Association between employment status and long-term changes in psychological stress across the COVID-19 pandemic period in Japan: NIPPON DATA2010

Abstract

Objectives

There is a growing interest in understanding the long-term impact of employment status on psychological stress. We aimed to explore the association between socioeconomic status and psychological stress over a long-term follow-up period across the COVID-19 pandemic, employing the Kessler 6-Item Psychological Distress Scale (K6).

Methods

We evaluated K6 scores from the 2021 follow-up survey of NIPPON DATA2010 using a self-administered questionnaire. The association between employment status and changes in K6 scores over 11 years was examined. Multiple regression analyses were used to estimate both crude and adjusted differences in K6 score changes across various socioeconomic factors including employment category, annual household income, marital status, and household size. Analyses were stratified by age, gender, and prefectural population size.

Results

This study included 1532 participants with an average age of 54.9 years. Over 11 years (2010–2021), participants in both gender and age groups showed increases in mean K6 scores (men: 2.79 to 3.06; women: 3.15 to 3.56; <65 years: 3.27 to 3.47; ≥65 years: 2.37 to 3.08). Nonemployed participants, particularly homemakers, showed significantly greater increases in K6 scores, compared with full-time employees, especially among women, younger individuals, and those in densely populated areas, with a significant interaction with age.

Conclusions

Nonemployed individuals, especially homemakers, experienced greater psychological stress over the past 11 years than did their fully employed counterparts. Public interventions, including strengthened social connections and telemental health services, may help mitigate these disparities, enhance mental well-being, and foster a sense of belonging.

Key points

• Kessler 6-Item Psychological Distress Scale (K6) scores increased over 11 years, irrespective of gender, age, or prefectural population size.

• Our prospective study showed long-term K6 score changes, highlighting the impact of employment status on psychological stress across the pandemic.

• Nonemployed individuals, particularly homemakers, experienced a significantly greater increase in K6 scores than full-time employees.

• Among homemakers, larger increases in K6 scores were more prevalent in women, younger individuals, and those residing in densely populated areas.

• Strengthening social connections and telemental health access may reduce psychological stress disparities, particularly during a social upheaval.

1. Introduction

The COVID-19 pandemic wreaked havoc on lifestyles and economies worldwide. Measures such as home confinement, social distancing, and temporary closure of various establishments, including workplaces and schools, dramatically altered people’s daily routines, resulting in weakened personal connections and heightened psychological strain. Furthermore, the substantial economic downturn stemming from the pandemic triggered feelings of financial insecurity.¹ Consequently, the prevalence globally of major depressive and anxiety disorders increased by approximately 27% and 25%, along with a 5% and 50% rise in disability-adjusted life years, respectively.² In Japan, suicide rates increased during the pandemic.^2,3

Previous research has underscored the association between socioeconomic status and psychological disorders over a relatively short time.^4-6 Sidorchuk et al⁴ demonstrated that temporary workers and unemployed individuals were approximately 1.3 and 3.1 times more likely to experience psychological distress measured by the General Health Questionnaire, respectively, compared with permanent workers. Unemployed people, retirees, homemakers, and students have also been found to be vulnerable to psychological distress,^4,6-8 primarily due to financial strain, social stigma, and diminished self-esteem.⁹ Conversely, Schonfeld and Mazzola¹⁰ suggested that self-employed people were more likely to face work-related stress, potentially increasing their vulnerability to cardiovascular diseases, compared with those with regular employment. Despite a substantial body of evidence supporting the association between socioeconomic status and psychological stress, the relationship remains complex and not fully understood. In particular, the long-term trajectory of psychological stress may vary according to socioeconomic status at a specific point in time.

The Kessler Psychological Distress Scale (K6)¹¹ is a well-validated instrument widely used globally to screen for nonspecific psychological distress, including symptoms associated with mood and anxiety disorders. Owing to its strong psychometric properties and brevity, K6 has been used for both population-based studies and clinical settings. The K6 has also been validated for use in the general Japanese population.¹² An increase in K6 scores has been associated with a higher risk of all-cause, cardiovascular, and suicide mortality in a dose–response fashion.¹³

The National Integrated Project for Prospective Observation of Non-communicable Disease and its Trends in the Aged (NIPPON DATA2010)¹⁴ is a cohort study of cardiovascular disease conducted in the general population of Japan that started in 2010. In a follow-up survey in 2021, a questionnaire on lifestyle changes during the COVID-19 pandemic and the K6 were administered. During the COVID-19 pandemic, a considerable number of people were exposed to stressful circumstances in their living environments. We analyzed the long-term changes in K6 scores of the NIPPON DATA2010 participants over 11 years, from 2010 to 2021. The aim of our study was to investigate longitudinal trends in psychological stress, measured by K6 scores, across different socioeconomic status groups over an 11-year follow-up period, and to assess whether distinct patterns were evident within specific employment status subgroups.

2. Methods

2.1. Study design

The NIPPON DATA2010¹⁴ is a prospective cohort study that combined data from the National Health and Nutrition Survey (NHNS2010)¹⁵ and the Comprehensive Survey of Living Conditions of the People on Health and Welfare (CSLC2010) conducted by the Ministry of Health, Labour and Welfare of Japan.¹⁶ Detailed information regarding the NIPPON DATA2010 has been documented elsewhere.¹⁴

A total of 8815 residents from 300 randomly selected districts across Japan participated in the dietary survey in the NHNS2010 (Figure 1). Among 7229 participants aged 20 years or older, 3873 individuals (1598 men and 2275 women) underwent blood tests and were invited to enroll in the NIPPON DATA2010. Of 2898 participants (1239 men and 1659 women) who consented to provide baseline data and be followed up, 7 were excluded because of data merging issues. The remaining 2891 participants (1236 men and 1655 women) provided baseline data. Of 2184 participants followed up in 2021, we included 1532 participants (619 men and 913 women) after excluding those with missing data in the 2021 lifestyle changes surveys or K6 scores from 2010 and 2021 (n = 652). All participants provided written informed consent. This study was approved by the Institutional Review Board of Shiga University of Medical Science (R2010-029) and Fukuoka University Clinical Research and Ethics Center (U21-09-001).

Figure 1.

Flow diagram of participant inclusion.

2.2. Baseline survey: socioeconomic factors in 2010

The baseline survey was conducted using self-administered questionnaires. Employment status in 2010 was categorized into 2 categories: individuals with and without paid jobs, and further categorized into 5 specific groups: full-time employees, those engaged in self-employed or family businesses, part-time employees (all classified as having paid jobs), homemaker, and other (those without paid jobs). The “other” category included retirees, long-term absentees owing to illness, students, and those who were not employed for unspecified reasons. Marital status and household size were categorized into 3 groups: married individuals, singles living with others, and singles living alone. Annual household income (JPY) was divided into 3 groups: <2 000 000, 2 000 000 to 6 000 000, or >6 000 000.

2.3. Baseline survey: other factors in 2010

Alcohol consumption in 2010 was categorized into 3 groups: current, former, and nondrinker. Similarly, smoking status was classified as current, former, and nonsmoker. Additionally, variables for diabetes, hypertension, and dyslipidemia were included in the models, as comorbid chronic diseases may act as confounders in the association between employment status and psychological stress.¹⁷ In the NHNS2010, blood samples were collected, and serum samples were separated and centrifuged after blood coagulation, with plasma samples collected into siliconized tubes containing sodium fluoride and shipped to a central laboratory (SRL, Tokyo, Japan). Serum total cholesterol levels were measured using enzymatic methods, and glycated hemoglobin (HbA1c) level was measured using latex agglutination inhibition assays with the standardized method of the Japan Diabetes Society (JDS). HbA1c values were converted into the National Glycohemoglobin Standardization Program (NGSP) values using the following formula: HbA1c (NGSP) (%) = 1.02 × HbA1c (JDS) (%) + 0.25.¹⁴ Information on blood chemistry data measurements and their performance is described elsewhere.¹⁸ Diabetes was defined as HbA1c ≥6.5%, casual blood glucose ≥200 mg/dL, or use of medication for diabetes. Hypertension was defined as systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or use of antihypertensive drugs. Dyslipidemia was defined as total cholesterol ≥220 mg/dL or use of cholesterol-lowering drugs. Comorbidities were dichotomized into 2 groups: yes and no.

2.4. K6 scores: baseline survey in 2010 and follow-up survey in 2021

K6 score was assessed at baseline (2010) and in the follow-up survey (2021) using the same self-administered questionnaire, the Japanese version of the K6 scale, among participants of the NIPPON DATA2010. The follow-up survey was conducted in the autumn of 2021. The K6 scale was calculated using the following question: “During the past 30 days, about how often did you feel: (1) nervous? (2) hopeless? (3) restless or fidgety? (4) so depressed that nothing could cheer you up? (5) that everything was an effort? and (6) worthless?—would you say all of the time, most of the time, some of the time, a little of the time, or none of the time?” Responses ranged from “none of the time” to “all of the time,” with each question scored from 0 (none of the time) to 4 (all the time). Total scores were calculated to yield a K6 score ranging from 0 to 24, with higher scores indicating higher psychological stress levels.

2.5. Statistical analysis

Baseline characteristics were presented by mean ± standard deviation (SD) for continuous variables and as percentages for categorical variables. Age groups were categorized as <65 years and ≥65 years, reflecting a substantial lifestyle shift typically associated with conventional retirement age in Japan. The prefectural population size of participants was categorized into 2 groups: <2 800 000 and ≥2 800 000, to almost evenly divide the participants into halves, using the Japanese census data of 2010.¹⁹ Continuous variables were compared using the Student t test, and categorical variables using the chi-squared test. For socioeconomic status subgroups with a limited number of participants (<10), data were presented as medians with interquartile ranges to account for potential skewness and to avoid misrepresenting the distribution. Crude mean changes in K6 scores between 2010 and 2021 were estimated using predictive margins. Differences in K6 score changes across socioeconomic factors were evaluated using partial derivatives of multiple linear regression models. The analyses were conducted using a crude model (Model 1), adjusted for age and gender (Model 2), and a fully adjusted model (Model 3), which accounted for age, gender, baseline K6 score, alcohol consumption, chronic comorbidities (any of hypertension, diabetes, and dyslipidemia), and prefectural population size at baseline. Finally, these characteristics were further stratified by age, gender, and prefectural population size of participants at baseline. P values were calculated to assess interactions between age, gender, or prefectural population size and socioeconomic factors. P values <.05 were considered statistically significant. All analyses were performed using Stata SE version 16 (Stata Statistical Software: Release 16, 2019; StataCorp, College Station, TX, USA) and R Statistical Software (v4.1.2; R Core Team, 2021).

3. Results

This study included 1532 participants (619 men, 913 women; Figure 1) with an average age of 54.9 years, of whom approximately 60% were employed and the majority married. Baseline characteristics are shown in Table 1. Women comprised approximately 60% of the participants. Men were older and more likely to have jobs than women were, and only 10% of men were engaged in part-time or homemaker roles. Men were more likely than women to be current drinkers and have chronic comorbidities. Approximately 60% of all participants engaged in paid jobs, with full-time employment being the most common job category. Nearly 1 in 4 participants were homemakers. The majority of the participants were married, and nearly half reported annual household income ranging from JPY 2 000 000 to JPY 6 000 000. Of all participants, current drinkers were more prevalent than former or nondrinkers, and over 60% reported having at least 1 chronic comorbidity: hypertension, diabetes, or dyslipidemia.

Table 1.

Baseline characteristics of the study participants (NIPPON DATA2010).

		All		Men		Women
		n	%	n	%	n	%	P value a
		1532	100.0	619	100.0	913	100.0
Age	Mean ± SD	54.9 ± 14.3		56.3 ± 14.0		54.0 ± 14.5		.002
Employment status	Individuals with paid jobsb	922	60.2	447	72.2	475	52.0	<.001
	Individuals without paid jobsc	610	39.8	172	27.8	438	48.0
	Full-time employee	558	36.4	283	45.7	275	30.1	<.001
	Self-employed/family-run business	220	14.4	117	18.9	103	11.3
	Part-time employee	144	9.4	47	7.6	97	10.6
	Homemaker	388	25.3	16	2.6	372	40.7
	Otherd	222	14.5	156	25.2	66	7.2
Marital status and household size	Married individuals	1229	80.2	524	84.7	705	77.2	<.001
	Singles living with someone	168	11.0	44	7.1	124	13.6
	Singles living alone	132	8.6	49	7.9	83	9.1
	Unknown	3	0.2	2	0.3	1	0.1
Household income, JPY/y	<2 000 000	203	13.3	71	11.5	132	10.5	.128
	2 000 000 to 6 000 000	825	53.9	353	57.0	472	51.7
	>6 000 000	364	23.8	151	24.4	213	23.3
	Unknown	140	9.1	44	7.1	96	10.5
Alcohol consumption	Current drinker	842	55.0	470	75.9	372	40.7	<.001
	Former drinker	21	1.4	11	1.8	10	1.1
	Nondrinker	665	43.4	136	22.0	529	57.9
	Unknown	4	0.3	2	0.3	2	0.2
Chronic comorbidities e	Yes	1000	65.3	452	73.0	548	60.0	<.001
	No	532	34.7	167	27.0	365	40.0
Prefectural population size in 2010	<2 800 000	740	48.3	310	50.1	482	52.8	.297
	≥2 800 000	792	51.7	309	49.9	431	47.2

Abbreviations: SD, standard deviation; JPY, Japanese yen.

^a P values were calculated using the Student’s t-test for continuous variables and the chi-squared test for categorical variables. Participants with unknown subgroup were excluded from the analysis.

^bFull-time employees, self-employed individuals, those running family-run businesses, and part-time workers.

^cHomemakers, retirees, individuals on long-term sick leave, students, and those not employed for unspecified reasons.

^dRetirees, individuals on long-term sick leave, students, and those not employed for unspecified reasons.

^eChronic comorbidities include hypertension, diabetes, or dyslipidemia.

Age-stratified analysis (Table S1) showed that younger participants (<65 years) were predominant. Women accounted for more than half of the participants in both age groups. In addition, younger participants were more likely to have paid jobs and report higher annual household income, but less likely to be current drinkers or have chronic comorbidities than older participants. Analysis stratified by prefectural population size (Table S2) showed that self-employment was less common in more densely populated areas, whereas the proportion of homemakers was higher. Participants with higher annual household income were more likely to reside in densely populated areas. In addition, single people living alone were disproportionately represented in these regions.

3.1. K6 score changes between 2010 and 2021

Table 2 shows the mean K6 scores during the follow-up period and the adjusted mean K6 score changes between 2010 and 2021 for all participants and those stratified by gender. A significant increase in K6 scores over the 11-year period was found (Model 3: 0.35; 95% CI, 0.16-0.54; P < .001), although no significant difference in K6 changes between gender was detected (Model 3: 0.23; 95% CI, −0.18 to 0.64; P = .276). Similarly, K6 score changes over 11 years were not significantly different between age groups (Model 3: 0.18; 95% CI, −0.26 to 0.62; P = .416) and the population size groups (−0.06; 95% CI, −0.43 to 0.32; P = .762) in fully adjusted models (Tables S3 and S4).

Table 2.

K6 score changes between 2010 and 2021 (NIPPON DATA2010).

		K6 scores		Crude mean changes in K6 scores between	Differences in K6 score changes
		2010	2021	2010 and 2021	Model 1 a		Model 2 b		Model 3 c
	n	Mean ± SD	Mean ± SD	(95% CI)	(95% CI)	P value d	(95% CI)	P value d	(95% CI)	P value d
All	1532	3.00 ± 3.29	3.36 ± 4.08	0.35 (0.15, 0.56)	0.35 (0.15, 0.56)	.001	0.35 (0.15, 0.56)	.001	0.35 (0.16, 0.54)	<.001
Men	619	2.79 ± 3.09	3.06 ± 4.03	0.27 (−0.05, 0.59)	Reference		Reference		Reference
Women	913	3.15 ± 3.42	3.56 ± 4.10	0.41 (0.14, 0.67)	0.13 (−0.28, 0.55)	.527	0.16 (−0.26, 0.57)	.454	0.23 (−0.18, 0.64)	.276

Abbreviations: SD, standard deviation; CI, confidence interval.

^aAdjusted for no covariate.

^bAdjusted for age and gender.

^cAdjusted for baseline K6 score, age, gender, alcohol consumption, chronic comorbidities, and prefectural population size in 2010.

^d P values were derived from multiple linear regression analyses.

3.2. Association between K6 score changes over 11 years and socioeconomic status

As shown in Table 3, a significant increase in K6 score was consistently observed among individuals without paid jobs, compared with those with paid jobs (Model 3: homemaker, 0.66; 95% CI, 0.11-1.21; P = .018; other: 0.79; 95% CI, 0.15-1.43; P = .016), compared with their fully employed counterparts. Other socioeconomic subgroups showed no significant differences in K6 score changes over the 11-year period. Gender-stratified analysis (Table 4) revealed that a significant increase in K6 scores among nonemployed participants compared with fully employed participants was observed only among women (homemaker: 0.70; 95% CI, 0.09-1.31; P = .024; other: 1.53; 95% CI, 0.48-2.57; P = .004), although the gender interaction was not statistically significant (P = .170). Age-stratified analysis (Table S5) showed that the significant increase in K6 scores among homemakers compared with full-time employees was restricted to younger participants (Model 3: 1.07; 95% CI, 0.44-1.70; P = .001). For the “other” category, no significant increase was observed (Model 3: 0.04; 95% CI, −0.90 to 0.99; P = .929), with a significant age interaction (P = .025). In the analysis stratified by prefectural population size (Table S6), the significant increase in K6 scores among homemakers compared with full-time employees was observed only in densely populated areas (1.10; 95% CI, 0.41-1.79; P = .002). In addition, a marginal increase in K6 scores for the “other” category was observed in densely populated areas (0.89; 95% CI, −0.03 to 1.61; P = .060), although the area interaction was not statistically significant (P = .197). No significant area interaction was observed for the association between other socioeconomic factors and K6 score changes over 11 years.

Table 3.

The association between K6 score changes and socioeconomic status (NIPPON DATA2010).

		K6 scores		Crude mean changes in K6 scores between 2010 and 2021	Differences in K6 score changes
		2010	2021		Model 1 a		Model 2 b		Model 3 c
	n	Mean ± SD	Mean ± SD	(95% CI)	(95% CI)	P value d	(95% CI)	P value d	(95% CI)	P value d
Employment status
Individuals with paid jobse	992	3.14 ± 3.36	3.18 ± 4.07	0.04 (−0.22, 0.30)	Reference		Reference		Reference
Individuals without paid jobsf	610	2.80 ± 3.18	3.62 ± 4.09	0.82 (0.50, 1.14)	0.77 (0.36, 1.19)	.001	0.77 (0.31, 1.22)	.001	0.77 (0.32, 1.23)	.001
Full-time employee	558	3.07 ± 3.18	3.17 ± 4.11	0.10 (−0.23, 0.44)	Reference		Reference		Reference
Self-employed/family-run business	220	3.18 ± 3.56	3.02 ± 4.27	−0.16 (−0.69, 0.37)	−0.26 (−0.89, 0.37)	.416	−0.29 (−0.93, 0.36)	.387	−0.09 (−0.69, 0.50)	.757
Part-time employee	144	3.33 ± 3.69	3.46 ± 3.56	0.13 (−0.53, 0.79)	0.03 (−0.71, 0.77)	.937	0.02 (−0.73, 0.77)	.958	0.16 (−0.53, 0.85)	.654
Homemaker	388	2.92 ± 3.16	3.71 ± 4.09	0.79 (0.39, 1.19)	0.69 (0.16, 1.21)	.010	0.66 (0.07, 1.26)	.029	0.66 (0.11, 1.21)	.018
Otherg	222	2.59 ± 3.22	3.46 ± 4.10	0.87 (0.34, 1.40)	0.77 (0.14, 1.39)	.017	0.72 (0.03, 1.41)	.041	0.79 (0.15, 1.43)	.016
Household income, JPY/y
<2 000 000	203	3.10 ± 3.26	3.38 ± 4.23	0.28 (−0.28, 0.84)	Reference		Reference		Reference
2 000 000 to 6 000 000	825	3.04 ± 3.35	3.52 ± 4.23	0.49 (0.21, 0.76)	0.21 (-0.41, 0.83)	.512	0.27 (−0.36, 0.89)	.402	0.18 (−0.40, 0.76)	.543
>6 000 000	364	2.81 ± 3.01	2.92 ± 3.65	0.12 (−0.30, 0.53)	−0.17 (−0.86, 0.53)	.641	−0.06 (−0.77, 0.64)	.857	−0.32 (−0.98, 0.33)	.332
Unknown	140	3.17 ± 3.67	3.44 ± 3.99	0.27 (−0.40, 0.94)	−0.01 (−0.88, 0.86)	.983	0.05 (−0.83, 0.92)	.918	0.01 (−0.80, 0.82)	.975
Marital status and household size
Married individuals	1229	2.93 ± 3.19	3.30 ± 4.00	0.37 (0.14, 0.59)	Reference		Reference		Reference
Singles living with someone	168	3.17 ± 3.83	3.63 ± 4.53	0.46 (−0.15, 1.07)	0.09 (−0.56, 0.74)	.787	0.21 (−0.47, 0.88)	.549	0.13 (−0.50, 0.75)	.684
Singles living alone	132	3.39 ± 3.43	3.51 ± 4.30	0.12 (−0.57, 0.81)	−0.25 (−0.97, 0.48)	.504	−0.32 (−1.05, 0.41)	.394	−0.03 (−0.71, 0.65)	.926
Unknown	3	9.0 (0.0, 9.0)h	6.0 (0.0, 6.0)h	—	—	—	—	—	—	—

Abbreviations: SD, standard deviation; CI, confidence interval; JPY, Japanese yen.

^aAdjusted for no covariate.

^bAdjusted for age.

^cAdjusted for baseline K6 score, age, alcohol consumption, chronic comorbidities, and prefectural population size in 2010.

^d P values were derived from multiple linear regression analyses.

^eFull-time employees, self-employed individuals, those running family-run business, and part-time workers.

^fHomemakers, retirees, individuals on long-term sick leave, students, and those not employed for unspecified reasons.

^gRetirees, individuals on long-term sick leave, students, and those not employed for unspecified reasons.

^hPresented as median with interquartile range, due to the skewed distribution associated with the small sample size in this category.

Table 4.

The association between K6 score changes and socioeconomic status according to gender (NIPPON DATA2010).

		K6 scores		Crude mean changes in K6 scores between	Differences in K6 score changes
		2010	2021	2010 and 2021	Model 1 a		Model 2 b		Model 3 c
	n	Mean ± SD	Mean ± SD	(95% CI)	(95% CI)	P value d	(95% CI)	P value d	(95% CI)	P value d	P interaction e
Men	619
Employment status
Individuals with paid jobsf	447	2.88 ± 3.09	3.06 ± 4.16	0.18 (−0.19, 0.56)	Reference		Reference		Reference
Individuals without paid jobsg	172	2.54 ± 3.09	3.05 ± 3.68	0.51 (−0.10, 1.11)	0.32 (−0.39, 1.04)	.375	0.15 (−0.66, 0.96)	.712	0.14 (−0.68, 0.96)	.736	.163
Full-time employee	283	2.90 ± 3.00	3.00 ± 4.16	0.10 (−0.37, 0.58)	Reference		Reference		Reference
Self-employed/family-run business	117	2.91 ± 3.30	3.14 ± 4.46	0.23 (−0.51, 0.97)	0.13 (−0.75, 1.00)	.774	0.03 (−0.88, 0.95)	.943	0.21 (−0.64, 1.06)	.628	.170
Part-time employee	47	2.68 ± 3.09	3.23 ± 3.42	0.55 (−0.61, 1.72)	0.45 (−0.80, 1.71)	.481	0.36 (−0.92, 1.64)	.581	0.41 (−0.78, 1.59)	.501
Homemaker	16	1.56 ± 2.10	2.25 ± 2.38	0.69 (−1.30, 2.68)	0.59 (−1.46, 2.63)	.575	0.38 (−1.75, 2.50)	.727	−0.06 (−2.09, 1.97)	.951
Otherh	156	2.64 ± 3.16	3.13 ± 3.79	0.49 (−0.15, 1.13)	0.38 (−0.41, 1.18)	.342	0.20 −-0.73, 1.14)	.671	0.42 (−0.45, 1.29)	.345
Household income, JPY/y											.746
<2 000 000	71	3.04 ± 3.38	2.80 ± 3.96	−0.24 (−1.18, 0.70)	Reference		Reference		Reference
2 000 000 to 6 000 000	353	2.76 ± 3.10	3.23 ± 4.20	0.47 (0.05, 0.90)	0.71 (−0.32, 1.75)	.177	0.77 (−0.27, 1.80)	.148	0.63 (−0.34, 1.60)	.201
>6 000 000	151	2.66 ± 2.94	2.73 ± 3.67	0.07 (−0.58, 0.71)	0.31 (−0.84, 1.45)	.600	0.41 (−0.75, 1.57)	.485	0.20 (−0.88, 1.29)	.714
Unknown	44	3.02 ± 3.05	3.23 ± 3.98	0.20 (−0.99, 1.40)	0.44 (−1.08, 1.97)	.568	0.56 (−0.98, 2.09)	.478	0.52 (−0.92, 1.96)	.478
Marital status and household size											.353
Married individuals	524	2.69 ± 2.99	2.91 ± 3.97	0.23 (−0.12, 0.57)	Reference		Reference		Reference
Singles living with someone	44	3.36 ± 3.44	3.91 ± 4.67	0.55 (−0.66, 1.75)	0.32 (−0.93, 1.57)	.615	0.65 (−0.68, 1.97)	.338	0.59 (−0.64, 1.83)	.344
Singles living alone	49	3.24 ± 3.67	3.86 ± 4.01	0.61 (−0.52, 1.75)	0.39 (−0.80, 1.58)	.523	0.46 (−0.73, 1.65)	.448	0.69 (−0.42, 1.81)	.220
Unknown	2	4.5 (0.0, 9.0)i	3.0 (0.0, 6.0)i	—	—	—	—	—	—	—
Women	913
Employment status
Individuals with paid jobsf	475	3.38 ± 3.58	3.29 ± 3.97	−0.09 (−0.45, 0.27)	Reference		Reference		Reference
Individuals without paid jobsg	438	2.90 ± 3.21	3.84 ± 4.23	0.94 (0.56, 1.32)	1.03 (0.51, 1.55)	<.001	1.05 (0.50, 1.60)	<.001	1.05 (0.50, 1.60)	<.001
Full-time employee	275	3.24 ± 3.35	3.35 ± 4.06	0.10 (−0.37, 0.58)	Reference		Reference		Reference
Self-employed/family-run business	103	3.50 ± 3.83	2.89 ± 4.06	−0.60 (−1.38, 0.17)	−0.70 (−1.61, 0.20)	.128	−0.69 (−1.62, 0.24)	.147	−0.46 (−1.32, 0.41)	.300
Part-time employee	97	3.64 ± 3.92	3.57 ± 3.64	−0.07 (−0.87, 0.73)	−0.17 (−1.10, 0.75)	.713	−0.17 (−1.10, 0.76)	.720	0.01 (−0.85, 0.87)	.982
Homemaker	372	2.98 ± 3.18	3.77 ± 4.14	0.79 (0.39, 1.20)	0.69 (0.07, 1.32)	.030	0.71 (0.05, 1.37)	.035	0.70 (0.09, 1.31)	.024
Otherh	66	2.48 ± 3.38	4.26 ± 4.69	1.77 (0.81, 2.74)	1.67 (0.59, 2.75)	.002	1.70 (0.57, 2.82)	.003	1.53 (0.48, 2.57)	.004
Household income, JPY/y
<2 000 000	132	3.13 ± 3.21	3.69 ± 4.35	0.56 (−0.13, 1.25)	Reference		Reference		Reference
2 000 000 to 6 000 000	472	3.25 ± 3.51	3.75 ± 4.24	0.50 (0.13, 0.87)	−0.06 (−0.84, 0.72)	.879	−0.02 (−0.81, 0.77)	.956	−0.06 (−0.78, 0.66)	.869
>6 000 000	213	2.91 ± 3.07	3.06 ± 3.64	0.15 (−0.39, 0.69)	−0.41 (−1.29, 0.47)	.360	−0.34 (−1.24, 0.56)	.459	−0.61 (−1.44, 0.22)	.147
Unknown	96	3.24 ± 3.94	3.54 ± 4.01	0.30 (−0.51, 1.11)	−0.26 (−1.32, 0.81)	.634	−0.22 (−1.29, 0.85)	.685	−0.24 (−1.22, 0.75)	.636
Marital status and household size
Married individuals	705	3.11 ± 3.32	3.59 ± 3.99	0.48 (0.18, 0.77)	Reference		Reference		Reference
Singles living with someone	124	3.10 ± 3.97	3.52 ± 4.49	0.43 (−0.28, 1.14)	−0.05 (−0.82, 0.72)	.903	0.05 (−0.73, 0.84)	.893	−0.04 (−0.77, 0.69)	.917
Singles living alone	83	3.47 ± 3.30	3.30 ± 4.47	−0.17 (−1.04, 0.70)	−0.64 (−1.56, 0.28)	.170	−0.78 (−1.72, 0.17)	.106	−0.51 (−1.38, 0.37)	.256
Unknown	1	9.0 (9.0, 9.0)i	6.0 (6.0, 6.0)i	—	—	—	—	—	—	—	—

Abbreviations: SD, standard deviation; CI, confidence interval; JPY, Japanese yen.

^aAdjusted for no covariate.

^bAdjusted for age.

^cAdjusted for baseline K6 score, age, alcohol consumption, chronic comorbidities, and prefectural population size in 2010.

^d P values were derived from multiple linear regression analyses.

^e P values for interaction between gender and socioeconomic status.

^fFull-time employees, self-employed individuals, those running family-run businesses, and part-time workers.

^gHomemakers, retirees, individuals on long-term sick leave, students, and those not employed for unspecified reasons.

^hRetirees, individuals on long-term sick leave, students, and those not employed for unspecified reasons.

ⁱPresented as median with interquartile range.

4. Discussion

In this study, we analyzed K6 survey data collected from the same population over an 11-year interval and found that nonemployed participants experienced significant increase in K6 scores, compared with full-time employees. Notably, when stratified by gender, this association was observed exclusively in women, although no significant gender interaction was detected. Age-stratified analysis further revealed that younger homemakers in the nonemployed group exhibited greater changes in K6 scores than their fully employed counterparts, with a significant age interaction. Additionally, stratification by prefectural population size showed that homemakers residing in more densely populated areas were more likely to experience increases in K6 scores over the 11-year period than full-time employees, although no significant area interaction was found.

These findings align with prior research linking employment status to psychological stress, particularly during the COVID-19 pandemic.^20,21 A survey of 1000 participants revealed that nonemployed individuals were more prone to depression, anxiety, or insomnia than employed counterparts.²¹ Several studies have also documented increased psychological stress and suicide rates among homemakers during the pandemic across various countries.^3,22,23 In India, homemakers accounted for a disproportionately high percentage of suicide deaths over the decade preceding the pandemic.²³ Homemakers have long been recognized as particularly vulnerable to psychological stress, owing to factors such as codependency, social isolation, and diminished self-esteem,⁶ even in the absence of a global crisis such as the pandemic. During the pandemic, stay-at-home measures likely exacerbated isolation and intensified the burden of unpaid household labor, including caregiving and efforts to avoid infection. Studies have also revealed an increase in domestic violence rates during this period,^3,24,25 potentially associated with increased alcohol consumption at home. Saravi et al²⁶ found that employed women reported better quality of life, vitality, social functioning, and mental health than did homemakers. Conversely, Dibaji et al²⁷ suggested that employed women faced higher levels of depression, possibly due to the dual responsibilities of paid work and unpaid caregiving. However, it is noteworthy that Dibaji et al’s analysis, in which mean stress scale scores were compared using a t test, may have been confounded by household income.

Our study also found that homemakers aged under 65 years exhibited a significant increase in K6 scores over 11 years, compared with their fully employed counterparts. However, homemakers aged 65 years or older did not show a significant increase, suggesting the potential differences in social context related to age at baseline. It is possible that older homemakers had more prior work experience before retirement, which could have enhanced their resilience in later life stages. A previous study indicated that women with prior work experience tend to report higher satisfaction post-retirement, benefitting from enhanced social connection and leisure activities, as they are no longer required to balance domestic duties and work commitments.²⁸ Moreover, previous studies have shown that workplace relationships can play a crucial role in a successful retirement, helping individuals build “new neighborhoods” and maintain social connections.²⁹ Social connection has consistently been identified as a key determinant of mental health, both before and during the pandemic.^30,31 As Kaplan⁶ emphasized, improving access to psychological support services is particularly important for addressing the mental health needs of homemakers. Thus, fostering stronger social networks may be crucial for mitigating the negative psychological impacts of life transitions, enhancing quality of life, and fostering a sense of belonging within society. Additionally, expanding access to telemental health services regardless of socioeconomic status could provide essential support to those vulnerable to isolation, particularly in times of social upheaval such as the pandemic.

Our findings also suggest that homemakers residing in urban areas (prefectural population size in 2010 ≥2 800 000) experience more pronounced long-term increases in K6 scores than those in rural areas. This aligns with previous studies highlighting urban–rural disparities, with urban residents more likely to report psychological stress during the pandemic.^32-34 Furusawa et al³² found that individuals living in urban areas were more likely to experience negative health outcomes during the pandemic, such as weight gain and reduced physical activity. A cross-sectional study³³ suggested that rural youths exhibited better mental health than their urban counterparts during the pandemic, potentially owing to the lower population density and more dispersed living conditions. Similarly, another study of over 20 000 women³⁴ showed that those living in urban areas reported higher negative impacts on mental health than women living in rural or small-town settings. Furthermore, the authors reported that women in urban areas without children were less likely to suffer from psychological stress than those with children, likely due to the added stress of parenting and financial strain.

The “other” category, comprising predominantly retirees, showed similar trends as homemakers in psychological stress over the 11-year follow-up. Evidence on the long-term psychological trajectory of retirees remains inconsistent, likely due to variation across different stages of post-retirement life.⁷ Cultural factors may also play a role; studies from Western countries have often reported reductions in stress following retirement, whereas research from developed Asian countries tends to highlight its adverse psychological effects.⁷

Employment is a key determinant of mental health, improving self-esteem, social relationships, and quality of life.³⁵ Our results suggest that individuals with paid jobs were more psychologically resilient during the pandemic compared with those without, although it is important to note that the K6 score does not directly measure psychological resilience. However, working also presents stressors; national surveys in the United Kingdom^36,37 showed that people with paid jobs experience mental disorders across all age groups, albeit at lower rates than among those without paid jobs.

Interestingly, our study indicated that individuals who were self-employed or operated family-run businesses did not experience significant long-term increases in K6 scores, despite the economic downturn caused by the pandemic-related lockdowns. This may be partly explained by the government’s uniform amount of subsidies provided to businesses, such as eateries, pubs, and restaurants, which were anticipated to face declines in sales due to reduced nighttime operations, irrespective of their business scales. These subsidies were particularly beneficial for smaller businesses owned by the self-employed, potentially shielding them from the economic impact of the pandemic.^3,38 Notably, the number of self-employed workers remained almost stable before and after the pandemic in Japan,³⁹ and suicide rates among self-employed individuals either remained steady or decreased, whereas suicide rates in other employment categories increased during the pandemic.^3,40

This study also found that part-time workers did not exhibit a significant long-term increase in K6 scores, although part-time workers, who disproportionately comprised women, were notably affected by the pandemic. Many service industrial sectors, including tourism and restaurants, faced closures that led to temporary layoffs or job losses for part-time workers. However, part-time workers constituted a relatively small percentage of our participants, which may explain the lack of a clear association between part-time employment and changes in K6 scores.

Strengths of our study include its large sample size, representative sampling from Japan, and a high response rate (93.2%). However, this study had some limitations. First, K6 scores, although a widely used screening tool, rely on self-reported data. Second, we could not account for changes in employment status or other factors during the 11-year study period, including the pandemic. Finally, although we sought to minimize bias, reverse causation cannot be entirely ruled out; individuals with psychological disorders may become unemployed, although mental illnesses account for only 1.4% of job departures in Japan.⁴¹

5. Conclusion

In conclusion, our findings suggest that nonemployed individuals of working age, whether voluntarily or involuntarily nonemployed, were more susceptible to experiencing psychological stress over the past 11 years compared with their fully employed counterparts, highlighting their distinct social backgrounds. Public interventions aimed at bolstering social connections and community cohesion and developing telemental health settings may be essential for addressing disparities in psychological well-being associated with employment status. These interventions could play a pivotal role in enhancing quality of life and fostering a sense of belonging, particularly among female homemakers.

The NIPPON DATA Research Group

Co-principal investigators: Katsuyuki Miura (Shiga University of Medical Science, Otsu Shiga), Akira Okayama (Research Institute of Strategy for Prevention, Tokyo), Tomonori Okamura (Keio University, Tokyo), and Takayoshi Ohkubo (Teikyo University, Tokyo). Past-principal investigator: Hirotsugu Ueshima (Shiga University of Medical Science, Otsu, Shiga, Japan).

Management committee: Yoshikazu Nakamura (Jichi Medical University, Shimotsuke, Tochigi), Aya Kadota (Shiga University of Medical Science, Otsu, Shiga), Takehito Hayakawa (Ritsumeikan University, Kyoto), Masaru Sakurai (Kanazawa Medical University, Uchinada, Ishikawa), and Naoyuki Takashima (Kyoto Prefectural University of Medicine, Kyoto).

Research members: Hirofumi Ohnishi (Sapporo Medical University, Sapporo, Hokkaido), Shigeyuki Saitoh (Nihon Iryou Daigaku Hospital, Sapporo, Hokkaido), Kiyomi Sakata, (Iwate Medical University, Morioka, Iwate), Masaki Ohsawa (Morioka Tsunagi Onsen Hospital, Morioka, Iwate), Atsushi Hozawa (Tohoku University, Sendai, Miyagi), Yukiko Okami (Gunma University, Maebashi, Gunma), Nobuo Nishi (St Luke’s International University, Tokyo), Yoshitaka Murakami (Toho University, Tokyo), Naoko Miyagawa (Keio University, Tokyo), Kei Asayama (Teikyo University, Tokyo), Takumi Hirata (Tokyo Metropolitan Institute for Geriatrics and Gerontology, Tokyo), Shigeru Inoue (Tokyo Medical University, Tokyo), Toshiyuki Ojima (Hamamatsu University School of Medicine, Hamamatsu, Shizuoka), Hiroshi Yatsuya (Nagoya University Graduate School of Medicine, Nagoya, Aichi), Hideaki Nakagawa (Kanazawa Medical University, Uchinada, Ishikawa), Yoshikuni Kita (Tsuruga Nursing University, Tsuruga, Fukui), Yasuyuki Nakamura, Naomi Miyamatsu, Akiko Harada, Keiko Kondo, Itsuko Miyazawa, Sayuki Torii, Kaori Kitaoka (Shiga University of Medical Science, Otsu, Shiga), Nagako Okuda (Kyoto Prefectural University, Kyoto), Katsushi Yoshita (Osaka Metropolitan University, Osaka), Yoshihiro Miyamoto, Makoto Watanabe (National Cerebral and Cardiovascular Center, Suita, Osaka), Akira Fujiyoshi, Aya Higashiyama (Wakayama Medical University, Wakayama), Takashi Hisamatsu (Okayama University, Okayama), Kazunori Kodama, Fumiyoshi Kasagi (Radiation Effects Research Foundation, Hiroshima), Yutaka Kiyohara (Hisayama Research Institute for Lifestyle Diseases, Hisayama, Fukuoka), Hisatomi Arima (Fukuoka University, Fukuoka), Toshiharu Ninomiya, Jun Hata (Kyushu University, Fukuoka), Koshi Nakamura (Ryukyu University, Nakagami, Okinawa).

Author contributions

Makiko Abe (Conceptualization [lead], Formal analysis [lead], Methodology [lead], Software [lead], Visualization [lead], Writing—original draft [lead]), Hisatomi Arima (Conceptualization [lead], Data curation [lead], Investigation [lead], Formal analysis [lead], Methodology [lead], Software [lead], Supervision, Writing—original draft [lead]), Nagako Okuda (Conceptualization [lead], Data curation [lead], Investigation [lead], Methodology [equal], Project administration [equal], Resources [equal], Supervision [lead], Writing—review & editing [lead]), Hirokazu Taniguchi (Methodology [supporting], Writing—review & editing [supporting]), Atsushi Satoh (Data curation [equal], Investigation [equal], Methodology [supporting], Writing—review & editing [supporting]), Nobuo Nishi (Conceptualization [equal], Data curation [lead], Investigation [lead], Methodology supporting], Resources [equal], Supervision [equal], Writing—review & editing [equal]), Naoki Aono (Conceptualization [supporting], Methodology [supporting], Validation [lead], Writing—review & editing [supporting]), Aya Higashiyama (Conceptualization [equal], Data curation [supporting], Investigation [supporting], Methodology [supporting], Writing—review & editing [equal]), Harumitsu Suzuki (Conceptualization [equal], Data curation [equal], Investigation [equal], Methodology [supporting], Writing—review & editing [equal]), Yukiko Okami (Data curation [equal], Investigation [equal], Methodology [supporting], Writing—review & editing [supporting]), Keiko Kondo (Data curation [equal], Investigation [equal], Methodology [supporting], Writing—review & editing [supporting]), Kaori Kitaoka (Data curation [equal], Investigation [equal], Methodology [supporting], Writing—review & editing [supporting]), Aya Kadota (Conceptualization [lead], Funding acquisition [equal], Methodology [equal], Administration [equal], Resources [equal], Supervision [equal], Writing—review & editing [equal]), Tomonori Okamura (Data curation [lead], Investigation [lead], Funding acquisition [equal], Project administration [equal], Resources [equal], Supervision [equal], Writing—review & editing [equal]), Takayoshi Ohkubo (Data curation [lead], Investigation [lead], Funding acquisition [equal], Project administration [equal], Resources [equal], Supervision [equal], Writing—review & editing [equal]), Akira Okayama (Conceptualization [lead], Data curation [lead], Investigation [lead], Project administration [equal], Resources [equal], Supervision [equal], Writing—review & editing [equal]), and Katsuyuki Miura (Conceptualization [lead], Data curation [lead], Investigation [lead], Funding acquisition [lead], Project administration [lead], Resources [lead], Supervision [lead], Writing—review & editing [equal]).

All the authors have read and agreed to the published version of the manuscript.

Supplementary material

Supplementary material is available at Journal of Occupational Health online.

Funding

This study was supported by a Grant-in-Aid from the Ministry of Health, Labour and Welfare under the auspices of the Japanese Association for Cerebro-cardiovascular Disease Control, a Research Grant for Cardiovascular Diseases (7A-2) from the Ministry of Health, Labour and Welfare, and Health and Labour Sciences Research Grants, Japan (Comprehensive Research on Aging and Health [H11-Chouju-046, H14-Chouju-003, H17-Chouju-012, H19-Chouju-Ippan-014] and Comprehensive Research on Life-Style Related Diseases including Cardiovascular Diseases and Diabetes Mellitus [H22-Junkankitou-Seishuu -Sitei-017, H25-Junkankitou-Seishuu-Sitei-022, H30-Junkankitou-Sitei-002, 21FA2002, 24FA2002]) and Fukuoka University (GW2324).

Conflicts of interest

The authors have no conflicts of interest regarding this article.

Data availability

Data were provided by the Ministry of Health, Labour and Welfare Japan and the authors do not have the right to share them.