alone in this area after dark.”
	Feel unsafe at night	Those who “strongly disagree” or “disagree” with the statement, “I feel safe when I am at home alone at night.”
Work-life balance	Difficult to concentrate on work because of family responsibilities	Those who find it difficult to concentrate at work because of family responsibilities “every day” or “several times a week”.
	Unhappy with amount of time spent on care for disabled adults	Those who report wishing they could spend either more time, or less time, caring for disabled or inform family member, neighbours or friends.
	Difficult to fulfill family responsibilities	Those who find it difficult to fulfil family responsibilities because of the amount of time spent on the job “every day” or “several times a week”.
	Working hours do not fit with family/social commitments	Those who report their working hours fit “rather not well” or “not well at all” with family or social commitments outside of work.
	Long hours in unpaid work	Work more than 60 hours per week, of which at least 30 hours are unpaid work. For this indicator, outcomes are only considered for the population considered to be a part of the labour force (those between 16 and 64 years-old).
	No time to do things I enjoy	Those who “strongly agree” or “agree” with the statement, “In my daily life, I seldom have time to do the things I really enjoy.”
	Unhappy with amount of time spent on childcare	Those who report wishing they could spend either more time, or less time, caring for children or grandchildren.
	Spend more time on childcare than partner	Report spending “more time than my partner does” looking after children.
	Too tired after work for household chore	Those who come home from work too tired to do some of the household jobs that need to be done “every day” or “several times a week”.
	Difficult to combine work with care	Report it is “very difficult” or “rather difficult” to combine paid work with care responsibilities.
	Unhappy with amount of time spent on hobbies	Those who report wishing they could spend either more time, or less time, on their own hobbies or interests.
Social connections and social capital	Contact family infrequently	Those who contact (see in person, or talk on the phone) family members or relatives less than 1-3 times a month, or never do.
	Contact friends infrequently	Those who contact (see in person, or talk on the phone) friends or neighbours less than 1-3 times a month, or never do.
	Dissatisfied with family life	Report a score of less than 5 on a scale from 1 (not at all satisfied with family life) to 10 (completely satisfied).
	Unable to get help from others	Those who answer "yes" to fewer than three of the following forms of help that one could receive from friends, family or a service provider/institution: (1) help around the house if you fall ill; (2) support if you need advice about a serious personal or family matter; (3) help when looking for a job; (4) support if you were feeling depressed and needed to talk to someone, and (5) if you needed to raise funds for an emergency situation.
	Do not feel close to people in neighbourhood	Those who “strongly disagree” or “disagree” with the statement, “I feel close to the people in the area where I live.”
	Live alone	Live alone (respondent is sole household member).
	Feel lonely	Report feeling lonely “all of the time”, “most of the time” or “more than half of the time” over the past two weeks.
	No trust in others	Report a score of less than 5 on a scale from 1 (you can’t be too careful) to 10 (most people can be trusted).
Civic engagement	Perceived corruption in the education sector	Report a score of greater than 5 on a scale from 1 (education sector is not at all corrupt) to 10 (completely corrupt).
	Feel left out of society	Those who “strongly agree” or “agree” with the statement, “I feel left out of society.”
	No trust in the police	Report a score of less than 5 on a scale from 1 (do not trust the police at all) to 10 (trust completely).
	Perceived corruption in hospitals	Report a score of greater than 5 on a scale from 1 (hospitals are not at all corrupt) to 10 (completely corrupt).
	No trust in the government	Report a score of less than 5 on a scale from 1 (do not trust the government at all) to 10 (trust completely).
	No trust in the legal system	Report a score of less than 5 on a scale from 1 (do not trust the legal system at all) to 10 (trust completely).
	Do not volunteer	Did not engage in unpaid voluntary work over the past 12 months.

Source: OECD categorisations based on the 2016 European Quality of Life Surveys (EQLS) (Eurofound, n.d.[36]) (database), https://www.eurofound.europa.eu/surveys/european-quality-of-life-surveys.

Cross-lagged panel model: Further information

This report contains figures depicting visual representations of cross-lagged panel models (CLPM), which are used to examine the reciprocal relationships between different dimensions of well-being and mental health outcomes (Mayer, 1986[37]; Selig and Little, 2013[38]). CLPMs are commonly used in behavioural and psychological research to infer longitudinal relations between variables (Saeri, Cruwys and Sibley, 2018[39]; Yu et al., 2015[40]).The aim of these analyses is to offer a more holistic perspective of the relationships between well-being and mental health by evaluating bidirectional relationships between different dimensions of well-being and population mental health outcomes, over time.

Data source

The CLPMs in this report use data from Understanding Society: The United Kingdom Household Longitudinal Study (UKHLS), which provides nationally representative, high-quality longitudinal panel data. UKHLS consists of a stratified and clustered general population sample (GPS) of randomly selected respondents from approximatively 30 000 households. The first wave was conducted from 2009-11, with respondents re-interviewed annually (the latest wave for which data were available for this report is wave 10, 2019-21). Only 52% of the GPS sample were still participating after six years. Attrition was reported to be greatest among the youngest age groups, men, the Black population, people on lower incomes and in the Greater London area. No strong association was found between the attrition rate and health status for the GPS sample (Lynn and Borkowska, 2018[41]). Data collection is conducted face-to-face via computer aided personal interviews with additional self-completion instruments. Comprehensive descriptions of the techniques and methodology used are published elsewhere (Berthoud et al., 2009[42]), as are sampling methodologies (ISER, n.d.[43]).

Different balanced panel sub-sample datasets were used to analyse the bidirectional relationships between mental health and different well-being indicators, depending on data availability for each pair of outcomes considered. That is, the maximum number of survey waves available for each mental health outcome and well-being indicator were used, resulting in different sample sizes for different models.

Outcome variables

Two contrasting indicators of mental health were examined to better understand the potential differences in well-being relationships to positive mental health vs. mental ill-health, both measured on a continuous scale. The General Health Questionnaire 12 (GHQ-12) is used to measure mental distress, ranging from 0 (the least distressed) to 12 (the most distressed). The short Warwick-Edinburgh Mental Well-Being Scale (SWEMWBS) captures positive mental health, with scores ranging from 7 (worst mental health outcome) to 35 (best possible outcome). Data for the GHQ-12 are available across all 10 waves, while data for the SWEMWBS are only available in waves 1, 4, 7 and 10.

Model design

CLPMs estimate the relationship between both mental health and a well-being indicator over time as well as the reciprocal temporal association between mental health and a well-being indicator, allowing for autoregressive and cross-lagged pathways. The models estimate the effect of one variable on the other, while controlling for levels of the outcome variable in the previous wave (“stability” effect). A simultaneous equation model that allows for autoregressive effects and cross-lagged effects between mental health outcomes ${(MH}_{it})$ and well-being indicators (${WB}_{it})$ at each measure point (t = 1, …, 10) may be written as:

${MH}_{it}=\mathrm{ }{\alpha }_t{MH}_{i,t-1}+\mathrm{ }{\beta }_t{WB}_{i,t-1}+\mathrm{ }{X}_i+{\upsilon }_{i,t}\mathrm{ }$ (1)

${WB}_{it}=\mathrm{ }{\delta }_t{WB}_{i,t-1}+\mathrm{ }{\gamma }_t{MH}_{i,t-1}+\mathrm{ }{X}_i+{\nu }_{i,t}\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }$ (2)

where t represents an occasion or survey wave, i represents an individual,$\mathrm{ }{\alpha }_t$ and ${\delta }_t$ are autoregressive parameters, ${\beta }_t$ and ${\gamma }_t$ are cross-lagged parameters, ${MH}_{i,t-1}$ and$\mathrm{ }{WB}_{i,t-1}\mathrm{ }$are the lags of one time unit for mental health and well-being outcomes, respectively, $X_i$ is a vector of control variables that vary over individuals but not over time, and ${\upsilon }_{i,t}$ and ${\nu }_{i,t}$ are the residuals (assumed to be normally distributed and correlated).

The autoregressive parameters are included to account for the stability of the constructs over time: the closer these autoregressive parameters are to one, the more stable the rank order of individuals is from one occasion to the next. Applying this to a mental health outcome, the closer the mental health autoregressive parameter is to one, the more that previous mental health outcomes influence current mental health. The cross-lagged parameters investigate reciprocal causal effects between well-being and mental health outcomes in this model. The relative effects of the mental health and well-being variables ${\gamma }_t$ on each other can be directly compared by standardising ${\beta }_t$ and ${\gamma }_t$, which can then be used to determine causal predominance, or comparing the impact of previous experience of mental ill-health on a current well-being outcome (say, unemployment) with the impact of previous unemployment on current mental ill-health.

The models shown in this report are the simple CLPMs described above, with standardised coefficients, for ease of interpretation. However CLPMs are estimated based on aggregating between- and within-person variance, therefore as a robustness check random-intercept cross-lagged panel models (RI-CLPM) to isolate the within-person from between-person variation in models with continuous dependent variables were also estimated. The findings were more-or-less in line with those of the CLPMs, and thus are not shown in this report.

Estimation and model fit

The two equations of the CLPM are estimated simultaneously and are adjusted for the complex sample design of UKHLS data including weighting, clustering and stratification at household level (residential addresses) (Lynn, 2009[44]).

First, unconstrained models (non-stationary models), allowing the magnitude of the autoregressive and cross-lagged effects to vary over survey waves were run, followed by a stationary model constraining lagged structural and autoregressive paths to be equal over each survey wave. In this report results from the stationary CLPM models are reported for simplicity, and because the improvement in model fit between stationary and non-stationary was often minimal.

In models with continuous dependent variables (i.e., household income, which is not shown in this report), full-information maximum likelihood (FIML) was used to reduce potential bias introduced by missing data. FIML is a more efficient way of dealing with missing data than listwise or pairwise deletion or similar response pattern imputation (Enders, Bandalos and And Bandalos, 2001[45]). In models with binary dependent variables (i.e., respondent smokes tobacco; respondent is unemployed), the Weighted Least Square Mean and Variance (WLSMV) is used.

The relative strengths of cross-lagged relationships are compared using standardised coefficients. In stationary models, the unstandardised coefficients are constrained to be equal over time, but standardised coefficients vary across waves. Consequently, when presenting results, we report standardised coefficients for the effects between the first two waves of data used.

Model fit was assessed by the Comparative Fit Index (CFI) and the Root-Mean-Square Error of Approximation (RMSEA). Fit is considered acceptable if CFI ≥0.90 and RMSEA ≤0.08, and good if CFI ≥0.95 and RMSEA ≤0.06. (Brown, 2015[46]; Hu and Bentler, 2009[47]; Chen et al., 2008[48]). All analyses were performed using Mplus and the R “MplusAutomation” package (Hallquist and Wiley, 2018[49]).

Assumptions and limitations to interpretation

As with any estimation, CLPMs require a number of assumptions in order to infer causality. For instance, it is assumed that time-invariant covariates have a constant effect across time, which may not necessarily be the case (Mund, Johnson and Nestler, 2021[50]). Additionally, neither models control for time-varying variables. A few additional points to be mindful of when interpreting or comparing results from the models in this report:

• Differences in the strength of cross-lagged effects across mental health outcomes could be due to the mental health outcomes used measuring different underlying constructs (for example, SWEMWBS also contains questions related to general life satisfaction and social connectedness).

• It may not be surprising to see attenuated effects or less/non-significant effects when using SWEMWBS as the mental health outcome since SWEMWBS data are only available every three years, meaning the model estimates the effects of three-year rather than one-year lags.