- Email: [email protected]
Self-reported job insecurity and health in the Whitehall II study: potential explanations of the relationship
ARTICLE IN PRESS
Social Science & Medicine 60 (2005) 1593–1602 www.elsevier.com/locate/socscimed
Self-reported job insecurity and health in the Whitehall II study: potential explanations of the relationship Jane E. Ferriea,, Martin J. Shipleya, Katherine Newmanb, Stephen A. Stansfeldc, Michael Marmota a
International Centre for Health and Society and Department of Epidemiology and Public Health, University College London Medical School, 1-19 Torrington Place, London WC1E 6BT, UK b Department of Public Policy, Kennedy School of Government, Harvard University, UK c Department of Psychiatry, Queen Mary University of London, UK Available online 30 November 2004
Abstract This paper examines the potential of demographic, personal, material and behavioural characteristics, other psychosocial features of the work environment and job satisfaction to explain associations between self-reported job insecurity and health in a longitudinal study of British white-collar civil servants. Strong associations were found between self-reported job insecurity and both poor self-rated health and minor psychiatric morbidity. After adjustment for age, employment grade and health during a prior phase of secure employment, pessimism, heightened vigilance, primary deprivation, financial security, social support and job satisfaction explained 68% of the association between job insecurity and self-rated health in women, and 36% in men. With the addition of job control, these factors explained 60% of the association between job insecurity and minor psychiatric morbidity, and just over 80% of the association with depression in both sexes. r 2004 Published by Elsevier Ltd. Keywords: Job insecurity; Health; Whitehall II study; UK
Introduction Most studies that have examined the effects of selfreported job insecurity on health have documented consistent adverse effects on measures of psychological morbidity. Evidence of adverse effects of self-reported job insecurity on other measures of morbidity is starting to accumulate, with reasonably consistent results being obtained for a number of health outcomes in both crosssectional and longitudinal studies (Platt, Pavis, & Corresponding author. Tel.: +0171-504-5643; fax: +0207-
813-0288. E-mail address: [email protected] (J.E. Ferrie). 0277-9536/$ - see front matter r 2004 Published by Elsevier Ltd. doi:10.1016/j.socscimed.2004.08.006
Akram, 1998; De Witte, 1999; Ferrie, 2001; Amick III et al., 1998). However, while evidence of the adverse effects of self-reported job insecurity on self-reported mental and physical morbidity is beginning to accumulate, work examining potential explanations of the association between job insecurity and health remains patchy (Platt et al., 1998; Kivimaki, Vahtera, Pennti, & Ferrie, 2000a; McDonough, 2000). Previous work in the Whitehall II study of British civil servants has shown self-reported job insecurity to be associated with poor self-rated health and minor psychiatric morbidity (Ferrie, Shipley, Stansfeld, & Marmot, 2002). A series of qualitative interviews with 38 British civil servants, whose current job was insecure, indicated a range of potential explanations of the job
ARTICLE IN PRESS 1594
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
insecurity–health relationship (Marmot, Ferrie, Newman, & Stansfeld, 2001). These included: demographic, personal, material and behavioural characteristics, other psychosocial features of the work environment and job satisfaction. The aim of this paper is to examine the potential of these factors to explain associations between self-reported job insecurity and health.
Methods Whitehall II study The target population for the Whitehall II study was all London-based office staff, aged 35–55, working in 20 Civil Service departments. With a response rate of 73%, the final cohort consisted of 10,308: 6895 men and 3413 women (Marmot et al., 1991). The true response rate is higher, however, because around 4% of those invited were not eligible for inclusion. Although mostly whitecollar, respondents covered a wide range of grades from office support to permanent secretary. Baseline screening (Phase 1) took place between late 1985 and early 1988. This involved a clinical examination and a self-administered questionnaire containing sections on demographic characteristics, health, lifestyle factors, work characteristics, social support, job satisfaction, life events and chronic difficulties. Successive phases of the study have alternated between collecting data by self-administered questionnaire only and collecting data via a clinical screening in addition to questionnaire completion. The most recent phase of data collection to include a clinical screening was completed between 1997 and 1999, Phase 5. Additional questions at Phase 5 collected detailed data on education, income, wealth, material deprivation and personal characteristics. Measures Job insecurity: Self-reported job insecurity was measured among participants still in paid employment at Phase 5 using the single item ‘How secure do you feel in your present job?’. Four response categories ranged from very insecure to very secure. Health outcomes: Self-reported health outcomes from the Phase 1 and 5 questionnaires include: self-rated health over the past 12 months (average, poor or very poor versus good or very good), presence of longstanding illness and minor psychiatric morbidity. The latter, assessed using the 30-item general health questionnaire (GHQ) (Goldberg, 1972), comprised GHQ total score and a subscale of the GHQ measuring depressive symptoms. For the GHQ total score each GHQ question, which asked about ‘the past few weeks’, was scored (0,0,1,1) and summed for analysis (Goldberg,
1972). The depression subscale was measured by four items derived by factor analysis. The four depressive symptom items loaded onto a single component in both sexes, (Stansfeld, Head, & Marmot, 1998) and comprised ‘thinking of yourself as a worthless person’, ‘felt life is entirely hopeless’, ‘felt life is not worth living’, ‘found at times you could not do anything because your nerves were too bad’. For the depression subscale, the four items were scored on a Likert scale from 0 to 3 and summed for analysis. The depression subscale had an alpha coefficient of 0.88. Both the GHQ-30 and depressive symptoms were used as continuous scores. Potential explanatory factors: The following factors from the Phase 5 questionnaire were examined for their potential to explain associations between self-reported job insecurity and health. Sociodemographic factors: Education was the highest qualification achieved at the time of questionnaire completion. Qualification categories were: no qualifications, ‘O’ level (exams usually taken at age 16), ‘A/S’ level, ONC/HND (exams usually taken at age 18, plus technical qualifications short of a degree), BA/BSc (first degree) and higher degree. Marital status was categorised as ‘married/co-habiting’, ‘single’, ‘divorced/separated’, and ‘widowed’. Personal characteristics: These included two subscales of the reactive responding scale: vigilance and emotional action. The vigilance scale measures the tendency of individuals to monitor the environment for threatening cues as a result of exposure to a high level of environmental demands coupled with urgency or danger. The emotional action scale measures the tendency to respond emotionally in demanding situations. In addition, a single item on optimism–pessimism, asking about expectations of positive or negative experiences over the next 5–10 years was included. This measure taps a dispositional dimension of hopelessness/negativity that has been shown to be a predictor of morbidity (Everson et al., 1996). Psychosocial work environment: Decision authority, skill discretion and job demands were adapted from the Job Content Instrument of Karasek (Karasek, 1979). Work social support comprised three components: support from colleagues, support from supervisors, and clarity and consistency of information from supervisors. Responses on a four-point scale from ‘often’ to ‘never/almost never’ were combined into summary scales and then divided into tertiles, high, moderate and low. A global measure of job satisfaction was obtained using eight items, each with four response categories. Responses were summed and divided into tertiles. Material factors: Primary deprivation was measured by summing responses to three questions, which asked about problems with housing, difficulties with the payment of bills, and how often the participant lacked sufficient money to afford the kind of food or clothing s/he or the family should have. Lack of access to common, desirable, but inessential items, such as a
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
dishwasher, CD player, home computer, holidays, savings, etc. were described as lifestyle deprivation (Whelan, 1992). The score was the sum of all items the participant would like to have but could not afford. The remaining material measures comprised personal income, household income and wealth. In addition, participants were asked to rate their sense of financial security over the next 10 years on a four-point scale from secure to insecure. Health-related behaviours: Two health-related behaviours were examined, alcohol consumption and smoking. Drinking over the recommended limits was defined as 15 or more units/week for women and 22 or more units/week for men, the recommended limits for safe drinking used in the UK General Household Survey (OPCS Social Survey Division, 1984). ‘Current smoker’ was any participant currently using either manufactured or hand-rolled cigarettes. Smokers were compared to never-smokers and ex-smokers combined. All analyses were adjusted for age at Phase 5 and socioeconomic position, measured as grade of employment at baseline. In addition, some analyses of selfreported health outcomes were adjusted for negative affect. This trait was assessed using the five negative affect items from Bradburn’s Affect Balance Scale (Bradburn, 1969). Study sample and statistical analysis Seventy-one percent (7270) of the 10,308 Whitehall II study participants at baseline completed the full questionnaire at Phase 5. A short-form questionnaire was completed via telephone interview by a further 560 participants, bringing the overall response rate to 76%. Of those who completed the full questionnaire, 4665 were still in paid employment and 2532 were out of the labour force (Box 1). Of the 4665 participants still in paid employment at Phase 5, the question on job security was completed by 4447 (1171 women and 3276 men), an item non-response of 5%. Responses to this question were dichotomised. Those who reported their current job as not very secure or very insecure, 30% of
1595
women and 25% of men, were compared with those whose job was secure or very secure. The aim of the analysis was to determine associations between job insecurity and health outcomes at Phase 5, adjusted for age, employment grade, and baseline value of the health outcome of interest. Associations between job insecurity and sociodemographic, personal, material, behavioural, job satisfaction and other psychosocial work characteristics were also determined and their potential to explain associations between job insecurity and health outcomes was assessed. For continuous variables, differences between those reporting job insecurity (exposure group) and those who felt secure in their employment (control group) at Phase 5 were assessed using analysis of covariance. Adjusted means and standard errors for both groups were produced by linear regression (GLM procedure in SAS). Results for continuous variables are presented by comparing the exposure group with the control group in terms of adjusted mean score differences for each variable of interest with 95% confidence intervals (95% CI) score difference. For dichotomous variables, logistic regression (LOGIST procedure in SAS) was used to compare the Phase 5 measures in both groups. Results for the dichotomous variables are presented in terms of odds ratios for job insecurity for each variable of interest with 95% CI. All analyses were adjusted sequentially for age and marital status at Phase 5, baseline (Phase 1) employment grade and baseline level of the variable of interest (basic model). All the outcomes were analysed separately in women and men as previous analyses of data from this cohort have observed sex differences in associations between attributed job insecurity and self-reported health outcomes (Ferrie, Shipley, Marmot, Stansfeld, & Davey Smith, 1995, 1998). Sex differences have also been observed in some of our potential explanatory factors, for example, education, marital status, income, wealth, financial security and health behaviours (Ferrie et al., 2001a; Marmot et al., 1991; Martikainen, Adda, Ferrie, Smith, & Marmot, 2003). Following the analysis of the effects of job insecurity on health outcomes, factors that might explain these
Box 1 Employment status at Phase 5 Employment status
Women
Men
In the Civil Service Working elsewhere Out of labour force Total
960 (44.8%) 285 (13.3%) 900 (42.0%) 2145 (100%)
2360 1060 1632 5052
a
Total (46.8%) (21.0%) (32.3%) (100%)
An additional 73 participants did not provide data on employment status.
3320 (46.1%) 1345 (18.7%) 2532 (35.2%) 7197a (100%)
ARTICLE IN PRESS 1596
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
Table 1 Health outcomes for insecurely employed women and men relative to those in secure employment at Phase 5 Health measures
N (women/men)
Women
Men a
Self-rated health av. or worse Longstanding illness
1155/3231 877/2497
Odds ratios (95% confidence intervals) 1.69 (1.17, 2.42) p=0.005 1.97 (1.53, 2.52) po0.001 1.27 (0.94, 1.72) p=0.12 1.05 (0.87, 1.28) p=0.61
General health questionnaire (30) score Depression score (subscale of GHQ)
1119/3192 1124/3196
Score difference (95% confidence intervals) 2.40 (1.64, 3.16) po0.001 1.97 (1.56, 2.38) po0.001 0.55 (0.30, 0.78) po0.001 0.46 (0.32, 0.59) po0.001
a Odds ratios and score differences for health outcomes are adjusted for age at phase 5, baseline employment grade, and baseline value of the outcome of interest.
associations were identified. The selected factors were ones which, when added singly to the basic model (described above), changed the coefficient for job insecurity by more than 10% in any one of the health outcomes in either sex. These factors, pessimism, vigilance, primary deprivation, financial insecurity, social support, job satisfaction, job control and household income, were therefore the ones which were associated, but not necessarily significantly so, with both job insecurity and with one of the health outcomes. Finally, all the selected factors for each health outcome were included together in the basic model for that outcome. Models were also adjusted separately for negative affect at baseline. Theoretically, as the exposure, the health outcomes and the potential explanatory factors are self-reports, negative affect may have produced overestimates of odds ratios and score differences in the analysis. Negative affect, which has been shown to be relatively stable over time (Roskies, Louis-Guerin, & Fournier, 1993), was not measured at follow-up and so the Phase 1 value was used. Ideally, all the analyses of self-reported health outcomes should have been adjusted for negative affect. However, the Affect Balance Scale was not included in the version of the questionnaire administered to the first 2193 participants at Phase 1 and so routine inclusion of this variable would have reduced the available health outcome data. To adjust for negative affect, the subset of individuals in whom negative affect was measured, 702 women and 2106 men, was used to fit models including the potential explanatory variables together with negative affect. Results are presented as unadjusted and adjusted (by the potential explanatory factors) score differences for continuous outcomes and odds ratios for dichotomous outcomes. The percentage changes between the adjusted and unadjusted score differences or between the log of the odds ratios for discrete variables are also presented (a log odds ratio value of zero indicates no association
between the explanatory factor and outcome). The datasets used in these analyses include only those respondents who have data for the health outcome of interest and the potential explanatory factors. Hence, the odds ratios or adjusted score differences presented in Table 2 may differ slightly from those presented in Table 1.
Results Health outcomes: Self-reported job insecurity was strongly associated with poor self-rated health (pp0.005) and the two measures of minor psychiatric morbidity, GHQ score and depression (pp0.001), in both sexes (Table 1). Although there was only a weak association with longstanding illness in the analyses presented in Table 1, the association among women was somewhat stronger, p=0.04, in the restricted dataset used for the analyses presented in Table 3. Potential explanatory factors: Self-reported job insecurity was associated with having fewer educational qualifications, especially for men, and with being single in women. High and moderate levels of pessimism were associated with self-reported job insecurity in both sexes, but a statistically significant association with emotional action was only seen in men. Level of vigilance was associated with job insecurity in both sexes. Compared to participants with high job control, moderate as well as low job control was strongly associated with selfreported job insecurity in both sexes, but high demands were only associated in women. In both sexes there were strong associations between self-reported job insecurity and low social support at work, and with both low and moderate job satisfaction. Self-reported job insecurity was strongly related to most material measures in both sexes, prevalence being higher among the more deprived and those with lower incomes and wealth. There was no association between job insecurity and alcohol or cigarette consumption (Table 2).
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
1597
Table 2 Odds ratio for reporting job insecurity for each category of the potential explanatory factora compared to the reference category Potential explanatory factorsb
Odds ratio (95% confidence intervals) Women N=1171
Men N=3276
No qualifications 4O=level 4A/S=level, ONC/HND BA/BSc
1.42 1.11 1.55 1.20
1.56 1.33 1.22 1.17
Single Divorced/separated Widowed
1.45 (1.03, 2.04) 0.82 (0.54, 1.24) 0.82 (0.42, 1.59)
1.11 (0.83, 1.49) 1.25 (0.86, 1.82) 0.81 (0.33, 2.01)
Personal characteristics Pessimism (reference category—optimistic [high, moderate and slight combined])
Slightly pessimistic Moderately pessimistic Highly pessimistic
1.28 (0.88, 1.85) 1.98 (1.36, 2.89) 4.05 (2.48, 6.60)
1.11 (0.87, 1.41) 1.53 (1.20, 1.95) 3.01 (2.20, 4.12)
Emotional action (reference category—unemotional)
Moderately emotional Highly emotional
0.78 (0.58, 1.06) 0.96 (0.69, 1.33)
0.72 (0.58, 0.88) 0.74 (0.61, 0.90)
Vigilance (reference category—low vigilance)
Moderate vigilance High vigilance
1.21 (0.84, 1.74) 1.98 (1.42, 2.75)
1.42 (1.15, 1.75) 1.84 (1.50, 2.25)
Work environment Decision latitude (job control) (reference category—high latitude)
Moderate latitude Low latitude
1.65 (1.15, 2.38) 2.14 (1.46, 3.14)
1.39 (1.14, 1.70) 2.51 (1.99, 3.17)
Skill discretion (reference category—high discretion)
Moderate discretion Low discretion
1.42 (0.98, 2.05) 1.68 (1.14, 2.47)
1.59 (1.30, 1.96) 2.52 (1.99, 3.19)
Job demands (reference category—low demands)
Moderate demands High demands
1.19 (0.84, 1.69) 1.53 (1.06, 2.21)
0.99 (0.79, 1.24) 1.17 (0.92, 1.49)
Social support at work (reference category—high support)
Moderate support Low support
1.13 (0.77, 1.65) 2.86 (2.04, 3.99)
1.22 (0.96, 1.54) 2.42 (1.96, 2.99)
Job satisfaction (reference category—high satisfaction)
Moderate satisfaction Low satisfaction
1.99 (1.41, 2.81) 4.02 (2.91, 5.55)
2.09 (1.68, 2.58) 3.84 (3.11, 4.74)
A little deprived Moderately deprived Most deprived
1.59 (1.12, 2.24) 2.42 (1.70, 3.44) 3.40 (2.13, 5.43)
1.27 (1.03, 1.55) 1.60 (1.28, 1.99) 2.61 (1.91, 3.56)
Lifestyle deprivation (reference category—least deprived)
A little deprived Moderately deprived Most deprived
1.38 (0.92, 2.08) 1.22 (0.78, 1.92) 1.48 (0.98, 2.24)
1.10 (0.86, 1.42) 1.31 (0.97, 1.76) 2.13 (1.62. 2.80)
Personal income (reference category—over, 50,000)
25–50,000 15–25,000 o15,000
0.65 (0.32, 1.34) 1.28 (0.58, 2.82) 1.50 (0.66, 3.38)
1.98 (1.45, 2.70) 2.48 (1.72, 3.56) 4.30 (2.75, 6.73)
Household income (reference category—over, 60,000)
40–60,000 20–40,000 o20,000
1.73 (1.00, 3.00) 2.06 (1.18, 3.61) 3.32 (1.76, 6.25)
1.14 (0.87, 1.48) 1.41 (1.05, 1.88) 2.90 (1.90, 4.45)
Wealth (reference category—over, 100,000)
40–100,000 o40,000
1.40 (1.01, 1.95) 1.45 (1.00, 2,12)
1.37 (1.10, 1.70) 1.43 (1.08, 1.88)
Financial security (10 years) (reference category—Secure)
Insecure
2.62 (1.92, 3.59)
4.28 (3.40, 5.39)
Sociodemographic factors Education (reference category—higher degree)
Marital status (reference category—married)
Material factors Primary deprivation (reference category—least deprived)
(0.74, (0.63, (0.89, (0.71,
2.69) 1.97) 2.68) 2.02)
(1.01, (0.99, (0.93, (0.90,
2.41) 1.77) 1.59) 1.52)
ARTICLE IN PRESS 1598
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
Table 2 (continued ) Potential explanatory factorsb
Odds ratio (95% confidence intervals)
Health-related behaviours Alcohol (reference category—not over recommended limits) Smoking (reference category—never and ex-smoker)
Women N=1171
Men N=3276
Over the recommended limits
0.92 (0.64, 1.30)
1.04 (0.88, 1.25)
Current smoker
0.80 (0.56, 1.13)
0.93 (0.71, 1.23)
a Odds ratios are adjusted for age at Phase 5 and baseline employment grade. Potential explanatory factors are sociodemographic factors, personal and work characteristics, job satisfaction, material factors and health-related behaviours. b Numbers of participants included in each analysis will vary due to the exclusion of those (generally few) with the potential explanatory factor missing.
Table 3 Effect of adjustment for potential explanatory factors on health outcomes for insecurely employed women and men at Phase 5 Health measure
Na
Adjustedb
Women Self-rated health (SF36)d Longstanding illnessd
941 709
Odds ratios (95% confidence intervals) p value 1.74 (1.16, 2.61) p=0.007 1.20 (0.76, 1.87) p=0.43 1.43 (1.02, 2.01) p=0.04 1.29 (0.88, 1.87) p=0.18
68 29
911 915
Score difference (95% confidence intervals) p value 2.30 (1.45, 3.15) po0.001 0.91 (0.03, 1.79) p=0.04 0.56 (0.29, 0.84) po0.001 0.09 ( 0.18, 0.37) p=0.51
60 83
3239
Odds ratios (95% confidence intervals) p value 2.15 (1.65, 2.80) po0.001 1.63 (1.22, 2.17) po0.001
36
2706 2706
Score difference (95% confidence intervals) p value 1.99 (1.53, 2.45) po0.001 0.81 (0.36, 1.27) po0.001 0.46 (0.32, 0.61) po0.001 0.08 ( 0.06, 0.23) p=0.25
59 82
GHQ (30) scoree GHQ depressiond Men Self-rated health (SF36)d GHQ (30) scoree GHQ depressione
Fully adjustedc
Change %
a Shows the number of subjects in the analysis who have no missing values for any of the explanatory factors that are specified in the adjusted model. b Odds ratios and score differences for health outcomes are adjusted for age at Phase 5, baseline employment grade, and baseline value of the outcome of interest. c Odds ratios and score differences for health outcomes in fully adjusted models are additionally adjusted as follows. d Pessimism, vigilance, primary deprivation, financial security, social support and job satisfaction. e Pessimism, vigilance, primary deprivation, financial security, social support, job satisfaction and job control.
Potential explanations of the relationship: The major potential explanatory factors were selected as described in the methods. The effects of adjusting the association between self-reported job insecurity and health for these factors are presented in Table 3. Only those health measures that showed a statistically significant inverse association with self-reported job insecurity in Table 1, plus longstanding illness in women, are shown. A combination of personal characteristics (pessimism and heightened vigilance), material factors (primary deprivation and financial security), and other psychosocial characteristics of the work environment (social support at work and job satisfaction) explained 68% of
the association between self-reported job insecurity and self-rated health in women and 36% in men. In addition, this combination of factors explained 29% of the association between job insecurity and longstanding illness in women. The same set of factors, with the addition of job control, explained 60% of the association between job insecurity and GHQ score and over 80% of the association with depression in both sexes. Job satisfaction explained the largest proportion of the association between job insecurity and self-rated health in women (31%), followed by pessimism (22%), while among men the two largest contributors were job satisfaction and financial insecurity, which both
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
explained 17%. Pessimism (28%) and social support at work (22%) explained the largest proportions of the association between job insecurity and longstanding illness in women. With respect to depression, pessimism explained the largest proportion of the association with job insecurity in women (41%), followed by job satisfaction (33%), while among men, job satisfaction explained the greatest proportion (34%), closely followed by financial insecurity (31%) and pessimism (30%). The proportion of the job insecurity–health association explained by each of the factors when entered into the final model show that the effects of the explanatory variables were partly independent and partly overlapping (data available on request). A further set of analyses in which the fully adjusted models were additionally adjusted for negative affectivity produced results similar to those presented in Table 3. The additional adjustment, in general, marginally strengthened the fully adjusted job insecurity–health relationships. These changes in effect ranged from an increase of 7% for GHQ depression to no change for self-rated health, both in women.
Discussion Methodological considerations: The major limitation of this study is that it is partially cross-sectional in design. Analysis of data collected on individuals at baseline screening, a phase of secure employment 11 years earlier, together with data collected at Phase 5 allowed us to control for the effects of previous health status on current job insecurity and for other confounding factors such as age and socioeconomic position. This enabled the study to separate morbidity associated with job insecurity from the effects of pre-existing morbidity. However, no such baseline data existed for most of the potential explanatory variables. This means that for associations between job insecurity and measures, such as pessimism, the available data do not allow direction of causation to be established. Even if this was not the case, causal associations between job insecurity, pessimism and depression are difficult to disentangle. Beliefs and emotional responses co-exist and are likely to interact. Removing the source of job insecurity is likely to improve mood and levels of optimism, but underlying anxiety, which may contribute to job insecurity, independent of objective work prospects, cannot be ruled out. Our measure of socioeconomic position, baseline employment grade, is not contemporaneous with our morbidity and risk factor measures at Phase 5. Baseline grade was chosen as it is available for all participants, and produced findings little different from analyses using last known grade. It was felt that using last known grade as the measure of socioeconomic position posed
1599
problems. Last known grade reflects mobility for those who remained in the Civil Service, but cuts short the trajectories of those who left the Civil Service to take up employment elsewhere, 29% of those in employment at Phase 5. Use of Phase 1 employment grade also minimised the effect of reverse causality, where the levels of morbidity at baseline may affect subsequent mobility and hence employment grade at Phase 5. The depression subscale of the GHQ is not an ideal measure of depression, but no other measure was available in the dataset at Phase 5. It is nevertheless a robust subscale, clearly emerging in factor analysis across gender and employment grade with good face validity for depressive content and plausible associations with known risk factors for depression (Stansfeld, Head, Fuhrer, Wardle, & Cattell, 2003). In an attempt to determine the effect of reporting bias on our findings we adjusted our analyses for negative affectivity, but found it had little influence. This is in contrast to Canadian work that found negative affectivity to be the single most important predictor of psychological distress, even after controlling for sociodemographic variables, perceived risk, and coping resources and strategies (Roskies et al., 1993). Although the trait of negative affectivity is relatively stable over time (Watson & Clark, 1984; Roskies et al., 1993), reliance on data measured 11 years earlier is not ideal. Furthermore, it has been suggested that measures of negative affectivity may be rather limited in their ability to address the overestimation of associations (Macleod, Smith, Heslop, Oliver, & Hart, 1999). Generalisability of findings: The participants, often a relatively homogeneous group working in one particular field or organisation, limit the generalisability of findings from most occupational cohort studies. Similarly, in this study, all respondents were in office-based, white-collar employment at baseline screening. However, in addition to specialists, the Civil Service employs large numbers of administrators and general office staff such as personal assistants, secretaries, personnel and welfare officers, clerks and messengers. These workers make it equivalent to many office-based settings both in the public and private sector. Also, by Phase 5, 29% of participants still in employment were working outside the Civil Service, a factor likely to increase generalisability to the wider working population. However, at a time of marked change in the labour market, it is important that findings generated by workers aged 44 and over are not assumed to apply to those at the beginning of their working life. Self-reported job insecurity and health: Relative to participants in secure employment, poor self-rated health and measures of minor psychiatric morbidity were significantly higher among workers of both sexes reporting job insecurity. Previous analyses using data from the Whitehall II cohort have demonstrated sex differences in associations between attributed job
ARTICLE IN PRESS 1600
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
insecurity and self-reported health outcomes (Ferrie, Shipley, Marmot, Stansfeld, & Smith, 1995, 1998). However, other analyses, which examined exposure to chronic self-reported job insecurity or change in job security found similar effects in both sexes (Ferrie et al., 2002). Analyses comparing the health effects of contractual job security with self-reported job security in a Finnish cohort similarly found that the different types of job insecurity had different associations with health, but found associations with health were similar in both sexes regardless of type of security (Virtanen, Vahtera, Kivimaki, Pentti, & Ferrie, 2002). Potential explanations of the association: Confirming the intimations of our qualitative findings (Marmot et al., 2001), we found a combination of personal characteristics, material factors, other psychosocial work characteristics and job satisfaction explained much of the association between job insecurity and ill health. Pessimism, vigilance, primary deprivation, future financial security, social support at work and job satisfaction explained 68% of the association between self-reported job insecurity and self-rated health in women and 36% in men. The same factors also explained 29% of the association between selfreported job insecurity and longstanding illness in women and, with the addition of job control, 60% of the association between job insecurity and GHQ score and over 80% of the association between job insecurity and depression in both sexes. Associations between personal characteristics and health outcomes are relatively well documented (DeNeve, 1999; Everson et al., 1996; Spiro III, Aldwin, Ward, & Mroczek, 1995). However, relatively little work has examined associations between personal characteristics and psychosocial characteristics of the work environment (Bosma, Stansfeld, & Marmot, 1998), including job insecurity, although one study found it to be significantly associated with neuroticism (Tivendell & Bourbonnais, 2000). Similarly, only few studies appear to have examined personal characteristics as potential explanations of associations between psychosocial work characteristics and health. Very little of the association between low job control and heart disease was explained by personal characteristics in the Whitehall II study (Bosma et al., 1998). The National Population Health Survey in Canada found that mastery and self-esteem explained half the cross-sectional association between job insecurity and self-rated health and 63% of the association between job insecurity and distress, but only a little of the association with use of medications (McDonough, 2000). Data from a longitudinal cohort study in Finland showed that hostile individuals were able to derive less benefit from psychosocial resources, such as support, network size and job control, during downsizing and were thus more vulnerable to disease (Vahtera, Kivimaki, Uutela, & Pentti, 2000).
One of the main explanations of associations between unemployment and health is loss of income, and anticipated financial problems have been found to be mediators of the relationship between unemployment and health problems (Leeflang, Klein-Hesselink, & SpruitI, 1992). Although job insecurity is likely to generate concern about future financial security, no previous research appears to have examined this as an explanation of the effect of job insecurity on health. However, data from a longitudinal cohort study in Finland showed that the effect of downsizing on sickness absence was greater among low income than high-income employees (Kivimaki, Vahtera, Pentti, & Ferrie, 2000a). In previous work we have examined the potential of other psychosocial work characteristics to explain associations between job insecurity attributed to workplace closure and major organisational change and measures of morbidity and cardiovascular risk factors. Loss of social support at work was a feature common to both situations and explained 11% of the association between attributed job insecurity and self-rated health and minor psychiatric morbidity (Ferrie et al., 2001a). Other evidence is beginning to accumulate that social support at work modifies associations between selfreported job insecurity and self-rated health (McDonough, 2000), and job insecurity and psychological symptoms (Dooley, Rook, & Catalano, 1987; Mohr, 2000; Kinnunen & Natti, 1994). However, neither support from colleagues, management nor unions modified the negative impact of job insecurity on psychological health among public transport workers in Australia (Dekker & Schaufeli, 1995). Little work has examined the role of job demands and control at work in the association between job insecurity and morbidity. Changes in job demands and control made a negligible contribution to the association between attributed job insecurity and self-reported measures of morbidity in previous work using Whitehall II data (Ferrie et al., 2001a, 2001b). Similarly, a study of South African gold miners found no evidence that job control moderated the adverse effect of job insecurity on psychological strain (Barling & Kelloway, 1996). However, a recent study in Finland found that decreased participation in decision making explained 19% of the association between sickness absence and downsizing (Kivimaki et al., 2000a). Further work on the same cohort has shown that job control also partially mediated the relationship between downsizing and selfrated health (Kivimaki et al., 2000b). Although the potential explanatory factors measured in this study explained some of the associations between job insecurity and self-reported health, between 17% and 71% of the association remained unexplained. While measurement imprecision might have underestimated the effect of the adjustments for the potential explanatory factors
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
in Table 3, probably the largest part of the residual associations are accounted for by pre-existing morbidity not adequately captured by self-rated health, longstanding illness or GHQ score, for example anxiety. We can only speculate why adjustment for the potential explanatory factors explained a similar proportion of the association between job insecurity and health for both women and men, for the GHQ measures but not for self-rated health. Possibly hostility, more frequently seen in men might explain part of the difference, alternatively, or in addition, reporting of self-rated health in men might focus more narrowly on somatic symptoms, whereas in women it may encompass more general sense of well-being. Well-being is more likely to be explained by a combination of personal characteristics and environmental factors. In conclusion, among participants in the Whitehall II study, there is strong evidence that self-reported job insecurity is associated with poor self-rated health, minor psychiatric morbidity and depression. A combination of pessimism, heightened vigilance, difficulty in paying bills and problems with housing, financial security, social support and job satisfaction explained much of the association between job insecurity and selfrated health. With the addition of job control, the same factors explained most of the association with minor psychiatric morbidity and depression. While direction of causation cannot be determined in this study, optimism, future financial security, social support at work, job control and job satisfaction are undoubted casualties when jobs are under threat. Organisational studies have shown these factors to be related to commitment, effort, morale and intention to quit, which have a direct bearing on the success of organisations (Burke & Cooper, 2000). At a time of increasing emphasis on the benefits of flexible labour markets to national economies, and increasing concern about the rise in absence from work, policies whose consequence is to increase job insecurity should take such findings into account. To this consideration must be added the detriment to individuals and organisations, which, in addition to personal misery, have economic implications of their own.
Acknowledgements The work presented in this article was supported by the United Kingdom Economic and Social Research Council (L128251046). The Whitehall II study has been supported by grants from the Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart Lung and Blood Institute (HL36310), US, NIH: National Institute on Aging (AG13196), US, NIH; Agency for Health Care Policy Research (HS06516); and the John
1601
D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. Jane Ferrie was supported by the Economic and Social Research Council (L128251046) during the preparation of this paper. Michael Marmot is supported by an MRC Research Professorship. Martin Shipley is supported by a grant from the British Heart Foundation. We also thank all participating Civil Service departments and their welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; the Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.
References Amick, B. C., III., Kawachi, I., Coakley, E. H., Lerner, D., Levine, S., & Colditz, G. A. (1998). Relationship of job strain and iso-strain to health status in a cohort of women in the United States. Scandinavian Journal of Work and Environmental Health, 24, 54–61. Barling, J., & Kelloway, E. K. (1996). Job insecurity and health: the moderating role of workplace control. Stress Medicine, 12, 253–259. Bosma, H., Stansfeld, S. A., & Marmot, M. G. (1998). Job control, personal characteristics, and heart disease. Journal of Occupational Health Psychology, 3, 402–409. Bradburn, N. M. (1969). The structure of psychological wellbeing. Chicago: Aldine. Burke, R., & Cooper, C. L. (2000). The organization in crisis: downsizing, restructuring and privatisation. Oxford: Blackwell. De Witte, H. (1999). Job insecurity and psychological wellbeing: review of the literature and exploration of some unresolved issues. European Journal of Work and Organisational Psychology, 8, 155–177. Dekker, S. W. A., & Schaufeli, W. B. (1995). The effects of job insecurity on psychological health and withdrawal: a longitudinal study. Australian Psychologist, 30, 57–63. DeNeve, K. M. (1999). Happy as an extraverted clam? The role of personality for subjective well-being. Current Directions in Psychological Research, 8, 141–144. Dooley, D., Rook, K., & Catalano, R. (1987). Job and non-job stressors and their moderators. Journal of Occupational Psychology, 60, 115–132. Everson, S. A., Goldberg, D. E., Kaplan, G. A., Cohen, R. D., Pukkala, E., Tuomilehto, J., & Salonen, J. T. (1996). Hopelessness and risk of mortality and incidence of myocardial infarction and cancer. Psychosomatic Medicine, 58, 113–121. Ferrie, J. E., Shipley, M. J., Marmot, M. G., Stansfeld, S., & Davey Smith, G. (1995). Health effects of anticipation of job change and non-employment: longitudinal data from the Whitehall II study. British Medical Journal, 311, 1264–1269.
ARTICLE IN PRESS 1602
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
Ferrie, J. E., Shipley, M. J., Marmot, M. G., Stansfeld, S., & Davey Smith, G. (1998). The health effects of major organisational change and job insecurity. Social Science & Medicine, 46, 243–254. Ferrie, J. E. (2001). Is job insecurity harmful to health? Journal Royal Society Medicine, 94, 71–76. Ferrie, J. E., Shipley, M. J., Marmot, M. G., Martikainen, P., Stansfeld, S. A., & Davey Smith, G. (2001a). Job insecurity in white-collar workers: towards an explanation of associations with health. Journal of Occupational Health Psychology, 6, 26–42. Ferrie, J. E., Martikainen, P., Shipley, M. J., Marmot, M., Stansfeld, S. A., & Davey Smith, G. (2001b). Employment status and health after privatisation in white collar civil servants: prospective cohort study. British Medical Journal, 322, 647–651. Ferrie, J. E., Shipley, M. J., Stansfeld, S. A., & Marmot, M. G. (2002). Effects of chronic job insecurity and change in job security on self-reported health, minor psychiatric morbidity, physiological measures and health-related behaviours in British civil servants: the Whitehall II study. Journal of Epidemiology and Community Health, 56, 450–454. Goldberg, D. P. (1972). The detection of psychiatric illness by questionnaire. London. Karasek, R. (1979). Job demands, job decision latitude, and mental strain: implications for job redesign. Administrative Science Quarterly, 24, 285–311. Kinnunen, U., & Natti, J. (1994). Job insecurity in Finland: antecedents and consequences. European Work and Organisational Psychologist, 4, 297–321. Kivimaki, M., Vahtera, J., Pentti, J., & Ferrie, J. E. (2000a). Factors underlying the effect of organisational downsizing on the health of employees: a longitudinal cohort study of changes in work, social relationships and health behaviours. British Medical Journal, 320, 971–975. Kivimaki, M., Vahtera, J., Pentti, J., Thomson, L., Griffiths, A., & Cox, T. (2000b). Downsizing, changes in work, and self-rated health of employees: a 7-year 3-wave panel study. Anxiety Stress and Coping, 14, 59–73. Leeflang, R. L. I., Klein-Hesselink, D. J., & SpruitI, P. (1992). Health effects of unemployment—I. Long-term unemployed men in a rural and an urban setting. Social Science & Medicine, 34, 341–350. Macleod, J., Davey Smith, G., Heslop, P., Oliver, S., & Hart, C. (1999). Always look on the bright side of life? The influence of reporting tendency when exposure and outcome measurements are based on self-report. Journal of Epidemiology and Community Health, 53, 660. Marmot, M., Ferrie, J. E., Newman, K., & Stansfeld, S. (2001). The contribution of job insecurity to socio-economic inequalities. Research FINDINGS 11 from the Health Variations Programme, Economic and Social Research Council. Marmot, M. G., Davey Smith, G., Stansfeld, S., Patel, C., North, F., Head, J., White, I., Brunner, E. J., & Feeney, A.
(1991). Health inequalities among British civil servants: the Whitehall II study. Lancet, 337, 1387–1393. Martikainen, P., Adda, J., Ferrie, J. E., Davey Smith, G., & Marmot, M. (2003). The effects of income and wealth on GHQ depression and poor self-rated health in white-collar women and men in the Whitehall II study. Journal of Epidemiology and Public Health, 57, 718–723. McDonough, P. (2000). Job insecurity and health. International Journal of Health Services, 30, 453–476. Mohr, G. (2000). The changing significance of different stressors after the announcement of bankruptcy: a longitudinal investigation with special emphasis on job insecurity. Journal of Organisational Behaviour, 21, 337–359. OPCS Social Survey Division (1984). General household survey 1982. London: OPCS Social Survey Division. Platt, S., Pavis, S., & Akram, G. (1998). Changing labour market conditions and health: a systematic literature review (1993–1998). Dublin. Roskies, E., Louis-Guerin, C., & Fournier, C. (1993). Coping with job insecurity: how does personality make a difference? Journal of Organisational Behaviour, 14, 617–630. Spiro, A., III., Aldwin, C. M., Ward, K. D., & Mroczek, D. K. (1995). Personality and the incidence of hypertension among older men: longitudinal data from the normative aging study. Health Psychology, 14, 563–569. Stansfeld, S., Head, J., & Marmot, M. (1998). Explaining social class differences in depression and well-being. Social Psychiatry and Psychiatric Epidemiology, 33, 1–9. Stansfeld, S. A., Head, J., Fuhrer, R., Wardle, J., & Cattell, V. (2003). Social inequalities in depressive symptoms and physical functioning in the Whitehall II study: exploring a common cause explanation. Journal of Epidemiology and Public Health, 57, 361–367. Tivendell, J., & Bourbonnais, C. (2000). Job insecurity in a sample of Canadian civil servants as a function of personality and perceived job characteristics. Psychological Reports, 87, 55–60. Vahtera, J., Kivimaki, M., Uutela, A., & Pentti, J. (2000). Hostility and ill health: role of psychosocial resources in two contexts of working life. Journal of Psychosomatic Research, 48, 89–98. Virtanen, P., Vahtera, J., Kivimaki, M., Pentti, J., & Ferrie, J. E. (2002). Employment security and health. Journal of Epidemiology and Public Health, 56, 569–574. Watson, D., & Clark, L. A. (1984). Negative affectivity: the disposition to experience aversive emotional states. Psychological Bulletin, 96, 465–490. Whelan, C. (1992). The role of income, life-style deprivation and financial strain in mediating the impact of unemployment on psychological distress: evidence from the Republic of Ireland. Journal Occupational Organisational Psychology, 65, 331–344.
Social Science & Medicine 60 (2005) 1593–1602 www.elsevier.com/locate/socscimed
Self-reported job insecurity and health in the Whitehall II study: potential explanations of the relationship Jane E. Ferriea,, Martin J. Shipleya, Katherine Newmanb, Stephen A. Stansfeldc, Michael Marmota a
International Centre for Health and Society and Department of Epidemiology and Public Health, University College London Medical School, 1-19 Torrington Place, London WC1E 6BT, UK b Department of Public Policy, Kennedy School of Government, Harvard University, UK c Department of Psychiatry, Queen Mary University of London, UK Available online 30 November 2004
Abstract This paper examines the potential of demographic, personal, material and behavioural characteristics, other psychosocial features of the work environment and job satisfaction to explain associations between self-reported job insecurity and health in a longitudinal study of British white-collar civil servants. Strong associations were found between self-reported job insecurity and both poor self-rated health and minor psychiatric morbidity. After adjustment for age, employment grade and health during a prior phase of secure employment, pessimism, heightened vigilance, primary deprivation, financial security, social support and job satisfaction explained 68% of the association between job insecurity and self-rated health in women, and 36% in men. With the addition of job control, these factors explained 60% of the association between job insecurity and minor psychiatric morbidity, and just over 80% of the association with depression in both sexes. r 2004 Published by Elsevier Ltd. Keywords: Job insecurity; Health; Whitehall II study; UK
Introduction Most studies that have examined the effects of selfreported job insecurity on health have documented consistent adverse effects on measures of psychological morbidity. Evidence of adverse effects of self-reported job insecurity on other measures of morbidity is starting to accumulate, with reasonably consistent results being obtained for a number of health outcomes in both crosssectional and longitudinal studies (Platt, Pavis, & Corresponding author. Tel.: +0171-504-5643; fax: +0207-
813-0288. E-mail address: [email protected] (J.E. Ferrie). 0277-9536/$ - see front matter r 2004 Published by Elsevier Ltd. doi:10.1016/j.socscimed.2004.08.006
Akram, 1998; De Witte, 1999; Ferrie, 2001; Amick III et al., 1998). However, while evidence of the adverse effects of self-reported job insecurity on self-reported mental and physical morbidity is beginning to accumulate, work examining potential explanations of the association between job insecurity and health remains patchy (Platt et al., 1998; Kivimaki, Vahtera, Pennti, & Ferrie, 2000a; McDonough, 2000). Previous work in the Whitehall II study of British civil servants has shown self-reported job insecurity to be associated with poor self-rated health and minor psychiatric morbidity (Ferrie, Shipley, Stansfeld, & Marmot, 2002). A series of qualitative interviews with 38 British civil servants, whose current job was insecure, indicated a range of potential explanations of the job
ARTICLE IN PRESS 1594
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
insecurity–health relationship (Marmot, Ferrie, Newman, & Stansfeld, 2001). These included: demographic, personal, material and behavioural characteristics, other psychosocial features of the work environment and job satisfaction. The aim of this paper is to examine the potential of these factors to explain associations between self-reported job insecurity and health.
Methods Whitehall II study The target population for the Whitehall II study was all London-based office staff, aged 35–55, working in 20 Civil Service departments. With a response rate of 73%, the final cohort consisted of 10,308: 6895 men and 3413 women (Marmot et al., 1991). The true response rate is higher, however, because around 4% of those invited were not eligible for inclusion. Although mostly whitecollar, respondents covered a wide range of grades from office support to permanent secretary. Baseline screening (Phase 1) took place between late 1985 and early 1988. This involved a clinical examination and a self-administered questionnaire containing sections on demographic characteristics, health, lifestyle factors, work characteristics, social support, job satisfaction, life events and chronic difficulties. Successive phases of the study have alternated between collecting data by self-administered questionnaire only and collecting data via a clinical screening in addition to questionnaire completion. The most recent phase of data collection to include a clinical screening was completed between 1997 and 1999, Phase 5. Additional questions at Phase 5 collected detailed data on education, income, wealth, material deprivation and personal characteristics. Measures Job insecurity: Self-reported job insecurity was measured among participants still in paid employment at Phase 5 using the single item ‘How secure do you feel in your present job?’. Four response categories ranged from very insecure to very secure. Health outcomes: Self-reported health outcomes from the Phase 1 and 5 questionnaires include: self-rated health over the past 12 months (average, poor or very poor versus good or very good), presence of longstanding illness and minor psychiatric morbidity. The latter, assessed using the 30-item general health questionnaire (GHQ) (Goldberg, 1972), comprised GHQ total score and a subscale of the GHQ measuring depressive symptoms. For the GHQ total score each GHQ question, which asked about ‘the past few weeks’, was scored (0,0,1,1) and summed for analysis (Goldberg,
1972). The depression subscale was measured by four items derived by factor analysis. The four depressive symptom items loaded onto a single component in both sexes, (Stansfeld, Head, & Marmot, 1998) and comprised ‘thinking of yourself as a worthless person’, ‘felt life is entirely hopeless’, ‘felt life is not worth living’, ‘found at times you could not do anything because your nerves were too bad’. For the depression subscale, the four items were scored on a Likert scale from 0 to 3 and summed for analysis. The depression subscale had an alpha coefficient of 0.88. Both the GHQ-30 and depressive symptoms were used as continuous scores. Potential explanatory factors: The following factors from the Phase 5 questionnaire were examined for their potential to explain associations between self-reported job insecurity and health. Sociodemographic factors: Education was the highest qualification achieved at the time of questionnaire completion. Qualification categories were: no qualifications, ‘O’ level (exams usually taken at age 16), ‘A/S’ level, ONC/HND (exams usually taken at age 18, plus technical qualifications short of a degree), BA/BSc (first degree) and higher degree. Marital status was categorised as ‘married/co-habiting’, ‘single’, ‘divorced/separated’, and ‘widowed’. Personal characteristics: These included two subscales of the reactive responding scale: vigilance and emotional action. The vigilance scale measures the tendency of individuals to monitor the environment for threatening cues as a result of exposure to a high level of environmental demands coupled with urgency or danger. The emotional action scale measures the tendency to respond emotionally in demanding situations. In addition, a single item on optimism–pessimism, asking about expectations of positive or negative experiences over the next 5–10 years was included. This measure taps a dispositional dimension of hopelessness/negativity that has been shown to be a predictor of morbidity (Everson et al., 1996). Psychosocial work environment: Decision authority, skill discretion and job demands were adapted from the Job Content Instrument of Karasek (Karasek, 1979). Work social support comprised three components: support from colleagues, support from supervisors, and clarity and consistency of information from supervisors. Responses on a four-point scale from ‘often’ to ‘never/almost never’ were combined into summary scales and then divided into tertiles, high, moderate and low. A global measure of job satisfaction was obtained using eight items, each with four response categories. Responses were summed and divided into tertiles. Material factors: Primary deprivation was measured by summing responses to three questions, which asked about problems with housing, difficulties with the payment of bills, and how often the participant lacked sufficient money to afford the kind of food or clothing s/he or the family should have. Lack of access to common, desirable, but inessential items, such as a
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
dishwasher, CD player, home computer, holidays, savings, etc. were described as lifestyle deprivation (Whelan, 1992). The score was the sum of all items the participant would like to have but could not afford. The remaining material measures comprised personal income, household income and wealth. In addition, participants were asked to rate their sense of financial security over the next 10 years on a four-point scale from secure to insecure. Health-related behaviours: Two health-related behaviours were examined, alcohol consumption and smoking. Drinking over the recommended limits was defined as 15 or more units/week for women and 22 or more units/week for men, the recommended limits for safe drinking used in the UK General Household Survey (OPCS Social Survey Division, 1984). ‘Current smoker’ was any participant currently using either manufactured or hand-rolled cigarettes. Smokers were compared to never-smokers and ex-smokers combined. All analyses were adjusted for age at Phase 5 and socioeconomic position, measured as grade of employment at baseline. In addition, some analyses of selfreported health outcomes were adjusted for negative affect. This trait was assessed using the five negative affect items from Bradburn’s Affect Balance Scale (Bradburn, 1969). Study sample and statistical analysis Seventy-one percent (7270) of the 10,308 Whitehall II study participants at baseline completed the full questionnaire at Phase 5. A short-form questionnaire was completed via telephone interview by a further 560 participants, bringing the overall response rate to 76%. Of those who completed the full questionnaire, 4665 were still in paid employment and 2532 were out of the labour force (Box 1). Of the 4665 participants still in paid employment at Phase 5, the question on job security was completed by 4447 (1171 women and 3276 men), an item non-response of 5%. Responses to this question were dichotomised. Those who reported their current job as not very secure or very insecure, 30% of
1595
women and 25% of men, were compared with those whose job was secure or very secure. The aim of the analysis was to determine associations between job insecurity and health outcomes at Phase 5, adjusted for age, employment grade, and baseline value of the health outcome of interest. Associations between job insecurity and sociodemographic, personal, material, behavioural, job satisfaction and other psychosocial work characteristics were also determined and their potential to explain associations between job insecurity and health outcomes was assessed. For continuous variables, differences between those reporting job insecurity (exposure group) and those who felt secure in their employment (control group) at Phase 5 were assessed using analysis of covariance. Adjusted means and standard errors for both groups were produced by linear regression (GLM procedure in SAS). Results for continuous variables are presented by comparing the exposure group with the control group in terms of adjusted mean score differences for each variable of interest with 95% confidence intervals (95% CI) score difference. For dichotomous variables, logistic regression (LOGIST procedure in SAS) was used to compare the Phase 5 measures in both groups. Results for the dichotomous variables are presented in terms of odds ratios for job insecurity for each variable of interest with 95% CI. All analyses were adjusted sequentially for age and marital status at Phase 5, baseline (Phase 1) employment grade and baseline level of the variable of interest (basic model). All the outcomes were analysed separately in women and men as previous analyses of data from this cohort have observed sex differences in associations between attributed job insecurity and self-reported health outcomes (Ferrie, Shipley, Marmot, Stansfeld, & Davey Smith, 1995, 1998). Sex differences have also been observed in some of our potential explanatory factors, for example, education, marital status, income, wealth, financial security and health behaviours (Ferrie et al., 2001a; Marmot et al., 1991; Martikainen, Adda, Ferrie, Smith, & Marmot, 2003). Following the analysis of the effects of job insecurity on health outcomes, factors that might explain these
Box 1 Employment status at Phase 5 Employment status
Women
Men
In the Civil Service Working elsewhere Out of labour force Total
960 (44.8%) 285 (13.3%) 900 (42.0%) 2145 (100%)
2360 1060 1632 5052
a
Total (46.8%) (21.0%) (32.3%) (100%)
An additional 73 participants did not provide data on employment status.
3320 (46.1%) 1345 (18.7%) 2532 (35.2%) 7197a (100%)
ARTICLE IN PRESS 1596
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
Table 1 Health outcomes for insecurely employed women and men relative to those in secure employment at Phase 5 Health measures
N (women/men)
Women
Men a
Self-rated health av. or worse Longstanding illness
1155/3231 877/2497
Odds ratios (95% confidence intervals) 1.69 (1.17, 2.42) p=0.005 1.97 (1.53, 2.52) po0.001 1.27 (0.94, 1.72) p=0.12 1.05 (0.87, 1.28) p=0.61
General health questionnaire (30) score Depression score (subscale of GHQ)
1119/3192 1124/3196
Score difference (95% confidence intervals) 2.40 (1.64, 3.16) po0.001 1.97 (1.56, 2.38) po0.001 0.55 (0.30, 0.78) po0.001 0.46 (0.32, 0.59) po0.001
a Odds ratios and score differences for health outcomes are adjusted for age at phase 5, baseline employment grade, and baseline value of the outcome of interest.
associations were identified. The selected factors were ones which, when added singly to the basic model (described above), changed the coefficient for job insecurity by more than 10% in any one of the health outcomes in either sex. These factors, pessimism, vigilance, primary deprivation, financial insecurity, social support, job satisfaction, job control and household income, were therefore the ones which were associated, but not necessarily significantly so, with both job insecurity and with one of the health outcomes. Finally, all the selected factors for each health outcome were included together in the basic model for that outcome. Models were also adjusted separately for negative affect at baseline. Theoretically, as the exposure, the health outcomes and the potential explanatory factors are self-reports, negative affect may have produced overestimates of odds ratios and score differences in the analysis. Negative affect, which has been shown to be relatively stable over time (Roskies, Louis-Guerin, & Fournier, 1993), was not measured at follow-up and so the Phase 1 value was used. Ideally, all the analyses of self-reported health outcomes should have been adjusted for negative affect. However, the Affect Balance Scale was not included in the version of the questionnaire administered to the first 2193 participants at Phase 1 and so routine inclusion of this variable would have reduced the available health outcome data. To adjust for negative affect, the subset of individuals in whom negative affect was measured, 702 women and 2106 men, was used to fit models including the potential explanatory variables together with negative affect. Results are presented as unadjusted and adjusted (by the potential explanatory factors) score differences for continuous outcomes and odds ratios for dichotomous outcomes. The percentage changes between the adjusted and unadjusted score differences or between the log of the odds ratios for discrete variables are also presented (a log odds ratio value of zero indicates no association
between the explanatory factor and outcome). The datasets used in these analyses include only those respondents who have data for the health outcome of interest and the potential explanatory factors. Hence, the odds ratios or adjusted score differences presented in Table 2 may differ slightly from those presented in Table 1.
Results Health outcomes: Self-reported job insecurity was strongly associated with poor self-rated health (pp0.005) and the two measures of minor psychiatric morbidity, GHQ score and depression (pp0.001), in both sexes (Table 1). Although there was only a weak association with longstanding illness in the analyses presented in Table 1, the association among women was somewhat stronger, p=0.04, in the restricted dataset used for the analyses presented in Table 3. Potential explanatory factors: Self-reported job insecurity was associated with having fewer educational qualifications, especially for men, and with being single in women. High and moderate levels of pessimism were associated with self-reported job insecurity in both sexes, but a statistically significant association with emotional action was only seen in men. Level of vigilance was associated with job insecurity in both sexes. Compared to participants with high job control, moderate as well as low job control was strongly associated with selfreported job insecurity in both sexes, but high demands were only associated in women. In both sexes there were strong associations between self-reported job insecurity and low social support at work, and with both low and moderate job satisfaction. Self-reported job insecurity was strongly related to most material measures in both sexes, prevalence being higher among the more deprived and those with lower incomes and wealth. There was no association between job insecurity and alcohol or cigarette consumption (Table 2).
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
1597
Table 2 Odds ratio for reporting job insecurity for each category of the potential explanatory factora compared to the reference category Potential explanatory factorsb
Odds ratio (95% confidence intervals) Women N=1171
Men N=3276
No qualifications 4O=level 4A/S=level, ONC/HND BA/BSc
1.42 1.11 1.55 1.20
1.56 1.33 1.22 1.17
Single Divorced/separated Widowed
1.45 (1.03, 2.04) 0.82 (0.54, 1.24) 0.82 (0.42, 1.59)
1.11 (0.83, 1.49) 1.25 (0.86, 1.82) 0.81 (0.33, 2.01)
Personal characteristics Pessimism (reference category—optimistic [high, moderate and slight combined])
Slightly pessimistic Moderately pessimistic Highly pessimistic
1.28 (0.88, 1.85) 1.98 (1.36, 2.89) 4.05 (2.48, 6.60)
1.11 (0.87, 1.41) 1.53 (1.20, 1.95) 3.01 (2.20, 4.12)
Emotional action (reference category—unemotional)
Moderately emotional Highly emotional
0.78 (0.58, 1.06) 0.96 (0.69, 1.33)
0.72 (0.58, 0.88) 0.74 (0.61, 0.90)
Vigilance (reference category—low vigilance)
Moderate vigilance High vigilance
1.21 (0.84, 1.74) 1.98 (1.42, 2.75)
1.42 (1.15, 1.75) 1.84 (1.50, 2.25)
Work environment Decision latitude (job control) (reference category—high latitude)
Moderate latitude Low latitude
1.65 (1.15, 2.38) 2.14 (1.46, 3.14)
1.39 (1.14, 1.70) 2.51 (1.99, 3.17)
Skill discretion (reference category—high discretion)
Moderate discretion Low discretion
1.42 (0.98, 2.05) 1.68 (1.14, 2.47)
1.59 (1.30, 1.96) 2.52 (1.99, 3.19)
Job demands (reference category—low demands)
Moderate demands High demands
1.19 (0.84, 1.69) 1.53 (1.06, 2.21)
0.99 (0.79, 1.24) 1.17 (0.92, 1.49)
Social support at work (reference category—high support)
Moderate support Low support
1.13 (0.77, 1.65) 2.86 (2.04, 3.99)
1.22 (0.96, 1.54) 2.42 (1.96, 2.99)
Job satisfaction (reference category—high satisfaction)
Moderate satisfaction Low satisfaction
1.99 (1.41, 2.81) 4.02 (2.91, 5.55)
2.09 (1.68, 2.58) 3.84 (3.11, 4.74)
A little deprived Moderately deprived Most deprived
1.59 (1.12, 2.24) 2.42 (1.70, 3.44) 3.40 (2.13, 5.43)
1.27 (1.03, 1.55) 1.60 (1.28, 1.99) 2.61 (1.91, 3.56)
Lifestyle deprivation (reference category—least deprived)
A little deprived Moderately deprived Most deprived
1.38 (0.92, 2.08) 1.22 (0.78, 1.92) 1.48 (0.98, 2.24)
1.10 (0.86, 1.42) 1.31 (0.97, 1.76) 2.13 (1.62. 2.80)
Personal income (reference category—over, 50,000)
25–50,000 15–25,000 o15,000
0.65 (0.32, 1.34) 1.28 (0.58, 2.82) 1.50 (0.66, 3.38)
1.98 (1.45, 2.70) 2.48 (1.72, 3.56) 4.30 (2.75, 6.73)
Household income (reference category—over, 60,000)
40–60,000 20–40,000 o20,000
1.73 (1.00, 3.00) 2.06 (1.18, 3.61) 3.32 (1.76, 6.25)
1.14 (0.87, 1.48) 1.41 (1.05, 1.88) 2.90 (1.90, 4.45)
Wealth (reference category—over, 100,000)
40–100,000 o40,000
1.40 (1.01, 1.95) 1.45 (1.00, 2,12)
1.37 (1.10, 1.70) 1.43 (1.08, 1.88)
Financial security (10 years) (reference category—Secure)
Insecure
2.62 (1.92, 3.59)
4.28 (3.40, 5.39)
Sociodemographic factors Education (reference category—higher degree)
Marital status (reference category—married)
Material factors Primary deprivation (reference category—least deprived)
(0.74, (0.63, (0.89, (0.71,
2.69) 1.97) 2.68) 2.02)
(1.01, (0.99, (0.93, (0.90,
2.41) 1.77) 1.59) 1.52)
ARTICLE IN PRESS 1598
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
Table 2 (continued ) Potential explanatory factorsb
Odds ratio (95% confidence intervals)
Health-related behaviours Alcohol (reference category—not over recommended limits) Smoking (reference category—never and ex-smoker)
Women N=1171
Men N=3276
Over the recommended limits
0.92 (0.64, 1.30)
1.04 (0.88, 1.25)
Current smoker
0.80 (0.56, 1.13)
0.93 (0.71, 1.23)
a Odds ratios are adjusted for age at Phase 5 and baseline employment grade. Potential explanatory factors are sociodemographic factors, personal and work characteristics, job satisfaction, material factors and health-related behaviours. b Numbers of participants included in each analysis will vary due to the exclusion of those (generally few) with the potential explanatory factor missing.
Table 3 Effect of adjustment for potential explanatory factors on health outcomes for insecurely employed women and men at Phase 5 Health measure
Na
Adjustedb
Women Self-rated health (SF36)d Longstanding illnessd
941 709
Odds ratios (95% confidence intervals) p value 1.74 (1.16, 2.61) p=0.007 1.20 (0.76, 1.87) p=0.43 1.43 (1.02, 2.01) p=0.04 1.29 (0.88, 1.87) p=0.18
68 29
911 915
Score difference (95% confidence intervals) p value 2.30 (1.45, 3.15) po0.001 0.91 (0.03, 1.79) p=0.04 0.56 (0.29, 0.84) po0.001 0.09 ( 0.18, 0.37) p=0.51
60 83
3239
Odds ratios (95% confidence intervals) p value 2.15 (1.65, 2.80) po0.001 1.63 (1.22, 2.17) po0.001
36
2706 2706
Score difference (95% confidence intervals) p value 1.99 (1.53, 2.45) po0.001 0.81 (0.36, 1.27) po0.001 0.46 (0.32, 0.61) po0.001 0.08 ( 0.06, 0.23) p=0.25
59 82
GHQ (30) scoree GHQ depressiond Men Self-rated health (SF36)d GHQ (30) scoree GHQ depressione
Fully adjustedc
Change %
a Shows the number of subjects in the analysis who have no missing values for any of the explanatory factors that are specified in the adjusted model. b Odds ratios and score differences for health outcomes are adjusted for age at Phase 5, baseline employment grade, and baseline value of the outcome of interest. c Odds ratios and score differences for health outcomes in fully adjusted models are additionally adjusted as follows. d Pessimism, vigilance, primary deprivation, financial security, social support and job satisfaction. e Pessimism, vigilance, primary deprivation, financial security, social support, job satisfaction and job control.
Potential explanations of the relationship: The major potential explanatory factors were selected as described in the methods. The effects of adjusting the association between self-reported job insecurity and health for these factors are presented in Table 3. Only those health measures that showed a statistically significant inverse association with self-reported job insecurity in Table 1, plus longstanding illness in women, are shown. A combination of personal characteristics (pessimism and heightened vigilance), material factors (primary deprivation and financial security), and other psychosocial characteristics of the work environment (social support at work and job satisfaction) explained 68% of
the association between self-reported job insecurity and self-rated health in women and 36% in men. In addition, this combination of factors explained 29% of the association between job insecurity and longstanding illness in women. The same set of factors, with the addition of job control, explained 60% of the association between job insecurity and GHQ score and over 80% of the association with depression in both sexes. Job satisfaction explained the largest proportion of the association between job insecurity and self-rated health in women (31%), followed by pessimism (22%), while among men the two largest contributors were job satisfaction and financial insecurity, which both
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
explained 17%. Pessimism (28%) and social support at work (22%) explained the largest proportions of the association between job insecurity and longstanding illness in women. With respect to depression, pessimism explained the largest proportion of the association with job insecurity in women (41%), followed by job satisfaction (33%), while among men, job satisfaction explained the greatest proportion (34%), closely followed by financial insecurity (31%) and pessimism (30%). The proportion of the job insecurity–health association explained by each of the factors when entered into the final model show that the effects of the explanatory variables were partly independent and partly overlapping (data available on request). A further set of analyses in which the fully adjusted models were additionally adjusted for negative affectivity produced results similar to those presented in Table 3. The additional adjustment, in general, marginally strengthened the fully adjusted job insecurity–health relationships. These changes in effect ranged from an increase of 7% for GHQ depression to no change for self-rated health, both in women.
Discussion Methodological considerations: The major limitation of this study is that it is partially cross-sectional in design. Analysis of data collected on individuals at baseline screening, a phase of secure employment 11 years earlier, together with data collected at Phase 5 allowed us to control for the effects of previous health status on current job insecurity and for other confounding factors such as age and socioeconomic position. This enabled the study to separate morbidity associated with job insecurity from the effects of pre-existing morbidity. However, no such baseline data existed for most of the potential explanatory variables. This means that for associations between job insecurity and measures, such as pessimism, the available data do not allow direction of causation to be established. Even if this was not the case, causal associations between job insecurity, pessimism and depression are difficult to disentangle. Beliefs and emotional responses co-exist and are likely to interact. Removing the source of job insecurity is likely to improve mood and levels of optimism, but underlying anxiety, which may contribute to job insecurity, independent of objective work prospects, cannot be ruled out. Our measure of socioeconomic position, baseline employment grade, is not contemporaneous with our morbidity and risk factor measures at Phase 5. Baseline grade was chosen as it is available for all participants, and produced findings little different from analyses using last known grade. It was felt that using last known grade as the measure of socioeconomic position posed
1599
problems. Last known grade reflects mobility for those who remained in the Civil Service, but cuts short the trajectories of those who left the Civil Service to take up employment elsewhere, 29% of those in employment at Phase 5. Use of Phase 1 employment grade also minimised the effect of reverse causality, where the levels of morbidity at baseline may affect subsequent mobility and hence employment grade at Phase 5. The depression subscale of the GHQ is not an ideal measure of depression, but no other measure was available in the dataset at Phase 5. It is nevertheless a robust subscale, clearly emerging in factor analysis across gender and employment grade with good face validity for depressive content and plausible associations with known risk factors for depression (Stansfeld, Head, Fuhrer, Wardle, & Cattell, 2003). In an attempt to determine the effect of reporting bias on our findings we adjusted our analyses for negative affectivity, but found it had little influence. This is in contrast to Canadian work that found negative affectivity to be the single most important predictor of psychological distress, even after controlling for sociodemographic variables, perceived risk, and coping resources and strategies (Roskies et al., 1993). Although the trait of negative affectivity is relatively stable over time (Watson & Clark, 1984; Roskies et al., 1993), reliance on data measured 11 years earlier is not ideal. Furthermore, it has been suggested that measures of negative affectivity may be rather limited in their ability to address the overestimation of associations (Macleod, Smith, Heslop, Oliver, & Hart, 1999). Generalisability of findings: The participants, often a relatively homogeneous group working in one particular field or organisation, limit the generalisability of findings from most occupational cohort studies. Similarly, in this study, all respondents were in office-based, white-collar employment at baseline screening. However, in addition to specialists, the Civil Service employs large numbers of administrators and general office staff such as personal assistants, secretaries, personnel and welfare officers, clerks and messengers. These workers make it equivalent to many office-based settings both in the public and private sector. Also, by Phase 5, 29% of participants still in employment were working outside the Civil Service, a factor likely to increase generalisability to the wider working population. However, at a time of marked change in the labour market, it is important that findings generated by workers aged 44 and over are not assumed to apply to those at the beginning of their working life. Self-reported job insecurity and health: Relative to participants in secure employment, poor self-rated health and measures of minor psychiatric morbidity were significantly higher among workers of both sexes reporting job insecurity. Previous analyses using data from the Whitehall II cohort have demonstrated sex differences in associations between attributed job
ARTICLE IN PRESS 1600
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
insecurity and self-reported health outcomes (Ferrie, Shipley, Marmot, Stansfeld, & Smith, 1995, 1998). However, other analyses, which examined exposure to chronic self-reported job insecurity or change in job security found similar effects in both sexes (Ferrie et al., 2002). Analyses comparing the health effects of contractual job security with self-reported job security in a Finnish cohort similarly found that the different types of job insecurity had different associations with health, but found associations with health were similar in both sexes regardless of type of security (Virtanen, Vahtera, Kivimaki, Pentti, & Ferrie, 2002). Potential explanations of the association: Confirming the intimations of our qualitative findings (Marmot et al., 2001), we found a combination of personal characteristics, material factors, other psychosocial work characteristics and job satisfaction explained much of the association between job insecurity and ill health. Pessimism, vigilance, primary deprivation, future financial security, social support at work and job satisfaction explained 68% of the association between self-reported job insecurity and self-rated health in women and 36% in men. The same factors also explained 29% of the association between selfreported job insecurity and longstanding illness in women and, with the addition of job control, 60% of the association between job insecurity and GHQ score and over 80% of the association between job insecurity and depression in both sexes. Associations between personal characteristics and health outcomes are relatively well documented (DeNeve, 1999; Everson et al., 1996; Spiro III, Aldwin, Ward, & Mroczek, 1995). However, relatively little work has examined associations between personal characteristics and psychosocial characteristics of the work environment (Bosma, Stansfeld, & Marmot, 1998), including job insecurity, although one study found it to be significantly associated with neuroticism (Tivendell & Bourbonnais, 2000). Similarly, only few studies appear to have examined personal characteristics as potential explanations of associations between psychosocial work characteristics and health. Very little of the association between low job control and heart disease was explained by personal characteristics in the Whitehall II study (Bosma et al., 1998). The National Population Health Survey in Canada found that mastery and self-esteem explained half the cross-sectional association between job insecurity and self-rated health and 63% of the association between job insecurity and distress, but only a little of the association with use of medications (McDonough, 2000). Data from a longitudinal cohort study in Finland showed that hostile individuals were able to derive less benefit from psychosocial resources, such as support, network size and job control, during downsizing and were thus more vulnerable to disease (Vahtera, Kivimaki, Uutela, & Pentti, 2000).
One of the main explanations of associations between unemployment and health is loss of income, and anticipated financial problems have been found to be mediators of the relationship between unemployment and health problems (Leeflang, Klein-Hesselink, & SpruitI, 1992). Although job insecurity is likely to generate concern about future financial security, no previous research appears to have examined this as an explanation of the effect of job insecurity on health. However, data from a longitudinal cohort study in Finland showed that the effect of downsizing on sickness absence was greater among low income than high-income employees (Kivimaki, Vahtera, Pentti, & Ferrie, 2000a). In previous work we have examined the potential of other psychosocial work characteristics to explain associations between job insecurity attributed to workplace closure and major organisational change and measures of morbidity and cardiovascular risk factors. Loss of social support at work was a feature common to both situations and explained 11% of the association between attributed job insecurity and self-rated health and minor psychiatric morbidity (Ferrie et al., 2001a). Other evidence is beginning to accumulate that social support at work modifies associations between selfreported job insecurity and self-rated health (McDonough, 2000), and job insecurity and psychological symptoms (Dooley, Rook, & Catalano, 1987; Mohr, 2000; Kinnunen & Natti, 1994). However, neither support from colleagues, management nor unions modified the negative impact of job insecurity on psychological health among public transport workers in Australia (Dekker & Schaufeli, 1995). Little work has examined the role of job demands and control at work in the association between job insecurity and morbidity. Changes in job demands and control made a negligible contribution to the association between attributed job insecurity and self-reported measures of morbidity in previous work using Whitehall II data (Ferrie et al., 2001a, 2001b). Similarly, a study of South African gold miners found no evidence that job control moderated the adverse effect of job insecurity on psychological strain (Barling & Kelloway, 1996). However, a recent study in Finland found that decreased participation in decision making explained 19% of the association between sickness absence and downsizing (Kivimaki et al., 2000a). Further work on the same cohort has shown that job control also partially mediated the relationship between downsizing and selfrated health (Kivimaki et al., 2000b). Although the potential explanatory factors measured in this study explained some of the associations between job insecurity and self-reported health, between 17% and 71% of the association remained unexplained. While measurement imprecision might have underestimated the effect of the adjustments for the potential explanatory factors
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
in Table 3, probably the largest part of the residual associations are accounted for by pre-existing morbidity not adequately captured by self-rated health, longstanding illness or GHQ score, for example anxiety. We can only speculate why adjustment for the potential explanatory factors explained a similar proportion of the association between job insecurity and health for both women and men, for the GHQ measures but not for self-rated health. Possibly hostility, more frequently seen in men might explain part of the difference, alternatively, or in addition, reporting of self-rated health in men might focus more narrowly on somatic symptoms, whereas in women it may encompass more general sense of well-being. Well-being is more likely to be explained by a combination of personal characteristics and environmental factors. In conclusion, among participants in the Whitehall II study, there is strong evidence that self-reported job insecurity is associated with poor self-rated health, minor psychiatric morbidity and depression. A combination of pessimism, heightened vigilance, difficulty in paying bills and problems with housing, financial security, social support and job satisfaction explained much of the association between job insecurity and selfrated health. With the addition of job control, the same factors explained most of the association with minor psychiatric morbidity and depression. While direction of causation cannot be determined in this study, optimism, future financial security, social support at work, job control and job satisfaction are undoubted casualties when jobs are under threat. Organisational studies have shown these factors to be related to commitment, effort, morale and intention to quit, which have a direct bearing on the success of organisations (Burke & Cooper, 2000). At a time of increasing emphasis on the benefits of flexible labour markets to national economies, and increasing concern about the rise in absence from work, policies whose consequence is to increase job insecurity should take such findings into account. To this consideration must be added the detriment to individuals and organisations, which, in addition to personal misery, have economic implications of their own.
Acknowledgements The work presented in this article was supported by the United Kingdom Economic and Social Research Council (L128251046). The Whitehall II study has been supported by grants from the Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart Lung and Blood Institute (HL36310), US, NIH: National Institute on Aging (AG13196), US, NIH; Agency for Health Care Policy Research (HS06516); and the John
1601
D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. Jane Ferrie was supported by the Economic and Social Research Council (L128251046) during the preparation of this paper. Michael Marmot is supported by an MRC Research Professorship. Martin Shipley is supported by a grant from the British Heart Foundation. We also thank all participating Civil Service departments and their welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; the Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.
References Amick, B. C., III., Kawachi, I., Coakley, E. H., Lerner, D., Levine, S., & Colditz, G. A. (1998). Relationship of job strain and iso-strain to health status in a cohort of women in the United States. Scandinavian Journal of Work and Environmental Health, 24, 54–61. Barling, J., & Kelloway, E. K. (1996). Job insecurity and health: the moderating role of workplace control. Stress Medicine, 12, 253–259. Bosma, H., Stansfeld, S. A., & Marmot, M. G. (1998). Job control, personal characteristics, and heart disease. Journal of Occupational Health Psychology, 3, 402–409. Bradburn, N. M. (1969). The structure of psychological wellbeing. Chicago: Aldine. Burke, R., & Cooper, C. L. (2000). The organization in crisis: downsizing, restructuring and privatisation. Oxford: Blackwell. De Witte, H. (1999). Job insecurity and psychological wellbeing: review of the literature and exploration of some unresolved issues. European Journal of Work and Organisational Psychology, 8, 155–177. Dekker, S. W. A., & Schaufeli, W. B. (1995). The effects of job insecurity on psychological health and withdrawal: a longitudinal study. Australian Psychologist, 30, 57–63. DeNeve, K. M. (1999). Happy as an extraverted clam? The role of personality for subjective well-being. Current Directions in Psychological Research, 8, 141–144. Dooley, D., Rook, K., & Catalano, R. (1987). Job and non-job stressors and their moderators. Journal of Occupational Psychology, 60, 115–132. Everson, S. A., Goldberg, D. E., Kaplan, G. A., Cohen, R. D., Pukkala, E., Tuomilehto, J., & Salonen, J. T. (1996). Hopelessness and risk of mortality and incidence of myocardial infarction and cancer. Psychosomatic Medicine, 58, 113–121. Ferrie, J. E., Shipley, M. J., Marmot, M. G., Stansfeld, S., & Davey Smith, G. (1995). Health effects of anticipation of job change and non-employment: longitudinal data from the Whitehall II study. British Medical Journal, 311, 1264–1269.
ARTICLE IN PRESS 1602
J.E. Ferrie et al. / Social Science & Medicine 60 (2005) 1593–1602
Ferrie, J. E., Shipley, M. J., Marmot, M. G., Stansfeld, S., & Davey Smith, G. (1998). The health effects of major organisational change and job insecurity. Social Science & Medicine, 46, 243–254. Ferrie, J. E. (2001). Is job insecurity harmful to health? Journal Royal Society Medicine, 94, 71–76. Ferrie, J. E., Shipley, M. J., Marmot, M. G., Martikainen, P., Stansfeld, S. A., & Davey Smith, G. (2001a). Job insecurity in white-collar workers: towards an explanation of associations with health. Journal of Occupational Health Psychology, 6, 26–42. Ferrie, J. E., Martikainen, P., Shipley, M. J., Marmot, M., Stansfeld, S. A., & Davey Smith, G. (2001b). Employment status and health after privatisation in white collar civil servants: prospective cohort study. British Medical Journal, 322, 647–651. Ferrie, J. E., Shipley, M. J., Stansfeld, S. A., & Marmot, M. G. (2002). Effects of chronic job insecurity and change in job security on self-reported health, minor psychiatric morbidity, physiological measures and health-related behaviours in British civil servants: the Whitehall II study. Journal of Epidemiology and Community Health, 56, 450–454. Goldberg, D. P. (1972). The detection of psychiatric illness by questionnaire. London. Karasek, R. (1979). Job demands, job decision latitude, and mental strain: implications for job redesign. Administrative Science Quarterly, 24, 285–311. Kinnunen, U., & Natti, J. (1994). Job insecurity in Finland: antecedents and consequences. European Work and Organisational Psychologist, 4, 297–321. Kivimaki, M., Vahtera, J., Pentti, J., & Ferrie, J. E. (2000a). Factors underlying the effect of organisational downsizing on the health of employees: a longitudinal cohort study of changes in work, social relationships and health behaviours. British Medical Journal, 320, 971–975. Kivimaki, M., Vahtera, J., Pentti, J., Thomson, L., Griffiths, A., & Cox, T. (2000b). Downsizing, changes in work, and self-rated health of employees: a 7-year 3-wave panel study. Anxiety Stress and Coping, 14, 59–73. Leeflang, R. L. I., Klein-Hesselink, D. J., & SpruitI, P. (1992). Health effects of unemployment—I. Long-term unemployed men in a rural and an urban setting. Social Science & Medicine, 34, 341–350. Macleod, J., Davey Smith, G., Heslop, P., Oliver, S., & Hart, C. (1999). Always look on the bright side of life? The influence of reporting tendency when exposure and outcome measurements are based on self-report. Journal of Epidemiology and Community Health, 53, 660. Marmot, M., Ferrie, J. E., Newman, K., & Stansfeld, S. (2001). The contribution of job insecurity to socio-economic inequalities. Research FINDINGS 11 from the Health Variations Programme, Economic and Social Research Council. Marmot, M. G., Davey Smith, G., Stansfeld, S., Patel, C., North, F., Head, J., White, I., Brunner, E. J., & Feeney, A.
(1991). Health inequalities among British civil servants: the Whitehall II study. Lancet, 337, 1387–1393. Martikainen, P., Adda, J., Ferrie, J. E., Davey Smith, G., & Marmot, M. (2003). The effects of income and wealth on GHQ depression and poor self-rated health in white-collar women and men in the Whitehall II study. Journal of Epidemiology and Public Health, 57, 718–723. McDonough, P. (2000). Job insecurity and health. International Journal of Health Services, 30, 453–476. Mohr, G. (2000). The changing significance of different stressors after the announcement of bankruptcy: a longitudinal investigation with special emphasis on job insecurity. Journal of Organisational Behaviour, 21, 337–359. OPCS Social Survey Division (1984). General household survey 1982. London: OPCS Social Survey Division. Platt, S., Pavis, S., & Akram, G. (1998). Changing labour market conditions and health: a systematic literature review (1993–1998). Dublin. Roskies, E., Louis-Guerin, C., & Fournier, C. (1993). Coping with job insecurity: how does personality make a difference? Journal of Organisational Behaviour, 14, 617–630. Spiro, A., III., Aldwin, C. M., Ward, K. D., & Mroczek, D. K. (1995). Personality and the incidence of hypertension among older men: longitudinal data from the normative aging study. Health Psychology, 14, 563–569. Stansfeld, S., Head, J., & Marmot, M. (1998). Explaining social class differences in depression and well-being. Social Psychiatry and Psychiatric Epidemiology, 33, 1–9. Stansfeld, S. A., Head, J., Fuhrer, R., Wardle, J., & Cattell, V. (2003). Social inequalities in depressive symptoms and physical functioning in the Whitehall II study: exploring a common cause explanation. Journal of Epidemiology and Public Health, 57, 361–367. Tivendell, J., & Bourbonnais, C. (2000). Job insecurity in a sample of Canadian civil servants as a function of personality and perceived job characteristics. Psychological Reports, 87, 55–60. Vahtera, J., Kivimaki, M., Uutela, A., & Pentti, J. (2000). Hostility and ill health: role of psychosocial resources in two contexts of working life. Journal of Psychosomatic Research, 48, 89–98. Virtanen, P., Vahtera, J., Kivimaki, M., Pentti, J., & Ferrie, J. E. (2002). Employment security and health. Journal of Epidemiology and Public Health, 56, 569–574. Watson, D., & Clark, L. A. (1984). Negative affectivity: the disposition to experience aversive emotional states. Psychological Bulletin, 96, 465–490. Whelan, C. (1992). The role of income, life-style deprivation and financial strain in mediating the impact of unemployment on psychological distress: evidence from the Republic of Ireland. Journal Occupational Organisational Psychology, 65, 331–344.