Table 2.

Characteristics of studies included in the review comparing the data to pre-recession periods and mental health outcomes, 2004–2014

Study	Setting	Study design	N Year Age	Socioeconomic determinants	Mental health outcomes	Associations/Effects	Strengths	Limitations
[28]	National population sample, Greece	Cohort	17,713 (2008–2013) Mean age 39.41y (SD = 8.83)	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Employment Status	Psychological Well-being Mental health (CES-D scale) Self-rated health	In the period 2008–2013 unemployed people faced more impaired health (3.21 vs 2.48, t = 8.34, p = 0.00) and mental health than did employed people (12.67 vs 9.39, t = 12.28, p =0.00). These health differences between unemployed and employed individuals were smaller in 2008–2009 than in 2010–2013.	Temporal order of exposures, confounders, and the outcome under consideration affected all participants at the same time, producing stronger causal conclusions. The results indicate a relationship between unemployment and health/mental health.	The impact of economic recessions varies across cultures and time periods; thus, the generalisability of the findings may be considerably limited by the uniqueness of the Greek situation.
[29]	National population sample, Italy	Cohort	37,782 (2006–2010) 15–64 y	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Employment Status	Psychological Well-being Self-rated health	Temporary workers, first job seekers and unemployed individuals all perceive their health as being worse than permanent workers do. The health inequalities between permanent workers and the unemployed rose, especially for males and young people, after the economic recession.	The temporal order of the exposures, confounders, and the outcome under consideration affected all the participants at the same time, producing stronger causal conclusions.	The impact of economic recessions varies across cultures and time periods; thus, generalising findings may be reasonably limited by the uniqueness of the Italian social system.
[31]	National population samples from Greece and Poland	Repeated cross-sectional Case–control	54,120 cases 136,952 controls (2006–2009)	Inter-time Variables Pre- and post-recession period	Psychological Well-being Self-reported health	Relative to the control population (Poland), Greece experienced a significantly bigger increase in the odds of poor health after the crisis (OR = 1.16; 95 % CI 1.04–1.29)	This study benefits from having a control group and cross-national design. The study is composed of comparable surveys across two time points, before and after the onset of the recession.	Its cross-sectional design removes the possibility of causal inference. Data is derived from 2006–2009 and the crisis started in 2008 so the long-term effects of the recession could not be investigated.
[30]	National population sample, Greece	Repeated cross-sectional	10,572 (2006) (2011) >18 years	Inter-time Variables Pre- and post- recession period	Psychological Well-being Self-reported health	Self-reported good health deteriorated from 71 % in 2006 to 68.8 % in 2011 (P < 0.05).	The study is composed of comparable surveys across two time periods (pre- and post-recession).	Its cross-sectional design removes the possibility of causal inference. It is limited to 2011.
[32]	National population sample, England, UK	Repeated cross-sectional	106,985 (1991) (2010) 25–64 y	Inter-time Variables Pre- and post- recession period	Psychological Well-being Mental health distress (GHQ-12)	Age-sex adj GHQ-12 caseness increased from 13.7 % (95 % CI:12.9–14.5 %) in 2008 to 16.4 % (95 % CI:14.9–17.9 %) in 2009 and 15.5 % (95 % CI:14.4–16.7 %) in 2010. Women only had a greater prevalence from 1991 until the recession, but men showed an increase over the period.	Uses a continuous measure of mental health symptoms. Large nationally representative dataset surveyed two times.	Its cross-sectional design removes the possibility of causal inference. Limited period of time; the long-term effects of the recession could not be investigated.
[33]	National population sample, Spain	Repeated cross-sectional	23,760 (2006) 16,616 (2012)	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Employment Status	Psychological Well-being Mental health distress (GHQ-12)	Results found an increase in the prevalence of poor mental health among men (prevalence ratio = 1.15, 95 % CI 1.04–1.26], especially among those aged 35–54 years, and a slight decrease for women between 2006/07 and 2011/12. There was a larger impact among the unemployed.	The study is composed of comparable surveys across two time points before and after the economic recession period. Representative sample	Its cross-sectional design removes the possibility of causal inference. Limited period of time (2012); the long-term effects of the recession could not be investigated.
[34]	Regional population sample, Working-age women, Stockholm, Sweden	Repeated cross-sectional	27,994 (2006) 22,639 (2010) 18–64 y	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Employment Status	Psychological Well-being Mental health distress (GHQ-12)	Mental distress increased among women of all types of employment status between 2006 and 2010, but more so among unemployed women, OR 2.65 (CI 95 % 2.17–3.23) in 2006 and OR 2.81 (CI 95 % 2.20–3.58) in 2010.	The study is composed of comparable surveys across multiple time points before and after the economic recession period.	Its cross-sectional design removes the possibility of causal inference. Data is derived from 2006–2010 and the crisis started in 2008, so the long term effects of the recession could not be investigated. The sample is composed only of women.
[35]	National population sample, Japan	Repeated cross-sectional	168,801 (1986–1989) 150,016 (1998–2001) 20–60 y	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Employment status/Income	Psychological Well-being Self-reported health	The OR for poor self-rated health (95 % CI) among middle-class people compared with the highest class was 1.02 (0.92–1.14) before the crisis and increased to 1.14 (1.02–1.29) after the crisis (p = 0.02). The association was stronger among males. Unemployed people were twice as likely to report poor health.	The study is composed of comparable surveys across multiple time points before and after the economic recession period, showing reports of increased poor health across all socioeconomic statuses.	Its cross-sectional design removes the possibility of causal inference. The study lacks individual-level information on job insecurity, work overload, or pay cuts that can work as confounders. The outcome was self-reported.
[36]	National population sample, Iceland	Cohort	9807 (2007) 5439 (2009)	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period Employment Status	Common Mental Disorders Psychological stress (PSS-4)	Age-adj stress levels increased between 2007 and 2009 (P = 0.004), only for women (P = 0.003). The OR for high stress levels increased only among women (OR = 1.37), especially those who were unemployed (OR = 3.38), students (OR = 2.01), with middle levels of education (OR = 1.65), or in the middle income bracket (OR = 1.59).	This study examines the longitudinal interrelations between employment status and socio-demographic in psychological stress levels during a period of extensive macroeconomic changes.	The impact of economic recessions varies across cultures and time periods; thus, the generalisability of the findings may be reasonably limited by the uniqueness of Icelandic culture, as well as the nature of the 2008 economic collapse in Iceland.
[37]	National population sample, Greece	Repeated cross-sectional	2197 (2008) 2256 (2011) 18–69 y	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Perceived financial strain	Common Mental Disorders Depression (SCID-I)	The odds of major depression were greater in 2011 than in 2008 (OR = 2.6, 95 % CI = 1.97–3.43). Financial strain independently and significantly predicts the presence of major depression (OR = 1.2, 95 % CI = 1.13–1.24).	The study is composed of comparable surveys across two time points before and after the period of the economic recession.	Its cross-sectional design removes the possibility of causal inference. Limited period of time; the long-term effects of the recession could not be investigated. Telephone survey.
[38]	National population sample, Greece	Repeated cross-sectional	2197 (2008) 2192 (2009) 18–69 y	Inter-time Variables Psychosocial/economic indicators Pre- and post- recession period Financial strain	Common Mental Disorders Major depressive episode–MDE (SCID-I)	The one-month prevalence of MDE in 2009 was found to be 6.8 %, compared to rates of 3.3 % in 2008 (p < 0.0001). Respondents facing serious economic hardship were at higher risk of developing an MDE.	Representative samples and comparable surveys across two time points before and during the period of the economic recession in Greece.	No causal inference can be made because of the cross-sectional nature of the study. The generalisability of the findings is limited by the uniqueness of the 2008 economic collapse in Greece. Limited period of time; the long-term effects of the recession could not be investigated.
[39]	National population sample, patients attending primary care centres, Spain	Repeated cross-sectional	7640 (2006–07) 5876 (2010–11)	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period Employment status, Mortgage payments	Common Mental Disorders Substance Disorders Depression Anxiety Somatoform Alcohol-related disorders	Since the pre-crisis period (2006), major depression increased by 19.4 %, anxiety by 8.4 %, somatoform disorders by 7.3 %, and alcohol-related disorders by 4.6 %. The risk of depression when unemployed was OR = 2.12, p < 0.001. The risk of depression resulting from mortgage payment difficulties was OR = 2.95, p < 0.001.	The study is composed of comparable surveys across multiple time points before and after the economic recession period.	Its cross-sectional design removes the possibility of causal inference. Limited period of time; the long-term effects of the recession could not be investigated. Only patients attending and able to access primary care were investigated.
[40]	Regional working population sample, Alberta, Canada	Repeated cross-sectional	3579 (2008–2009)	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period	Common Mental Disorders Major depressive disorder (MDD) Dysthymia Anxiety (CIDI)	The 12-month prevalence of major depressive disorder (MDD) before September 1, 2008; between September 1, 2008, and March 1, 2009; and between March 1, 2009, and October 30, 2009, was 5.1, 6.8, and 7.6 % (P = 0.03), respectively. The lifetime prevalence of dysthymia reported during the 3 periods was 0.4, 0.7, and 1.5 % (P = 0.006), respectively. No changes in the 12-month prevalence of social phobia, panic disorder, and generalized anxiety disorder were found over time.	This study examines changes in the population prevalence of common mental disorders before and during the period of the economic recession.	No causal inference can be made because of the cross-sectional nature of the study. The effects of socioeconomic variables were not adjusted. The long-term effects of the recession could not be investigated.
[41]	National population sample, Hong Kong, China	Repeated cross-sectional	3016 (2007) 2011 (2009) 15–65 y	Inter-time Variables Psychosocial/economic indicators Pre- and post-economic crisis period Socioeconomic/Employment Status	Common Mental Disorders Major depressive episode (MDE)	The 12-month prevalence of MDE was significantly higher in 2009 (/12.5 %) than 2007 (8.5 %). The prevalence of MDE in the unemployed group increased from 2007 (14.6 %) to 2009 (17.8 %).	The study is composed of comparable surveys across two time points before and after the period of the economic recession.	Because of population characteristics and cultural norms concerning response to economic adversity, the findings may not generalise to other societies. Its cross-sectional design removes the possibility of causal inference. The long-term effects of the recession could not be investigated.
[42]	National population sample, USA	Cohort	2,050,431 (2006–2007; 2008–2009) >18 y	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period Employment status	Substance-Disorders Alcohol use	The prevalence of any alcohol use significantly declined from 52.0 % in 2006–2007 to 51.6 % in 2008–2009. There was an increase in the prevalence of frequent bingeing, from 4.8 % in 2006–2007 to 5.1 % in 2008–2009 (P < 0.01). Unmarried non-Black men under 30 years who recently became unemployed were at the greatest risk of frequent bingeing.	Large representative sample. Longitudinal measures on changes of alcohol use prevalence during a period of economic recession.	The generalisability of the findings may be reasonably limited to the country’s own policy regarding alcohol use and the social welfare system. Limited period of time; the long-term effects of the recession could not be investigated.
[43]	Community sample, province and city of Buenos Aires, Argentina	Cross-sectional	1000 (2002) 18–65 y	Inter-time Variables During recession period	Substance-Disorders Patterns of drinking behaviour Drinking-related problems (Genacis)	During the economic crisis people drank more at home or at friends’ homes. A large number of respondents also reported that people had changed to cheaper or lower-quality alcoholic drinks.	Brings evidence on how the economic crisis is possibly resulting in increased drinking of cheaper or lower-quality alcoholic drinks.	No causal inference can be made because of the cross-sectional design. The validity of self-reports of sensitive behaviours, such as alcohol consumption. Because of the uniqueness of the Argentinean economic collapse and societal characteristics, the generalisability of the findings may be reasonably limited.
[44]	National population sample, Sweden	Cohort	4,224,210 (1992–1996)	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period Employment status	Suicidal Behaviours Excess mortality effects (suicide)	During the recession there was no additional risk of mortality as a result of suicide. During the post-recession period, there was an additional risk of mortality through suicide for unemployed men (HR = 1.43; 95 % CI = 1.31, 1.56) but not unemployed women.	This study examines longitudinal changes in suicide mortality during a period of mass unemployment in Sweden. Reports post-recessionary increases on suicide among unemployed men, considering possible time-lagged effects.	This research study fails to determine if mental health declined as a result of unemployment, or loss of benefits or income over time. The generalisability of the findings may be limited by the uniqueness of the Swedish welfare system and its economic recession.
[45]	National population sample, Greece	Repeated cross-sectional	2192 (2009) 2256 (2011)	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period Financial strain (Index of Personal Economic Distress)	Suicidal Behaviours Suicidal ideation Suicide attempts	The rate of suicidal ideation increased from 5.2 % in 2009 to 6.7 % in 2011 (χ 2 = 3.92, df = 1, p = 0.04). The increase was significant in men (7.1 vs. 4.4 %, χ 2 = 6.41, df = 1, p = 0.011) and those aged 55–64 years (7.2 vs. 1.9 %, χ 2 = 14.41, df = 1, p < 0.001), while it decreased in those younger than 24 years (4.9 vs. 13.9 %, χ 2 = 15.83, df = 1, p < 0.001). Suicidal ideation increased among psychotropic medication users (22.7 vs. 4.5 %, χ 2 = 11.10, df = 1, p < 0.001) and those asking for mental healthcare (17.3 vs. 8.3 %, χ 2 = 13.36, df = 1, p < 0.001). No unemployed respondent reported a suicide attempt in 2009, while the proportion was 4.4 % in 2011 (χ 2 = 4.12, df = 1, p = 0.042).	Representative sample. This study provides evidence on the increase in the prevalence of suicidal ideation and reported suicide attempts in a country facing a deep economic recession.	Reported suicidal ideation and suicide attempts can be susceptible to recall bias or to reluctance on the part of respondents to disclose such sensitive information. No causal inference can be made because of the cross-sectional nature of the study. Limited period of time; the long-term effects of the recession could not be investigated. The generalisability of the findings may be reasonably limited by the uniqueness of the 2008 economic collapse in Greece.
[46]	National population sample, South Korea	Repeated cross-sectional	27,745 (1998) 27,413 (2001) 25,487 (2005) 3335 (2007) >19 y	Inter-time Variables Psychosocial/economic indicators Pre- and post-recession period Socioeconomic inequalities	Common Mental Disorders Suicidal Behaviours Depression (doctor-diagnosed) Suicidal ideation Suicide attempts	The pro-rich inequalities in the prevalence of depression, suicidal ideation and suicide attempts doubled between 1998 and 2007. The CI for depression decreased from −0.126 (SE: 0.068) in 1998 to −0.278 (SE: 0.068) in 2001 and stayed constant. The CI for suicidal ideation fell gradually: −0.138 (SE: 0.012) in 1998 and −0.250 (SE: 0.028) in 2007. The CI for suicide attempts increased from −0.221 (SE: 0.062) in 1998 to −0.175 (SE: 0.075) in 2001 and −0.179 (SE: 0.089) in 2005, and in 2007 to −0.400 (SE: 0.116).	Nationally representative survey data sets The study is composed of comparable surveys across several time points before and after the period of the economic recession.	Reported suicidal ideation and suicide attempts can be susceptible to recall bias or to reluctance on the part of respondents to disclose such sensitive information. Depressed individuals in lower income groups might have been under-represented because of financial difficulties in seeking professional help. No causal inference can be made as it is a cross-sectional study.
[47]	National Population sample, Spain	Ecological study	(2005–2010)	Inter-time Variables Pre- and post-recession period	Suicidal Behaviours National suicide rates	An 8.0 % increase was found in the suicide rate above the trend since the financial crisis (95 % CI: 1.009–1.156; P = 0.03). Stratified analyses suggested that the association between the crisis and suicide rates is greatest in males and those of working age.	Uses stratified analyses and adjusted for seasonal fluctuations.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide.
[48]	Regional population sample, Andalusia, Spain	Ecological study	24,380 (2003–2012)	Inter-time Variables Pre- and post-recession period	Suicidal Behaviours Hospital records on suicide attempts	Compared to the historical trends prior to the onset of the crisis, between 2008 and 2012 there were 4989 more suicide attempts (95 % CI: 1985–8013): 2017 (95 % CI: 87–3987) in men and 2972 (95 % CI: 1878–4075) in women. In men, an association between unemployment and suicidal behaviour was found.	First ad hoc study of the impact of the recession on suicide attempts in Spain based on hospital records in a large population sample.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide.
[49]	Regional population sample, Andalusia, Spain	Ecological study	1975–2012	Inter-time Variables Pre- and post-recession period	Suicidal Behaviours Regional suicide rates	Suicide rates have increased since 1975.in recent decades, an upward trend has been observed in young people (15 to 44 y), an annual percentage rate change of 1.21 (95%CI: 0.7–1.7) for men and 0.93 (95 % CI: 0.4–1.4) for women.	Regional trend analysis of the variation in suicide rates.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide.
[50]	National population sample, Italy	Ecological study	(1980–2010) >15 y	Inter-time Variables Pre- and post-recession period	Suicidal Behaviours National suicide rates	The suicide rate for men involved in the labour force increased by 12 % in 2010 compared with that in 2006. The suicide rate declined for women of all ages and for men younger than 25 and older than 65 years of age.	Examined trends in the total official suicide rate before and after the onset of the recession.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide. There were no economic variables involved in the analyses.
[51]	National population sample, England, UK	Ecological study	2008–2010	Inter-time Variables Macroeconomic indicators Pre- and post-recession period Regional unemployment rate	Suicidal Behaviours National suicide rates	During 2008 and 2010, there were 846 more (95 % CI: 818–877) suicides among men and 155 (121–189) more suicides among women than would have been expected on the basis of historical trends. The 10 % increase in men’s unemployment was significantly associated with an increase of 1.4 % (0.5–2.3 %) in suicides.	Examined trends in the total suicide rate before and after the onset of the recession and in relation to unemployment rates.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide.
[52]	National population sample, England and Wales, UK	Ecological study	(2001–2011) 16–64 y	Inter-time Variables Pre- and post-recession period	Suicidal Behaviours National Suicide Rates	The downward trend in the suicide rate for men stopped for men aged 16–34 years in 2006 (95 % CI Quarter 3 (Q3) 2004, Q3 2007 for 16–24-year-olds & Q1 2005, Q4 2006 for 25–34-year-olds). The suicide rate in 35–44-year-old men reversed from a downward to an upward trend in early 2010 (95 % CI Q4 2008, Q2 2011). No clear evidence of an association between trends in female suicide rates and indicators of economic recession was found.	Used age- and sex-specific trends in suicide in the years before and after the economic recession of 2008 in relation to a variety of indicators of recession effects. Excluded accidental deaths	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide.
[53]	Cross-national population samples, EU, Canada and USA	Ecological study	2001–2011	Inter-time Variables Pre- and post-recession period	Suicidal Behaviours National suicide rates	In the EU, there was a rise in the suicide rate of 6.5 % above past trends in 2009. In Canada, suicides rose by 4.5 % between 2007 and 2009. In the USA, suicides rose by 4.8 % between 2007 and 2010.	Cross-national analysis. Examined trends in the total official suicide rate before and after the onset of the recession.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide. Suicide rates may vary across nations for cultural reasons.
[54]	National population sample, USA	Ecological study	1999–2010	Inter-time Variables Macroeconomic indicators Pre- and post-recession period State-level unemployment rate	Suicidal Behaviours National suicide rate	The suicide rate accelerated after the onset of the recession. There were an additional 0.51 deaths per 100,000 per year (95 % CI 0•28–0•75) in 2008–10 – an additional 1580 suicides per year (95 % CI 860–2300). A 1 % rise in unemployment is associated with a 0.99 % increase in the suicide rate (95 % CI 0 · 60–1 · 38, p < 0 · 0001)	Examined trends in the total suicide rate before and after the onset of the recession and in relation to unemployment rates.	Because of its ecological nature, the quality of the data is not assessable and no implications can be drawn regarding causality. The results should be interpreted with caution as other variables, independent of economic conditions, may be involved in the precipitation of suicide.