Beruflich Dokumente
Kultur Dokumente
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
Watanabe et al JOEM Volume 59, Number 3, March 2017
employees were (a) employees who worked in the sampled work- scale was measured as the ordinal variable, classified into seven
sites, and (b) full-time or dispatched employees. There were no categories (10 to 29, 30 to 49, 50 to 99, 100 to 299, 300 to 999,
exclusion criteria among either worksites or employees. First, 1000 to 4999, and more than or equal to 5000 employees). The
worksite-level questionnaires were sent to those responsible for measurement of industry was based on the ‘‘Japan Standard Industrial
‘‘industrial safety and health’’ at the worksites. Second, individual Classification,’’37 classified into 20 categories. Because worksites that
employees were randomly sampled from the same worksites as were conducted improvements of their work environments could also adopt
sampled initially, and were asked to respond to employee-level other approaches for improving work-related stress, other mental
questionnaires. For the employee-level data collection, the number health measures at the worksites were measured using 15 dichotomous
of employees sampled depended on the number employed at the items. The responsible persons at the worksites indicated which, if any,
worksites (5, 10, 15, or 20 employees were sampled from worksites of the following items their worksites conducted: (1) location of the
employing 10 to 29, 30 to 299, 300 to 499, or more than or equal to safety and health committee for mental health care, (2) planning and
500 employees, respectively). The MHLW calculated the survey action for solutions to mental health problems, (3) selection of persons
weights to represent the total population of Japanese employees, who were responsible for mental health care, (4) educating and
accounting for non-responses, and sampling probability.38 giving information to employees, (5) educating and giving infor-
mation to managers, (6) educating and giving information to health
Variables and Measurements care staff at the worksites, (7) mental health care during consul-
We measured whether the worksites conducted improvements tation after a health checkup, (8) facilitation of the off-site oppor-
of the work environment as an exposure variable, employee-level tunities for mental health care, (9) facilitation of the on-site
work-related stress as an outcome variable, and other confounding opportunities for mental health care, (10) stress check using
factors as covariates. All variables were measured by self-reported questionnaires, (11) support for sick leave employees to return
questionnaires by company representatives (usually human resource to work, (12) mental health care utilizing community occupational
personnel) of the worksites or the employees. health centers, (13) mental health care utilizing prefectural occu-
pational health promotion centers, (14) mental health care utilizing
Improvement of the Work Environment medical institutions, and (15) mental health care utilizing other
Each worksite representative was asked if his/her worksite off-site institutions.
conducted any measure for the mental health of employees. If he/she As employee-level confounders, sex, age, employment status
responded positively, then he/she was further asked about the (regular, contract, part-time, dispatched, and daily), job type, shift type
improvements. Improvement of the work environment was measured (day, night, rotation shift, alternate, and others), and working hours per
by asking a dichotomous question in the worksite-level questionnaire: day were measured. Job type was broken into seven categories, (1)
‘‘Does your worksite conduct assessment and improvements of the sales, (2) services, (3) transport or operation, (4) stationary, construc-
work environment?’’ When the company representative chose tion, or wrapping, (5) security, (6) management, and (7) other.
the ‘‘yes,’’ the company was coded as conducting improvements of
the work environment. Improvement of the work environment could Analysis
include organizational-level interventions at work, such as changing A multilevel multinominal logistic regression analysis39
material conditions (eg, vibrations, noise, chemical substances, or was conducted because of the two-level information and hier-
ergonomics), time-related conditions (eg, work speed, shifts, dead- archical data in this study. We used a sample which consisted of
lines, pace of work, or breaks), psychosocial factors, and processes employees nested in worksites, and data which consisted of
and procedures of work (eg, methods of work, order of tasks, team worksite-level and employee-level information. If we had applied
organization, or structure of the hierarchy).4 conventional regression analyses for worksite-level variables, we
would have not kept the assumption of independency of sampling
Work-Related Stress data.40 We transformed individual work-related stress into four
Each employee was asked if he/she had any work-related categories based on the number of work-related stresses chosen,
stress: ‘‘Do you feel strong anxiety, worry, and distress related to and used this as the outcome: 0 (none), 1 (one), 2 (two), 3 (three).
your work and work life?’’ If he/she agreed, he/she was then asked Of the covariates, each confounding variable was transformed
to choose up to three items from a list of 11 items of cause-specific into a dummy variable, except for other mental health measures
work-related stress: (1) qualitative workload, (2) quantitative work- at the worksites: worksite scale (10 to 49 [reference group]),
load, (3) suitability for work, (4) human relationships in the work- industry (manufacturing [reference group]), sex (men [reference
place, (5) promotion and salary increase, (6) personnel reshuffling, group]), age (less than or equal to 29 [reference group]), employ-
(7) stability of employment, (8) future prospects of the company, (9) ment status (regular [reference group]), job type (not manager
job after retirement and old age, (10) experience with accidents and [reference group]), shift type (daytime [reference group]),
disasters, and (11) other. The number of work-related stresses was and working hours per day (less than 8 [reference group]).
used as outcomes: 0 (none), 1 (one), 2 (two), 3 (three). Although the Other mental health measures at the worksites were transformed
limit on the number was three, respondents who selected three types into a numerical variable—the number of kinds of mental
of stress might have had more than three items; in fact, some health measures.
employees went against the instructions and chose more than three We estimated an unconditional model (model 1), a crude
kinds of work-related stress. Our interpretation was that respondents conditional model (model 2), and an adjusted conditional model
with three or more types of work-related stresses had a severe level (model 3) in the analysis. An equation for the adjusted conditional
of work-related stress because individuals who suffered from model (model 3) was explained as follows.
multiple sources of work-related stress were supposed to have a Level 1 (employee-level)
greater risk of mental and physical problems, as the iso-strain model !
Prob R i j ¼ m
suggests that the combination of multiple work-related stressors log ¼ hmi j
Prob R i j ¼ 0½none
causes health problems.10
¼ b0 jðmÞ þ b1 jðmÞ ðWomenÞij þ b2 jðmÞ ðAge 3039Þij þ b3 jðmÞ
ðAge 4049Þij þ b4 jðmÞ ðAge 5059Þij þb5 jðmÞ ðAge 60Þij þ b6 jðmÞ
Covariates ðNot regular employmentÞij þ b7 jðmÞ ðManagerÞij þ b8 jðmÞ
As worksite-level confounders, we measured worksite scale, ðNot daytime shiftÞij þ b9 jðmÞ ðWorking hours 810Þij þ b10 jðmÞ
industry, and other mental health measures at the worksites. Worksite ðWorking hours 10Þij
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
JOEM Volume 59, Number 3, March 2017 Improvement of the Work Environment and Stress
Level 2 (worksite-level)
b0 j ¼ g 00ðmÞ þg 01ðmÞ ðWorksite scale 5099Þj þ g 02ðmÞ ðWorksite scale 100299Þj
þg 03ðmÞ ðWorksite scale 300999Þj þ g 04ðmÞ ðWorksite scale 1;000Þj
þg 05ðmÞ ðIndustry serviceÞj þ g 06ðmÞ ðIndustry othersÞj þ g 07ðmÞ
ðthe N of other mental health measuresÞj þ g 08ðmÞ
ðimprovement of the work environmentÞj þ m0 jðmÞ
bq jðmÞ ¼ g q0ðmÞ;
mo j½m e N 0; t 00½m
Level 2 (worksite-level)
b0 j ¼ g 00 þ g 01 ðWorksite scale 5099Þj þg 02 ðWorksite scale 100299Þj þg 03
ðWorksite scale 300 999Þj þg 04 ðWorksite scale 1;000Þj þ g 05
ðIndustry serviceÞJ þ g 06 ðIndustry othersÞj þ g 07
ðthe N of other mental health measuresÞj þ g 08
ðimprovement of the work environmentÞj þ m0 j
bq j ðmÞ ¼ g q0 ðmÞ;
q ¼ 1:::10
ðmo j ½mÞ N ð0; t 00 ½mÞ
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
Watanabe et al JOEM Volume 59, Number 3, March 2017
N (%)
Employee—Level Variables Total (N ¼ 9908) Men (N ¼ 6356) Women (N ¼ 3552) Missing (%)
for one, Exp [g] ¼ 0.91 for two, Exp [g] ¼ 0.93 for three). The P ¼ 0.889). The adjusted conditional model (model 3) also indicated
number of other mental health measures was not significantly a slightly negative, but non-significant relationship (g08 ¼ 0.04,
associated with work-related stress among the total sample (Exp 95%CI, 0.13, 0.06, P ¼ 0.452).
[g] ¼ 1.00 for one, Exp [g] ¼ 1.00 for two, Exp [g] ¼ 1.00 for three). The subgroup analyses on stratified sex indicated different
Excluding this variable from the model did not alter the findings: the results from that of the total sample (Table 4). Among 6356 men
associations between improvement of work environment and work- nested in 986 worksites, improvement of the work environment
related stress were non-significant (Exp [g] ¼ 0.96 [95%CI, 0.82, showed a significant and negative association with having three or
1.12] for one, Exp [g] ¼ 0.93 [95%CI, 0.79, 1.10] for two, Exp more kinds of work-related stress (Exp [g] ¼ 0.83, 95%CI, 0.70,
[g] ¼ 0.94 [95%CI, 0.80, 1.11] for three) (data available upon 0.98, P ¼ 0.032 in the crude conditional model [model 2]; Exp
request). [g] ¼ 0.80, 95%CI, 0.65, 0.99, P ¼ 0.036 in the adjusted conditional
Table 3 shows results of the sensitivity analysis, treating model [model 3]). The number of other mental health measures was
work-related stress as a continuous variable. In the unconditional not significantly associated with work-related stress (Exp [g] ¼ 1.00
model (model 1), a random effect of the intercepts (t00) of work- for one, Exp [g] ¼ 1.00 for two, Exp [g] ¼ 1.00 for three). Excluding
related stress was significant (t00 ¼ 0.15, 95%CI, 0.12, 0.18, the variable from the model did not alter the findings: the associ-
P < 0.001). In the crude conditional model (model 2), improvement ations between improvement of work environment and work-related
of the work environment showed a non-significant association with stress were not significant (Exp [g] ¼ 0.91 [95%CI, 0.77, 1.09] for
the number of work-related stress (g08 ¼ 0.01, 95%CI, 0.08, 0.09, one, Exp [g] ¼ 0.91 [95%CI, 0.75, 1.10] for two, Exp [g] ¼ 0.81
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
JOEM Volume 59, Number 3, March 2017 Improvement of the Work Environment and Stress
TABLE 2. Relationships Between Improvement of the Work Environment and Work-Related Stress Among Employees (Total
Sample)
Model 2 (Crude Conditional Model) Nj ¼ 1026, Ni ¼ 9908 Model 3 (Adjusted Conditional Model) Nj ¼ 1026, Ni ¼ 9887
For One (Compared For Two (Compared For Three (Compared For One (Compared For Two (Compared For Three (Compared
With None) With None) With None) With None) With None) With None)
Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI)
Employee–level
Gender
Men – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
Women (g10) – – – 1.33 (1.16, 1.54) 1.30 (1.14, 1.49) 1.07 (0.94, 1.21)
Age group
29 yrs – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
30–39 yrs (g20) – – – 1.11 (0.93, 1.33) 1.28 (1.07, 1.54) 1.29 (1.10. 1.52)
40–49 yrs (g30) – – – 1.09 (0.90, 1.32) 1.44 (1.20, 1.72) 1.51 (1.28, 1.77)
50–59 yrs (g40) – – – 0.99 (0.80, 1.21) 1.30 (1.07, 1.58) 1.15 (0.96, 1.37)
60 yrs (g50) – – – 0.81 (0.60, 1.09) 0.64 (0.46, 0.87) 0.45 (0.34, 0.60)
Employment status
Regular – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
Not regular (g60) – – – 0.87 (0.72, 1.05) 0.89 (0.75, 1.05) 0.83 (0.70, 0.98)
Job type
Not manager – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
Manager (g70) – – – 1.11 (0.93, 1.32) 0.90 (0.77, 1.06) 0.92 (0.80, 1.06)
Shift type
Daytime – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
Not daytime (g80) – – – 1.15 (0.93, 1.42) 1.22 (1.00, 1.49) 1.27 (1.07, 1.52)
Working hours per day
<8 hr – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
8–10 hr (g90) – – – 1.13 (0.98, 1.31) 1.32 (1.15, 1.51) 1.45 (1.28, 1.65)
10 hr (g100) – – – 1.72 (1.34, 2.20) 2.27 (1.83, 2.81) 3.23 (2.65, 3.94)
Worksite–level
Worksite scale
10–49 – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
50–99 (g01) – – – 1.08 (0.82, 1.42) 1.12 (0.83, 1.51) 1.25 (0.96, 1.63)
100–299 (g02) – – – 1.11 (0.86, 1.42) 1.41 (1.10, 1.80) 1.25 (1.00, 1.57)
300–999 (g03) – – – 1.05 (0.83, 1.34) 1.44 (1.14, 1.82) 1.11 (0.88, 1.40)
1000 (g04) – – – 1.05 (0.83, 1.34) 1.29 (1.01, 1.65) 1.05 (0.83, 1.32)
Industry
Manufacturing – – – 1.00 (reference) 1.00 (reference) 1.00 (reference)
Services (g05) – – – 0.96 (0.79, 1.18) 0.78 (0.64, 0.96) 0.92 (0.75, 1.13)
Others (g06) – – – 0.94 (0.82, 1.90) 0.82 (0.70, 0.95) 0.90 (0.77, 1.04)
The N of other mental – – – 1.00 (0.98, 1.03) 1.01 (0.98, 1.03) 1.00 (0.98, 1.03)
health measures (g07)
Improvement of the work environment
No 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Yes (g08) 0.98 (0.84, 1.14) 0.99 (0.84, 1.16) 0.99 (0.84, 1.16) 0.94 (0.79, 1.12) 0.91 (0.76, 1.10) 0.93 (0.77, 1.12)
Intercept (g00) 0.99 (1.06, 0.92) 0.76 (0.84, 0.69) 0.30 (0.38, 0.23) 1.25 (1.52, 0.98) 1.31 (1.58, 1.05) 0.80 (1.04, 0.55)
AIC 26038.15 25607.39
McFadden pseudo R2 0.00 0.02
AIC, Akaike information criteria; CI, confidence interval; Exp, exponential; Ni, number of workers; Nj, number of worksites. Robust maximum likelihood estimation was used
using Mplus 7.4.
P < 0.05.
P < 0.01
P < 0.001.
[95%CI, 0.67, 0.97] for three) (data available upon request). [model 2]). However, the association was not significant after adjust-
When work-related stress was treated as a continuous variable, ing for the covariates (Exp [g] ¼ 1.29, 95%CI, 0.96, 1.74, P ¼ 0.096 in
improvement of the work environment had a marginally significant the adjusted conditional model [model 3]). The number of other
negative relationship with individual work-related stress in the mental health measures was not significantly associated with work-
crude model (model 2, g08 ¼ 0.08, 95%CI, 0.17, 0.01, related stress (Exp [g] ¼ 1.02 for one, Exp [g] ¼ 1.02 for two, Exp
P ¼ 0.075) and in the adjusted conditional model (model 3, [g] ¼ 1.01 for three) (data available upon request). The association
g08 ¼ 0.10, 95%CI, 0.20, 0.00, P ¼ 0.059). between improvement of the work environment and work-related
Among 3552 women nested in 905 worksites, improvement of stress remained even after excluding the variable from the model (Exp
the work environment showed a significant and positive association [g] ¼ 1.05 [95%CI, 0.80, 1.38] for one, Exp [g] ¼ 1.00 [95%CI, 0.75,
with having three or more kinds of work-related stress (Exp [g] ¼ 1.41, 1.33] for two, Exp [g] ¼ 1.35 [95%CI, 1.04, 1.76] for three) (data
95%CI, 1.09, 1.82, P ¼ 0.009 in the crude conditional model available upon request). When work-related stress was treated as a
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
Watanabe et al JOEM Volume 59, Number 3, March 2017
TABLE 3. Sensitivity Analysis: Relationships Between Improvement of the Work Environment and Work-Related Stress Among
Employees (Total Sample)
Model 1 (Unconditional Model 2 (Crude Conditional Model 3 (Adjusted Conditional
Model) Nj ¼ 1026, Ni ¼ 9908 Model) Nj ¼ 1026, Ni ¼ 9908 Model) Nj ¼ 1026, Ni ¼ 9887
Employee–level
Gender (REF: men)
Women (g10) – – 0.04 (0.02, 0.10) 0.212
Age group (REF: 29 years)
30–39 yrs (g20) – – 0.13 (0.05, 0.21) 0.001
40–49 yrs (g30) – – 0.21 (0.13, 0.29) <0.001
50–59 yrs (g40) – – 0.07 (0.02, 0.16) 0.106
60 yrs (g50) – – 0.34 (0.46, 0.23) <0.001
Employment status (REF: regular)
Not regular (g60) – – 0.09 (0.16, 0.01) 0.034
Job type (REF: not manager)
Manager (g70) – – 0.04 (0.11, 0.03) 0.249
Shift type (REF: daytime)
Not daytime (g80) – – 0.13 (0.04, 0.22) 0.003
Working hours (REF: <8 hr)
8–10 hr (g90) – – 0.19 (0.13, 0.25) <0.001
10 hrs (g100) – – 0.56 (0.47, 0.65) <0.001
Worksite–level
Worksite scale (REF: 10–49)
50–99 (g01) – – 0.13 (0.01, 0.26) 0.066
100–299 (g02) – – 0.13 (0.02, 0.24) 0.022
300–999 (g03) – – 0.09 (0.03, 0.21) 0.130
1000 (g04) – – 0.05 (0.06, 0.17) 0.361
Industry (REF: manufacturing)
Services (g05) – – 0.07 (0.17, 0.03) 0.170
Others (g06) – – 0.08 (0.16, 0.01) 0.029
The N of other mental health measures (g07) – – 0.00 (0.01, 0.02) 0.570
Improvement of the work environment (REF: No)
Yes (g08) – 0.01 (0.08, 0.09) 0.889 0.04 (0.13, 0.06) 0.452
Intercept (g00) 1.36 (1.32, 1.39) <0.001 1.36 (1.32, 1.39) <0.001 1.10 (0.99, 1.23) <0.001
Random intercept (t00) 0.15 (0.12, 0.18) <0.001 0.15 (0.12, 0.18) <0.001 0.12 (0.10, 0.15) <0.001
Residual variance (eij) 1.44 (1.41, 1.47) <0.001 1.44 (1.41, 1.47) <0.001 1.39 (1.36, 1.43) <0.001
AIC 32,399.26 32,401.24 31,995.68
AIC, Akaike information criteria; CI, confidence interval; Ni, number of workers; Nj, number of worksites; REF, reference group in the analyses; Robust maximum likelihood
estimation was used using Mplus 7.4.
continuous variable, improvement of the work environment work-related stress, as compared with having no work-related stress.
showed a significant and positive association with work-related This result indicated that the prevalence of having severe work-
stress among women, in the crude conditional model (model 2, related stress could be reduced by approximately 20% by improving
g08 ¼ 0.15, 95%CI, 0.02, 0.28, P ¼ 0.021); it was not significant in the work environment (Exp [g] ¼ 0.80). This result was in line with
the adjusted model (model 3, g08 ¼ 0.09, 95%CI, 0.05, 0.24, previous intervention studies reporting that improving the work
P ¼ 0.219). environment could reduce work-related stress among men.33,42
However, since we used a representative sample of worksites, most
DISCUSSION of the improvements of the work environment were supposed to be
Conducting improvements in the work environment was not ‘‘everyday’’ activities, rather than those based on well-designed
significantly associated with employee-level work-related stress in the projects often involving outside experts as seen in the publi-
total sample. However, sex-stratified analyses indicated contrasting cations.19,43 It is important to note that the present study suggests
results for men and women. Among men, a significant negative that everyday activities that improve the work environment are
association was observed between improvement of the work environ- associated with reducing work-related stress of men nationwide
ment and severe levels (ie, three or more sources) of work-related in Japan.
stress. The association was not significant among women. Improve- On the other hand, worksite-level improvement of the work
ment of the work environment may be associated with the reduction of environment was not significantly associated with work-related
work-related stress among men, but not women, in Japan. These stress for women workers. Rather, a non-significant, but slightly
results were the first findings that monitored the effectiveness of positive, association was observed, after adjusting for the other
improving the work environment on work-related stress at the national covariates. As suggested by previous findings,27– 33 the effect of
level, and suggested that we should consider and address, the differ- improving the work environment, in its current form as conducted in
ence in its effectiveness between men and women in the real world. Japan, might be less clear for women. A sex difference in the effect
Among men, a significant negative relationship was observed of improving the work environment could be caused by several
between improvement of the work environment and having severe factors. One possible reason for this result was the difference in job
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
JOEM Volume 59, Number 3, March 2017 Improvement of the Work Environment and Stress
TABLE 4. Relationships Between Improvement of the Work Environment and Work-Related Stress Among Workers
(Gender-stratified)
Work-Related Stress as a Categorical Variable
Model 2 (Crude Conditional Model) Nj ¼ 986, Ni ¼ 6356 Model 3 (Adjusted Conditional Model)y Nj ¼ 986, Ni ¼ 6346
For One (Compared For Two (Compared For Three (Compared For One (Compared For Two (Compared For Three (Compared
With None) With None) With None) With None) With None) With None)
Men Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI)
No 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Yes (g08) 0.96 (0.80, 1.14) 0.97 (0.81, 1.17) 0.83 (0.70, 0.98) 0.92 (0.76, 1.12) 0.91 (0.74, 1.12) 0.80 (0.65, 0.99)
AIC 16,576.92 16,307.15
Model 2 (Crude Conditional Model) Nj U 905, Ni U 3552 Model 3 (Adjusted Conditional Model)y Nj U 904, Ni U 3541
For One (Compared For Two (Compared For Three (Compared For One (Compared For Two (Compared For Three (Compared
With None) With None) With None) With None) With None) With None)
Women
Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI) Exp (g) (95%CI)
No 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
Yes (g08) 1.07 (0.82, 1.41) 1.07 (0.79, 1.38) 1.41 (1.09, 1.82) 0.99 (0.72, 1.37) 0.94 (0.68, 1.30) 1.29 (0.96, 1.74)
AIC 9,426.97 9,327.52
Yes (g08) 0.08 (0.17, 0.01) 0.075 0.10 (0.20, 0.00) 0.059
AIC 20,962.03 20,696.35
AIC, Akaike information criteria; CI, confidence interval; Exp, exponential; Ni, number of workers; Nj, number of worksites. Robust maximum likelihood estimation was used
using Mplus 7.4.
y
Adjusted by age, employment status, job type, shift type, and working hours, worksite scale, industry, and the number of other mental health cares.
P < 0.05.
P < 0.01.
type categories between men and women. For example, Dahl- environment should be reviewed from a sex equality perspective and
Jorgensen and Saksvik28 argued that the effects of their organiz- then improved if necessary.
ational intervention might be decreased due to the lack of stability in There are several limitations in the study. Given that this
the workforce where participants were predominantly part-time study design was cross-sectional, a causal relationship could not be
women, and when they were not planning to stay in the job long addressed. A selection bias due to the low valid response rate
term. On the other hand, Mikkelsen and Saksvik30 suggested that (59.2% for the worksite-level surveys and 56.7% for the
worksite supervisors, who had an important role in the organiz- employee-level surveys) might underestimate the relationship
ational stress reduction program, were predominantly men, and between the improvement of the work environment and work-
would consider men a major target population if their subordinates related stress. For instance, worksites that conducted few mental
had been predominantly men. Women, therefore, are more likely to health measures, and whose employees felt severe work-related
be a minority in the worksites, and receive less benefit from stress, can be reluctant to participate in the survey. Because
improvement of the work environment. The other possible reason measurements of the exposure, outcome, and covariates were all
is that women were less affected by workplace psychosocial factors self-reported, and were not statistically validated, some systematic
than men did. For instance, a previous intervention study reported bias and random error were incorporated in the values. The coef-
that the intervention improved the psychosocial work environment, ficients may be confounded by factors that we could not measure in
but did not improve depression among nurses in Japan.27 The the study. Most plausible moderator/confounders can be types of
finding challenges the current national strategy promoting improve- improvement of the work environment: employee-participatory
ment of the work environment as a measure of worksite mental approach and supervisor-directed approach. Although both types
health in Japan.16 Programs and procedures for improving the work could be effective on work-related mental health,19,33 several
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
Watanabe et al JOEM Volume 59, Number 3, March 2017
previous studies30,44,45 recommended the employee-participatory 23. Karanika-Murray M, Biron C, editors. Derailed Organizational Interventions
for Stress and Well-Being: Confessions of Failure and Solutions for Success.
approach because employees in the workplace understand both the Dordrecht Heidelberg, New York London: Springer; 2015.
problems and their solutions.19 Many other confounders, which 24. Cox T, Karanika M, Griffiths A, Houdmont J. Evaluating organizational-level
determine the effects of improving the work environment, were also work stress interventions: beyond traditional methods. Work Stress. 2007;
suggested, such as lack of support from employers and manag- 21:348–362.
ers,46,47 and external conditions that were beyond their control.48 25. Egan M, Bambra C, Petticrew M, Whitehead M. Reviewing evidence on
These limitations need to be addressed in future research. complex social interventions: appraising implementation in systematic
reviews of the health effects of organizational-level workplace interventions.
J Epidemiol Community Health. 2009;63:4–11.
REFERENCES 26. Kobayashi Y, Kaneyoshi A, Yokota A, Kawakami N. Effects of a worker
1. Hurrell Jr JJ, McLaney MA. Exposure to job stress—a new psychometric participatory program for improving work environments on job stressors and
instrument. Scand J Work Environ Health. 1988;14(suppl 1):27–28. mental health among workers: a controlled study. J Occup Health. 2008;
2. Stansfeld S, Candy B. Psychosocial work environment and mental health—a 50:455–470.
meta-analytic review. Scand J Work Environ Health. 2006;32:443–462. 27. Uchiyama A, Odagiri Y, Ohya Y, Takamiya T, Inoue S, Shimomitsu T. Effect
3. Eguchi H, Tsuda Y, Tsukahara T, Washizuka S, Kawakami N, Nomiyama T. on mental health of a participatory intervention to improve psychosocial
The effect of workplace occupational mental health and related activities on work environment: a cluster randomized controlled trial among nurses. J
psychological distress among workers: a multilevel cross-sectional analysis. Occup Health. 2013;55:173–183.
J Occup Environ Med. 2012;54:939–947. 28. Dahl-Jorgensen C, Saksvik PO. The impact of two organizational interven-
4. Montano D, Hoven H, Siegrist J. Effects of organisational-level interventions tions on the health of service sector workers. Int J Health Sev. 2005;35:
at work on employees’ health: a systematic review. BMC Public Health. 529–549.
2014;14:135; doi: 10.1186/1471-2458-14-135. 29. Schrijnemaekers VJJ, van Rossum E, Candel MJJM, et al. Effects of
5. Semmer NK. Job stress interventions and the organization of work. Scand J emotion-oriented care on work-related outcomes of professional caregivers
Work Environ Health. 2006;32:515–527. in homes for elderly persons. J Gerontol B Psychol Sci Soc Sci. 2003;
58:S50–S57.
6. Wall TD, Clegg CW. A longitudinal field study of group work redesign. J
Occup Behav. 1981;2:31–49. 30. Mikkelsen A, Saksvik SO. Impact of a participatory organizational
intervention on job characteristics and job stress. Int J Health Serv.
7. Evans GW, Johansson G, Rydstedt L. Hassles on the job: a study of job 1999;29:871–893.
intervention with urban bus drivers. J Occup Behav. 1999;20:199–208.
31. Bourbonnais R, Brisson C, Vient A, Vézina M, Abdous B, Gaudet M.
8. Heaney CA, Proce FH, Rafferty J. Increasing coping resources at work: a Effectiveness of a participative intervention on psychosocial work factors
field experiment to increase social support, improve work team functioning, to prevent mental health problems in a hospital setting. Occup Environ Med.
and enhance employee mental health. J Organ Behav. 1995;16:335–352. 2006;63:335–342.
9. Karasek RA. Job demands, job decision latitude, and mental strain; implica- 32. Pryce J, Albertsen K, Nielsen K. Evaluation of an open-rota system in a
tions for job redesign. Adm Sci Q. 1979;24:285–308. Danish psychiatric hospital: a mechanism for improving job satisfaction and
10. Johnson JV, Hall EM. Job strain, workplace social support, and cardiovas- work-life balance. J Nurs Manag. 2006;14:282–288.
cular disease: a cross-sectional study of a random sample of the Swedish 33. Kawakami N, Araki S, Kawashima K, Masumoto T, Hayashi T. Effects
working population. Am J Public Health. 1988;78:1336–1342. of work-related stress reduction on depressive symptoms among
11. Siegrist J. Adverse health effects of high-effort/low-reward conditions. J Japanese blue-collar workers. Scand J Work Environ Health. 1997;
Occup Health Psychol. 1996;1:27–41. 23:54–59.
12. Leka S, Griffiths A, Cox T. Work Organization and Stress—Protecting 34. Shimizu Y, Makino S, Takata T. Employee stress status during the past
Workers Health Series No 3. Geneva: World Health Organization; 2004 , decade (1982–1992) based on a nation-wide survey conducted by the
Available at: http://www.who.int/occupational_health/publications/pwh3rev. Ministry of Labour in Japan. Ind Health. 1997;35:441–450.
pdf. Accessed October 31, 2016. 35. Tabuchi T, Colwell B. Disparity and trends in secondhand smoke exposure
13. International Labour Organization. Stress Prevention at Work Checkpoints— among Japanese employees, particularly smokers vs. non-smokers. PLoS
Practical Improvements for Stress Prevention in the Workplace. Geneva: ILO; One. 2016;11:e0152096.
2012, Available at: http://www.ilo.org/wcmsp5/groups/public/@ed_prote 36. Vandenbroucke JP, von Elm E, Altman DG, et al. For the STROBE
ct/@protrav/@safework/documents/instructionalmaterial/wcms_177108.pdf. Initiative. Strengthening the reporting of observational studies in
Accessed October 31, 2016. epidemiology (STROBE): explanation and elaboration. PLos Med. 2007;
14. UK Health Safety Executive. How to Tackle Work-Related Stress—A Guide 4:e297.
for Employers on Making the Management Standards Work. UK Health and 37. Ministry of Internal Affairs and Communications. Japan Standard Industrial
Safety Executive; 2009, Available at: http://www.hse.gov.uk/pubns/indg430. Classification. Japan: Ministry of Internal Affairs and Communications;
pdf. Accessed October 31, 2016. 2007. Available at: http://www.soumu.go.jp/toukei_toukatsu/index/seido/
15. Leka S, Jain A, Cox T, Kortum E. The development of the European sangyo/19index.htm. Accessed October 31, 2016.
framework for psychosocial risk management: PEIMA-EF. J Occup Health. 38. Tabuchi T, Hoshino T, Nakayama T. Are partial workplace smoking bans as
2011;53:137–143. effective as complete smoking bans? A national population-based study of
16. Kawakami N, Tsutsumi A. The stress check program: a new national policy smoke-free policy among Japanese employees. Nicotine Tob Res.
for monitoring and screening psychosocial stress in the workplace in Japan. J 2016;18:1265–1273.
Occup Health. 2016;58:1–6. 39. Raudenbush SW, Bryk AS. Hierarchical Linear Models: Applications and
17. Lamontagne AD, Keegei T, Louie AM, Ostry A, Landsbergis PA. A system- Data Analysis Methods. 2nd ed. Thounsand Oaks Calif: Sage Publications;
atic review of the job-stress intervention evaluation literature, 1990-2005. Int 2002.
J Occup Environ Health. 2007;13:268–280. 40. Ickes W, Duck S, editors. The Social Psychology of Personal Relationships.
18. Egan M, Bambra C, Thomas S, Petticrew M, Whitehead M, Thomson H. The New York: Wiley & Sons; 2000.
psychosocial and health effects of workplace reorganization. 1. A systematic 41. Muthén LK, Muthén BO. Mplus User’s Guide. Seventh ed. Los Angeles, CA:
review of organizational-level interventions that aim to increase employee Muthén & Muthén; 1998–2012 , Available at: http://www.statmodel.com/
control. J Epidemiol Commun Health. 2007;61:945–954. ugexcerpts.shtml. Accessed June 6, 2016.
19. Tsutsumi A, Nagami M, Yoshikawa T, Kogi K, Kawakami N. Participatory 42. DeJoy DM, Wilson MG, Vandenberg RJ, McGrath-Higgins AL, Griffin-
intervention for workplace improvements on mental health and job perform- Blake CS. Assessing the impact of healthy work organization intervention. J
ance among blue-collar workers: a cluster randomized controlled trial. J Occup Organ Psychol. 2010;83:139–165.
Occup Environ Med. 2009;51:554–563.
43. Migind K, Borg V, Flyvholm M, Sell L, Jepsen KF. A study of the
20. Yoshimura K, Kawakami N, Tsutsumi A, et al. Cost-benefit analysis of implementation process of an intervention to prevent work-related skin
primary prevention programs for mental health at the workplace in Japan problems in wet-work occupations. Int Arch Occup Environ Health.
[in Japanese]. San Ei Shi. 2013;55:11–24. 2006;79:66–74.
21. Reynolds S. Interventions: what works, what doesn’t? Occup Med (Lond). 44. Anderzen I, Arnetz BB. The impact of a prospective survey-based
2000;50:315–319. workplace intervention program on employee health, biologic stress
22. van der Klink JJ, Blonk RW, Schene AH, van Dijk FJ. The benefits of markers, and organizational productivity. J Occup Environ Med. 2005;
interventions for work-related stress. Am J Public Health. 2001;91:270–276. 47:671–682.
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited
JOEM Volume 59, Number 3, March 2017 Improvement of the Work Environment and Stress
45. Bourbonnais R, Brisson C, Vézina M. Long-term effects of an 47. Logan MS, Ganster DC. An experimental evaluation of a control
intervention psychosocial work factors among healthcare professionals in a intervention to alleviative job related stress. J Manage. 2005;31:
hospital setting. Occup Environ Med. 2011;68:479–486. 90–107.
46. Mattila P, Elo A, Kuosma E, Kylä-Stälä E. Effect of participative work 48. Rasmussen K, Glasscock DJ, Hansesen ON, Castensen O, Jepsen JF, Nielsen
conference on psychosocial work environment and well-being. Eur J Work KJ. Worker participation in change processes in a Danish industrial setting.
Organ Psychol. 2006;15:459–476. Am J Ind Med. 2006;49:767–779.
Copyright © 2017 American College of Occupational and Environmental Medicine. Unauthorized reproduction of this article is prohibited