- Email: [email protected]
Future uncertainty and socioeconomic inequalities in health: the Whitehall II study
ARTICLE IN PRESS
Social Science & Medicine 57 (2003) 637–646
Future uncertainty and socioeconomic inequalities in health: the Whitehall II study Jane E. Ferriea,*, Martin J. Shipleya, Stephen A. Stansfeldb, George Davey Smithc, Michael Marmota a
International Centre for Health and Society, Department of Epidemiology and Public Health, University College London Medical School, London, UK b Department of Psychiatry, Queen Mary University of London, UK c Department of Social Medicine, University of Bristol, UK
Abstract Over the past 20 years, socioeconomic inequalities in mortality have widened, while job security and financial security have decreased. This paper examines the Whitehall II study, a longitudinal study of white-collar British civil servants. In the Whitehall II cohort socioeconomic gradients in morbidity and cardiovascular risk factors at Phase 5 (1997–99) were generally steeper than at Phase 1 (1985–88). We examine the contribution of job and financial insecurity to these at Phase 5 in 6770 women and men, all of whom were white-collar civil servants at Phase 1. Steep, inverse employment grade gradients were observed for all health measures at Phase 5, except cholesterol and systolic blood pressure in women. Gradients in the sub-population of non-employed participants tended to be steeper than gradients for participants in employment, although, with the exception of self-rated health and General Health Questionnaire (GHQ) score in men, differences were non-significant. Steep gradients in job insecurity were observed among employed participants (pp0.01), and in financial insecurity among both employed and non-employed participants (pp0.001), particularly non-employed men. With the exception of depression, adjustment for job insecurity had little effect on the employment grade gradients in morbidity. However, financial insecurity contributed substantially to gradients in selfrated health, longstanding illness, and depression in both employed and non-employed men, and additionally to GHQ score and diastolic blood pressure in the latter. Adjustment for financial insecurity in non-employed women substantially attenuated gradients in self-rated health, GHQ score and depression. These findings imply that the specific effects of job insecurity in this cohort may be less important than the more general effects of financial insecurity in determining inequalities in health. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Job insecurity; Financial insecurity; Health inequalities; Employment status; Cross-sectional; White-collar; UK
Introduction Over the past 20 years, against a backdrop of overall declines in mortality rates, analyses of population data *Corresponding author. Department of Epidemiology and Public Health, University College London Medical School, 1-19 Torrington Place, London WC1E 6BT, UK. Tel.: +44-171504-5643; fax: +44-171-813-0242. E-mail address: [email protected] (J.E. Ferrie).
have shown that the social gradient in mortality in the United Kingdom (UK) has increased (Acheson, 1998, Shaw, Dorling, Gordon, & Davey Smith, 1999). However, although socioeconomic gradients in various measures of morbidity are well documented (Acheson, 1998, Marmot, et al., 1991, Erens & Primatesta, 1999), there has been relatively little research examining trends in these measures over time. Most of the evidence to date has concentrated mainly on traditional risk factors for cardiovascular disease and indicates that socioeconomic differences are static or narrowing (Bartley,
0277-9536/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 4 0 6 - 9
ARTICLE IN PRESS 638
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
Fitzpatrick, Firth, & Marmot, 2000, Iribarren, Luepker, McGovern, Arnett, & Blackburn, 1997, Vartiainen et al., 1998, Bennet, 1995, Osler et al., 2000). Steep socioeconomic gradients in measures of morbidity and cardiovascular risk factors have long been demonstrated in the Whitehall II study, a longitudinal cohort of white-collar British civil servants, in which socioeconomic position is assessed by grade of employment (Marmot et al., 1991). Over 11 years follow-up of the Whitehall II cohort up to 1998, inequalities in health between those in the lowest employment category and those in the highest have widened slightly and marked increases in the gradient have been documented in minor psychiatric morbidity in both sexes and cholesterol in men (Ferrie, Shipley, Davey Smith, Stansfeld, & Marmot, 2002). Additionally during the 1980s and 1990s in the UK, patterns of employment, job security and welfare provision associated with the social order since the second world war have undergone and continue to undergo major change. The future, for many people, is less certain than it used to be (Hutton, 1994, 1995, Sennett, 1998). Two markers of future uncertainty have previously received some attention in relation to health; job insecurity and financial insecurity. Self-perceived job insecurity has been shown to be associated with increased mental and physical ill-health (Platt, Pavis, & Akram, 1998, De Witte, 1999, Ferrie, 2001) and there is evidence from Finland that it is inversely associated with socioeconomic position (Kinnunen & Natti 1994, Lynch, Kaplan, & Salonen, 1997). In addition, job insecurity attributed to workplace closure and downsizing has been shown to be associated with increased psychological morbidity, self-reported ill-health, long spells of sickness absence and health service use (Ferrie, 2001). A considerable body of work has documented the deleterious effect of unemployment-related financial strain on health, particularly psychological health (Warr, Banks, & Ullah, 1985, Kessler, Blake Turner, & House, 1987, Ensminger & Celentano 1988, Whelan, 1992). However, little work has examined the wider role of financial insecurity. Research to date shows poor financial security to be associated with minor psychiatric morbidity (Romans, Walton, McNoe, Herbison, & Mullen, 1993, Jackson, Iezzi, & Lafreniere, 1997) and weight gain (Gerace & George, 1996). Data on financial insecurity and on job insecurity for the whole cohort were first collected during Phase 5 (1997–99) of the Whitehall II study. The aims of this paper are to determine socioeconomic gradients in job and financial insecurity in the Whitehall II cohort, and the contribution of these to socioeconomic inequalities in morbidity and cardiovascular risk factors at Phase 5.
Methods Participants 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). Sub–samples of nonresponders indicated that approximately 4% of those invited were not eligible for inclusion, suggesting that the true response rate was higher. Although mostly white–collar, respondents covered a wide range of grades from office support to permanent secretary. Data collection Baseline screening (Phase 1) took place between late 1985 and early 1988. This involved a clinical examination and a self-administered questionnaire. Since the baseline screening there have been five data collection phases, which have alternated between a self-administered questionnaire only and a clinical examination and questionnaire. As at previous odd-numbered phases, the Phase 5 (1997–99) clinical examination included, among other measures, height, weight, blood pressure, and serum cholesterol and the questionnaire contained sections on demographic characteristics, health, lifestyle factors, work characteristics, social support, life events and chronic difficulties. Measures Standard items drawn from the Phase 5 questionnaire include personal details: age and employment status; selfreported health: self–rated health and presence of longstanding illness. Self-rated health was measured using a single item from the SF36 ‘In general would you say your health is; excellent, very good, good, fair, poor.’ (Ware, Kosinski, & Keller, 1994). For analysis, health self-rated as fair or poor was compared with good, very good and excellent. Mental health was assessed using the 30–item General Health Questionnaire (Goldberg, 1972) and comprised GHQ score and depression, measured using a 4-item depression subscale of the GHQ, derived by principal components analysis. The depression items were as follows: ‘Have you recently - been thinking of yourself as a worthless person, - felt that life is entirely hopeless, felt that life isn’t worth living, - found at times you couldn’t do anything because your nerves were too bad? As for all GHQ questions, each item was scored on a likert scale from 0 to 3 and summed for analysis. In principal components analysis these four items loaded on a single component in both women and men. (Stansfeld, Head, & Marmot, 1998)
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
New items drawn from the Phase 5 questionnaire were perceived job insecurity and financial insecurity. A single item ‘How secure do you feel in your present job?’ offered 4 response categories ranging from very insecure to very secure. Responses to the question ‘Thinking of the next ten years, how financially secure do you feel?’ were collected using a 4 category scale ranging from secure to insecure. Grade of employment and negative affectivity were determined from the Phase 1 questionnaire. Negative affectivity was assessed using the five negative affect items from Bradburn’s Affect Balance Scale (Bradburn, 1969). The Phase 1 measure was used as negative affectivity has not been measured subsequently. Grade of employment was determined by asking all participants to give their Civil Service grade title. On the basis of salary the Civil Service identified twelve nonindustrial grades which, in order of decreasing salary, comprised seven ‘‘unified grades’’, senior executive officer (SEO), higher executive officer (HEO), executive officer (EO), clerical officer and clerical assistant. Other professional and technical staff were assigned by the Civil Service to one of these grades on the basis of salary. For analysis, unified grades 1–6 were combined into one group and the bottom two clerical grades into another, thus producing six categories; category 1 represents the highest status jobs and category 6 the lowest. At the Phase 5 screening examination, blood pressure in millimetres of mercury (mmHg) was measured twice with the participant seated after a 5 min rest, using a Hawksley random–zero sphygmomanometer. Blood was taken and serum cholesterol concentration in millimoles/ litre (mmol/l) measured using the cholesterol oxidase/ peroxidase colorimetric method (BCL kit). Weight in kilograms (kg) and height in metres (m) were recorded. Body Mass Index (BMI) was calculated from these two measures as kg/m2. Further details of these standard repeat measures have been reported previously (Rose et al., Jarrett, 1977, Marmot et al., 1991). Study sample and statistical analysis Seventy-one per cent (7270) of the 10,308 Whitehall II study participants, at baseline, completed the full
639
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%. In addition to those who failed to respond to invitations for the Phase 5 data collection, non–responders included participants who had died and those who could not be traced. Of respondents to the Phase 5 data collection 4665 were still in paid employment and 2532 were out of the labour force (Box 1). In all, 4447 completed the question on job security, 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 were compared with those whose job was secure or very secure. All respondents to Phase 5 were asked about financial security, 6658 completed this question, an item non-response rate of 8%. Responses to the question on financial security were dichotomised. Those who were secure or fairly secure being compared with those who were insecure or fairly insecure. Initial analyses determined employment grade gradients, based on Phase 1 grade, for all morbidity measures at Phase 5 using least-squares regression for continuous outcomes and logistic regression for dichotomous outcomes. These analyses produce a regression line showing the increase in morbidity for a unit increase in the employment grade. We have used this employment grade gradient to estimate the odds ratio (OR) or difference (Diff) in morbidity (and 95% confidence interval) between respondents in the lowest grade (Grade 6) and respondents in the highest grade (Grade 1), and presented these in the Tables. ORs above 1.0 and Diffs above 0.0 indicate an inverse grade gradient, that is, greater morbidity among the lower grades, while ORs below 1.0 and Diffs below 0.0 indicate a positive grade gradient, that is, greater morbidity among the higher grades. Different grade gradients for women and men had already been demonstrated at baseline in the Whitehall II cohort (Marmot et al., 1991) so analyses were conducted separately by sex. All analyses were adjusted for age categorised into 5-year age groups. Data were analysed using SAS version 6.12 for Windows (SAS Institute Inc., 1990). Phase 5 participants were then divided into the employed and the non-employed. Age-adjusted employ-
Box 1 Employment status at Phase 5 Employment status
Women
Men
In the Civil Service Working Elsewhere Non-employed Total
960 285 900 2145
2360 1060 1632 5052
(44.8%) (13.3%) (42.0%) (100%)
Total (46.8% ) (21.0%) (32.3%) (100%)
3320 1345 2532 7197
(46.1%) (18.7%) (35.2%) (100%)
ARTICLE IN PRESS 640
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
ment grade gradients for these two sub-populations were determined in the same way as for the whole Phase 5 population. P-values for the difference in the gradients between the sub-populations were determined by calculating the difference between the regression coefficients for the gradient for each population and testing this against a null hypothesis of no difference. In addition, age-adjusted gradients were determined for job insecurity (employed only), and financial insecurity. Finally, the contributions of job insecurity and financial insecurity at Phase 5 to employment grade gradients in morbidity and cardiovascular risk factors were assessed by calculating the percentage reduction in the employment grade coefficients when adjusted for the measures of insecurity. Health measures included in these analyses were restricted to those for which there was evidence of a gradient (po0.05) in the sub-population under investigation. Item non-response to the insecurity questions weakened the grade gradient for some of these measures, but they were retained in the analysis. The contributions of the insecurity measures were determined by adding these terms into the models for each morbidity or risk factor measure. As some of the health and both security measures were self-reported there was the possibility that negative affect or reporting bias might have explained, to some degree, any association between insecurity and health. To take account of this the gradients for self-rated health, longstanding illness, GHQ score, depression, job insecurity and financial insecurity were determined with and without adjustment for negative affectivity. Ideally, all the analyses should have been adjusted for negative affect. However, the Affect Balance Scale (Bradburn, 1969) 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.
Results Data on current employment status was provided by 7197 of those who completed the Phase 5 questionnaire—(Box 1). Overall, just under half of the respondents were still working in the Civil Service. The majority who had left were non-employed, but a sizable minority were employed elsewhere. Employment grade gradients in morbidity and cardiovascular risk factors: In the whole Phase 5 population steep inverse employment grade gradients were seen for all measures in both sexes (p=0.04 to po0.001), with the exception of serum cholesterol and systolic blood pressure in women (Table 1) When divided into two subpopulations, employed and non-employed participants, grade gradients tended to be steeper among nonemployed women and men compared to employed
participants, with the exception of self-rated health and diastolic pressure in women. For two measures the difference in the gradients was considerable, self-rated health (p=0.03) and GHQ score (p=0.009) in men. Employment grade gradients in job insecurity and financial insecurity: Steep inverse employment grade gradients were seen for job insecurity amongst employed women (p=0.006) and men (p=0.01) and for financial insecurity in all participants whether employed or nonemployed (pp0.001) (Table 2). Among non-employed men the gradient in financial insecurity was much steeper OR 41.6 (18.2, 95.1) than among employed men OR 6.5 (4.4, 9.6), test of difference po0.001. Morbidity and risk factor gradients adjusted for job insecurity: In this population, job insecurity at Phase 5 contributed little to the grade gradients in morbidity and cardiovascular risk factors in either sex. The exception was depression in which the gradient was attenuated by 19% in women and 9% in men—(Table 3). For selfrated health in both sexes, BMI in women and longstanding illness in men job insecurity attenuated associations by roughly 5%. Morbidity and risk factor gradients adjusted for financial insecurity: Adjustment for financial insecurity at Phase 5 among those in employment resulted in attenuation of the gradients in self-rated health and depression by 13% and 14%, respectively, in women, and 33% and 49%, respectively, in men (Table 4). Additionally among men, the gradient in longstanding illness was attenuated by 25%. The gradient in BMI among employed women was only marginally affected by adjustment for financial insecurity, 6%, and adjustment of the gradient in diastolic blood pressure increased the slope by 11%. Among non-employed participants adjustment for financial insecurity attenuated associations with self-rated health, GHQ score and depression by 24%, 75% and 75% among women and by 37%, 42% and 37% among men (Table 5). In addition, associations with longstanding illness and diastolic blood pressure were attenuated, respectively, by 35% and 14% among men. Adjustment of the gradients in self-reported morbidity and insecurity for negative affectivity tended to increase the slope very slightly for all measures among women (table available on request). In men, adjustment attenuated the gradient for all the self-reported morbidity measures. The degree of attenuation ranged from 4% for self-rated health to 18% for GHQ score, but had a negligible effect on the p-values for the gradients in all cases. Adjustment for negative affect produced negligible attenuation of the gradient in job insecurity (2%) or financial insecurity (0%) in either sex. These findings make it unlikely that reporting bias can explain the steep grade gradients in the self-reported morbidity measures and either insecurity measure.
OR OR Diff Diff Diff Diff Diff Diff
Men Self-rated health (SF36) [n=4967] Longstanding illness [n=5046] GHQ score [n=4920] Depression [n=4918] Cholesterol mmol/L [n=4543] Systolic blood pressure mmHg [n=4573] Diastolic blood pressure mmHg [n=4573] Body mass index kg/m2 [n=3943] 3.02 1.33 0.95 0.61 0.59 1.99 1.61 0.39
3.74 1.30 1.26 0.54 0.09 0.99 1.57 2.16
(2.2, 4.1) po0.001 (1.1, 1.6) p=0.004 (0.5, 6.0) po0.001 (0.5, 0.8) po0.001 ( 1.1, -0.1) p=0.03 (0.4, 3.6) p=0.02 (0.6, 2.7) p=0.003 (0.1, 0.8) p=0.04
(2.4, 5.8) po0.001 (1.0, 1.7) p=0.07 (0.4, 2.2) p=0.006 (0.3, 0.9) po0.001 (-0.9, 0.7) p=0.82 (-1.6, 3.6) p=0.45 (0.1, 3.1) p=0.04 (1.4, 3.0) po0.001
All participants OR or Diff (95% CI)
2.38 1.33 0.51 0.53 0.48 1.80 1.23 0.39
4.10 1.08 0.95 0.49 0.37 0.44 1.93 2.06
(1.6, 3.4) po0.001 (1.0, 1.7) p=0.02 (-0.2, 1.2) p=0.12 (0.3, 0.8) po0.001 ( 1.2, 0.2) p=0.14 ( 0.1, 3.7) p=0.06 ( 0.1, 2.5) p=0.06 ( 0.1, 0.9) p=0.11
(2.3, 7.3) po0.001 (0.8, 1.5) p=0.67 ( 0.2, 2.1) p=0.10 (0.1, 0.9) po0.001 ( 1.3, 0.6) p=0.45 ( 2.6, 3.5) p=0.78 (0.1, 3.8) p=0.03 (1.1, 3.1) po0.001
Employed OR or Diff (95% CI)
p-value for test of difference between OR (or Diff) for the employed versus the non-employed.
OR OR Diff Diff Diff Diff Diff Diff
Women Self-rated health (SF36) [n=2098] Longstanding illness [n=2135] GHQ score [n=2053] Depression [n=2060] Cholesterol mmol/l [n=1834] Systolic blood pressure mgHg [n=1865] Diastolic blood pressure mgHg [n=1865] Body mass index kg/m2 [n=1656]
a
OR/Diff
Health measure
Table 1 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade
4.67 1.46 1.87 0.73 0.90 2.36 2.52 0.46
4.01 1.56 1.89 0.65 0.57 2.51 0.92 2.40
(2.8, 7.7) po0.001 (1.0, 2.0) p=0.03 (1.1, 2.7) po0.001 (0.5, 1.0) po0.001 ( 1.9, 0.1) p=0.06 ( 0.7, 5.4) p=0.13 (0.6, 4.4) p=0.01 ( 0.2, 1.1) p=0.17
(2.1, 8.0) po0.001 (0.9, 2.6) p=0.08 (0.4, 3.4) p=0.01 (0.2, 1.2) p=0.01 ( 0.9, 2.1) p=0.44 ( 2.2, 7.2) p=0.29 ( 1.8, 3.6) p=0.50 (1.0, 3.8) po0.001
Non-employed OR or Diff (95% CI)
p=0.03 p=0.63 p=0.009 p=0.26 p=0.47 p=0.76 p=0.27 p=0.86
p=0.98 p=0.35 p=0.57 p=0.71 p=0.29 p=0.46 p=0.54 p=0.70
p-valuea
ARTICLE IN PRESS
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646 641
ARTICLE IN PRESS Not applicable Po0.001
Discussion
p-value for test of difference between OR (or Diff) for the employed versus the non-employed.
Methodological considerations
a
Not applicable 41.62 (18.2, 95.1) po0.001 [1555] 1.43 (1.1, 1.9) p=0.01 [3272] 6.52 (4.4, 9.6) po0.001 [3204] Not applicable 9.75 (6.9, 13.8) po0.001 [4759]
Not applicable 6.32 (2.4, 16.9) po0.001 [786]
Women Job insecurity Financial insecurity Men Job insecurity Financial insecurity
Not applicable 5.34 (3.3, 8.6) po0.001 [1899]
1.78 (1.2, 2.7) p=0.006 [1171] 5.21 (3.0, 8.9) po0.001 [1113]
p-valuea Non-employed OR (95% CI) p value [n] Employed OR (95% CI) p value [n] All participants OR (95% CI) p value [n] Security measure
Table 2 Age-adjusted ORs for measures of insecurity at Phase 5 between the lowest versus the highest employment grade
Not applicable P=0.73
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
642
As employment grade at Phase 1 was used to determine the gradients, our measure of socioeconomic position 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 analyses using last known grade produced findings little different from those using grade at Phase 1 (data not shown). We felt that use of last known grade as the measure of socioeconomic position posed problems. It 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 participants in employment at Phase 5. Use of Phase 1 grade also minimised the effect of reverse causality where the levels of morbidity at baseline may affect subsequent mobility and hence grade at Phase 5. The generalisability of the findings presented in this paper is limited by the fact that all participants in the Whitehall II study were white-collar civil servants on entry to the study. However, as can be seen from Box 1, those who remained in the Civil Service were a minority at the time of the Phase 5 data collection. With early retirement reaching a peak in the late 1990s the Whitehall II cohort at Phase 5 is probably fairly representative of workforces aged 45-65, originally in long-term, secure, white-collar employment in other large organisations, such as the financial services or the health service.
Findings Steep employment grade gradients were observed for all measures at Phase 5, except cholesterol and systolic pressure in women. Gradients in the sub-population of non-employed women and men tended to be steeper than gradients for those in paid employment, although differences were not significant, with the exception of self-rated health (p=0.03) and GHQ score (p=0.009) in men. Socioeconomic inequalities in health are a persistent feature of both Whitehall cohorts (Marmot Shipley, & Rose, 1984, Marmot et al., 1991), as in the general population (Acheson, 1998). However, in contrast to the first Whitehall study, in which mortality differentials based on employment grade tended to decline postretirement (Marmot, & Shipley, 1996, van Rossum, Shipley, van de Mheen, Grobbee, & Marmot, 2000), employment grade gradients in morbidity and cardiovascular risk factors tend to be steeper among nonemployed participants. This would seem to indicate that employment grade is a marker for factors whose influence extend beyond the workplace.
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
643
Table 3 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade in employed participants, unadjusted and adjusted for job insecurity - women and men Health measure [n] Women Self-rated health (SF36) [1157] Depression [1133] Diastolic blood pressure [1132] Body mass index [996] Men Self-rated health (SF36) [3239] Longstanding illness [3269] Depression [3213] a
Attenuation %a
OR or Diff
Unadjusted OR or Diff (95% CI) p value
Adjusted for job insecurity OR or Diff (95% CI) p value
OR
4.35 (2.4, 8.0) po0.001
4.08 (2.2, 7.5) po0.001
4
Diff Diff
0.43 (0.1, 0.8) p=0.02 1.77 (-0.2, 3.7) p=0.07
0.35 (0.0, 0.7) p=0.65 1.79 (-0.2, 3.7) p=0.07
19 1
Diff
2.01 (1.0, 3.0) po0.001
2.12 (1.1, 3.2) po0.001
5
OR
2.24 (1.5, 3.3) po0.001
2.14 (1.5, 3.2) po0.001
6
OR
1.31 (1.0, 1.7) p=0.03
1.29 (1.0, 1.6) p=0.04
6
Diff
0.51 (0.3, 0.7) po0.001
0.46 (0.3, 2.0) po0.001
9
% attenuation calculated from the unadjusted and adjusted regression coefficients.
Table 4 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade in employed participants, unadjusted and adjusted for financial insecurity—women and men Health measure [n] Women Self-rated health (SF36) [1105] Depression [1089] Diastolic blood pressure [1019] Body mass index [903] Men Self-rated health (SF36) [3177] Longstanding illness [3201] Depression [3158] a
Attenuation %a
OR or Diff
Unadjusted OR or Diff (95% CI) p value
Adjusted for financial insecurity OR or Diff (95% CI) p value
OR
3.92 (2.1, 7.3) po0.001
3.28 (1.8, 6.1) po0.001
13
Diff Diff
0.52 (0.2, 0.9) p=0.006 1.84 (-0.1, 3.8) p=0.06
0.35 ( 0.1, 0.7) p=0.07 2.04 (0.1, 0.4) p=0.04
14 11
Diff
1.95 (0.9, 3.0) po0.001
1.83 (0.8, 2.9) po0.001
6
OR
2.23 (1.5, 3.3) po0.001
1.71 (1.1, 2.6) p=0.009
33
OR
1.22 (1.0, 1.6) p=0.12
1.16 (0.9, 1.5) p=0.24
25
Diff
0.49 (0.3, 0.7) po0.001
0.25 (0.1, 0.5) p=0.03
49
% attenuation calculated from the unadjusted and adjusted regression coefficients.
Steep grade gradients were observed in job insecurity for both sexes in paid employment (pp0.01). Job losses in the Civil Service between 1995 and 1998 were inversely related to grade of employment (Government Statistical Service, 1998), however, with the exception of depression, adjustment for job insecurity had little effect on gradients in morbidity. Little work has examined the association between job insecurity and socioeconomic position, although there is some evidence of an
association in two Finnish cohorts (Kinnunen & Natti, 1994; Lynch, et al., 1997). Every study of perceived job insecurity in the literature has documented consistent adverse effects on all measures of psychological morbidity (Dooley, Rook, & Catalano, 1987, Burchell; 1994, Dekker & Schaufeli; 1995, Kuhnert, Sims, & Lahey, 1989; Domenighetti, D’Avanzo, & Bisig, 1999; De Witte, 1999; Kasl, & Cobb, 1980). In addition, a significant association between perceived job insecurity
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
644
Table 5 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade in non-employed participants, unadjusted and adjusted for financial insecurity—women and men Health measure [n] Women Self-rated health (SF36) [776] GHQ score [767] Depression [773] Body mass index [581] Men Self-rated health (SF36) [1547] Longstanding illness [1554] GHQ score [1529] Depression [1529] Diastolic blood pressure [1368] a
Attenuation %a
OR or Diff
Unadjusted OR or Diff (95% CI) p value
Adjusted for financial insecurity OR or Diff (95% CI) p value
OR
3.18 (1.5, 6.5) p=0.002
2.42 (1.2, 5.0) p=0.02
23.8
Diff Diff Diff
1.54 (-0.0, 3.1) p=0.05 0.48 (-0.1, 1.0) p=0.08 1.92 (0.5, 3.4) p=0.01
0.39 ( 1.1, 1.9) p=0.61 0.12 ( 0.4, 0.6) p=0.65 1.85 (0.4, 3.4) p=0.01
75.0 74.7 3.4
OR
3.77 (2.2, 6.4) po0.001
2.30 (1.3, 4.0) p=0.004
37.3
OR
1.33 (0.9, 1.9) p=0.11
1.20 (0.8, 1.7) p=0.31
34.5
Diff Diff Diff
1.79 (1.0, 2.6) po0.001 0.69 (0.4, 1.0) po0.001 2.97 (1.0, 5.0) p=0.003
1.04 (0.2, 1.9) p=0.02 0.44 (0.1, 0.8) p=0.005 2.57 (0.55, 4.6) p=0.01
42.2% 36.5 13.6
% attenuation calculated from the unadjusted and adjusted regression coefficients.
and minor psychiatric morbidity has been demonstrated previously in this cohort (Ferrie et al., 2001). Steep grade gradients were observed for financial insecurity among employed and non-employed women and men (pp0.001). This result is unsurprising as grade of employment determined income at Phase 1 in the Whitehall II study, and financial security is largely determined by income and wealth. On the 1st of April 1998 there were 16 salary bands for white-collar civil servants, ranging from the band d1–d5,000 to the band d75,001 and over. Almost 93% of white-collar civil servants earned between d5,000 and d25,000 and the median salary of d15,480 was 2.1% higher than in April 1997. Over the same time retail prices rose by 4.0% (Government Statistical Service, 1998). Financial insecurity contributed considerably to gradients in self-rated health, longstanding illness, and depression in both employed and non-employed men, as well as to GHQ score and diastolic blood pressure in non-employed men only. Adjustment for financial insecurity in non-employed women substantially attenuated gradients in self-rated health, GHQ score and depression. The findings relating to GHQ score and depression reflect those from previous studies which have shown financial insecurity to be associated with poorer mental health outcomes (Romans et al., 1993, Jackson et al., 1997). Although financial insecurity attenuated the gradient in diastolic pressure among non-employed men, it generally contributed little to gradients in the physiological measures. Little other work appears to have looked at associations between financial insecurity and physiological measures.
These findings provide evidence in this cohort that, particularly for men, financial insecurity, rather than job insecurity, seems to be the more central anxiety influencing risk of self-reported morbidity in whitecollar workers aged 45 and over. In a cohort of this age job insecurity in the absence of financial insecurity may be accepted philosophically, endorsing the long-held finding of Eisenberg and Lazarsfield (1938) that having a job in itself is not as important as having a feeling of economic security. Furthermore, financial insecurity among non-employed women and men seems to be a very important determinant of inequalities in health.
Acknowledgements 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 D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. JF was supported by the Economic and Social Research Council (L128251046) during the preparation of this paper. MM is supported by an MRC Research Professorship. MS is supported by a grant from the British Heart Foundation.
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
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 Acheson, D. (1998). Independent inquiry into inequalities in health (report). London: The Stationery Office. Bartley, M., Fitzpatrick, R., Firth, D., & Marmot, M. (2000). Social distribution of cardiovascular disease risk factors: Change among men in england 1984–1993. Journal of Epidemiology and Community Health, 54, 806–814. Bennet, S. (1995). Cardiovascular risk factors in Australia: Trends in socioeconomic inequalities. Journal of Epidemiology and Community Health, 49, 363–372. Bradburn, N. M. (1969). The structure of psychological wellbeing. Chicago: Aldine. Burchell, B. (1994). The effects of labour market position, job insecurity, and unemployment on psychological health. In D. Gallie, C. Marsh, & C. Vogler (Eds.), Social change and the experience of unemployment (pp. 188–212). Oxford: Oxford University Press. 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. Domenighetti, G., D’Avanzo, B., Bisig, B. (1999). Health effects of job insecurity among employees in the Swiss general population. Lausanne: Universite de Lausanne. Dooley, D., Rook, K., & Catalano, R. (1987). Job and non-job stressors and their moderators. Journal of Occupational Psychology, 60, 115–132. Eisenberg, P., & Lazarsfeld, P. F. (1938). The psychological effects of unemployment. Psychological Bulletin, 35, 358–390. Ensminger, M. E., & Celentano, D. D. (1988). Unemployment and psychiatric distress: Social resources and coping. Social Science & Medicine, 27, 239–247. Erens, B., Primatesta, P. (1999) Health Survey for England cardiovascular disease ‘98. London: The Stationery Office. Ferrie, J. E. (2001). Is job insecurity harmful to health? Journal of the Royal Society of Medicine, 94, 71–76. Ferrie, J. E., Martikainen, P., Shipley, M. J., Marmot, M., Stansfeld, S. A., & Davey Smith, G. (2001). 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., Davey Smith, G., Stansfeld, S. A., & Marmot, M. (2002). Change in health inequalities among British civil servants: The Whitehall II study. Journal of Epidemiology and Community Health, 56, 922–926.
645
Gerace, T. A., & George, V. A. (1996). Predictors of weight increases over 7 years in fire fighters and paramedics. Preventive Medicine, 25, 593–600. Goldberg, D. P. (1972). The detection of psychiatric illness by questionnaire. London: Oxford University Press. Government Statistical Service. (1998). Civil service statistics 1998. London: Cabinet Office. Hutton, W. (1994). Bad times for the good life. Guardian, August 2, 2–3. Hutton, W. (1995). The state we’re in. London: Jonathon Cape. Iribarren, C., Luepker, R. V., McGovern, P. G., Arnett, D. K., & Blackburn, H. (1997). Twelve-year trends in cardiovascular disease risk factors in the Minnesota Heart Survey. Archives Internal Medicine, 157, 873–881. Jackson, T., Iezzi, A., & Lafreniere, K. (1997). The impact of psychosocial features of employment status on emotional distress in chronic pain and healthy comparison samples. Journal of Behavioural Medicine, 20, 241–256. Kasl, S. V., & Cobb, S. (1980). The experience of losing a job: Some effects on cardiovascular functioning. Psychotherapy and Psychosomatics, 34, 88–109. Kessler, R. C., Blake Turner, J., & House, J. S. (1987). Intervening processes in the relationship between unemployment and health. Psychological Medicine, 17, 949–961. Kinnunen, U., & Natti, J. (1994). Job insecurity in Finland: Antecedents and consequences. European Journal of Work and Organisational Psychology, 4, 297–321. Kuhnert, K. W., Sims, R. R., & Lahey, M. A. (1989). The relationship between job security and employee health. Group and Organization Studies, 14, 399–410. Lynch, J. W., Kaplan, G. A., & Salonen, J. T. (1997). Why do poor people behave poorly? Variation in adult health behaviours and psychosocial characteristics by stages of the socioeconomic lifecourse. Social Science & Medicine, 44, 809–819. 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. Marmot, M. G., & Shipley, M. J. (1996). Do socioeconomic differences in mortality persist after retirement? 25 Year follow up of civil servants from the first Whitehall II study. British Medical Journal, 313, 1177–1180. Marmot, M. G., Shipley, M. J., & Rose, G. (1984). Inequalities in death—specific explanations of a general pattern? Lancet(, i), 1003–1006. Osler, M., Gerdes, L. U., Davidsen, M., Bronnum-Hansen, H., Madsen, M., Jorgensen, T., & Schroll, M. (2000). Socioeconomic status and trends in risk factors for cardiovascular disease in the Danish MONICA population, 1982–1992. Journal of Epidemiology and Community Health, 54, 108–113. Platt, S., Pavis, S., Akram, G. (1998). Changing labour market conditions and health: a systematic literature review (1993 – 1998). Dublin: European Foundation for the Improvement of Living and Working Conditions. Romans, S. E., Walton, V. A., McNoe, B., Herbison, G. P., & Mullen, P. E. (1993). Otago Women’s Health Survey 30month follow-up. I: Onset patterns of non-psychotic psychiatric disorder. British Journal of Psychiatry, 163, 733–738.
ARTICLE IN PRESS 646
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
Rose, G., Hamilton, P. S., Keen, H., Reid, D. D., McCartney, P., & Jarrett, R. J. (1977). Myocardial ischaemia, risk factors and death from coronary heart-disease. Lancet(, i), 105–109. SAS Institute Incorporated. (1990). SAS language reference version 6 first edition. Cary, NC: SAS Institute Inc. Sennett, R. (1998). The corrosion of character: The personal consequences of work in the new capitalism. New York: W.W. Norton and Company. Shaw, M., Dorling, D., Gordon, D., & Davey Smith, G. (1999). The widening gap: Health inequalities and policy in britain. Bristol: Policy Press. Stansfeld, S., Head, J., & Marmot, M. (1998). Explaining social class differences in depression and well-being. Social Psychiatry and Psychiatric Epidemiology, 33, 1–9. van Rossum, C. T. M., Shipley, M. J., van de Mheen, H., Grobbee, D. E., & Marmot, M. G. (2000). Employment grade differences in cause specific mortality. A 25 year follow up of civil servants from the first Whitehall study.
Journal of Epidemiology and Community Health, 54, 178–184. Vartiainen, E., Pekkanen, J., Koskinen, S., Jousilahti, P., Salomaa, V., & Puska, P. (1998). Do changes in cardiovascular risk factors explain the increasing socioeconomic difference in mortality from ischaemic heart disease in Finland? Journal of Epidemiology and Community Health, 52, 416–419. Ware, J.E., Kosinski, M., Keller, S. D. (1994). SF-36 Physical and mental summary scales: a user’s manual. Boston, MA: The Health Institute. Warr, P., Banks, M. H., & Ullah, P. (1985). The experience of unemployment among black and white urban teenagers. British Journal of Psychology, 76, 75–87. 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 of Occupational and Organisational Psychology, 65, 331–344.
Social Science & Medicine 57 (2003) 637–646
Future uncertainty and socioeconomic inequalities in health: the Whitehall II study Jane E. Ferriea,*, Martin J. Shipleya, Stephen A. Stansfeldb, George Davey Smithc, Michael Marmota a
International Centre for Health and Society, Department of Epidemiology and Public Health, University College London Medical School, London, UK b Department of Psychiatry, Queen Mary University of London, UK c Department of Social Medicine, University of Bristol, UK
Abstract Over the past 20 years, socioeconomic inequalities in mortality have widened, while job security and financial security have decreased. This paper examines the Whitehall II study, a longitudinal study of white-collar British civil servants. In the Whitehall II cohort socioeconomic gradients in morbidity and cardiovascular risk factors at Phase 5 (1997–99) were generally steeper than at Phase 1 (1985–88). We examine the contribution of job and financial insecurity to these at Phase 5 in 6770 women and men, all of whom were white-collar civil servants at Phase 1. Steep, inverse employment grade gradients were observed for all health measures at Phase 5, except cholesterol and systolic blood pressure in women. Gradients in the sub-population of non-employed participants tended to be steeper than gradients for participants in employment, although, with the exception of self-rated health and General Health Questionnaire (GHQ) score in men, differences were non-significant. Steep gradients in job insecurity were observed among employed participants (pp0.01), and in financial insecurity among both employed and non-employed participants (pp0.001), particularly non-employed men. With the exception of depression, adjustment for job insecurity had little effect on the employment grade gradients in morbidity. However, financial insecurity contributed substantially to gradients in selfrated health, longstanding illness, and depression in both employed and non-employed men, and additionally to GHQ score and diastolic blood pressure in the latter. Adjustment for financial insecurity in non-employed women substantially attenuated gradients in self-rated health, GHQ score and depression. These findings imply that the specific effects of job insecurity in this cohort may be less important than the more general effects of financial insecurity in determining inequalities in health. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Job insecurity; Financial insecurity; Health inequalities; Employment status; Cross-sectional; White-collar; UK
Introduction Over the past 20 years, against a backdrop of overall declines in mortality rates, analyses of population data *Corresponding author. Department of Epidemiology and Public Health, University College London Medical School, 1-19 Torrington Place, London WC1E 6BT, UK. Tel.: +44-171504-5643; fax: +44-171-813-0242. E-mail address: [email protected] (J.E. Ferrie).
have shown that the social gradient in mortality in the United Kingdom (UK) has increased (Acheson, 1998, Shaw, Dorling, Gordon, & Davey Smith, 1999). However, although socioeconomic gradients in various measures of morbidity are well documented (Acheson, 1998, Marmot, et al., 1991, Erens & Primatesta, 1999), there has been relatively little research examining trends in these measures over time. Most of the evidence to date has concentrated mainly on traditional risk factors for cardiovascular disease and indicates that socioeconomic differences are static or narrowing (Bartley,
0277-9536/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 4 0 6 - 9
ARTICLE IN PRESS 638
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
Fitzpatrick, Firth, & Marmot, 2000, Iribarren, Luepker, McGovern, Arnett, & Blackburn, 1997, Vartiainen et al., 1998, Bennet, 1995, Osler et al., 2000). Steep socioeconomic gradients in measures of morbidity and cardiovascular risk factors have long been demonstrated in the Whitehall II study, a longitudinal cohort of white-collar British civil servants, in which socioeconomic position is assessed by grade of employment (Marmot et al., 1991). Over 11 years follow-up of the Whitehall II cohort up to 1998, inequalities in health between those in the lowest employment category and those in the highest have widened slightly and marked increases in the gradient have been documented in minor psychiatric morbidity in both sexes and cholesterol in men (Ferrie, Shipley, Davey Smith, Stansfeld, & Marmot, 2002). Additionally during the 1980s and 1990s in the UK, patterns of employment, job security and welfare provision associated with the social order since the second world war have undergone and continue to undergo major change. The future, for many people, is less certain than it used to be (Hutton, 1994, 1995, Sennett, 1998). Two markers of future uncertainty have previously received some attention in relation to health; job insecurity and financial insecurity. Self-perceived job insecurity has been shown to be associated with increased mental and physical ill-health (Platt, Pavis, & Akram, 1998, De Witte, 1999, Ferrie, 2001) and there is evidence from Finland that it is inversely associated with socioeconomic position (Kinnunen & Natti 1994, Lynch, Kaplan, & Salonen, 1997). In addition, job insecurity attributed to workplace closure and downsizing has been shown to be associated with increased psychological morbidity, self-reported ill-health, long spells of sickness absence and health service use (Ferrie, 2001). A considerable body of work has documented the deleterious effect of unemployment-related financial strain on health, particularly psychological health (Warr, Banks, & Ullah, 1985, Kessler, Blake Turner, & House, 1987, Ensminger & Celentano 1988, Whelan, 1992). However, little work has examined the wider role of financial insecurity. Research to date shows poor financial security to be associated with minor psychiatric morbidity (Romans, Walton, McNoe, Herbison, & Mullen, 1993, Jackson, Iezzi, & Lafreniere, 1997) and weight gain (Gerace & George, 1996). Data on financial insecurity and on job insecurity for the whole cohort were first collected during Phase 5 (1997–99) of the Whitehall II study. The aims of this paper are to determine socioeconomic gradients in job and financial insecurity in the Whitehall II cohort, and the contribution of these to socioeconomic inequalities in morbidity and cardiovascular risk factors at Phase 5.
Methods Participants 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). Sub–samples of nonresponders indicated that approximately 4% of those invited were not eligible for inclusion, suggesting that the true response rate was higher. Although mostly white–collar, respondents covered a wide range of grades from office support to permanent secretary. Data collection Baseline screening (Phase 1) took place between late 1985 and early 1988. This involved a clinical examination and a self-administered questionnaire. Since the baseline screening there have been five data collection phases, which have alternated between a self-administered questionnaire only and a clinical examination and questionnaire. As at previous odd-numbered phases, the Phase 5 (1997–99) clinical examination included, among other measures, height, weight, blood pressure, and serum cholesterol and the questionnaire contained sections on demographic characteristics, health, lifestyle factors, work characteristics, social support, life events and chronic difficulties. Measures Standard items drawn from the Phase 5 questionnaire include personal details: age and employment status; selfreported health: self–rated health and presence of longstanding illness. Self-rated health was measured using a single item from the SF36 ‘In general would you say your health is; excellent, very good, good, fair, poor.’ (Ware, Kosinski, & Keller, 1994). For analysis, health self-rated as fair or poor was compared with good, very good and excellent. Mental health was assessed using the 30–item General Health Questionnaire (Goldberg, 1972) and comprised GHQ score and depression, measured using a 4-item depression subscale of the GHQ, derived by principal components analysis. The depression items were as follows: ‘Have you recently - been thinking of yourself as a worthless person, - felt that life is entirely hopeless, felt that life isn’t worth living, - found at times you couldn’t do anything because your nerves were too bad? As for all GHQ questions, each item was scored on a likert scale from 0 to 3 and summed for analysis. In principal components analysis these four items loaded on a single component in both women and men. (Stansfeld, Head, & Marmot, 1998)
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
New items drawn from the Phase 5 questionnaire were perceived job insecurity and financial insecurity. A single item ‘How secure do you feel in your present job?’ offered 4 response categories ranging from very insecure to very secure. Responses to the question ‘Thinking of the next ten years, how financially secure do you feel?’ were collected using a 4 category scale ranging from secure to insecure. Grade of employment and negative affectivity were determined from the Phase 1 questionnaire. Negative affectivity was assessed using the five negative affect items from Bradburn’s Affect Balance Scale (Bradburn, 1969). The Phase 1 measure was used as negative affectivity has not been measured subsequently. Grade of employment was determined by asking all participants to give their Civil Service grade title. On the basis of salary the Civil Service identified twelve nonindustrial grades which, in order of decreasing salary, comprised seven ‘‘unified grades’’, senior executive officer (SEO), higher executive officer (HEO), executive officer (EO), clerical officer and clerical assistant. Other professional and technical staff were assigned by the Civil Service to one of these grades on the basis of salary. For analysis, unified grades 1–6 were combined into one group and the bottom two clerical grades into another, thus producing six categories; category 1 represents the highest status jobs and category 6 the lowest. At the Phase 5 screening examination, blood pressure in millimetres of mercury (mmHg) was measured twice with the participant seated after a 5 min rest, using a Hawksley random–zero sphygmomanometer. Blood was taken and serum cholesterol concentration in millimoles/ litre (mmol/l) measured using the cholesterol oxidase/ peroxidase colorimetric method (BCL kit). Weight in kilograms (kg) and height in metres (m) were recorded. Body Mass Index (BMI) was calculated from these two measures as kg/m2. Further details of these standard repeat measures have been reported previously (Rose et al., Jarrett, 1977, Marmot et al., 1991). Study sample and statistical analysis Seventy-one per cent (7270) of the 10,308 Whitehall II study participants, at baseline, completed the full
639
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%. In addition to those who failed to respond to invitations for the Phase 5 data collection, non–responders included participants who had died and those who could not be traced. Of respondents to the Phase 5 data collection 4665 were still in paid employment and 2532 were out of the labour force (Box 1). In all, 4447 completed the question on job security, 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 were compared with those whose job was secure or very secure. All respondents to Phase 5 were asked about financial security, 6658 completed this question, an item non-response rate of 8%. Responses to the question on financial security were dichotomised. Those who were secure or fairly secure being compared with those who were insecure or fairly insecure. Initial analyses determined employment grade gradients, based on Phase 1 grade, for all morbidity measures at Phase 5 using least-squares regression for continuous outcomes and logistic regression for dichotomous outcomes. These analyses produce a regression line showing the increase in morbidity for a unit increase in the employment grade. We have used this employment grade gradient to estimate the odds ratio (OR) or difference (Diff) in morbidity (and 95% confidence interval) between respondents in the lowest grade (Grade 6) and respondents in the highest grade (Grade 1), and presented these in the Tables. ORs above 1.0 and Diffs above 0.0 indicate an inverse grade gradient, that is, greater morbidity among the lower grades, while ORs below 1.0 and Diffs below 0.0 indicate a positive grade gradient, that is, greater morbidity among the higher grades. Different grade gradients for women and men had already been demonstrated at baseline in the Whitehall II cohort (Marmot et al., 1991) so analyses were conducted separately by sex. All analyses were adjusted for age categorised into 5-year age groups. Data were analysed using SAS version 6.12 for Windows (SAS Institute Inc., 1990). Phase 5 participants were then divided into the employed and the non-employed. Age-adjusted employ-
Box 1 Employment status at Phase 5 Employment status
Women
Men
In the Civil Service Working Elsewhere Non-employed Total
960 285 900 2145
2360 1060 1632 5052
(44.8%) (13.3%) (42.0%) (100%)
Total (46.8% ) (21.0%) (32.3%) (100%)
3320 1345 2532 7197
(46.1%) (18.7%) (35.2%) (100%)
ARTICLE IN PRESS 640
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
ment grade gradients for these two sub-populations were determined in the same way as for the whole Phase 5 population. P-values for the difference in the gradients between the sub-populations were determined by calculating the difference between the regression coefficients for the gradient for each population and testing this against a null hypothesis of no difference. In addition, age-adjusted gradients were determined for job insecurity (employed only), and financial insecurity. Finally, the contributions of job insecurity and financial insecurity at Phase 5 to employment grade gradients in morbidity and cardiovascular risk factors were assessed by calculating the percentage reduction in the employment grade coefficients when adjusted for the measures of insecurity. Health measures included in these analyses were restricted to those for which there was evidence of a gradient (po0.05) in the sub-population under investigation. Item non-response to the insecurity questions weakened the grade gradient for some of these measures, but they were retained in the analysis. The contributions of the insecurity measures were determined by adding these terms into the models for each morbidity or risk factor measure. As some of the health and both security measures were self-reported there was the possibility that negative affect or reporting bias might have explained, to some degree, any association between insecurity and health. To take account of this the gradients for self-rated health, longstanding illness, GHQ score, depression, job insecurity and financial insecurity were determined with and without adjustment for negative affectivity. Ideally, all the analyses should have been adjusted for negative affect. However, the Affect Balance Scale (Bradburn, 1969) 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.
Results Data on current employment status was provided by 7197 of those who completed the Phase 5 questionnaire—(Box 1). Overall, just under half of the respondents were still working in the Civil Service. The majority who had left were non-employed, but a sizable minority were employed elsewhere. Employment grade gradients in morbidity and cardiovascular risk factors: In the whole Phase 5 population steep inverse employment grade gradients were seen for all measures in both sexes (p=0.04 to po0.001), with the exception of serum cholesterol and systolic blood pressure in women (Table 1) When divided into two subpopulations, employed and non-employed participants, grade gradients tended to be steeper among nonemployed women and men compared to employed
participants, with the exception of self-rated health and diastolic pressure in women. For two measures the difference in the gradients was considerable, self-rated health (p=0.03) and GHQ score (p=0.009) in men. Employment grade gradients in job insecurity and financial insecurity: Steep inverse employment grade gradients were seen for job insecurity amongst employed women (p=0.006) and men (p=0.01) and for financial insecurity in all participants whether employed or nonemployed (pp0.001) (Table 2). Among non-employed men the gradient in financial insecurity was much steeper OR 41.6 (18.2, 95.1) than among employed men OR 6.5 (4.4, 9.6), test of difference po0.001. Morbidity and risk factor gradients adjusted for job insecurity: In this population, job insecurity at Phase 5 contributed little to the grade gradients in morbidity and cardiovascular risk factors in either sex. The exception was depression in which the gradient was attenuated by 19% in women and 9% in men—(Table 3). For selfrated health in both sexes, BMI in women and longstanding illness in men job insecurity attenuated associations by roughly 5%. Morbidity and risk factor gradients adjusted for financial insecurity: Adjustment for financial insecurity at Phase 5 among those in employment resulted in attenuation of the gradients in self-rated health and depression by 13% and 14%, respectively, in women, and 33% and 49%, respectively, in men (Table 4). Additionally among men, the gradient in longstanding illness was attenuated by 25%. The gradient in BMI among employed women was only marginally affected by adjustment for financial insecurity, 6%, and adjustment of the gradient in diastolic blood pressure increased the slope by 11%. Among non-employed participants adjustment for financial insecurity attenuated associations with self-rated health, GHQ score and depression by 24%, 75% and 75% among women and by 37%, 42% and 37% among men (Table 5). In addition, associations with longstanding illness and diastolic blood pressure were attenuated, respectively, by 35% and 14% among men. Adjustment of the gradients in self-reported morbidity and insecurity for negative affectivity tended to increase the slope very slightly for all measures among women (table available on request). In men, adjustment attenuated the gradient for all the self-reported morbidity measures. The degree of attenuation ranged from 4% for self-rated health to 18% for GHQ score, but had a negligible effect on the p-values for the gradients in all cases. Adjustment for negative affect produced negligible attenuation of the gradient in job insecurity (2%) or financial insecurity (0%) in either sex. These findings make it unlikely that reporting bias can explain the steep grade gradients in the self-reported morbidity measures and either insecurity measure.
OR OR Diff Diff Diff Diff Diff Diff
Men Self-rated health (SF36) [n=4967] Longstanding illness [n=5046] GHQ score [n=4920] Depression [n=4918] Cholesterol mmol/L [n=4543] Systolic blood pressure mmHg [n=4573] Diastolic blood pressure mmHg [n=4573] Body mass index kg/m2 [n=3943] 3.02 1.33 0.95 0.61 0.59 1.99 1.61 0.39
3.74 1.30 1.26 0.54 0.09 0.99 1.57 2.16
(2.2, 4.1) po0.001 (1.1, 1.6) p=0.004 (0.5, 6.0) po0.001 (0.5, 0.8) po0.001 ( 1.1, -0.1) p=0.03 (0.4, 3.6) p=0.02 (0.6, 2.7) p=0.003 (0.1, 0.8) p=0.04
(2.4, 5.8) po0.001 (1.0, 1.7) p=0.07 (0.4, 2.2) p=0.006 (0.3, 0.9) po0.001 (-0.9, 0.7) p=0.82 (-1.6, 3.6) p=0.45 (0.1, 3.1) p=0.04 (1.4, 3.0) po0.001
All participants OR or Diff (95% CI)
2.38 1.33 0.51 0.53 0.48 1.80 1.23 0.39
4.10 1.08 0.95 0.49 0.37 0.44 1.93 2.06
(1.6, 3.4) po0.001 (1.0, 1.7) p=0.02 (-0.2, 1.2) p=0.12 (0.3, 0.8) po0.001 ( 1.2, 0.2) p=0.14 ( 0.1, 3.7) p=0.06 ( 0.1, 2.5) p=0.06 ( 0.1, 0.9) p=0.11
(2.3, 7.3) po0.001 (0.8, 1.5) p=0.67 ( 0.2, 2.1) p=0.10 (0.1, 0.9) po0.001 ( 1.3, 0.6) p=0.45 ( 2.6, 3.5) p=0.78 (0.1, 3.8) p=0.03 (1.1, 3.1) po0.001
Employed OR or Diff (95% CI)
p-value for test of difference between OR (or Diff) for the employed versus the non-employed.
OR OR Diff Diff Diff Diff Diff Diff
Women Self-rated health (SF36) [n=2098] Longstanding illness [n=2135] GHQ score [n=2053] Depression [n=2060] Cholesterol mmol/l [n=1834] Systolic blood pressure mgHg [n=1865] Diastolic blood pressure mgHg [n=1865] Body mass index kg/m2 [n=1656]
a
OR/Diff
Health measure
Table 1 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade
4.67 1.46 1.87 0.73 0.90 2.36 2.52 0.46
4.01 1.56 1.89 0.65 0.57 2.51 0.92 2.40
(2.8, 7.7) po0.001 (1.0, 2.0) p=0.03 (1.1, 2.7) po0.001 (0.5, 1.0) po0.001 ( 1.9, 0.1) p=0.06 ( 0.7, 5.4) p=0.13 (0.6, 4.4) p=0.01 ( 0.2, 1.1) p=0.17
(2.1, 8.0) po0.001 (0.9, 2.6) p=0.08 (0.4, 3.4) p=0.01 (0.2, 1.2) p=0.01 ( 0.9, 2.1) p=0.44 ( 2.2, 7.2) p=0.29 ( 1.8, 3.6) p=0.50 (1.0, 3.8) po0.001
Non-employed OR or Diff (95% CI)
p=0.03 p=0.63 p=0.009 p=0.26 p=0.47 p=0.76 p=0.27 p=0.86
p=0.98 p=0.35 p=0.57 p=0.71 p=0.29 p=0.46 p=0.54 p=0.70
p-valuea
ARTICLE IN PRESS
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646 641
ARTICLE IN PRESS Not applicable Po0.001
Discussion
p-value for test of difference between OR (or Diff) for the employed versus the non-employed.
Methodological considerations
a
Not applicable 41.62 (18.2, 95.1) po0.001 [1555] 1.43 (1.1, 1.9) p=0.01 [3272] 6.52 (4.4, 9.6) po0.001 [3204] Not applicable 9.75 (6.9, 13.8) po0.001 [4759]
Not applicable 6.32 (2.4, 16.9) po0.001 [786]
Women Job insecurity Financial insecurity Men Job insecurity Financial insecurity
Not applicable 5.34 (3.3, 8.6) po0.001 [1899]
1.78 (1.2, 2.7) p=0.006 [1171] 5.21 (3.0, 8.9) po0.001 [1113]
p-valuea Non-employed OR (95% CI) p value [n] Employed OR (95% CI) p value [n] All participants OR (95% CI) p value [n] Security measure
Table 2 Age-adjusted ORs for measures of insecurity at Phase 5 between the lowest versus the highest employment grade
Not applicable P=0.73
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
642
As employment grade at Phase 1 was used to determine the gradients, our measure of socioeconomic position 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 analyses using last known grade produced findings little different from those using grade at Phase 1 (data not shown). We felt that use of last known grade as the measure of socioeconomic position posed problems. It 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 participants in employment at Phase 5. Use of Phase 1 grade also minimised the effect of reverse causality where the levels of morbidity at baseline may affect subsequent mobility and hence grade at Phase 5. The generalisability of the findings presented in this paper is limited by the fact that all participants in the Whitehall II study were white-collar civil servants on entry to the study. However, as can be seen from Box 1, those who remained in the Civil Service were a minority at the time of the Phase 5 data collection. With early retirement reaching a peak in the late 1990s the Whitehall II cohort at Phase 5 is probably fairly representative of workforces aged 45-65, originally in long-term, secure, white-collar employment in other large organisations, such as the financial services or the health service.
Findings Steep employment grade gradients were observed for all measures at Phase 5, except cholesterol and systolic pressure in women. Gradients in the sub-population of non-employed women and men tended to be steeper than gradients for those in paid employment, although differences were not significant, with the exception of self-rated health (p=0.03) and GHQ score (p=0.009) in men. Socioeconomic inequalities in health are a persistent feature of both Whitehall cohorts (Marmot Shipley, & Rose, 1984, Marmot et al., 1991), as in the general population (Acheson, 1998). However, in contrast to the first Whitehall study, in which mortality differentials based on employment grade tended to decline postretirement (Marmot, & Shipley, 1996, van Rossum, Shipley, van de Mheen, Grobbee, & Marmot, 2000), employment grade gradients in morbidity and cardiovascular risk factors tend to be steeper among nonemployed participants. This would seem to indicate that employment grade is a marker for factors whose influence extend beyond the workplace.
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
643
Table 3 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade in employed participants, unadjusted and adjusted for job insecurity - women and men Health measure [n] Women Self-rated health (SF36) [1157] Depression [1133] Diastolic blood pressure [1132] Body mass index [996] Men Self-rated health (SF36) [3239] Longstanding illness [3269] Depression [3213] a
Attenuation %a
OR or Diff
Unadjusted OR or Diff (95% CI) p value
Adjusted for job insecurity OR or Diff (95% CI) p value
OR
4.35 (2.4, 8.0) po0.001
4.08 (2.2, 7.5) po0.001
4
Diff Diff
0.43 (0.1, 0.8) p=0.02 1.77 (-0.2, 3.7) p=0.07
0.35 (0.0, 0.7) p=0.65 1.79 (-0.2, 3.7) p=0.07
19 1
Diff
2.01 (1.0, 3.0) po0.001
2.12 (1.1, 3.2) po0.001
5
OR
2.24 (1.5, 3.3) po0.001
2.14 (1.5, 3.2) po0.001
6
OR
1.31 (1.0, 1.7) p=0.03
1.29 (1.0, 1.6) p=0.04
6
Diff
0.51 (0.3, 0.7) po0.001
0.46 (0.3, 2.0) po0.001
9
% attenuation calculated from the unadjusted and adjusted regression coefficients.
Table 4 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade in employed participants, unadjusted and adjusted for financial insecurity—women and men Health measure [n] Women Self-rated health (SF36) [1105] Depression [1089] Diastolic blood pressure [1019] Body mass index [903] Men Self-rated health (SF36) [3177] Longstanding illness [3201] Depression [3158] a
Attenuation %a
OR or Diff
Unadjusted OR or Diff (95% CI) p value
Adjusted for financial insecurity OR or Diff (95% CI) p value
OR
3.92 (2.1, 7.3) po0.001
3.28 (1.8, 6.1) po0.001
13
Diff Diff
0.52 (0.2, 0.9) p=0.006 1.84 (-0.1, 3.8) p=0.06
0.35 ( 0.1, 0.7) p=0.07 2.04 (0.1, 0.4) p=0.04
14 11
Diff
1.95 (0.9, 3.0) po0.001
1.83 (0.8, 2.9) po0.001
6
OR
2.23 (1.5, 3.3) po0.001
1.71 (1.1, 2.6) p=0.009
33
OR
1.22 (1.0, 1.6) p=0.12
1.16 (0.9, 1.5) p=0.24
25
Diff
0.49 (0.3, 0.7) po0.001
0.25 (0.1, 0.5) p=0.03
49
% attenuation calculated from the unadjusted and adjusted regression coefficients.
Steep grade gradients were observed in job insecurity for both sexes in paid employment (pp0.01). Job losses in the Civil Service between 1995 and 1998 were inversely related to grade of employment (Government Statistical Service, 1998), however, with the exception of depression, adjustment for job insecurity had little effect on gradients in morbidity. Little work has examined the association between job insecurity and socioeconomic position, although there is some evidence of an
association in two Finnish cohorts (Kinnunen & Natti, 1994; Lynch, et al., 1997). Every study of perceived job insecurity in the literature has documented consistent adverse effects on all measures of psychological morbidity (Dooley, Rook, & Catalano, 1987, Burchell; 1994, Dekker & Schaufeli; 1995, Kuhnert, Sims, & Lahey, 1989; Domenighetti, D’Avanzo, & Bisig, 1999; De Witte, 1999; Kasl, & Cobb, 1980). In addition, a significant association between perceived job insecurity
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
644
Table 5 Age-adjusted ORs or Diffs in morbidity at Phase 5 between the lowest versus the highest employment grade in non-employed participants, unadjusted and adjusted for financial insecurity—women and men Health measure [n] Women Self-rated health (SF36) [776] GHQ score [767] Depression [773] Body mass index [581] Men Self-rated health (SF36) [1547] Longstanding illness [1554] GHQ score [1529] Depression [1529] Diastolic blood pressure [1368] a
Attenuation %a
OR or Diff
Unadjusted OR or Diff (95% CI) p value
Adjusted for financial insecurity OR or Diff (95% CI) p value
OR
3.18 (1.5, 6.5) p=0.002
2.42 (1.2, 5.0) p=0.02
23.8
Diff Diff Diff
1.54 (-0.0, 3.1) p=0.05 0.48 (-0.1, 1.0) p=0.08 1.92 (0.5, 3.4) p=0.01
0.39 ( 1.1, 1.9) p=0.61 0.12 ( 0.4, 0.6) p=0.65 1.85 (0.4, 3.4) p=0.01
75.0 74.7 3.4
OR
3.77 (2.2, 6.4) po0.001
2.30 (1.3, 4.0) p=0.004
37.3
OR
1.33 (0.9, 1.9) p=0.11
1.20 (0.8, 1.7) p=0.31
34.5
Diff Diff Diff
1.79 (1.0, 2.6) po0.001 0.69 (0.4, 1.0) po0.001 2.97 (1.0, 5.0) p=0.003
1.04 (0.2, 1.9) p=0.02 0.44 (0.1, 0.8) p=0.005 2.57 (0.55, 4.6) p=0.01
42.2% 36.5 13.6
% attenuation calculated from the unadjusted and adjusted regression coefficients.
and minor psychiatric morbidity has been demonstrated previously in this cohort (Ferrie et al., 2001). Steep grade gradients were observed for financial insecurity among employed and non-employed women and men (pp0.001). This result is unsurprising as grade of employment determined income at Phase 1 in the Whitehall II study, and financial security is largely determined by income and wealth. On the 1st of April 1998 there were 16 salary bands for white-collar civil servants, ranging from the band d1–d5,000 to the band d75,001 and over. Almost 93% of white-collar civil servants earned between d5,000 and d25,000 and the median salary of d15,480 was 2.1% higher than in April 1997. Over the same time retail prices rose by 4.0% (Government Statistical Service, 1998). Financial insecurity contributed considerably to gradients in self-rated health, longstanding illness, and depression in both employed and non-employed men, as well as to GHQ score and diastolic blood pressure in non-employed men only. Adjustment for financial insecurity in non-employed women substantially attenuated gradients in self-rated health, GHQ score and depression. The findings relating to GHQ score and depression reflect those from previous studies which have shown financial insecurity to be associated with poorer mental health outcomes (Romans et al., 1993, Jackson et al., 1997). Although financial insecurity attenuated the gradient in diastolic pressure among non-employed men, it generally contributed little to gradients in the physiological measures. Little other work appears to have looked at associations between financial insecurity and physiological measures.
These findings provide evidence in this cohort that, particularly for men, financial insecurity, rather than job insecurity, seems to be the more central anxiety influencing risk of self-reported morbidity in whitecollar workers aged 45 and over. In a cohort of this age job insecurity in the absence of financial insecurity may be accepted philosophically, endorsing the long-held finding of Eisenberg and Lazarsfield (1938) that having a job in itself is not as important as having a feeling of economic security. Furthermore, financial insecurity among non-employed women and men seems to be a very important determinant of inequalities in health.
Acknowledgements 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 D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. JF was supported by the Economic and Social Research Council (L128251046) during the preparation of this paper. MM is supported by an MRC Research Professorship. MS is supported by a grant from the British Heart Foundation.
ARTICLE IN PRESS J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
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 Acheson, D. (1998). Independent inquiry into inequalities in health (report). London: The Stationery Office. Bartley, M., Fitzpatrick, R., Firth, D., & Marmot, M. (2000). Social distribution of cardiovascular disease risk factors: Change among men in england 1984–1993. Journal of Epidemiology and Community Health, 54, 806–814. Bennet, S. (1995). Cardiovascular risk factors in Australia: Trends in socioeconomic inequalities. Journal of Epidemiology and Community Health, 49, 363–372. Bradburn, N. M. (1969). The structure of psychological wellbeing. Chicago: Aldine. Burchell, B. (1994). The effects of labour market position, job insecurity, and unemployment on psychological health. In D. Gallie, C. Marsh, & C. Vogler (Eds.), Social change and the experience of unemployment (pp. 188–212). Oxford: Oxford University Press. 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. Domenighetti, G., D’Avanzo, B., Bisig, B. (1999). Health effects of job insecurity among employees in the Swiss general population. Lausanne: Universite de Lausanne. Dooley, D., Rook, K., & Catalano, R. (1987). Job and non-job stressors and their moderators. Journal of Occupational Psychology, 60, 115–132. Eisenberg, P., & Lazarsfeld, P. F. (1938). The psychological effects of unemployment. Psychological Bulletin, 35, 358–390. Ensminger, M. E., & Celentano, D. D. (1988). Unemployment and psychiatric distress: Social resources and coping. Social Science & Medicine, 27, 239–247. Erens, B., Primatesta, P. (1999) Health Survey for England cardiovascular disease ‘98. London: The Stationery Office. Ferrie, J. E. (2001). Is job insecurity harmful to health? Journal of the Royal Society of Medicine, 94, 71–76. Ferrie, J. E., Martikainen, P., Shipley, M. J., Marmot, M., Stansfeld, S. A., & Davey Smith, G. (2001). 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., Davey Smith, G., Stansfeld, S. A., & Marmot, M. (2002). Change in health inequalities among British civil servants: The Whitehall II study. Journal of Epidemiology and Community Health, 56, 922–926.
645
Gerace, T. A., & George, V. A. (1996). Predictors of weight increases over 7 years in fire fighters and paramedics. Preventive Medicine, 25, 593–600. Goldberg, D. P. (1972). The detection of psychiatric illness by questionnaire. London: Oxford University Press. Government Statistical Service. (1998). Civil service statistics 1998. London: Cabinet Office. Hutton, W. (1994). Bad times for the good life. Guardian, August 2, 2–3. Hutton, W. (1995). The state we’re in. London: Jonathon Cape. Iribarren, C., Luepker, R. V., McGovern, P. G., Arnett, D. K., & Blackburn, H. (1997). Twelve-year trends in cardiovascular disease risk factors in the Minnesota Heart Survey. Archives Internal Medicine, 157, 873–881. Jackson, T., Iezzi, A., & Lafreniere, K. (1997). The impact of psychosocial features of employment status on emotional distress in chronic pain and healthy comparison samples. Journal of Behavioural Medicine, 20, 241–256. Kasl, S. V., & Cobb, S. (1980). The experience of losing a job: Some effects on cardiovascular functioning. Psychotherapy and Psychosomatics, 34, 88–109. Kessler, R. C., Blake Turner, J., & House, J. S. (1987). Intervening processes in the relationship between unemployment and health. Psychological Medicine, 17, 949–961. Kinnunen, U., & Natti, J. (1994). Job insecurity in Finland: Antecedents and consequences. European Journal of Work and Organisational Psychology, 4, 297–321. Kuhnert, K. W., Sims, R. R., & Lahey, M. A. (1989). The relationship between job security and employee health. Group and Organization Studies, 14, 399–410. Lynch, J. W., Kaplan, G. A., & Salonen, J. T. (1997). Why do poor people behave poorly? Variation in adult health behaviours and psychosocial characteristics by stages of the socioeconomic lifecourse. Social Science & Medicine, 44, 809–819. 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. Marmot, M. G., & Shipley, M. J. (1996). Do socioeconomic differences in mortality persist after retirement? 25 Year follow up of civil servants from the first Whitehall II study. British Medical Journal, 313, 1177–1180. Marmot, M. G., Shipley, M. J., & Rose, G. (1984). Inequalities in death—specific explanations of a general pattern? Lancet(, i), 1003–1006. Osler, M., Gerdes, L. U., Davidsen, M., Bronnum-Hansen, H., Madsen, M., Jorgensen, T., & Schroll, M. (2000). Socioeconomic status and trends in risk factors for cardiovascular disease in the Danish MONICA population, 1982–1992. Journal of Epidemiology and Community Health, 54, 108–113. Platt, S., Pavis, S., Akram, G. (1998). Changing labour market conditions and health: a systematic literature review (1993 – 1998). Dublin: European Foundation for the Improvement of Living and Working Conditions. Romans, S. E., Walton, V. A., McNoe, B., Herbison, G. P., & Mullen, P. E. (1993). Otago Women’s Health Survey 30month follow-up. I: Onset patterns of non-psychotic psychiatric disorder. British Journal of Psychiatry, 163, 733–738.
ARTICLE IN PRESS 646
J.E. Ferrie et al. / Social Science & Medicine 57 (2003) 637–646
Rose, G., Hamilton, P. S., Keen, H., Reid, D. D., McCartney, P., & Jarrett, R. J. (1977). Myocardial ischaemia, risk factors and death from coronary heart-disease. Lancet(, i), 105–109. SAS Institute Incorporated. (1990). SAS language reference version 6 first edition. Cary, NC: SAS Institute Inc. Sennett, R. (1998). The corrosion of character: The personal consequences of work in the new capitalism. New York: W.W. Norton and Company. Shaw, M., Dorling, D., Gordon, D., & Davey Smith, G. (1999). The widening gap: Health inequalities and policy in britain. Bristol: Policy Press. Stansfeld, S., Head, J., & Marmot, M. (1998). Explaining social class differences in depression and well-being. Social Psychiatry and Psychiatric Epidemiology, 33, 1–9. van Rossum, C. T. M., Shipley, M. J., van de Mheen, H., Grobbee, D. E., & Marmot, M. G. (2000). Employment grade differences in cause specific mortality. A 25 year follow up of civil servants from the first Whitehall study.
Journal of Epidemiology and Community Health, 54, 178–184. Vartiainen, E., Pekkanen, J., Koskinen, S., Jousilahti, P., Salomaa, V., & Puska, P. (1998). Do changes in cardiovascular risk factors explain the increasing socioeconomic difference in mortality from ischaemic heart disease in Finland? Journal of Epidemiology and Community Health, 52, 416–419. Ware, J.E., Kosinski, M., Keller, S. D. (1994). SF-36 Physical and mental summary scales: a user’s manual. Boston, MA: The Health Institute. Warr, P., Banks, M. H., & Ullah, P. (1985). The experience of unemployment among black and white urban teenagers. British Journal of Psychology, 76, 75–87. 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 of Occupational and Organisational Psychology, 65, 331–344.