Association Between Employment Status and Short Sleep Duration

Among Middle-Aged Japanese: the Survey on Time Use

and Leisure Activities

Mitsuya Maeda

1

& Ronald Filomeno

1

& Yumi Kawata

1

& Tomoyo Sato

1

& Koutatsu Maruyama

1,2

Ai Ikeda

#

International Society of Behavioral Medicine 2019

regularly employed,”“self-employed,” and “other.” Authors also defined a sleep duration shorter than 5 h per night as SSD.

Multivariable-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) of SSD using regular employment as the reference

group were calculated using the logistic regression model, adjusting for potential confounding factors, i.e., age, marital status,

education level, communication device use, and the amount of time spent on various daily activities.

Results The prevalence of SSD was 2.8% (1639/58,308) in men and 3.2% (1976/62,475) in women. The multivariable-adjusted

OR (95% CI) of SSD for the self-employed was 0.78 (0.65–0.92) in men and 1.78 (1.43–2.21) in women after adjustment for

potential confounding factors, i.e., age, marital status, education level, and communication device use. Further adjustment for the

time spent on daily activities revealed that the OR (95% CI) of SSD for the self-employed was 0.78 (0.65–0.92) in men and 1.89

(1.52–2.36) in women.

Conclusion Self-employed women had a higher prevalence of SSD. By contrast, self-employed men had a lower prevalence of

NHNSJ National Health and Nutrition Survey, Japan

OR Odds ratio

SSD Short sleep duration

STULA Survey on Time Use and Leisure Activities

US United States

Sleep duration is an important lifestyle factor that is closely

related to health [1]. Short sleep duration (SSD) is associated

with increased risk of mortality, diabetes, cardiovascular dis-

ease, coronary heart disease, and obesity [2]. The Center for

over the last three decades, resulting in a loss of 3 to 4 h of

sleep per week for both male and female workers [5]. The

potential reasons for this decline include changes in various

social environments, i.e., industrialization, urbanization, com-

puterization, and the increase in shift work [6]. Many high-

income countries have experienced these social environment

changes; however, the influence on sleep duration has been

inconsistent. A previous review showed that the average sleep

duration for adults had increased in seven countries, including

Bulgaria, Poland, Canada, France, Britain, Korea, and the

ment, lower levels of income, and workers with nighttime or

shift work [8–10]. A previous study sho wed that among

Latinos living in the US, individuals with full-time employ-

ment tended to have SSD compared to the unemployed [8]. In

Korea, workers with nighttime or shift work had a higher

prevalence of SSD than subjects working during the daytime

[10]. Cultural factors specific to Japanese workers include

long working hours, which may more strongly impact sleep

duration. Indeed, Japanese males work 10 h longer and fe-

males work 7 h longer per week than Americans [5], and the

crease their operational efficiency [11]. Previous studies

among Japanese middle-aged populations have shown

significant differences between people employed by

others versus those that are self-employed in risk for mor-

tality [12] and ce r eb r ova s c ul a r disease [13]. These differ-

ences could be explained by education level, marital sta-

tus, flexibility of working hours, workload, and autono-

my. How ever, there are few studies on the relationship

regularly employed are longer than those of the non-

regularly employed and (2) t he regularly employed cannot

generally determine their own working hours.

Additionally, authors hypothesize that self-employed

women have a higher prevalence of SSD. This hypothesis

is motivated by large gender gap in the amount of time

spent on household work in Japan.

The Survey on Time Use and Leisure Activities (STULA) is a

policies aimed at maintaining a vital aging society, improving

the childcare environment, facilitating gender equality, and

enabling the use of information and communication technol-

ogy in daily life, while considering the current social environ-

ment. The STULA employs a stratified two-stage sampling

method with the Enumeration Districts (the units of the inves-

tigation area of the Population Census, which consists of

about 50 households) of the 2000 Population Census as the

first-stage sampling unit and households as the second-stage

sampling unit. The full details of the STULA are described on

the MIAC to use the data under Article 36 of the statistics law

of Japan on October 27 in 2015.

This survey was conducted according to the guidelines of

the Declaration of Helsinki, and verbal informed consent was

obtained from each individual participant. Under the Statistics

Act, the MIAC anonymized individual-level data collected

from the STULA and provided the authors with the datasets

for this study. In accordance with the Ethical Guidelines of

Epidemiological Research established by the Ministry of

Education, Culture, Sports, Science and Technology and the

household income level (n = 1858), marital status (n =185),

education level (n = 1035), and communication device use

(n = 1011), the final sample used in this analysis was com-

prised of 58,308 male participants and 62,475 female partici-

pants aged 30–59 years.

720 min), and commuting (≥ 30 min) was lower for the self-

employed than for the regularly employed (12.5%, 4.1%,

8.1%, and 30.2% re spec tively for the self-employed, a nd

15.1%, 7.6%, 9.3%, and 48.9% respectively for the regularly

employed).

For women, the proportion of participants who spent time

on household affairs (≥ 285 min), nursing care (≥ 1 min), and

working (≥ 720 min) was higher for the self-employed than

for the regularly employed (19.6%, 4.8%, and 2.7% respec-

tively for the self-employed and 14.3%, 3.2%, and 2.1% re-

adjusted OR (95% CI) of SSD for the self-employed was 1.79

(1.44–2.22) in women. Even after adjusting for potential con-

founding factors (i.e., marital status, education level, and com-

munication device use), the significance remained. The

multivariable-adjusted OR (95% CI) of SSD for self-

employed men was 0.78 (0.65–0.92). On the other hand, the

OR for self-employed women was 1.78 (1.43–2.21).

Further adjustment for daily activity duration resulted in

the following ORs (95% CIs) of SSD for self-employment

with regular employment as the reference group: 0.78 (0.65–

The primary focus of this study was to describe the variation

in the prevalence of SSD by employment status using the

STULA database. Self-employed women had a higher preva-

differed significantly. Further adjustment for time spent on

other daily activities revealed that females with non-regular

employment and “other” employment had a higher prevalence

of SSD with regular employment as the reference group. In

addition, the unemployed had a significantly lower prevalence

of SSD with regular employment as the reference group, but

the significance diminished even after adjusting for time spent

on other daily activities.

Authors hypothesized that the regularly employed would

have a higher prevalence of SSD for both men and women. In

hypothesized that self-employed women would have a signif-

icantly higher prevalence of SSD with regular employment as

the reference group, and the results of this survey confirmed

this hypothesis. One explanation for this finding was the large

gender gap in the amount of time spent on household work,

with wives devoting many hours to both gainful employment

and household work, especially among the self-employed

couples [28].

The distribution of time spent on daily activities differed

among the five employment statuses in men and women.

higher odds of long sleep duration (≥ 11-h sleep) compared to

private sector employees on weekdays because they had more

flexibility in their work hours [24]. Our findings for men were

consistent with this US study. Ohida et al. indicated that the

highest percentage of SSD (≤ 6-h sleep) was found in the 40–

49 age group among women due to domestic responsibilities,

such as taking care of children and housekeeping [9]. Doi et al.

noted that being married or living with a family was protective

against excessive daytime sleepiness for men, but women

were more likely to suffer from excessive daytime sleepiness

Mizunoya used data from the 2001 STULA survey for mar-

ried couples with one or more children and found that hus-

bands worked over 9 h per weekday, while only about one

tenth participated in household work and that wives devoted

many hours to both gainful employment and household work

[28].

Authors also confirmed that short duration sleepers worked

significantly more hours than non-short sleepers, regardless of

sex (data are not shown). The proportion of self-employed

women working ≥ 720 min was 2.7%, and this value was

Honjo et al. indicated that self-employed women had a

higher risk of death than women employed full-time. The

hypothesized benefits of being self-employed, such as job

discretion and choice of working hours, may not be beneficial

for self-employed females in Japan [12]. Fujino et al. indicated

#

0.64 1.00 1.02 0.78 0.76 0.89 1.00 1.05 1.78 1.23

#

that self-employed men showed a significantly lower risk of

death from cerebrovascular disease compared to regularly

employed men after adjusting for possible confounding fac-

tors, but no association was identified among women [13].

Although authors could not directly compare these results to

those of the present study since the reference groups were

different, both studies indicated that self-employed women

were less likely to have the health-related advantages that their

male counterparts enjoyed.

Above all, this finding suggests that the reasons that self-

employed women was significantly higher after adjusting for

confounding factors. Honjo et al. indicated that part-time

workers were more likely to experience harmful health effects

compared to full-time workers among middle-aged Japanese

women; this association was stronger among less-educated

and unmarried women [12].

In the present study, the self-reported prevalence of SSD in

people aged 30–59 years (2.8% in men and 3.2% in women,

respectively) was lower than the prevalence found in the 2006

National Health and Nutrition Survey, Japan (NHNSJ)

8000). Furthermore, participants in the STULA recorded the

time spent on 20 types of life activities, e.g., sleep duration, in

15-min intervals for 2 days in a similar manner to keeping a

diary. By contrast, participants of the NHNSJ reported their

average sleep duration for the past month. The discrepancy

between the results may be due to diffe rences in th e

questioning methods. Campanini et al. indicated that sleep

diaries and actigraphy showed moderate or good agreement

when assessing several sleep parameters in a highly educated

population [30]. A Japanese study noted that sleep diaries

Wetter et al. indicated that self-reported data on sleep status

show at least moderate agreement with data from laboratory

studies [32]. Lauderdale et al. reported longer self-reported

sleep durat ions compared with sleep durations determined

by actigraphy, suggesting that there is at least some degree

of error or variance in the estimation of self-reported sleep

duration [33].

Given that the “sleep” category for the STULA included a

wide range of terms that may not involve physiological sleep

(e.g., falling asleep, dozing off, napping, getting up, waking

up), it is suggested that the STULA sleep times overestimated

the actual physiological sleep obtained by respondents per

24 h. This may have contributed to the longer estimated sleep

durations and to the lower prevalence of SSD in people aged

30–59 years in this study compared to the results of other

population-based studies.

Furthermore, the STULA asked individuals to report only

may have fewer differences than other statistics on working

hours, e.g., the Basic Research of the Work Structure (MIAC)

and Labor Force Survey (MIAC) [35]. Additionally, the sam-

ple size of this survey is large. Thus, data for this study are

more representative of the general population than the data of

other population-based surveys.

Authors analyzed participants aged 30–59 years of age.

One of the reasons why we considered those ages as represen-

tative of adults was that the proportions of the population

attending school were 20.7%, 1.8%, and 0.5% or under in

and 5 min to 33 min, although the average sleep duration for

all participants was 7 h and 42 min in the 2006 STULA [14].

The last reason was that most of the studies reported a signif-

icant association between SSD and mortality from cardiovas-

cular disease among individuals aged more than 30 years old

[16]. Authors also confirmed the association between employ-

ment status and SSD within the participants aged 20–59 years

old, and the only difference between this group and the par-

ticipants aged 30–59 years old was that non-regularly

employed men had a higher prevalence of SSD; however,

sents a national sample. This study is one of a few large-scale

investigations to examine the association between employ-

ment status and the prevalence of SSD where the sampling

represents the Japanese population. To date, only a few studies

have examined the association between employment statuses,

especially for the self-employed, and sleep duration in the

Asian region.

There are several limitations in the present study. First,

because the study design was cross-sectional, authors could

not determine any causal relationship between self-

employment and SSD. Authors cannot be certain that sleep

duration is reduced by an increase d working time or vice

versa.

Second, there was no information on health status (even

height and weight) or history of chronic diseases in the

quantity as well as quality, authors could not consider phar-

maceutical status in our analyses due to a lack of information.