Anticipation of Job Loss or Job Change and Cardiovascular Risk Factors : a Study of Retiring Self-Defense Officials in Japan

Self-defense officials in Japan are to retire at the age of early 50s. This unique situation prompted the authors to investigate whether preexisting morbid conditions cause any difficulty in finding a post-retirement job and whether anticipation of job loss or job change, as measured by the status of a post-retirement job and months remaining until retirement, was related to biological cardiovascular risk factors. The subjects were 2,228 male self-defense officials who received a preretirement health examination at three Self-Defense Forces Hospitals from 1991 to 1992; the period in time remaining until retirement ranged from 1-40 months (median 12 months), and 62% had one year or less until the retirement. The defined preexisting illnesses included a wide range of chronic, non-communicable diseases. Overall, the preexisting illness was unrelated to the determination of a post-retirement job. In men having 6 months or less until retirement, however, the security of post-retirement job was less frequent when they had the preexisting illness, especially cardiovascular diseases. In 1,839 men excluding those with the preexisting illness, the period until retirement was not adversely related to obesity, blood pressure, serum lipids, serum uric acid, or glucose intolerance whether the post-retirement job had been secured or not. The findings suggest that the preexisting illness decreases the chance of obtaining a post-retirement job, but do not provide any evidence that anticipation of job loss or job change due to early retirement exerts an adverse effect on biological cardiovascular risk factors. J Epidemiol, 1997 ; 7 : 161-166.

Health effects of unemployment have been a matter of concern in many industrialized countries.Unemployment has been related to increased risk of coronary heart disease although it is inconclusive whether the association is causal or due to health-related selection to unemployment1-3) Unemployment was shown to be associated with self-reported hypertension's or raised blood pressure 5).Potential adverse effects on health may emerge in the early process of unemployment before actual job loss, namely the anticipation phase of job loss or job change.A limited number of studies reported that threatening unemployment or job change was associated with self-rated deterioration in health condition 6>, increased numbers of medical consultations7), and even elevation in serum total cholesterol (TC) concentrations 8).
The preretirement health examination program of the Japan Self-Defense Forces (SDF) provides a unique opportunity of investigating the relation between anticipation of job loss or job change and cardiovascular risk factors.The retirement age of self-defense officials is early 50s, which is much earlier than that of the average employees in Japan, and thus fording a next job after retirement seems indispensable for them to live another quarter of a century and mostly to maintain their family.Early retirement without securing a post-retirement job or with inevitable job change would result in psychological stress which may exert adverse effects on health.The present study examined whether preexisting morbid conditions caused diffi-culty in finding a next job and whether anticipation of job loss or job change, as measured by status of the next job and the length in time until retirement, was related to cardiovascular risk factors.

SUBJECTS AND METHODS
Study subjects consisted of 2,228 male self-defense officials admitted consecutively to the SDF Fukuoka and Kumamoto Hospitals between January 1991 and December 1992 and to the SDF Sapporo Hospital from April to December 1992 for a preretirement health examination, which is part of a nationwide program offering free comprehensive medical examinations in a five-day admission for those retiring from the SDF.The three hospitals cover most of those in Kyushu and Hokkaido.The age ranged from 49 to 55 years old, and 91% of the subjects were at the age of 51 or 52 years.
Details of the health examination have been described elsewhere 9 -12).In brief, fasting venous blood was drawn for the determinations of serum biochemicals including TC, triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and uric acid.At each of the three hospital laboratories, serum concentrations of TC, TG, and uric acid were enzymatically determined with auto analyzers, and HDL-C was assayed using different precipitation methods 9-11).These biochemical measurements were not standardized among the hospitals, but coefficients of variance of the measurements did not vary much with hospitals: TC 17%, TG (natural logarithm) 10-11%, HDL-C 24-27%, and uric acid 20-23%.Blood pressure was determined by ward nurses on the right arm after at least 5 minute rest on the bed.A single reading on the first day of admission was recorded for the present study 10).A 75-g oral glucose tolerance test was a routine procedure, and subjects were classified into normal, impaired, and diabetic as regards glucose tolerance in accordance with the World Health Organization criteira13).Besides body mass index (BMI), waist-hip circumference ratio (WHR) was determined as an index of obesity10,12).Current medication and past medical history were ascertained by physicians or nurses.
A self-administered questionnaire queried the status of postretirement job and the number of months left until retirement as well as lifestyle characteristics such as smoking, alcohol drinking, physical activity, and dietary habits 9 -12) The questionnaire included no specific questions on stress asociated with job loss or job change.The subjects were arbitrarily classified into three groups with respect to the remaining period until retirement (more than 12 months, 7 to 12 months, and 6 months or less).It was assumed that anticipation of job loss or job change was greater when the remaining months were shorter.The period until retirement among those having secured a next job is a surrogate measure of anticipation of job change.The analysis regarding to the period until retirement was thus done for men with a post-retirement job and those without separately.Alcohol consumption among current drinkers, recreational exercise and the consumption of a limited number of food and beverage items were ascertained on average over the past one year.
Preexisting cardiovascular and non-cardiovascular illnesses were defined arbitrarily as follows; cardiovascular morbid conditions included a prior history of cerebrovascular diseases (n=11), myocardial infarction (n=10) or coronary angioplasty (n=1), known valvular heart disease or myocardial disease (n=18), angina pectoris on medication (n=28), known diseases of the aorta and peripheral artery (n=3), and hypertension on medication (n=21 1).Non-cardiovascular conditions included a prior history of malignant neoplasm (n=26), diabetes mellitus under dietary or drug treatment (n=60), hyperthyroidism (n=5), chronic bronchitis, asthma or bronchiectasis (n=8), ulcerative colitis (n=2), chronic hepatitis or liver cirrhosis (n=27), nephritis, nephrotic syndrome or chronic renal failure (n=6), collagen diseases (n=3), hyperlipidemia on medication (n=19), and hyperuricemia on medication (n=29).Several men had two or more of these preexisting morbid conditions.A total of 389 men with the defined preexisting illnesses were excluded in the analysis on the relation of the status of post-retirement job and the period until retirement to blood pressure, serum lipids, serum uric acid, and glucose intolerance.
Chi-square test with Yates' correction was used to assess the difference in proportions.Adjustment for covariates in the analysis of dichotomous variables was made by the Mantel-Haenszel method or multiple logistic regression analysis.The trend for proportions in relation to the period until retirement was tested by the Mantel-Haenszel method or logistic regression models assigning ordinal values.Analysis of covariance was used for the comparison of adjusted means.Multiple linear regression analysis was used to assess a decreasing or increasing trend in adjusted means by assigning ordinal values to the three categories of the period until retirement.Hospital and high rank (see below) were always taken into account in the analysis regarding biological cardiovascular risk factors.P values less than 0.05 (two-sided) were regarded as statistically significant.All the computations were done by using the SAS statistical package (SAS Institute, Inc., Cary, North Carolina).
Table 1 shows the association between the preexisting illness and the status of post-retirement job.Overall, the presence of the preexisting illness was not materially associated with the status of post-retirement job (p = 0.66).This was also the case in the analysis neglecting hypertension, hyperuricemia, and hyperlipidemia each under drug treatment; proportions of men having secured a post-retirement job were 4.2% (8/192) among those with the preexisting morbid conditions and 5 .6%(114/2,036) among those without (p = 0.50).In men who had 6 months or less until retirement, however, the security of postretirement job was less frequent among those with the preexisting illness (p = 0.12), especially with cardiovascular diseases (p = 0.05), although the difference was not statistically signifi-Table 1. Percentages of men having secured a post-retirement job according to the presence of preexisting illness and period until retirement (n = 2 ,228).
Values in parentheses are numbers of men having secured a post-retirement job in the numerator and total numbers in the denominator.
Including men with concomittent non-cardiovascular diseases.
Table 2. Adjusted means (SE) of obesity indices, blood pressure, serum lipids, and serum uric acid according to the period until retiremnt.
BP: blood pressure, a Based on analysis of covariance controlling for hospital and high rank .b Based on multiple linear regression analysis using an ordinal variable for the remaining months until retirement.
One man was excluded because of an extrordinarily high value (1,049 mg/dl). cant.
The SDF rank was strongly related to the status of postretirement job; proportions of men having secured a postretirement job were 5.0% (71/1,420) for the low ranks, 4.5% (25/554) for the middle ranks, and 10.2% (26/254) for the high ranks (p = 0.002).Thus the high rank was always controlled for in the subsequent analyses.
As shown in Table 2, the remaining period until retirement was not materially related to obesity levels, blood pressure, serum lipids, and serum uric acid whether the post-retirement job had been secured or not.In general, the remaining period (treated as an ordinal variable) showed no measurable interaction with either hospital or high rank.There were only suggestive interactions between the period until retirement and high rank as regards body mass index (p = 0.06) and wait-hip ratio (p = 0.08).For ease of presentation, we therefore chose to pre-sent results based on multivariate models without the interaction term.
Table 3 compares obesity levels, blood pressure, serum lipids, and serum uric acid between men with the next job secured and those without.The period remaining until retirement was also controlled for in this analysis.The status of post-retirement job did not show any material interaction with either the remaining period or hospital or rank.Men having secured a next job had higher levels of TG (p = 0.02) and uric acid (p = 0.06) than those not having secured the next job.
The frequency of glucose intolerance, diabetes mellitus and impaired glucose tolerance combined, did not vary with either the status of post-retirement job or the remaining period until retirement.Proportions of glucose intolerance, directly adjusted for hospital and high rank using the whole subjects as standard, were 15.2% (n = 17) among men with a post-retirement job and 17.8% (n = 298) among those without (p = 0.97).The adjusted proportions according to the remaining months in the latter group were 17.6% (n = 122), 17.5% (n = 152), and 15.4% (n = 24) for the intervals until retirement of 13+, 7-12, and <6 months, respectively (trend p = 0.39).The numbers were too small to calculate such adjusted proportions in men having secured a next job, but the trend was statistically null (p = 0.98).
Table 4 shows selected lifestyle characteristics according to the period until retirement for men with a post-retirement job and those without separately.Recreational exercise tended to be more frequent among men having the shortest period until retirement when they had secured a next job.No association was noted for smoking, alcohol drinking, green-tea use or coffee use.When the comparison was made between men with a secured post-retirement job and those without, daily users of green tea were slightly more frequent in the latter (76.7% versus 84.8%, p = 0.04).
Results shown in Tables 2 and 3 did not change after additional adjustment for cigarette smoking (never, former, and current smokers consuming <20 or 20+ cigarettes per day), alcohol use (never, former, current drinkers consuming <30, 30-59 or 60+ ml per day), and leisure-time physical activity (times per week), green tea (cups per day) and coffee (cups per day).Fully adjusted means of TG (natural logarithm) were 4.86 among men with a secured job and 4.74 among those without (p = 0.02).The corresponding values for uric acid were 6.16 and 5.90, respectively (p = 0.04).There was virtually no change in the association with glucose intolerance after adjustment for the above-mentioned factors by multiple logistic regression analysis.

DISCUSSION
Only 5% of men had determined a post-retirement job, and more than 80% still had more than 6 months until retirement.It would thus be difficult to find any measurable effect of the preexisting illness on the determination of a post-retirement job in the whole subjects.In men whose retirement was impending, however, the preexisting illness showed a tendency of conferring a poor prospect of reemployment.The finding suggests that ill health is a source of selection to unemployment.It should be noted that our study cannot address the relation between ill health and long-term unemployment.A longitudinal study of retired men of the SDF is warranted to understand health-related selection to unemployment and reemployment among men of advanced middle age.
Our study did not demonstrate any adverse association between anticipation of job loss or job change and biological cardiovascular risk factors.A six-year follow-up study of middle-aged shipyard workers in Sweden 10 showed a significantly greater increase in serum TC in the group threatened with unemployment due to plant closure than the control group , whereas there was no differential change in body weight, blood pressure, serum TG, and blood glucose between the study and control groups.In a study of male employees experiencing a permanent plant closing in the United States, Kasl and Cobb14) reported an approximately 10% decrease in serum TC one year later when most of the study subjects were in a stable reemployment.In that study14), however, TC levels in the study group were almost the same as those in the control group in the periods of job loss anticipation and actual unemployment, and body weight also decreased in parallel with the change of serum TC.
While unemployment, especially long-term unemployment, may be related to sustained hypertension4,5), it is not clear whether anticipation of job loss or job change results in a measurable elevation in blood pressure.As compared with blood pressure in a stable employment, no increase was observed for blood pressure in male employees of a shipyard threatened with unemployment in Sweden8) and in male and female employees of a brokerage firm undergoing massive layoffs in the United States15).Our reported finding on blood pressure is thus in agreement with these observations.However, these findings including ours cannot exclude a transient, short-term effect of anticipation of job loss or job change.A temporary, statistically insignificant increase in diastolic, not systolic, blood pressure (3.5 nunHg) was noted among 13 men who had been notified of a transfer to another company in the previous week, and their blood pressure returned to the baseline level several months later15).Among men having experienced unemployment, Kasl and Cobb14) also reported a small decrease in blood pressure 6-12 months after the plant closure, i.e. in the phase of reemployment.But, again, the investigators did not measure the baseline blood pressure and did not directly address an elevation in blood pressure associated with job loss anticipation 14).
Unexpected was the observation that serum levels of TG and uric acid were higher among those having secured a postretirement job than those not having.The difference was not explained by the different length in time until retirement between the two groups.The findings were not ascribed to lifestyle covariates such as smoking, alcohol use, and physical activity.Additional adjustment for BMI and WHR did not change the difference (data not shown).High consumption of alcohol has been related to raised levels of both serum TG and uric acid 9.17).Our measure of alcohol consumption was the average intake in the past year.We did not specifically ascertain the recent consumption of alcohol.Men having secured a post-retirement job may have had higher alcohol consumption in the recent period than usual.The findings may be simply due to chance because not a few statistical tests were performed.
We did not directly evaluate subjective perception of psy-chological stress associated with job change or job loss.It is suggested that more stress seems to be experienced by the older work force when obliged to face unemployment and job change because of their poor reemployment prospect and reduced adaptability 1).Financial strain associated with unemployment is implicated being a primary factor in negative health effects among the unemployed16).It is thus of particular relevance to mention unique aspects of our study subjects.
Besides being relatively young at the time of retirement, they have expected the retirement as scheduled in the SDF.
Because of job security in the past decades, the SDF officials may be affordable financially even in the state of temporary unemployment.Although we have no relevant data, the support for reemployment may be better organized and more systematic for men retiring from the SDF than for other employees in Japan.Furthermore, unemployment rate in Japan has been much lower than the figures in Western countries; for instance, the rates in 1991 were 2.1% in Japan, 6.7% in the United States, and 8.0% in England and Wales18).Taken together these aspects unique to the SDF officials and the low rate of unemployment in Japan in general, psychological stress associated with job loss or job change due to the early retirement may not cause much stress leading to deterioration in biological cardiovascular risk factors in our study population.It is possible that stress associated with job loss or job change may differ as regards job type before retirement.Information on job type was not available, however.

Table 3 .
Adjusted means (SE) of obesity indices, blood pressure, serum lipids, and serum uric acid for men with post-retirement job and those withouta.
BP: blood pressure.a Based on analysis of covariance controlling for hospital, high rank, and remaining months until retirement (13+, 7-12, or <_ 6 months).b One man was excluded because of an extrordinarily high value (1,049 mg/dl).

Table 4 .
Lifestyle characteristics according to the remaining period until retirement.
Values are percentages.a Based on the Mantel-Haenszel test , b At least once per week .