How does unemployment affect self-assessed health? A systematic review focusing on subgroup effects

Norström, Fredrik; Virtanen, Pekka; Hammarström, Anne; Gustafsson, Per E; Janlert, Urban

doi:10.1186/1471-2458-14-1310

BMC Public Health

Table 1 Summary of characteristics for review articles

From: How does unemployment affect self-assessed health? A systematic review focusing on subgroup effects

Country	Year	Study design	Ages	#	Health measure	OR^a	Gender	Age	Educ	MarS	Other factors	Method
Australia [17]	2001	CS	All^b	7,682	Part of SF-36	n/a	Sep	Part	Part, Sep	Part	Ch, Geo, HHI, Other	LinReg (non-trivial)
Australia [26]	2001–2009	Long	20–55	7,176	MHI-5	n/a	Sep	Part	-	Part	Other	Path
Australia [32]	2001–2009	Long	15–64	37,369	MHI-5	n/a	Sep	Part	Part, Sep	Part	CM, HHI, SN, Other	LinReg (non-trivial)
Belgium [12]	1992–2002	Long	≤65	5,790	HDLFGDS	Age-based	Part	Sep	Part	Part^c	HHI	Log
Brazil [16]	2002–2003	CS	15–64	6,426	SRH-5	1.7/1.3	Part	Part	Part	Part	Alc, HA, LC, Sm, Other	Log
Canada [9]	1994 & 1996	CS2	18–55	6,096	CIDI, DS	Age-based	Part	Sep	Part	Part	HA, Other	LinReg & Log
China [25]	2005	CS	20–49	8,075	HCTP	-/1.4	Part	Part, Sep	Part	Part	Alc, BMI, CM, Ethn, Geo, LC, PA, Sm, SN	Log (non-trivial)
Croatia [38]	1997–1999 & 2003	CS	19–57	4,139	SF-36	n/a	Part	Part	Part	-	-	ANCOVA
Europe [2]	2004	CS	50–64	11,462	SRH-5	-/2.1	Part	Part	Part	Part	Alc, BMI, Geo, PA, Sm, Other	Log
Europe [6]	2002–2004	CS	25–60	37,499	SRH-5	Strata	Sep	Part	-	-	Geo	Log
Europe [13]	1994–2002	Long	20–65	39,042 PY	Other	-/1.3	Part, Sep	Sep	Part	-	Geo, HHI	Log (non-trivial)
Europe [33]	1994/1995	CS	25–49	4,650	SRH-5	Strata^d	Women	Part	-	Part	Ch, Geo, HHI	Log
Europe [34]	1994–1995	CS	25–49	6,449	SRH-3, SRH-5	Strata	Sep	Part	Part	Part	Ch, Geo	Log
Europe [36]	2008/2009	CS	First work < 30	5,746	EURO-D, SRH-5	Strata	Sep	Part	Part	-	Ethn, HA,LC, RU, SES, Other	Log
Finland [7]	1996–2001	Long	All^b	20,599	SRH-5	n/a	Part	Part	Part	-	-	DiD, Prop
Finland [41]	1998	CS	20–54	15,468	BDI, SRH-5	Sex-based	Sep	Part	Part	Part	Alc, BMI, Sm, SN, Other	Log
Germany [23]	2009	CS	30–59	10,387	HRQOL-4	n/a	Sep	Part	Part	-	HHI, SN	MCDR
Germany [35]	1991–2008	Long	≤58	23,734	HS, MCSS	n/a	Part, Sep	Part, Sep	Part	Part	Ch, Geo, RU, SES, Other	LinReg (non-trivial
Great Britain [8]	1991–2008	Long	≥16	1,642	GHQ	n/a	Part	Part	Part	Part	HHI, LocUn, Other	LinReg (non-trivial)
Great Britain [10]	2003	CS	16–64	1,281	GHQ, SRH-5	1.5/1.7	Part	Part	Part	-	Geo	Log
Great Britain [14]	1991–2007	Long	16–65	10,494	GHQ	n/a	-	Part	Part	Part	HA, HHI, LC, Other	Reg
Great Britain [15]	1991–2009	Long	16–64	107,035	GHQ, Other	n/a	Part, Sep	Part, Sep	Part	Part	Ethn, HHI, LC, LocUn, Sm, Other	IV, LinReg
Great Britain [28]	2001	CS	25–59	698,880	SRH-3	n/a	Sep	Part	Part	-	LC	GLM
Great Britain [29]	1978–2004	CS2	25–59	264,660	SRH-3	Sex-based	Sep	Part	Part	-	LC, Other	Log
Great Britain [39]	1991–2009	Long	16–64	8,784	GHQ	n/a	Men	Part	-	Part	Ch, SES, Other	Reg
Norway [27]	1997–2002	Long	18–66	3,663	SCL	n/a	Part	Part	Part	Part	LocUn	Reg
Poland [22]	Not presented	CS^e	25–64	968	EQ-VAS	2.7/1.5	Part, Sep	Part	Part	-	HHI, PA, Sm	Log
Slovakia [4]	1998 & 2002	CS2	~19–22	844	RAND, SRH-5, Well-being	n/a	Sep	-	-	-	CM, SN	Reg
Spain [3]	1994	CS	25–64	3,881	GHQ	Sex-based	Sep	Part	-	Sep	Ch, SES, Other	Log
Spain [30]	2006	CS	25–64	8,515	GHQ	n/a	Sep	Sep	-	-	SES, Other	Prev
Sweden [1]	1983–1989, 1992–1997	CS	16–64	59,571	SRH-5	1.8 & 2.4 /1.9 & 2.7^f	Part, Sep	Part, Sep	Part, Sep	Part, Sep	Ethn, Geo, HA	MNLog
Sweden [5]	2001–2007	Long	20–59	12,605	GHQ	Strata	Part, Sep	Part, Sep	-	Part	Ch, HA, HHI, SES	Log
Sweden [18]	2007	Long	42	916	DepS, DS, SRH-3	Sex-based	Sep	Same age	-	-	HA	Log
Sweden [19]	1997	CS	18–24	3,453	Other	n/a	Sep^g	Sep^g	Sep^g	-	CM, SN	ANOVA, t-test
Sweden [20]	1997	CS	25–64	4,149	Other	n/a	Sep^g	Sep^g	Sep^g	-	CM, SN	ANOVA, t-test
Sweden [21]	1995	Long	30	1,044	DepS, SRH-3	Sex-based	Sep	Same age	-	-	CM, SN, Other	Log
Sweden [24]	1999–2000	CS	18–64	5,180	GHQ	Strata	Sep	Part	Part	-	CM, Geo, SN	Log
Sweden [31]	1995	Long	30	864	SRH-3	Sex-based	Sep	Same age	-	-	Alc, Ch, CM, HA, SES, SN, Other	Log
The Netherlands [37]	2003	CS	16–65	2,057	SF-36, SRH-5	-/2.6^d	Part	Part	Part	Part	Ethn	Log, LinReg
USA [11]	Many	Long	Varies^h	9,108	CES-D, SRH-5	n/a	Part	Partⁱ	Part	Part	HHI, RU	LinReg
USA [40]	1999, 2001, 2003	Long	Unspecified	8,125; 16,724 PY	SRH-5	n/a	Part	Part	Part	Part	Ethn, HHI, RU, SES, Other	MNLog

Explanation of short forms:
Country refers to country where study was performed (Europe refers to studies where at least two European countries participated). Year refers to the year(s) the study was performed. Study design refers to the study design (CS = cross-sectional, CS2 = two cross sections of the same individuals, Long = longitudinal).
Ages = Age (in years) for participants; # = number of individuals in study population (PY = Person years).
Health measures BDI = 21-item version of Beck’s Depression Inventory (validated scale), CES-D = Centre for Epidemiological Studies Depression Scale (validated scale), CIDI = Composite International Diagnostic Interview, DepS = depressive symptoms, DS = distress scale, EURO-D = European collaboration (validated scale for depression), EQ-VAS = EuroQol 5-dimensions visual analogue scale (validated scale),GHQ = General Health Questionnaire (validated scale), HCTP = health compared with peers of same age, HDLFGDS = Health Daily Living Form Global Depression Scale (validated scale), HRQOL-4 = Four-item Health Days Core Module from the Center for Disease Control (validated scale), HS = health satisfaction, Index = author-created index, MCSS = Mental Component Summary Scale (part of SF-36), MHI = Mental Health Index (part of SF-36), RAND = SF-36 questions with a different scoring algorithm, SCL = Hopkins’ Symptom Check List (validated scale), SF-36 = Short Form 36 (validated scale), SRH = self-rated health (3 or 5 groups/alternatives in questionnaire))
OR = Odds ratio presented in the paper. The first number is the crude OR and the second is the OR from the multivariate model with the most variables if non-stratified estimates are presented (n/a = not applicable, − = odds ratio not estimated for uncontrolled multivariate model) MarS = Marital status used in analysis; Sex = Sex involved in the analysis; Age = Age used in the analysis.
Other factors = Other factors included as part of the analyses or in separate analyses (Alc = high alcohol consumption, BMI = body mass index, Ch = children in the household, CM = cash margin/financial stress, Ethn = ethnicity or other similar difference in personal characteristics, Geo = geographical comparisons within and between countries, HA = health aspects such as any chronic medical condition or long-standing illness, HHI = household or individual income, LC = living conditions, LocUn = local or regional unemployment, PA = physical activity, RU = reason for unemployment, SES = socioeconomic status based on work, Sm = smoking, SN = social networks/social).
Method = Statistical method used for analysing the relation between health and unemployment (ANOVA = analysis of variance, DiD = difference in difference (similar to linear regression), GLM = generalized linear models, IV = instrumental variables, Log = logistic regression, LinReg = linear regression, MCDR = multivariate count data regression, MNLog = multinomial logistic regression, OL = ordinary logit, Prop = propensity scores, Reg = regression technique other than linear and logistic).
For all parts of the matrix: Sep = separate analyses, Part = only included in the statistical model (odds ratio or similar not always presented).
^aFor all presented odds ratios in this column, health were significantly poorer for unemployed than employed individuals.
^bInformation about age is not explicitly stated.
^cVariable “family composition” is not explained in the paper. We assume that the authors refer to marital status.
^dUnemployment is added as a controlling factor, but in theory the odds ratio is calculated identically as if unemployment was the main exposure.
^ePaper states that it has longitudinal variables but does not describe which year(s) the data collection is based on and the analysis does not indicate that it is based on longitudinal data.
^fOdds ratios presented for the periods 1983–1989 and 1992–1997 in two separate analyses.
^gCompares within groups of unemployed and employed but gives no direct comparison measurement and no significances for between-group comparisons.
^hTwo different populations are used in the article.
ⁱAge was only available for one of the studies because the other had only individuals of the same age. No factors are included in both studies other than household income.

Back to article page

ISSN: 1471-2458

Contact us

Submission enquiries: bmcpublichealth@biomedcentral.com
General enquiries: ORSupport@springernature.com