- Jane E Ferrie, senior research fellow ()a,
- Pekka Martikainen, senior research fellowb,
- Martin J Shipley, senior lecturer in medical statisticsa,
- Michael G Marmot, directora,
- Stephen A Stansfeld, professor of psychiatryc,
- George Davey Smith, professor of clinical epidemiologyd
- a International Centre for Health and Society, Department of Epidemiology and Public Health, University College London Medical School, London WC1E 6BT
- b Department of Sociology, PO Box 18, 00014 University of Helsinki, Finland
- c Department of Psychiatry, Basic Medical Sciences Building, Queen Mary, University of London, London E1 4NS
- d Department of Social Medicine, University of Bristol, Canynge Hall, Bristol BSS 2PR
- Correspondence to: J Ferrie
- Accepted 16 January 2001
Objectives: To determine whether employment status after job loss due to privatisation influences health and use of health services and whether financial strain, psychosocial measures, or health related behaviours can explain any findings.
Design: Data collected before and 18 months after privatisation.
Setting: One department of the civil service that was sold to the private sector.
Participants: 666 employees during baseline screening in the department to be privatised.
Main outcome measures: Health and health service outcomes associated with insecure re-employment, permanent exit from paid employment, and unemployment after privatisation compared with outcomes associated with secure re-employment.
Results: Insecure re-employment and unemployment were associated with relative increases in minor psychiatric morbidity (mean difference 1.56 (95% confidence intervals interval 1.0 to 2.2) and 1.25 (0.6 to 2.0) respectively) and having four or more consultations with a general practitioner in the past year (odds ratio 2.04 (1.1 to 3.8) and 2.39 (1.2 to 4.7) respectively). Health outcomes for respondents permanently out of paid employment closely resembled those in secure re-employment, except for a substantial relative increase in longstanding illness (2.25; 1.1 to 4.4). Financial strain and change in psychosocial measures and health related behaviours accounted for little of the observed associations. Adjustment for change in minor psychiatric morbidity attenuated the association between insecure re-employment or unemployment and general practitioner consultations by 26% and 27%, respectively.
Conclusions: Insecure re-employment and unemployment after privatisation result in increases in minor psychiatric morbidity and consultations with a general practitioner, which are possibly due to the increased minor psychiatric morbidity.
Traditionally the public sector in the United Kingdom was immune to the pressures of the marketplace, and among its main attractions were job security, a career, and good conditions of service. However, much of this changed during the 1980s, when the United Kingdom led the way among industrialised countries in moves away from planned public ownership and provision.1 Privatisation of the first public service occurred in 1984. By the end of 1997 most public utilities had been privatised, and currently privatisation is being introduced into education, health care, transport, and central and local government. The future privatisation of the executive functions of government came on to the agenda with the introduction of the “Next Steps” programme in August 1988. Early in the restructuring, one of the 20 departments participating in the Whitehall II study, the Property Services Agency, was sold to the private sector.
Whitehall II is an ongoing study of the health of civil servants, and baseline data were collected before any indication of major restructuring. It is thus ideally placed to address some of the methodological limitations of previous studies of the effects of workplace closure on health. Rumours of the forthcoming privatisation reached the work force two to three years before the sale, and during this “anticipation” phase there was a deterioration in self reported health both compared with baseline and, crucially, compared with that reported in other departments in the Whitehall II study.2 By the “pre-termination phase,” immediately before the sale, both self reported morbidity and physiological risk factors had increased relative to those seen in respondents in the control departments.3 These increases in morbidity were not explained by changes in other psychosocial work characteristics or changes in health related behaviours.4
We examined the effects on health and general practitioner consultations of employment status 18 months after the privatisation and whether any associations could be explained by changes in financial strain, psychosocial measures, and health related behaviours.
The privatisation of the Property Services Agency, which was responsible for the design, construction, and maintenance of all government buildings, was complete by the end of 1993. Between April 1990 and July 1991 the agency was split into six separate businesses. Most of the Whitehall II respondents in this department at baseline were in projects division, the design and construction side, which was sold to Tarmac plc in December 1992.5 After privatisation all employees lost their original jobs.
The Whitehall II study
The target population for the Whitehall II study was all office staff based in London who were working in 20 civil service departments between late 1985 and early 1988. With a response rate of 73%, the final cohort consisted of 10 308 participants (6895 men and 3413 women). Although mostly white collar (office) workers, respondents covered a wide range of grades. The baseline screening of the cohort involved a clinical examination and a self administered questionnaire that contained sections on demographic characteristics, health, lifestyle, and work.6
Property Services Agency study
A study specifically designed to investigate effects of the privatisation started in 1994. The study population was all 666 (153 women and 513 men) Whitehall II respondents who were working in the agency at baseline screening. In addition to using baseline data, we gathered follow up data by self administered questionnaire 18 months after privatisation, eight to nine years after baseline screening. We have used the baseline survey and data from the follow up questionnaire.
Personal details—Items drawn from the baseline and follow up questionnaires include age, marital status, civil service employment grade at baseline, and employment status 18 months after privatisation.
Health outcomes and consultations with general practitioner—Self reported health outcomes at baseline and follow up included health over the past year rated as average, fair, or poor versus good or very good; presence of longstanding illness; number of symptoms in the past fortnight (from a checklist of 17); number of health problems in the past year; and minor psychiatric morbidity assessed with the 12 item general health questionnaire.7 In the follow up questionnaire we also asked about the number of general practitioner consultations in the preceding 12 months.
Exposure measures—We determined employment status 18 months after privatisation from responses to the follow up questionnaire. From the answer to “How secure do you feel in your present job?” we divided employed respondents into two groups: those who were “secure” or “very secure” and those who were “not very secure” or “very insecure.” We divided those not in paid employment according to their response to the question “Would you like to find another job?” and classified those seeking work as unemployed and those not seeking work as permanently out of paid employment. The final four categories were secure re-employment, insecure re-employment, unemployment, and permanently out of paid employment.
Explanatory factors—We assessed negative affectivity with the five negative items from Bradburn's affectivity balance scale.8 Financial strain was assessed by combining responses to two questions from Pearlin and Schooler's list of chronic strains (scores ranged from 0 to 8).9 Questions covered problems with paying bills and buying the kind of food and clothing the respondent thought she or he and the family should have. For psychosocial measures we investigated perception of low ability to influence health (external locus of control) and two or more adverse life events in the past year, versus 0 or 1, in all employment groups. For those in employment we examined four other psychosocial work characteristics. Decision authority, skill discretion, and job demands were adapted from the job content instrument of Karasek.10 Social support at work comprised three components: support from colleagues, support from supervisors, and clarity and consistency of information from supervisors. All the questions required responses on a four point scale from “often” to “never/almost never.” We divided each scale into thirds, and, for analysis, change from third at baseline to a more adverse or beneficial third by follow up formed the explanatory factor. Those who experienced adverse change were compared with those who experienced no change or beneficial change and vice versa. We investigated three behaviours related to health: alcohol consumption over the recommended limits, smoking, and exercise. We measured all explanatory factors, except negative affectivity, at baseline and follow up.
Our overall aim was to determine whether change in morbidity between baseline and follow up differed between respondents in the four categories of employment after privatisation. In the absence of a control group who had not experienced privatisation we used participants in the most favourable category in the labour market (secure re-employment) as the reference group.
Sex differences for all measures were small (analyses not shown) so we combined the sexes for further analysis. Initially, we compared baseline characteristics of respondents who comprised the employment groups after privatisation. For continuous variables we used linear regression (general linear models procedure in SAS) to produce adjusted means with 95% confidence intervals, with adjustment consecutively for age, employment grade, marital status, and the baseline level of the variable of interest. Results for continuous variables compare the exposure groups with the reference group in terms of adjusted mean differences. For dichotomous variables we used logistic regression (LOGIST procedure in SAS) with results presented as odds ratios and 95% confidence intervals.
After the analysis of morbidity measures and general practitioner consultations we identified potential mediators of increases in these outcomes. These were explanatory factors which had changed between baseline and follow up, relative to the reference group, at conventional levels of significance (P<0.05). Such factors were included in the final model for each health outcome. Models were also adjusted separately for negative affectivity, which is characterised by a disposition to overreport negative events and experience chronically high levels of distress. The datasets used in these analyses include only respondents who have data for the health outcome of interest and the potential mediator(s), hence odds ratios or differences may differ slightly.
Ethical approval for the Whitehall II study was obtained from the University College London Medical School committee on the ethics of human research.
Of the 666 respondents in the Property Services Agency at baseline, 541 (81%) responded to the follow up questionnaire. Non-responders were younger than responders, and a smaller proportion did vigorous exercise at baseline. The 539 respondents who provided usable data were categorised by employment status 18 months after privatisation (table 1). Less than 10% of respondents in the study population were re-employed by Tarmac plc.
In general, respondents with less favourable employment outcomes had greater morbidity and poorer psychosocial profiles and health related behaviours at baseline (table 2). Results of tests of heterogeneity between the groups were not significant for psychosocial factors and health related behaviours but were significant for health measures except longstanding illness. However, analyses of health outcomes after privatisation adjusted for the baseline values of all the health measures and all the potential explanatory variables were similar to the results presented in table 2.
Health outcomes and general practitioner consultations
Insecure re-employment and unemployment —After adjustment for baseline measures, morbidity was greater among insecurely re-employed or unemployed respondents than among securely re-employed respondents. For minor psychiatric morbidity and consulting a general practitioner four or more times in the past year differences were significant (table 3).
Permanent exit from paid employment—Among respondents permanently out of paid employment outcomes for health self rated as average or worse and number of symptoms in the past fortnight compared favourably with the reference group (secure re-employment). There was little difference in number of health problems in the past year, but the relative difference in longstanding illness was significant. Levels of minor psychiatric morbidity were similar in the two groups, and although general practitioner use was considerably raised, the relative difference was not significant (table 3).
Potential explanatory factors
All the less favourable employment outcomes were associated with a relative increase in financial strain, which was significant in the unemployed. Relative to securely re-employed respondents, those in insecure re-employment generally experienced adverse changes in other psychosocial work characteristics (table 4). Overall, health related behaviours among those with less favourable employment outcomes were better than among the securely re-employed, including an increase in vigorous exercise among unemployed respondents. However, there was a considerable relative increase in smoking among respondents permanently out of paid employment.
Table 5 gives details of health effects adjusted for potential explanatory factors. Adjustment for negative affectivity had a negligible effect on the relation between permanent exit from paid employment and longstanding illness. The only potential mediator that attenuated the association between insecure re-employment and minor psychiatric morbidity was adverse change in decision authority (6%). Financial strain attenuated the association between unemployment and minor psychiatric morbidity by 9%.
Adjustment for minor psychiatric morbidity attenuated the association between employment status and general practitioner consultations in the past year by 26% among respondents in insecure re-employment and by 27% among the unemployed. Financial strain attenuated the relation between unemployment and general practitioner consultations by 9%, but adjustment for increased exercise strengthened the association by 11%. The effect of adjustment for all the potential mediators and negative affectivity together shows that these effects are partially independent and partially overlapping. Adjustment for negative affectivity had no effect on the association between permanent exit from paid employment and longstanding illness (adjusted odds ratio 2.28, fully adjusted 2.28 (1.2 to 4.5), log % change 0.4%).
This large study of employment after privatisation shows that insecure re-employment and unemployment are both associated with increases in minor psychiatric morbidity and that being permanently out of paid work is associated with longstanding illness. These results cannot be explained by changes in financial strain, psychosocial factors, or health related behaviours.
Many studies on workplace closure have been limited by their inability to collect data from a period of secure employment before any rumour of job loss. Whitehall II has personal details and data on health status, psychosocial measures, and health related behaviours that were collected before privatisation of the Property Services Agency was anticipated. History of ill health is often the strongest predictor of subsequent morbidity. Comparison of data for individuals from the baseline screening, a phase of secure employment, with those collected 18 months after privatisation enabled us to determine changes related to loss of secure employment separately from the effects of previous health status and other demographic factors.
The combination of data from a time of secure employment and the longitudinal design allowed us to use an analytic strategy equivalent to an intention to treat analysis used in clinical trials. Thus, the analyses included all participants who were in the agency at baseline and from whom data were collected at follow up. This means the cohort was entirely unselected and included respondents who left the agency before any rumour of privatisation and those who left or transferred to another department during the process but for reasons other than privatisation. Inclusion of respondents who had little or no exposure to privatisation, or were relatively unaffected by it, results in conservative estimates of effects but avoids overstating its impact.
The study's weakest points are the absence of a control group and potential selection into re-employment. However, use of the securely re-employed, who also went through the privatisation, as the reference group is likely to result in further underestimation of effects. Furthermore, adjustment for all the morbidity measures and potential explanatory factors at baseline had little effect on health outcomes after privatisation (data not shown), indicating that selective re-employment is unlikely to explain our findings fully.
The generalisability of findings from most occupational studies is limited by the participants, often a relatively homogeneous group working in one particular specialty or organisation. Similarly, in this study, in addition to being white collar, many of the respondents were specialised professional and technical staff in the construction industry. However, the agency also employed a considerable number of administrators and general office staff such as personal assistants, secretaries, personnel and welfare officers, clerks, and messengers, who make it equivalent to many office based settings in the public and private sector.
Self reported morbidity
Respondents who found secure re-employment after the sale of the agency had the best self reported health, while those who were insecurely re-employed or unemployed had the worst outcomes for most measures. Among those permanently out of paid employment self reported health outcomes were similar to those for respondents in secure re-employment, except for longstanding illness, which was much higher than in any other group. Longstanding illness has been shown in other studies to be associated with permanent exit from paid employment (mostly people who were permanently sick or had taken early retirement), particularly at times of high unemployment.11
Minor psychiatric morbidity
Most studies of workplace closure have compared mental health in unemployed people with that in re-employed people.12–15 With one exception14 such comparisons show that re-employed people have better mental health than unemployed people, although long term unemployment narrows or eliminates this difference. The problem with such comparisons is that differences may be due to the selective re-employment of those with better mental health. 16 17 After the privatisation in this study, however, re-employed people were divided into those in secure re-employment and those in insecure re-employment. This division showed that change in minor psychiatric morbidity was significant among the insecurely re-employed compared with those in secure re-employment. This relative increase is commensurate with Burchell's finding that increased depression scores in unemployed men are not reduced by re-employment in an insecure job.18 Compared with satisfactory re-employment, unsatisfactory re-employment after closure among male steel workers19 and car workers15 increased depression scores, while scores for the unemployed fell in between.19 Perceived job insecurity has also been associated with an increased risk of minor psychiatric morbidity in cross sectional studies in different occupational groups, predominantly white collar workers.20–22
Unemployment 18 months after privatisation was associated with minor psychiatric morbidity. Two recent reviews of the effect of unemployment on health concluded that longitudinal studies show that unemployment is associated with deteriorating mental health, 23 24 although it is unclear how long such effects persist.24 The mean score on the general health questionnaire for respondents permanently out of paid employment was almost the same as that for the securely re-employed. A workplace closure study among car workers showed that one year after closure securely re-employed people and retired people had similarly low levels of depression,15 as did older ship builders who accepted early retirement on full pay.25
General practitioner consultations
Eighteen months after privatisation there was a strong positive association between those with less favourable employment outcomes and number of consultations with a general practitioner. This association was significant for the insecurely re-employed and the unemployed, which were also the employment outcomes associated with greater levels of self reported morbidity. Other studies which have data on this outcome have shown insecure re-employment 26 27 and unemployment28–31 to be associated with increased number of consultations with a general practitioner. Adjustment for minor psychiatric morbidity showed that over a quarter of the increase among the insecurely re-employed and the unemployed is attributable to increased minor psychiatric morbidity.
Explanations based on psychosocial factors, financial strain, and negative affectivity
Financial strain was associated with unemployment and explained 9% of the association between unemployment and increased minor psychiatric morbidity. Most other work has shown that relations between unemployment and psychological symptoms become weaker or disappear after adjustment for financial hardship32 and that general health questionnaire scores are dependent on proportional change in family income.33 However, Whelan has shown that although lack of household heat, food, and clothing and increased debt have a large role in mediating the impact of unemployment on minor psychiatric morbidity, unemployment itself continues to have a substantial independent effect.34
Adverse changes in decision authority explained 6% of the association between minor psychiatric morbidity and insecure re-employment after privatisation. A recent Finnish study found that adjustment for decreased participation in decision making explained 19% of the association between major versus minor downsizing and medically certified sickness absence,35 but a study among miners found that job control did not moderate the adverse effect of job insecurity on psychological strain.36 Adjustment for negative affectivity had little influence on our findings, although respondents who report their employment as being insecure may also give adverse reports about other aspects of their life, and measures of negative affectivity may be rather limited in their ability to address this issue.37
Explanations based on health related behaviours
None of the studies on workplace closure have reported data on exercise. Cross sectional studies have found that unemployed people report levels of physical activity comparable with those for employed people. 38 39 However, a cross sectional population study in Sweden found that those unemployed for one year or more had raised levels of physical activity compared with men who had experienced little unemployment.40 A study of male construction workers in Finland found a relative increase in exercise among those who were unemployed for over 24 months.41 Exercise data from this study seem to indicate that respondents who were not employed were spending some of their increased spare time in physical activity. Indeed, most sports and leisure facilities in the United Kingdom have special rates for unemployed and retired people. Adjustment for exercise showed that general practitioner consultations among unemployed people would have been greater by 11% had this group not taken up exercise.
All our findings suggest that employment status after privatisation has a direct effect on minor psychiatric morbidity and longstanding illness. In addition to this increase in individual morbidity, the loss of secure public sector employment adds to NHS costs through increased consultations with general practitioners, which our results show are partly related to the increased minor psychiatric morbidity associated with privatisation.
What is already known on this topic
Epidemiological evidence points to greater morbidity and more consultations with a general practitioner among those who remain unemployed after job loss
Re-employed people have better mental health than unemployed people
Most studies have failed to differentiate between secure employment and insecure re-employment
What this study adds
Insecure re-employment and unemployment increase minor psychiatric morbidity and the number of consultations with a general practitioner
Adjustment for change in minor psychiatric morbidity attenuated the association with general practitioner consultations by over 25%
Adjustment for financial strain, change in other psychosocial work characteristics, and health related behaviours accounted for only a small proportion of observed change
We 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.
Contributors: JEF organised the data collection at follow up, carried out the analysis, and wrote the original and successive drafts of the paper. PM helped to interpret the data and commented on all drafts of the paper. MJS advised on the analysis and drafts of the paper. MGM designed and directs the Whitehall II study. SAS commented on drafts of the paper. GDS designed the substudy presented in this paper and commented on all drafts of the paper. JEF will act as guarantor for the study.
Funding Economic and Social Research Council (R000235083). Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart Lung and Blood Institute (RO1-HL36310), US, NIH; National Institute on Aging (RO1-AG13196), US, NIH; Agency for Health Care Policy Research (RO1-HS06516); and the John D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socioeconomic Status and Health. PM is also supported by the Academy of Finland (grant 48600) and the Signe and Ane Gyllenberg Foundation. MJS is supported by the British Heart Foundation. MGM is a Medical Research Council research professor. GDS was a Wellcome Fellow in Clinical Epidemiology when baseline data for this study were collected.
Competing interests None declared.