Job insecurity and incident coronary heart disease: the Whitehall II prospective cohort study

Abstract

Objective

This study uses a prospective design to examine the association between self-reported job insecurity and incident coronary heart disease; an association which has been little investigated previously.

Methods

Participants were 4174 British civil servants (1236 women and 2938 men), aged 42 to 56 with self-reported data on job insecurity and free from coronary heart disease at baseline (1995-6). These participants were followed until 2002-4, an average of 8.6 years, for incident fatal coronary heart disease, clinically verified incident non-fatal myocardial infarction, or definite angina (a total of 168 events).

Results

Cox proportional hazard models adjusted for socio-demographic characteristics showed job insecurity to be associated with a 1.42-fold (95% CI, 1.05-1.93) risk of incident coronary heart disease compared with secure employment. Adjustment for physiological and behavioral cardiovascular risk factors had little effect on this estimate; 1.38 (1.01-1.88).

Conclusion

This study suggests that job insecurity may adversely affect coronary health.

Introduction

In the UK the government plans to cut 880,000 public-sector jobs by 2017.[1] Economic crises across Europe in the wake of the banking crisis and ongoing high levels of job insecurity and unemployment in the US indicate that this is not a problem limited to the UK. With knock on effects on local economies and further job losses in those sections of the private sector dependent on public sector money, many workers are experiencing a high degree of job insecurity.

Evidence strongly suggests that unemployment is associated with increased physical and psychological morbidity and increased mortality.[2,3] However, evidence of a link between job insecurity, or the threat of unemployment, and health outcomes, with the exception of psychological well-being and some self-reported health outcomes, [4-8] remains very limited. Given that self-reported job insecurity is a strong psychosocial stressor,[9] surprising little work has examined its association with coronary heart disease (CHD), the classic objective health outcome in much stress-related research. [10]

The three population studies to date have produced mixed findings. A study in men provided no evidence of an association between self-reported job insecurity and ischemic heart disease after adjustment for somatic and behavioral coronary risk factors.[11] After similar adjustment, a study in 36,910 women did show an association between self-reported job insecurity and non-fatal myocardial infarction (MI), but no association with total CHD or fatal CHD over a 2-year follow-up, or with any outcome after 4 years.[12] Lastly, an association between self-reported job insecurity and cardiovascular disease observed in analyses adjusted for age, race and randomization among 22,086 women in an aspirin and Vitamin E placebo-controlled trial was attenuated on adjustment for education and income.[13] To address this equivocal evidence we undertook a prospective analysis of self-reported job insecurity and incident CHD in a cohort of white-collar workers, approximately two-thirds men.

Methods

Study population

The target population for the Whitehall II study was all London-based office staff aged 35-55 in 20 civil service departments in 1985. Of these, 10,308 enrolled; a response rate of 73%. Data collection at enrolment, Phase 1 (1985-1988) involved a clinical examination, which obtained physiological measures, and a self-administered questionnaire covering socio-economic factors, health, work and lifestyle. Subsequent data collection phases have alternated between a questionnaire (even-numbered phases), and questionnaire plus clinical examination (odd-numbered phases).[14] This study uses data from Phase 4 (1995-1996) when participants were aged 42-56 and self-reported job insecurity was included for the first time.

Exposure, outcome and covariates

Self-reported job insecurity was measured using the question ‘How secure is your present job?’ Response categories ‘Very insecure’ and ‘Insecure’ were collapsed to form the ‘insecure’ exposure group and ‘Very secure’ and ‘Secure’ to form the reference group. Follow-up of incident fatal CHD, clinically verified incident non-fatal myocardial infarction, or definite angina until Phase 7 (2002-2004) provided a mean follow-up for CHD events of 8.6 years (S.D. 1.9). Incident CHD comprised coronary death, first non-fatal MI, or first definite angina. The British National Health Service Central Registry provided information regarding the date and cause of all deaths. These deaths were classified as coronary either if International Classification of Diseases Ninth Revision (ICD9) codes 410 to 414, or ICD10 codes I20 to I25, were cited on the death certificate. Non-fatal MI was defined using the WHO MONICA Project criteria [15] and ascertained using data from Whitehall II study resting electrocardiograms (ECGs) recorded every 5 years; and ECGs and cardiac enzyme levels obtained from records during hospitalisation for acute myocardial infarction. Definite angina was defined by clinical records, abnormalities on ECG or coronary angiogram, and nitrate medication use, but excluded self-reports not clinically verified. [16] Two trained experts classified cardiac events and reached agreement on inconsistencies. Covariates included age, sex, marital status, occupational grade level, and prevalent diabetes (measured at Phase 4); systolic and diastolic blood pressure, cholesterol and body mass index (BMI), smoking, alcohol consumption, daily fruit and vegetable use, hours/week of moderate or vigorous exercise (measured at Phase 3, 1991-1994).[14] The role of mental health in the association between job insecurity and CHD was examined using the SF-36 mental health score (a 5-item scale assessing psychological wellbeing).[17]

Sample selection

Out of the 5411 participants still employed by the Civil Service at Phase 4, 4922 responded to the question on self-reported job insecurity. Of these, 4316 had complete data for the covariates. Removal of 142 participants with prevalent CHD left 4174 participants in the analyses.

Statistical analysis

Associations between self-reported job insecurity and baseline characteristics were examined using a chi-squared test for heterogeneity. For continuous measures, differences between insecure and secure groups were assessed using univariate analysis of variance. After confirmation that the proportional hazards assumption were met (time-dependent interaction term between job insecurity and log of the follow-up period for CHD was non-significant p=0.49), Cox proportional hazard models with follow-up period as the time scale were used to examine the association between job insecurity and incident CHD. Secure employees formed the reference category used to calculate hazard ratios and their 95% confidence intervals. Models were serially adjusted for covariates to examine whether they affected the association. As there was no interaction between sex and job insecurity in relation to CHD (p=0.17), women and men were combined.

Results

Forty percent of participants reported that their job was insecure.Table 1 presents the distribution of baseline covariates by self-reported job security. Insecure participants were slightly younger than secure participants, more likely to be lower in the occupational hierarchy and have a higher BMI, and less likely to eat fruit and vegetables daily. Otherwise, self-reported job security was evenly distributed across the covariates measured. Table 2 shows the association between self-reported job insecurity and incident CHD. Altogether there were 35,896 person-years of follow-up during which 168 new CHD events occurred over 8.6 years, an event rate of 4.7/1000 person-years. In the age-adjusted model (Model 1), job insecurity was associated with incident CHD HR 1.41, (95% CI 1.04–1.91), compared to secure employment. Cumulative additional adjustment for socioeconomic factors (Model 2), physiological measures (Model 3) and behavioral factors (Model 4) had little effect on this association HR 1.38, 95% CI 1.01–1.88 (Model 4). To examine whether mental health is a potential mediator of any association between job insecurity and CHD, we compared the SF-36 mental health score between participants in insecure and secure jobs. SF36 scores among the insecure indicated poorer mental health in this group (mean 68.5 vs. 76.3, P<0.001). Adjustment for the SF-36 mental health score further attenuated the association between job insecurity and CHD (HR 1.26, 95% CI 0.91-1.73) (Model 5).

Table 1. Characteristics of the participants by self-reported job insecurity at baseline.

Characteristics		Job insecurity at baseline
	All (n=4,174)	Secure (n=2,506)	Insecure (1,668)	P-value*
Age (years), Mean (S.D.)	50.3 (4.9)	50.5 (5.1)	50.1 (4.7)	0.009
Sex, (n %)				0.10
Men	2,938 (70)	1,788 (71)	1,150 (69)
Women	1,236 (30)	718 (29)	518 (31)
Marital status, (n %)				0.27
Married/cohabiting	3,254 (79)	1,968 (79)	1,286 (77)
Non-married/cohabiting	920 (21)	538 (21)	382 (23)
Occupational grade level, (n %)				<0.001
I highest	789 (19)	514 (21)	275 (16)
II	924 (22)	597 (24)	327 (20)
III	591 (14)	345 (14)	246 (15)
IV	699 (17)	395 (16)	304 (18)
V	567 (14)	302 (12)	265 (16)
VI lowest	604(14)	353(14)	251(15)
Prevalent diabetes, (n %)				0.89
No	4,095 (98)	2,458 (98)	1,637 (98)
Yes	79 (2)	48 (2)	31 (2)
Systolic BP**(mmHg), Mean (S.D.)	119.5 (12.9)	119.6 (12.9)	119.2 (13.0)	0.31
Diastolic BP (mmHg), Mean (S.D.)	79.2 (9.1)	79.3 (9.2)	79.1 (9.1)	0.60
Cholesterol (mmol/L), Mean (S.D.)	6.4 (1.2)	6.4 (1.1)	6.4 (1.2)	0.47
Body mass index (kg/m2), Mean (S.D.)	25.0 (3.6)	24.9 (3.6)	25.2(3.7)	0.010
Smoking, (n %)				0.33
Never	2,209 (53)	1,343 (54)	866 (52)
Ex	1,412 (34)	846 (34)	566 (34)
Current	553 (13)	317 (13)	236 (14)
Alcohol use (units/week), (n %)				0.08
0	745 (18)	420 (17)	325 (19)
0–14 (women)/–21 (men)	2,706 (65)	1648 (66)	1058 (63)
> 14/21	723 (17)	438 (17)	285 (17)
Daily fruit and vegetable use				0.002
Yes	2,566 (61)	1,588 (63)	978 (59)
No	1,608 (39)	918 (37)	690 (41)
Moderate/vigorous exercise (hrs/week)				0.95
<1.5	1,359 (33)	815 (33)	544 (33)
≥1.5	2,815 (67)	1,691 (67)	1,124 (67)

^*P-value for difference in Chi-square test and univariate analysis of variance

^**BP = blood pressure

Table 2. Association between self-reported job insecurity at baseline and incident CHD (fatal CHD, non-fatal MI, or definite angina) during the follow-up.

	CHD
Self-reported Job insecurity at baseline	Events (N)	Participants (N)	Person-years	Rate/1000 Person-years	Model 1 HR (95% CI)	Model 2 HR (95% CI)	Model 3 HR (95% CI)	Model 4 HR (95% CI)	Model 5 HR (95% CI)
All	168	4174	35843.6	4.7
Secure	90	2506	21591.5	4.2	1.00	1.00	1.00	1.00	1.00
Insecure	78	1668	14252.1	5.5	1.41 (1.04-1.91)	1.42 (1.05-1.93)	1.40 (1.03-1.90)	1.38 (1.01-1.88)	1.26 (0.91-1.73)

HR=Hazard ratio. CI=Confidence interval.

Model 1: Adjusted for age.

Model 2: Additionally adjusted for sex, marital status, and occupational grade.

Model 3: Additionally adjusted for diabetes, systolic and diastolic blood pressure, and cholesterol.

Model 4: Additionally adjusted for body mass index, smoking, alcohol consumption, fruit and vegetable use, and exercise level.

Model 5: Additionally adjusted for mental health.

Discussion

This prospective study provides evidence that self-reported job insecurity is associated with a 40% excess risk of new-onset coronary heart disease. This association was also observed after taking account of socioeconomic factors and other well-known major risk factors for heart disease. Our finding stands in contrast to previous studies of self-reported job insecurity and CHD which have provided little evidence of an association in either sex.[11-13] On the other hand, the size of the effect observed is identical to that for other work-based psychosocial stressors, such as job strain, where a 40% excess in incident coronary events has been reported in the most recent meta-analyses of published studies.[18] It is possible of course that the association between job insecurity and coronary heart disease may be mediated through other stressful work characteristics, such as job strain. In an earlier report from the Whitehall II study, perceived job insecurity was associated with low decision latitude, low skill discretion, high job demands (in women), low social support at work, and low job satisfaction.[19] Further research using repeat measurements of work characteristics is needed to examine whether work characteristics act as mediators or as confounders of the job insecurity-CHD association. Our examination of the role of mental health in the association between job insecurity and CHD found a degree of attenuation after adjustment for SF-36 mental health score. This finding is consistent with the role of mental health as a confounder of the association between job insecurity and CHD but does not preclude the notion that job insecurity may adversely affect mental health which, in turn, is associated with increased coronary disease risk.

A limitation in our data was reliance on a heterogeneous CHD endpoint, which included self-reported angina. Re-running the analyses separately for definite angina and fatal CHD/non-fatal MI gave a hazard ratio of 1.46 (1.02-2.08) for the association between job insecurity and angina in the fully adjusted model. The corresponding hazard ratio for fatal CHD and non-fatal MI as an outcome was 1.40 (0.86-2.27). Thus the magnitude of the effect was similar in both analyses, but the latter analysis, with only 68 events, lacked statistical power. All self-reported instances of angina were confirmed clinically. While some misclassification due to undiagnosed angina is possible, previous studies have shown angina to be a strong predictor of future cardiovascular events.[20,21] Larger studies that can examine more specific CHD endpoints, such as fatal and non-fatal myocardial infarction, stable and non-stable angina, and first myocardial infarction with and without the ST segment on ECG (STEMI and non-STEMI) are needed to provide a more nuanced understanding of the potential adverse consequences of job insecurity for heart health.[22-25]

Sample attrition between recruitment (1985-1988) and baseline for the present study (1995-1996) may have introduced non-response bias, as non-response has been associated with increased mortality in the Whitehall II study.[26] Certainly there were differences between the Whitehall II baseline population (67% men, 23% in the lowest occupational grade) and the sample used in the present analyses (70% men, 14% in the lowest occupational grade). However, as non-response is more likely among those in insecure rather than secure employment, the observed associations might, if anything, underestimate the job insecurity-CHD relation.

Other limitations of this study which should be noted when interpreting the results are; unmeasured or imprecisely measured risk factors that may result in residual confounding; a single measure of job insecurity at baseline will not capture ongoing exposure; covariates modeled as time independent may change over the follow-up period; and findings from a white-collar cohort may not be generalisable to blue-collar workers and private sector employees. These limitations are counterbalanced by important strengths; a prospective design, large population including both sexes, clinically verified event data, and a suitable follow-up period.

Abbreviations

BP

Blood Pressure

CI

Confidence Interval

CHD

Coronary Heart Disease

HR

Hazard Ratio

MI

Myocardial Infarction