The health effects of major organisational change and job insecurity

Pergamon PII: S0277-9536(97)00158-5 Soc. Sci. Med. Vol. 46, No. 2, pp. 243-254, 1998 © 1998 ElsevierScienceLtd. All rights reserved Printed in Great Britain 0277-9536/98 $19.00 + 0.00 THE HEALTH EFFECTS OF MAJOR ORGANISATIONAL CHANGE A N D JOB INSECURITY J A N E E. F E R R I E , I* M A R T I N J. SHIPLEY,' M I C H A E L G. M A R M O T , 2 S T E P H E N S T A N S F E L D 3 and G E O R G E D A V E Y S M I T H 3 ~Department of Epidemiology and Public Health, University College London Medical School, 1-19 Torrington Place, London WCIE 6BT, U.K., qnternational Centre for Health and Society, University College London Medical School, 1-19 Torrington Place, London WCIE 6BT, U.K. and 3Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, U,K. Abstract--Since August 1988 an increasing proportion of the executive functions of government in the United Kingdom have been devolved to executive agencies. Transfer to an executive agency involves a period of uncertainty during which the options of elimination or transfer to the private sector are considered, followed by a marked change in management style and further periods of uncertainty when the agency's function is reconsidered for transfer to the private sector. This paper examines the effects of this major organisational change and consequent job insecurity on the health status of a cohort of 7419 white-collar civil servants by comparing groups either exposed to or anticipating exposure to this stressor, with controls experiencing no change. Compared with controls, men both already working in and anticipating transfer to an executive agency experienced significant increases in health self-rated as "average or worse", longstanding illness, adverse sleep patterns, mean number of symptoms in the fortnight before questionnaire completion, and minor psychiatric morbidity. Significant relative increases in body mass index were seen in both exposure groups while exposure to agency status was also associated with significant relative increases in blood pressure. Health-related behaviours, where they differed between exposure and control groups, tended to favour those in the exposure groups. Compared with controls, women in both exposure groups reported small increases in most self-reported morbidity measures and most clinical measurements, accompanied by slight beneficial changes in some health-related behaviours and small adverse changes in others. Significant relative increases were seen in mean number of symptoms, and ischaemia among women anticipating exposure and in body mass index among those exposed to agency status. Policy makers should be aware of the wider consequences of job insecurity when considering the efficiency of changes in employment policy. ~) 1998 Elsevier Science Ltd. All rights reserved Key words--job insecurity, health, white-collar, longitudinal INTRODUCTION In the United Kingdom a combination of recent recessions, public spending cuts, technological innovation, and increased remuneration to shareholders has resulted in both public and private organisations subcontracting work and/or shedding labour (Hutton, 1995). The informal social contract governing employment relations has also changed. A tranche of legislation has weakened trade unions and a climate has been created in which employers are more willing to use the leeway the legal system allows. In this new climate a flexible labour market is seen as competitive and efficient (Beatson, 1995). However, the less acceptable corollary of this flexibility is job insecurity. In the U.K. in the year to autumn 1995, 278,000 women and 484,000 men were made redundant (MacErlean, 1996). While a few find redundancy opens up positive opportunities (Fineman, 1987), many who go on to find another job, end up in positions characterised by *Author for correspondence. less security, lower income and poorer fringe benefits (Hartley et al., 1991). In the 1980s this exodus from the labour force was mostly blue-collar, but an increasing number of those facing insecurity, redundancy and job change in the 1990s are whitecollar employees traditionally accustomed to longterm, secure employment. Many countries over the 1980s turned towards privatisation and away from planned public ownership and provision (Hutton, 1995). The U.K. was no exception. Privatisation of the first public service came in 1984 with the sale of British Telecommunications. In 1986, a report Using Private Enterprise in Government recommended the examination of civil service functions to determine whether they could be abolished or transferred to the private sector. Implementation of this recommendation began in August 1988 through the " N e x t Steps" programme, which has aimed to separate executive and policy functions of government and transfer the executive functions to clearly designated units called agencies. The " N e x t Steps" programme is essentially a strategy for meeting the 243 244 Jane E. Ferrie et al. goals of cutting public expenditure, cutting staff numbers and improving efficiency (National Union of Civil and Public Servants, 1991). The transfer of a particular government function to an executive agency involves a period of uncertainty during which the options of elimination or transfer to the private sector are considered, followed by a marked change in management style and further periods of uncertainty when the agency's function is reconsidered for transfer to the private sector. Since 1988, 386,000 (74%) civil servants, have been moved into 130 executive agencies (Next Steps Team, 1997) whose progression toward commercial viability and privatisation are being accelerated (H.M.G., 1996). In an examination of the health and psychosocial effects of restructuring in government Dr Elizabeth McCloy, Chief Executive of the then Civil Service Occupational Health and Safety Agency (McCloy, 1995), identified several stressors generated by change to agency status: delayering and the collapse o f hierarchies, which to the individual may appear as a threat to status or promotion prospects; increased delegation o f responsibility, which for some has involved forgoing responsibility and for others has involved premature delegation of responsibility before training has been given and relevant skills acquired; team working, which requires increased personal interaction and may involve team performance-related pay with its attendant anxieties of individual remuneration dependent on the diligence of others; and a marked increase in work hours with no expectation of time off in lieu, which resulted in the genuinely sick coming to work and in guilt and feelings of inadequacy among those who do take leave. Very little has been published about the health effects of major change within an organisation. Studies of change in labour market status usually deal with larger transitions, such as moves in and out of the workforce. In one study among managers and executives, change of job between branches of the same company was associated with negligible change in number of coronary events while transfer between departments was associated with a small decrease in events (Hinkle et al., 1968). However, Theorell found that when eight work items were considered (change to a different line of work; retirement from work; major change in work schedule; increased responsibility; decreased responsibility; trouble with boss; trouble with colleagues; and unemployment for more than one month), 41% of his myocardial infarction group reported such changes during the year preceding onset of disease, while only 17% of the matched controls reported such changes (Theorell, 1974). An investigation of occupational stress in the Tecumseh community health study showed that although high levels of self-reported job stress at one point in time did not predict mortality, those who reported high levels of job tensions and pressures at two time points were at a significantly increased risk (House et al., 1986). In a longitudinal study of job insecurity Heaney showed that chronic job insecurity significantly decreased job satisfaction and increased physical symptoms over and above the effects of job insecurity at any single point in time (Heaney et al., 1994). The lack of data on the relationship between major organisational change and health does not apply to other labour market positions, whose effects on health have long been established. Those in employment are known to be healthier than those not working (the so-called "healthy worker effect" Carpenter, 1987). Numerous studies have documented a higher prevalence of physical and psychological morbidity, and excess mortality among the non-employed, particularly the unemployed, in studies using macro-economic time series or aggregated data (Morris and Titmuss, 1944; Moser et al., 1984; Platt and Kreitman, 1990; Catalano and Dooley, 1983; Brenner and Mooney, 1983), and in studies of individuals (Haynes et al., 1978; Morris et al., 1994; Kessler et al., 1987, 1988; Studnicka et al., 1991). Several mechanisms through which unemployment may effect health have been given serious consideration: poverty, unemployment as a stressful life event, health-related behaviour and the effect of a spell of unemployment on subsequent work patterns (Bartley, 1994). Threat of redundancy has also been shown to have adverse effects on self-reported physical and psychological morbidity, sickness absence, health service use, and clinical measurements among individuals (Cobb and Kasl, 1977; Beale and Nethercott, 1985; Hartley et al., 1991; Burchell, 1994; Arnetz et al., 1988; Owens, 1966; Layton, 1987; Ferrie et al., submitted). Possible mechanisms for these effects must, by default, be limited to stress or adverse changes in health-related behaviour. Evidence from the empirical literature to date indicate that the decline in physical and mental health during a phase of anticipated job loss are not attributable to changes in health-related behaviours (Arnetz et al., 1988; Mattiasson et al., 1990; Ferrie et al., 1995). A major consideration when examining the effect of labour market status change and job insecurity on health is the contribution of existing ill health to the outcome. Other factors, such as early life experiences, may also increase the likelihood of unemployment in adulthood (Montgomery et al., 1996). Opportunities to study the effect of job insecurity or change in labour market status on health in situations where no plausible selection process is involved occur infrequently. An example is studies of workplace closure in which all respondents are made redundant regardless of personal characteristics such as poor health or depression, which might in other situations increase the individual's risk of job loss. Even in these studies, those who feel that their opportunities post-closure are limited may experience anticipation of redundancy as a Health effects of organisational change and job insecurity more stressful life event. However, where it is possible to control for the contribution of morbidity existing before any indication of change is apparent one aspect of this selection issue can be resolved. Such studies are rarely feasible as they require the collection of data before change is anticipated. The decision to transfer a particular government function to an executive agency is dependent on the nature of that function and independent of the health of the employees who perform it. The Whitehall II study (Marmot et al., 1991), an ongoing, prospective study of the health of male and female civil servants, for which baseline data were collected considerably before the implementation of the "Next Steps" programme, has thus provided the opportunity to study the health effects of major organisational change and job insecurity in a situation in which the contribution of health selection to the outcome is minimised. We have previously reported data from the Whitehall II study in relation to the privatisation of one complete civil service department, the Property Services Agency (PSA) (Ferrie et al., 1995; Ferrie et al., submitted). The "anticipation phase" for PSA was approximately three years prior to the sale when change was an acknowledged, but distant event of uncertain outcome. "Pre-termination" for PSA was approximately three months prior to the sale when respondents were aware of their fate and a considerable number had already left the PSA labour force. By this time, changes in management style similar to those accompanying change to agency status were in place in preparation for sale to the private sector. Results from the PSA study show increases in self-reported morbidity measures during the anticipation (Ferrie et al., 1995) and pre-termination phases (Ferrie et al., submitted) among respondents in PSA compared to respondents in other departments. During the anticipation phase several of these relative increases were significant at conventional levels. However, by the pre-termination phase the relative increase in self-reported morbidity was less, but deterioration in clinical measurements was clearly seen. Conventionally significant relative increases were found in cholesterol concentration, ischaemia and Body Mass Index (BMI) among PSA men and systolic and diastolic blood pressure and BMI among PSA women. Compared with controls the slight improvement in health-related behaviours observed among PSA men during the anticipation phase was repeated during pre-termination, while the reverse was observed among the PSA women. The present study prospectively examined the effects of job insecurity and major organisational change on health and health behaviours throughout the Whitehall II cohort of civil servants. Some members of the cohort were unexposed to change, others were anticipating change and exposed to the job insecurity this entailed, while others were 245 experiencing major changes in management style and the continued job insecurity that accompanied executive agency status. Based on previous research in this area it was hypothesised that: anticipation of transfer to an executive agency would have significant adverse effects on selfreported morbidity and psychological health, which would be potentiated by the continuing insecurity associated with experience of working in an executive agency; while anticipation of transfer to an executive agency may have small adverse effects on clinical measurements, experience of working in an executive agency would be associated with larger adverse effects on clinical measurements, given the time lag involved in producing changes in such measures; and any effects on health of major organisation change and job insecurity would not be purely attributable to changes in health-related behaviours. PARTICIPANTS, CONTEXT AND METHODS Participants The target population for the Whitehall II study was all London-based office staff working in 20 civil service departments between 1985 and 1988. With a response rate of 73%, the final cohort consisted of 10,308:6895 men and 3413 women (Marmot et al., 1991). Although mostly white-collar, the respondents covered a wide range of grades from office support staff to permanent secretary. As of 1 August 1992 annual salaries ranged from £87,620 for a permanent secretary to £7387 for the lowest paid office support grade. Baseline screening, Phase 1, took place between late 1985 and early 1988. This involved a clinical examination, in which, among other measurements, height, weight, blood pressure, electrocardiogram (ECG) and serum cholesterol were determined. A self-administered questionnaire which contained sections on demographic characteristics, health, lifestyle factors, work characteristics, social support, life events and chronic difficulties was completed by each respondent. In 1989/1990, Phase 2, the same questionnaire data were collected by post and between 1992-1993 a further round of clinical screening and questionnaire data collection was completed, Phase 3. Context and methods The Conservative manifesto of 1979 was vague about civil service reform, and their "strategy" appeared to be merely a belief that the civil service was too large and badly managed. However, an anti-public sector stance remained central to government policy and the government became 246 Jane E. Ferrie et al. renowned for its hostility to the public sector in general and the civil service in particular (Hennessy, 1990). Civil servants were considered to have a "soft life" and the pursuit of efficiency, effectiveness, economy and value for money were used to justify the widespread introduction of private sector techniques such as; downsizing, contracting out, competitive tendering, and performance-related pay (National Union of Civil and Public Servants, 1991). Civil service employment conditions which used to be comparatively good now compare unfavourably with parts of the private sector. Considerable changes in terms and conditions of service and in working practices have taken place over the last fifteen years, but the rate of change has accelerated since 1988/1989 and the introduction of the "Next Steps" programme. Consequently many civil servants have seen extensive changes to their job descriptions, including increases in workload and pace of work. Against this background a question was introduced into the Phase 3 questionnaire to differentiate those whose work had been transferred to an executive agency from those who were anticipating such a move, and those who did not expect their work to be so transferred. The statement "change of your department into an agency" was followed by the options: (1) has happened, (2) is planned, (3) not certain what will happen, and (4) is not planned. For the purpose of analysis, respondents selecting options (2) and (3) were combined and are described as those anticipating change. Respondents selecting option (1) are described as those exposed to change and those selecting option (4) are the control group. All respondents to the baseline screening of the Whitehall II study were invited to participate at Phase 3, whether still in the civil service, working elsewhere or not working. However, the question "change of your department to an agency" only applied to those working in the civil service at time of screening. While it is possible that a few of these respondents may have experienced a period of unemployment between Phases 1 and 3, the vast majority of respondents to this question will have remained continuously employed in the civil service. Measures Items drawn from the Phase l and 3 questionnaires cover personal details: age, marital status and current grade of employment; health: the London School of Hygiene cardiovascular questionnaire on angina pectoris and possible myocardial infarction, hours of sleep on an average week night, self-rated health over the past 12 months, presence of longstanding illness, presence of seventeen different symptoms in the previous 14 days; minor psychiatric morbidity, assessed using the 30 item General Health Questionnaire; health-related behaviours: current smoking habits, alcohol consumption in last seven days, and exercise patterns. Heavy drinking was defined as more than 21 units of alcohol per week for men and more than 14 units/week for women. An increase in General Health Questionnaire (GHQ) score represents an increase in minor psychiatric morbidity. All those scoring 0 4 on the G H Q are considered "non-cases" and those scoring 5 + , "GHQ cases" (Stansfeld and Marmot, 1992). Negative affectivity was measured by the Negative Affect subscale of the Affect Balance Scale (Bradburn, 1969). At the screening examination, blood pressure in millimetres of mercury (mmHg) was measured twice with a Hawksley random zero sphygmomanometer while the participant remained seated after a five minute rest. Weight (wt) in kilogrammes and height (ht) in metres were recorded and BMI was calculated from these two measures as wt/ht 2. Blood was taken and the serum cholesterol concentration in millimoles/litre was measured using the cholesterol oxidase/peroxidase colorimetric method (BCL kit). ECGs were recorded with the "Mingorec" system (Siemens) on magnetic tape. Tapes were analysed at Professor Peter Mcfarlane's laboratory (Department of Medical Cardiology, University of Glasgow) where Minnesota codes were assigned by computer. The criterion for probable ischaemia on ECG was the presence of Q waves with Minnesota codes 1-1 to 1-2; possible ischaemia included any Q wave with codes 1-1 to 1-3, S-T or T waves codes 4-1 to 4-4 or 5-1 to 5-3, or left-bundle branch block code 7-1-1. All ECG tracings designated "ischaemia probable or possible" by computer were independently coded by an experienced coder. For an ECG to be considered to show possible or probable ischaemia, it had to be labelled as such by both the computer and the human coder. Where the human coder did not confirm the computer designation the ECG was not classified as abnormal. Angina was diagnosed as pain located over the sternum or in both left chest and left arm, that comes on exertion, that causes the person to stop, and that goes away in 10 minutes or less. Respondents were judged to suffer ischaemia either if their ECG was abnormal and/or angina was diagnosed. Further details on these measures have been reported previously (Marmot et al., 1991). Study sample and statistical analysis A total of 8354 respondents participated in Phase 3 of the Whitehall II study with a non-response rate of 24% for women and 16% for men. In addition to those who failed to respond to invitations for Phase 3 data collection, non-responders included participants who had died and those who could not be traced. Of those who responded, 7149 completed the question "change of your department to an agency", 86% of the men and 85% of the women. The majority of those omitting this question had already left the civil service. Health effects of organisational change and job insecurity The aim of the analysis was to compare changes between Phase 1 and Phase 3 in the various measures between those anticipating change, those exposed to change, and controls. The analysis of this form of data has received attention in recent years in the statistical literature (Plewis, 1985; Anderson et al., 1980) much of it concerning whether simple differences can be used or whether adjustment is required for baseline values. Adjustment for baseline values using analysis of covariance seems to provide a more general method of analysis. Fortunately, for continuous variables, the analysis adjusting for baseline values gives identical results irrespective of whether the baseline (Phase 1) to Phase 3 difference or just the Phase 3 value is used as the outcome. We feel that the results are easier to comprehend if direct comparison of Phase 3 values is made between the two groups. Using the Phase 3 values as the outcome also allows for continuous and discrete variables to be analysed and presented in a similar manner. For continuous variables, therefore, differences between the groups at Phase 3 follow up were assessed using analysis of covariance with adjustment being made for age, employment grade and baseline (Phase 1) level of the variable of interest. For dichotomous variables, logistic regression was used to compare the Phase 3 measures in those anticipating and those exposed to change with the controls in terms of odds ratios and to adjust for age, employment grade and the baseline level of the variable of interest. Adjustment was made for age and employment grade to ensure that changes between baseline and Phase 3 could not be attributed to any heterogeneity in the age and employment grade structure between the groups being compared. In the case of ischaemia, respondents positive at baseline were excluded and new cases at Phase 3 analysed. Age adjusted proportions were calculated by direct standardisation using five-year age groups with the total population as the standard. Differences in these proportions were assessed using Cochran-MantelHaenszel tests of association. Analyses were conducted separately in men and women. RESULTS On completing the Phase 3 questionnaire, 53.3% of the men reported themselves as working in departments where transfer to agency status was not planned, 25% reported transfer to agency status was possible or probable and 21.7% were already working in executive agencies. Among women, the equivalent figures were 54.1%, 29.7% and 16.2%, respectively. Table 1 shows age, grade, marital status, selfreported health status, clinical measurements, and health-related behaviour variables for control and exposure groups at baseline. For men, the only sig- 247 nificant differences between the control and exposure groups are for mean number of symptoms and percentage of non-drinkers. Among the women, however, the differences between the groups are generally more marked. In both sexes these differences reflect the higher percentage of clerical/ support staff in the group anticipating change. Symptom score, longstanding illness, marital status and body mass index are all related to grade. However, as all subsequent analyses are adjusted for grade and baseline values of the variable of interest, these differences at Phase 1 are controlled for in the Phase 3 findings. Self-reported health status measures from the Phase 3 questionnaire are presented in Table 2. Comparisons between control and exposure groups are shown adjusted for age, employment grade and self-reported morbidity at baseline. Compared to controls, men in both exposure groups show increases in all self-reported morbidity measures. The relative increase is significant for health selfrated as average or worse, GHQ caseness, and mean number of symptoms in the last fortnight. A significant relative increase in longstanding illness and the percentage of men sleeping <5 hrs/night is seen among those anticipating change and sleeping >9hrs/night among men exposed to change. Compared to control women, a less marked adverse trend is seen for self-reported morbidity in both exposure groups. The relative increase in mean number of symptoms in the last fortnight is significant among women anticipating change. This adverse trend fails to hold for sleeping <5 hrs/night in both exposure groups and for health self-rated as average or worse among women exposed to change. Clinical measurements from the Phase 3 screening examination are presented in Table 3. Adverse changes are seen in blood pressure and BMI for men in both exposure groups compared to control men. For BMI this relative increase is significant in both exposure groups but for systolic and diastolic blood pressure the relative increase is only significant for men exposed to agency status. Adverse relative changes are seen in all clinical measures among women in both exposure groups, except for diastolic blood pressure among women anticipating agency status. The relative increases in the percentage of women with ischaemia among those anticipating change and BMI among those exposed to change are significant. Health-related behaviours as reported in the Phase 3 questionnaire are presented in Table 4. Among men and women in both exposure groups most health-related behaviours differ little from those seen in the control group. Exceptions to this among men are the proportions of smokers, which show relative decreases in both exposure groups, significantly so among those exposed to change. In women the exceptions are the significant relative increase in the percentage taking no exercise and Jane E. Ferrie et al. 248 ~; " C~ A A V V o J 2 o C~ C~ c5 A A AA AAAA ,o .o o. o. ~. ~ ~. o0 i: o~ ~ a _ ~ !~ ~,~ ,,4 Z "~ 0 Z H e a l t h effects o f o r g a n i s a t i o n a l c h a n g e a n d j o b insecurity 249 Table 2. Age adjusted percentages and means for self-reported health status measures for men and women according to change status during Phase 3 Change to agency status Sex Not planned Possible Has happened O R a (95% CI) Possible OR S (95% C1) Has happened Number M F Self-rated health average or worse (%) M F 19.1 30.7 24.8 34.1 22.5 31.5 1.28"* (1.07 1.53) 1.02 (0.81-1.28) 1.30"*(1.08 1.57) 1.00 (0.76-1.32) Longstanding illness (%) M F 31.3 30.0 34.9 34.2 34.4 34.3 1.22" (1.00 1.48) 1.20 (0.92-1.56) 1.20 (0.99-1.46) 1.16 (0.85-1.59) G H Q caseness I%) M F 17.8 26.1 24.0 25.7 24.7 24.7 1.48"**(1.24-1.76) 1.05 (0.83 1.33) 1.56"**(1.30-1.86) 1.03 (0.77 1.37) Sleep 5 hrs or less (%) M F 3.4 6. l 4.9 5.9 3.3 4.2 1.42" (1.00 2.03) 0.97 (0.62-1.50) 1.24 (0.83-1.86) 0.63 (0.34 1.16) Sleep 9 hrs or more (%) M F 2.1 2.3 2.3 4.0 3.4 3.3 1.15 (0.72 1.85) 1.65 (0.93 2.91) 1.65' (1.06--2.56) 1.23 (0.59-2.56) 2.78 3.58 Adj diffb (SE) Possible 0.44***(0.08) 0.43"**(0.13) Adj diffb (SE) Change has happened 0.38***(0.08) 0.27 (0.15) Mean number of symptoms 2641 1185 M F 1240 650 2.49 3.37 1077 356 2.98 3.69 aOdds ratios compared with no change, adjusted for age, grade and baseline values. hDifferences compared with no change, adjusted for age, grade and baseline values. *P < 0.05, **P < 0.01. ***P < 0.001. the relative increase in smoking, both among women anticipating change. As the exposure categories and many of the outcomes are self-reports, theoretically, information bias may have produced overestimates of odds ratios and differences in the analysis, for example, if the exposure groups contained more respondents with a tendency to over-report negative events (negative affectivity). The analyses, therefore, were repeated adjusting for negative affect. Results adjusted in this way differ little from those pre- sented. Similarly, when Property Services Agency going privatisation during II study, were removed results remain the same. respondents from the (PSA), which was underPhase 3 of the Whitehall from the analysis, the DISCUSSION Throughout the following discussion it should be borne in mind that the control group were also exposed to the general changes that took place in Table 3. Age adjusted percentages and means for physiological measurements for men and women according to change status during Phase 3 Sex Number M F Ischaemia b (%) M F Change to agency status Not planned Possible Has happened 2641 1185 7.3 7.5 1240 650 7.4 11.1 OR a (95% CI) Possible OR" (95% CI) Has happened 1077 356 8.2 9.5 1.03 (0.73-1.47) 1.14 (0.79-1.64) 1.58" (1.05 2.37) IA7 (0.68 2.01) Mean systolic blood pressure ImmHg) M F 121.6 116.9 121.8 118.0 123.2 117.9 Adj difff (SE) Possible 0.30 (0.36) 0.42 (0.55) Mean diastolic blood pressure (mmHg) M F 80.7 76.4 81.0 76.8 82.2 77.0 0.30 (0.26) -0.14 (0.39) 1.46"**(0.28) 0.49 (0.48) Mean cholesterol concentration (mmol/l) M F -0.01 (0.03) 0.04 (0.04) 0.00 (0.03) 0.03 (0.05) Mean Body Mass Index (ht/wt 2) M F 0.17"**(0.05) 0.08 (0.10) 0.23***(0.05) 0.31 (0.12) 6.51 6.41 25.0 25.3 6.47 6.55 25.3 26.1 aOdds ratios compared with no change, adjusted for age, grade and baseline values. bExcluding ischaemia positives at baseline. CDifferences compared with no change, adjusted for age, grade and baseline values. *P < 0.05, **P < 0.01, ***P < 0.001. 6.49 6.48 25.3 26.1 Adj difff (SE) Has happened 1.97"**(0.38) 0.63 (0.67) 250 Jane E. Ferric et al. Table 4. Age adjusted percentages for health-related behaviours among men and women according to change status during Phase 3 Sex Not planned Change to agency status Possible Has happened 2641 1185 1240 650 O R S (95% CI) Possible O R a (95% CI) Has happened Number M F t 077 356 Non-drinker (%) M F 14.4 26.3 16.2 33.2 15.8 30.2 1.06 (0.85-1.31) 1.14 (0.89-1.45) 1.16 (0.92-1.46) 1.03 (0.76 1.40) Heavy drinker (%) M F 18.8 10.5 17,6 8.0 20.1 9.2 0.94 (0.76-1.17) 0.92 (0,60-1.39) 1,13 (0.91-1.41) 1.19 (0.74 1.93) Current smoker (%) M F 13.0 14.6 12.0 19.7 11.8 18.1 0,77 (0.54-1.09) 0.68* (0.47-0.99) 1.46 (0.91-2.33) 1.13 (0.64-1.98) No exercise (%) M F 14.8 31.8 15.1 38.5 15.0 34,1 0,96 (0.78-1.18) 1.27" (1.03 1.57) 1.08 (0.87-1.33) 1.11 (0.86-1.45) aOdds ratios compared with no change, adjusted for age, grade and baseline values. *P < 0.05. the civil service between 1979 and the time of the Phase 3 questionnaire, with the exception of transfer to an executive agency. This may have reduced differences in health between respondents in the control and exposure groups. Further, transfer of functions to executive agencies had been taking place gradually since 1988. It is thus possible that some respondents may have been exposed to the stressor in question, either anticipation or experience of agency status, for a considerable length of time. If, as in the case of unemployment and psychological morbidity, (Iversen and Sabroe, 1987; Warr and Jackson, 1987; Morrell et al., 1994) lengthy exposure attenuates the relationship between stressor and outcome, this is likely to have minimised further the estimated effects of anticipation and experience of transfer to an executive agency. Thus, our study probably gives a minimum estimate of the effects on health produced by the job insecurity induced by the changes occurring to the work of the respondents. Findings from the anticipation phase of the PSA study (Ferric et al., 1995) and studies of the health effects of insecure employment (Hartley et al., 1991; Burchell, 1994) led us to hypothesise that anticipation of transfer to an executive agency would have significant adverse effects on self-reported morbidity and psychological health. The hypothesis was confirmed among men, who showed significant adverse changes in all but one of the self-reported health status measures. Women, however, showed only slight adverse changes except for a significant increase in the number of symptoms. The hypothesis that experience of working in an executive agency would potentiate effects on self-reported health measures was generally not supported by the data. Adverse effects on self-reported health measures among men and women already working in an executive agency were not very different from those seen amongst respondents anticipating transfer. This finding is commensurate with Burchell's finding that, among men, the increased GHQ scores associated with unemployment are not affected by re-employment in an insecure job (Burchell, 1994) but does not confirm the evidence that chronic job insecurity has a significantly more potent effect on physical symptoms than acute job insecurity (Heaney et al., 1994). In general, the results for the self-reported health status measures are consistent with findings from other studies. Considerable excess physical morbidity has been reported prior to workplace closure (Cobb and Kasl, 1977; Jacobsen, 1972; Beale and Nethercott, 1985; Owens, 1966; Ferric et al., 1995). Compared to the control group a significantly greater percentage of men anticipating exposure to agency status were found to be sleeping <5 hrs/ night, while a significantly greater percentage of those exposed to change, as for PSA men during the pre-termination phase, were found to be sleeping 9 hrs or more. These sleep patterns have been shown to be associated with increased morbidity (Belloc and Breslow, 1972) and mortality (Belloc, 1973; Breslow and Enstrom, 1980). Sleep disturbance has been reported in other studies both among blue-collar men (Mattiasson et al., 1990) and women (Arnetz et al., 1988) anticipating redundancy. Both anticipation of organisational change associated with uncertainty about future job security, and exposure to agency status were related to significant risk of GHQ caseness in men. Job insecurity and threat of job loss have been related to increased psychological disorder, anxiety and depression in workplace closure studies, (Arnetz et al., 1988; Mattiasson et al., 1990; Iversen and Sabroe, 1988) while removal of this threat has been associated with reduction in psychological symptoms (Jenkins et al., 1982). Largely supporting our hypotheses, differences in clinical measurements between controls and respondents anticipating change to agency status were generally very small, while among those already working in an executive agency, apart from cholesterol, they were considerable. No studies of workplace closure over the last 30 years have reported on clinical measurements during the rumour phase Health effects of organisational change and job insecurity prior to confirmation of closure. However, consultations for chronic illness, particularly sustained hypertension and myocardial infarction, increased significantly during this phase in one study (Beale and Nethercott, 1988) and hospital admissions for cardiovascular diseases increased slightly in another (Iversen et al., 1989). Results from other workplace closure studies have demonstrated significant effects on clinical measurements in the pre-termination phase (Cobb and Kasl, 1977; Arnetz et al., 1988; Mattiasson et al., 1990; Ferric et al., submitted) in line with the finding by House and colleagues that chronic, rather than acute job stress significantly increases mortality (House et al., 1986). Varying findings with respect to blood pressure prior to workplace closure are reported in the literature. A study among blue-collar workers, mostly female, found blood pressure to be slightly lower prior to factory closure (Arnetz et al., 1988), while another among blue-collar men found a relative rise in both systolic and diastolic blood pressure (Kasl and Cobb, 1970). Neither of these studies had stable employment data. However, Mattiasson, who did have stable employment data, found a relative drop of 2 mmHg in systolic and 0.1 mmHg in diastolic blood pressure (Mattiasson et al., 1990). Schnall, who made similar comparisons with secure employment data, found a slight increase in diastolic and a significant decrease in systolic pressure among white-collar men and women during a period of widespread layoffs (Schnall et al., 1992). However, over two-thirds of respondents were lost to follow up and the blood pressure results were adjusted for BMI which could be seen as overadjustment, since increased BMI could be in the causal pathway between job disruption and blood pressure. Among men exposed to agency status we found significant relative increases in both systolic and diastolic blood pressure. The relative increase in BMI was significant among men in both exposure groups and among women experiencing change to agency status. The British Regional Heart study found men who experienced nonemployment unrelated to illness were significantly more likely to gain more than 10% in weight than men who remained continuously employed (Morris et al., 1992). The profile of adverse changes in both self-report and clinical measures would seem to agree with TheoreU's finding of a relationship between work changes and myocardial infarction (Theorell, 1974). Self-reported health status has been shown to predict mortality over a period of years, even after adjustment for physical ill health at baseline (Idler and Kasl, 1991). As in the PSA study (Ferrie et al., 1995; Ferric et al., submitted), results across the two exposure groups indicate an initial response during anticipation of transfer to an executive agency that manifests itself in increases in selfreported morbidity. Changes in clinical measure- 251 ments, in which a lag phase would be expected, manifest themselves later, among respondents exposed to working in an executive agency. Differences in the health-related behaviours between respondents in the control and exposure groups, on the whole, confirmed our hypothesis that the adverse effects on health cannot be attributed to changes in the behaviours measured. Data on potentially health damaging behaviour prior to workplace closure are scarce and mostly relate to men. Changes in alcohol consumption and smoking in a factory closure study among blue-collar men showed no difference between subjects and controls (Mattiasson et al., 1990) and journalists facing redundancy reported no significant changes in alcohol consumption (Jenkins et al., 1982). In the PSA study, if anything, slight improvements were found in health-related behaviours among men both during the anticipation and pre-termination phases (Ferric et al., 1995; Ferrie et al., submitted). These results reflect findings that loss of employment is not associated with increased smoking or drinking (Morris et al., 1994) or decreased physical activity (Grayson, 1993) but is associated with increased mortality and increased morbidity, even after controlling for selection out of the work force due to ill health (Morris et al., 1994). The results show very consistent differences between men and women. Men anticipating transfer to, and men already working in an executive agency exhibit considerably greater increases in selfreported physical and psychological morbidity than women. Arber has shown that self-rated health for women is less closely related to their own occupational class and employment status than it is for men. Factors outside the workplace, including partner's class and employment status (if married or cohabiting) and the material conditions in which the household lives are also important (Arber, 1997). Domestic responsibilities and caring have also been shown to be important determinants of women's self-reported morbidity (Macran et al., 1996; Elstad, 1996). Burchell found that re-employment in an insecure job, for women reduced GHQ scores elevated during a spell of unemployment but did not reduce scores for men (Burchell, 1994). It is possible that women use health damaging behaviours to help them to cope with adverse situations, as has been shown in the case of smoking (Graham, 1987: Jacobson et al., 1989). Thus, the small adverse changes in health-related behaviours seen among women facing job insecurity in the present study and the PSA study may explain why women exhibit changes in clinical measurements. The 7149 respondents who provided the data for these analyses represent just over half the original target population for the Whitehall II study. Normally this would represent a serious problem of bias as non-responders are generally less healthy and have higher levels of self-damaging behaviour Jane E. Ferrie et al. 252 than responders (Alderson, 1976). However, there is no reason to believe (Hennessy, 1990; National Union of Civil and Public Servants, 1991; Next Steps Team, 1996, 1997) that health factors played any role in decisions regarding timing of the transfer of particular functions of government to executive agencies and there is no other obvious reason why non-responders should be concentrated in any of the three groups analysed. Inclusion of nonresponders would, thus, have increased the level of adverse indices in all groups but left unchanged the differences between them. We are confident, therefore, in the generalisability of our results across the civil service and feel that the similarity to many of the changes taking place in other workplaces makes it likely that the results will apply more generally. Our results have implications for public health practice. Several of the factors which have contributed to current increasing levels of job insecurity look set to continue into the foreseeable future. The present study suggests that job insecurity has adverse effects on health. As these are likely to have repercussions for health service use as well as for employee well-being and organisational efficiency it is important that public health professionals push for these costs to be taken into account when the returns of the flexible labour market are counted. Care Policy Research (5 R01 HS06516), Institute for Work and Health, Toronto, Canada, and the John D. and Catherine T. MacArthur Foundation Research Network on Successful Midlife Development. REFERENCES Alderson, M. (1976) An Introduction to Epidemiology. Macmillan, London and Basingstoke. Anderson, S., Auquier, A., Oakes, D., Vandele, W. and Weisberg, H, I. (1980) Statistical Methods for Comparative Studies: Techniques for Bias Reduction. Wiley, New York. Arber, S. (1997) Comparing inequalities in women's and men's health: Britain in the 1990s. Social Science & Medicine 44, 773-787. Arnetz, B. B., Brenner, S., Hjelm, R., Levi, L. and Petterson, I. (1988) Stress Reactions in Relation to Threat of Job Loss and Actual Unemployment: Physiological, Psychological and Economic Effects of Job Loss and Unemployment. Stress Research Reports, No. 206, Karolinska Institute, Stockholm. Bartley, M. (1994) Unemployment and ill health: understanding the relationship. Journal of Epidemiology and Community Health 48, 333-337. Beale, N. and Nethercott, S. (1985) Job-loss and family morbidity: a study of a factory closure. Journal of the Royal College of General Practitioners 35, 510-514. Beale, N. and Nethercott, S. (1988) The nature of unemployment morbidity. 2. Description. Journal of the Royal College of General Practitioners 38, 200-202. Beatson, M. (1995) Labour market flexibility. Employment Department Research Series 48. CONCLUSION Among civil servants anticipation of major organisational change results in an increase selfreported morbidity, significant for most measures among men, and small increases in clinical measurements. For civil servants already working in executive agencies self-reported morbidity is similar to that exhibited by those anticipating transfer, but there are larger and more consistent adverse changes in clinical measurements, particularly among men. These adverse trends cannot be explained by changes in the health damaging behaviours measured among the men and probably not among the women, despite a slight relative increase in such behaviours among the latter. More importantly, this study has addressed the issue of preexisting morbidity by having controlled for health status during a period of stable employment well before any indication of change. Acknowledgements--We thank all participating civil ser- vice departments and their welfare and personnel officers, the Civil Service Occupational Health Service, the Civil Service Central Monitoring Service, Dr Elizabeth McCloy, and all participating civil servants. Our thanks are also extended to two anonymous referees for their helpful comments on the first draft of this paper. The work presented in this paper was supported by the Economic and Social Research Council (R000235083). The Whitehall II study is funded by grants from the Medical Research Council, Health and Safety Executive, Department of Health, British Heart Foundation, National Heart, Lung and Blood Institute (2 R01 H136310-04), Agency for Health Belloc, N. (1973) Relationship of physical health status and health practices. Preventive Medicine 2, 67-81. Belloc, B. and Breslow, L. (1972) Relationship of physical health status and health practices. Preventive Medicine 1, 409~,21. Bradburn, N. M. (1969) The Structure of Psychological Well-being. Aldine, Chicago. Brenner, M. H. and Mooney, A. (1983) Unemployment and health in the context of economic change. Social Science & Medicine 17, 1125-1138. Breslow, L. and Enstrom, J. (1980) Persistence of health habits and their relationship to mortality. Preventive Medicine 9, 469-483. Burchell, B. (1994) The effects of labour market position, job insecurity, and unemployment on psychological health. In Social Change and the Experience of Unemployment, eds D. Gallie, C. Marsh and C. Vogler, pp. 188-212. Oxford University Press, Oxford. Carpenter, L. (1987) Some observations on the healthy worker effect. British Journal Industrial Medicine 44, 289-291. Catalano, R. and Dooley, D. (1983) Health effects of economic instability: a test of economic stress hypothesis. Journal of Health and Social Behavior 24, 46-60. Cobb, S. and Kasl, S. V. (1977) Termination: The consequences of Job Loss. DHEW-NIOSH Publication No. 77-224, National Institutes for Occupational Safety and Health, Cincinnati. Elstad, J. I. (1996) Inequalities in health related to women's marital, parental, and employment status--a comparison between the early '70s and late '80s, Norway . Social Science & Medicine 42, 75-89. Ferrie, J., Shipley, M. J., Marmot, M. G., Stansfeld, S. Davey Smith, G. (1995) Health effects of anticipation of job change and non-employment: longitudinal data from the Whitehall II study. British Medical Journal 311, 1264-1269. Ferrie, J., Shipley, M., Marmot, M. G., Stansfeld, S. and Davey Smith, G. (submitted) An uncertain future: the Health effects of organisational change and job insecurity 253 health effects of threats to employment security in MacErlean, N. (1996) Down and out Britain. Observer. 17 white-collar men and women. American Journal of March, p. 1. Macran, S., Clarke, L. and Joshi, H. (1996) Women's Public Health. Fineman, S. (1987) The middle class: unemployed and health: dimensions and differentials. Social Science & underemployed. In Unemployment: Personal and Social Medicine 42, 1203-1216. Consequences. ed. S. Fineman, pp. 74-93. Tavistock, Marmot, M. G., Davey Smith, G., Stansfeld, S., Patel, C., North, F., Head, J., White, I., Brunner, E. J. and London. Feeney, A. (1991) Health inequalities among British Graham, H. (1987) Women's smoking and family health. civil servants: the Whitehall II study. Lancet 337, 1387 Social Science & Medicine 25, 47-56. 1393. Grayson, J. P. (1993) Health, physical activity level and employment status in Canada. International Journal of Mattiasson, I., Lindeg~irde, F., Nilsson. J. ~,. and Theorell, T. (1990) Threat of unemployment and cardioHealth Services 23, 743-761. vascular risk factors: longitudinal study of quality of H.M.G. (1996) Government Opportunities 1(7). Business sleep and serum cholesterol concentrations in men Information Publications Ltd, Glasgow. threatened with redundancy. British Medical Journal Hartley, J., Jacobson, D., Klandermans, B. and Van 301,461-466. Vuuren, T. (1991) Job Insecurity: Coping with jobs at McCloy, E. (1995) Re-inventing Government. The Human risk. Sage, London. Factors and Health Impacts. Work. Stress and Health Haynes, S. G., McMichael, A. J. and Tyroler, H. A. (1978) '95--Conference of the American Psychological Survival after early and normal retirement. Journal of Association. Sept. 14-16. Cassette, Sound Images, Gerontology 33, 269-278. Englewood, CO. Heaney, C., Israel, B. and House, J. (1994) Chronic job Montgomery, S. M., Bartley, M. J., Cook, D. G. and insecurity among automobile workers: effects on job satWadsworth, M. E. J. (1996) Health and social precurisfaction and health. Social Science & Medicine 38, sors of unemployment in young men in Britain. 1431-1437. Journal of Epidemiology and Community Health 50, Hennessy, P. (1990) Whitehall. Fontana Press, London. 415-422. Hinkle, L. E., Whitney, L. H., Lehman, E. W., Dunn, J., Morrell, S., Taylor, R., Quine, S., Kerr, C. and Western, Benjamin, B., King, R., Plakun, A. and Flehinger, J. (1994) A cohort study of unemployment as a cause of B. (1968) Occupation, education, and coronary heart psychological disturbance in Australian youth. Social disease. Science 161, 238-246. Science & Medicine 38, 1553-1564. House, J. S., Strecher, V., Metzner, H. L. and Robbins, C. A. (1986) Occupational stress and health among men Morris, J. K., Cook, D. G. and Shaper, A. G. (1992) Non-employment and changes in smoking, drinking and and women in the Tecumseh community health study. body weight. British Medical Journal 304, 536-541. Journal of Health and Social Behavior 27, 62-77. Morris, J. K., Cook, D. G. and Shaper, A. G. (1994) Loss Hutton, W. (1995) The State We're In. Jonathon Cape, of employment and mortality. British Medical Journal London. 308, 1135-1139. Idler, E. L. and Kasl, S. (1991) Health perceptions and survival--do global evaluations of health-status really Morris, J. N. and Titmuss, R. M. (1944) Health and social change: recent history of rheumatic heart disease. predict mortality? Journal of Gerontology 46, 55S-65S. Medical Officer 2, 69-87. Iversen, L. and Sabroe, S. (1987) Plant closures, unemployment, and health: Danish experiences from the Moser, K. A., Fox, A. J. and Jones, D. R. (1984) Unemployment and mortality in the OPCS declining ship-building industry. In Unemployment, Longitudinal Study. Lancet II, 1324-1329. Social Vulnerability, and Health in Europe, eds D. National Union of Civil and Public Servants (1991) Schwefel, P. Svensson and H. Zollner, pp. 31-47. Agenda for the Future." The Civil Service towards the Springer, Berlin. Twenty First Century. National Union of Civil and Iversen, L. and Sabroe, S. (1988) Psychological well-being Public Servants, London. among unemployed and employed people after a comNext Steps Team (1996) Next Steps Briefing Note. pany closedown: a longitudinal study. Journal of Social February, Cabinet Office, London. Issues 44, 141-152. Iversen, L., Sabroe, S. and Damsgaard, M. T. (1989) Next Steps Team (1997) Next Steps Briefing Note. March, Cabinet Office, London. Hospital admissions before and after shipyard closure. Owens, C. (1966) Sick leave among railwaymen threatened British Medical Journal 299, 1073-1076. by redundancy: a pilot study. Occupational Psychology Jacobsen, K. (1972) Afskedigelse og sygelighed. Ugeskrift 40, 43-52. for laeger 134, 352-354. Jacobson, B., Amos, A. and Aghi, M. (1989) World No Platt, S. D. and Kreitman, N. (1990) Long term trends in parasuicide and unemployment in Edinburgh 1968Tobacco Day: a challenge for women's health. Lancet 1987. Social Psychiatry and Psychiatric Epidemiology 25, 334, 1193-1194. 56-61. Jenkins, R., MacDonald, A., Murray, J. and Strathdee, G. (1982) Minor psychiatric morbidity and the threat of Plewis, I. (1985) Analysing Change." Measurement and Explanation Using Longitudinal Data. Wiley, Chichester. redundancy in a professional group. Psychological Schnall, P. L., Landsbergis, P. A., Pieper, C. F., Schwartz, Medicine 12, 799-807. J., Dietz, D., Gerin, W., Schlussel, Y., Warren, K. and Kasl, S. V. and Cobb, S. (1970) Blood pressure changes in Pickering, T. G. (1992) The impact of anticipation of men undergoing job loss: a preliminary report. job loss on psychological distress and worksite blood Psychosomatic Medicine 32, 19-38. pressure. American Journal of Industrial Medicine 21, Kessler, R. C., House, J. S. and Turner, J. B. (1987) 417-432. Unemployment and health in a community sample. Stansfeld, S. A. and Marmot, M. G. (1992) Social Journal of Health and Social Behavior 28, 51-59. class and minor psychiatric disorder in British civil Kessler, R. C., Turner, J. B. and House, J. S. (1988) servants: a validated screening survey using the Effects of unemployment on health in a community surGeneral Health Questionnaire. Psychological Medicine vey: main, modifying and mediating effects. Journal of 22, 739-749. Social Issues 4, 69-85. Layton, C. (1987) Levels of state anxiety for males facing Studnicka, M., Studnicka-Benke, A., Wogerbauer, G., Rastetter, D., Wenda, R., Gathmann, P. and Ringel, redundancy and subsequently reporting to be employed E. (1991) Psychological health, self-reported physical or unemployed. Perceptual and Motor Skills 65, 53-54. 254 Jane E. Ferrie et al. health and health service use. Risk differential observed after one year of unemployment. Social Psychiatry and Psychiatric Epidemiology 26, 86-91. Theorell, T. (1974) Life events before and after the onset of a premature myocardial infarction. In Stressful Life Events, eds B. S. Dohrenwend and B. P. Dohrenwend, pp. 101-117. Wiley, London. Warr, P. and Jackson, P. (1987) Adapting to the unemployed role: a longitudinal investigation. Social Science & Medicine 25, 1219-1224.