Does an increase in working hours affect mortality risk? The relationship between working hours and mortality among the older population

Abstract

BACKGROUND:

Population aging, caused by an increase in life expectancy and decrease in fertility rates, has created changes and challenges in various spheres, including the labor market. Though health deteriorates with age, more and more older adults choose to stay in the labor force and work into late life.

OBJECTIVE:

Understanding the effects of various work conditions on the health of older workers is crucial for designing policies and interventions to ensure healthy late life and maintain a productive workforce. To contribute to this endeavor, this study investigates the relationship between long working hours (LWH) and mortality among older populations.

METHODS:

The study uses the Cox proportional hazards regression model to investigate data from the Health and Retirement Survey (HRS) between the years 1992–2016, a longitudinal nationally representative dataset from the United States.

RESULTS:

The results indicate that working 50 hours or more per week is not associated with an increased risk of mortality, for the full sample (1.45 [95% CI: 0.86, 2.45]), for both genders (females 0.51 [95% CI: 0.06, 4.28], males 1.45 [95% CI: 0.81, 2.61]), and for immigrants (female immigrants 0.55 [95% CI: 0.06, 4.75], male immigrants 1.44 [95% CI: 0.79, 2.62]).

CONCLUSIONS:

This analysis confirms and extends the findings of earlier studies by taking into consideration the potential impact of many demographic, socioeconomic, work-related and health-related factors.

Keywords

Immigrants population aging work conditions

1 Introduction

Population aging, caused by an increase in life expectancy and decrease in fertility rates, is creating many changes and challenges in social, economic, and political life. From the perspective of the labor market, this phenomenon changes the level and composition of the labor force. It is also associated with transformations in pension systems, wage structures, employment, and productivity [1, 2], among others. Beyond these economic consequences, an increase in the number of older workers and participation of “more experienced adults” in the labor force has repercussions for health systems and outcomes. Indeed, research shows that work conditions have various impacts on the health of older workers [3, 4], who already face age-related physical and cognitive deterioration [5]. Hence, identifying the relationships between various work conditions and health outcomes among older workers is central for policy design to ensure healthy late life and a productive workforce. Accordingly, this paper strives to contribute to this end by investigating the relationship between long working hours (LWH) and mortality among older workers.

In fact, LWH is a much-debated labor concern with clear scientific basis and applicable policy considerations [6, 7]. Working time was one of the earliest labor issues regulated at national and international level, as its adverse health and familial effects were recognized in the 1800s [8]. Though many OECD countries’ percentage of the population working 40 hours or more per week declined steadily between 1995 and 2017 [9], long working hours remain an issue for some countries. For instance, in 2017, more than 20% of the workforce in Japan, Republic of Korea, Mexico, and Turkey averaged at least 50 hours of weekly working time [10]. In the same year, 16.3% of workers worked 50 hours and more per week in the United States [11].

Accordingly, a large portion of the workforce in many countries potentially faces adverse health outcomes from LWH. Indeed, LWH is associated with many chronic diseases, physical conditions, and psychological disorders, including cardiovascular diseases [12], musculoskeletal disorders [13, 14], diabetes [15], hypertension [16], metabolic syndrome [17], depression [18], stress [19], sleep disruption [20], and suicidal ideation [21]. LWH is correlated with scores of other adverse outcomes, including fatigue [22, 23], poor self-assessed health [24], all-cause mortality [25, 26], obesity [27, 28], and work-related injuries [29, 30], among many others. Moreover, people working long work hours are more likely to consume alcohol and/or smoke [27, 28].

Taking into account the various health outcomes of long working hours, some of which could prove fatal in old age [31], and the steadily increasing number of older people in the workforce, it is important to identify if a relationship exists between LWH and mortality among older adults, to evade premature deaths through focused interventions. Accordingly, the present study draws on responses to the United States Health and Retirement Survey, a nationally representative longitudinal data set, in order to investigate this relationship using Cox regression survival analyses. This study also considers gender and immigration background, self-assessed health, and number of diseases, as well as overnight hospital stay. To the best of our knowledge, no previous study has investigated the relationship between LWH and mortality for older age working adults. The present study is a first in this respect. Following a literature review on LWH, mortality risk, and immigrants’ health, this article introduces the data and methodology, presents and discusses the results, and finally concludes with implications for future research and policy.

2 Literature review

2.1 Long working hours (LWH) and mortality risk

The various health outcomes associated with long working hours have been investigated greatly [12 , 32–35]. While only a couple of studies focus on all-cause mortality, their findings are inconsistent and contradictory due to different sampling characteristics, varying thresholds for long work hours, and unique industry-specific circumstances. For instance, based on their study of full-time office staff aged 39–61 from London’s civil service departments, Virtanen et al. [36] suggest that employees working 1, 2, and 3–4 hours overtime per day have no significant elevated risk for all-cause mortality compared to those who do not work overtime. On the other hand, Nylen, Voss, and Floderus [25] combined data from the Swedish twin registry with self-reported overtime figures and found gendered differences in mortality risk. Their results demonstrated that for women, up to five hours of overtime did not affect the risk of death, while over five hours increased the risk of death. However, men demonstrated a reduced risk of mortality for a maximum of five hours of overtime per week and no effect on mortality for working over five hours of overtime. This result is in line with empirical evidence provided by Holtermann et al. [37], whose study of employed men aged 40–59 in Denmark demonstrated that working 41 to 45 hours per week or more was not a significant increased risk factor for all-cause mortality among men. Furthermore, in a well-known study carried out in Northern Ireland using data set of full-time employees aged 20–64, O’reilly and Rosato [26] indicated no evidence of an increased mortality risk associated with long working hours (more than 55 hours per week). In a recent study, Hannerz and Soll-Johanning [38] followed a cohort of full-time employees aged 20–64 in Denmark and showed that, while working 41–48 hours per week was statistically significantly associated with decreased all-cause mortality rates, there was no significant effect of overtime work (48 hours or more per week).

Two recent studies have linked long working hours and mortality in the United States. In a meta-analysis, Goh et al. [39] put forth that, as workplace stressors, long work hours and overtime work increase mortality by 20%. Furthermore, Goh, Pfeffer, and Zenios [40] established a model linking long work hours to increased mortality and incremental health care costs. However, neither of these studies were empirical in nature.

Filling a critical gap in the literature, the present study extends the investigation of the relationship between long work hours and mortality to older adults, specifically those 50 and older. Moreover, it tests the existence of this relationship on the basis of nationally representative data for the United States.

2.2 Immigrants’ health and mortality in the United States

The United States experienced rapid growth in its foreign-born population throughout the 21st century. More than 44.7 million immigrants lived in the United States in 2018, representing 13.7 percent of the country’s population. Both immigrants and US-born older adults represent 16% in their respective populations [41]. The immigrant population has a significant influence on American society demographically, economically, fiscally, and politically. [42, 43]. Prior studies have concluded that adult and older immigrants in the United States enjoy better health and lower mortality than the ones born in the United States. [44 –49]. The proposed explanation for these health advantages depends on two hypotheses. The first, referred to as the “healthy-migrant effect,” proposes that immigrants are healthier before migration than people born in their host country, as well as people in their home country who do not migrate [50 –52]. The second, the “salmon-bias effect,” asserts that immigrants might return to their homeland when they get ill or old in order to die in their country of origin [53 –56]. On the other hand, research indicates that this advantage could decrease and converge with native-born levels over time due to limited socioeconomic opportunities [42 , 58], adoption of unhealthy lifestyles [44 , 59–61], exposure to stress [62, 63], discrimination [64, 65], and adverse occupational conditions, including extended work hours [66 –68]. Given that a large body of this literature focuses on explaining the factors that affect immigrants’ health and mortality, the present study also investigates the effect of working 50 hours per week or more on older immigrants’ all-cause mortality.

3 Data and methodology

This study relies on the Health and Retirement Survey (HRS), which is conducted by the University of Michigan. HRS is a longitudinal panel study that surveys a representative sample of older Americans. Beginning in 1992, HRS has followed 26,000 individuals on a biennial basis and collected information on respondents’ socioeconomic characteristics, health status, and work histories. The most recent available survey data was collected in 2016. The RAND user-friendly HRS version was used for the present analysis.

The statistical method used in this study is based on survival analysis. This type of analysis is useful for explaining factors that contribute to mortality risk. The Cox proportional hazards regression model states that the hazard rate for the j^th subject in the data is: $\begin{matrix} h (t | x_{1 j}, x_{2 j}, \dots, x_{kj}) = \\ h_{0} (t) \exp (β_{1} x_{1 j} + β_{2} x_{2 j} + \dots + β_{k} x_{kj}) \end{matrix}$ (1)

Table 1

Summary statistics, Health and Retirement Study (1992-2016)

	Overall sample						Immigrants
Explanatory variables at baseline	Full sample		Women		Men		Full sample		Women		Men
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Long working hours (50+)	0.06	0.239	0.03	0.181	0.10	0.298	0.06	0.240	0.03	0.182	0.10	0.299
Age	68.93	10.441	69.04	10.884	68.76	9.780	69.01	10.429	69.17	10.882	68.79	9.752
White	0.80	0.398	0.79	0.407	0.82	0.385	0.82	0.388	0.80	0.398	0.83	0.372
Income	9.66	1.374	9.52	1.392	9.85	1.323	9.71	1.277	9.57	1.286	9.91	1.238
Married	0.61	0.489	0.51	0.500	0.74	0.437	0.60	0.489	0.51	0.500	0.74	0.438
Number of cohabitants	2.18	1.169	2.11	1.196	2.28	1.123	2.12	1.084	2.05	1.117	2.22	1.027
Number of children	3.23	2.177	3.21	2.177	3.26	2.178	3.19	2.146	3.18	2.151	3.21	2.140
Based_self-reported health	1.97	0.954	1.99	0.957	1.93	0.947	1.95	0.939	1.97	0.939	1.92	0.939
Self-reported health	2.92	1.133	2.93	1.134	2.90	1.132	2.89	1.128	2.90	1.127	2.88	1.131
Change in health status	0.27	0.445	0.28	0.450	0.26	0.438	0.27	0.444	0.28	0.448	0.26	0.437
Obesity	0.28	0.447	0.29	0.452	0.26	0.438	0.28	0.448	0.29	0.453	0.26	0.440
Based_body mass index	24.60	4.780	24.27	5.139	25.06	4.180	24.64	4.809	24.29	5.170	25.11	4.204
Hospitalization	0.28	0.449	0.27	0.446	0.29	0.452	0.28	0.451	0.28	0.448	0.29	0.454
Number of diseases	2.04	1.480	2.07	1.479	1.99	1.482	2.07	1.482	2.09	1.481	2.03	1.482
Health limits work	0.31	0.462	0.31	0.464	0.30	0.459	0.31	0.464	0.32	0.465	0.31	0.461
Alcohol consumption	0.47	0.499	0.41	0.492	0.56	0.496	0.47	0.499	0.42	0.493	0.56	0.497
Cigarette consumption	0.14	0.347	0.13	0.337	0.15	0.360	0.14	0.351	0.14	0.343	0.16	0.362
Stress	0.16	0.362	0.14	0.348	0.18	0.380	0.16	0.362	0.14	0.349	0.17	0.379
Physical effort	0.10	0.298	0.08	0.278	0.12	0.323	0.10	0.297	0.08	0.277	0.12	0.321
Tenure	4.04	9.143	3.18	7.720	5.25	10.722	4.13	9.318	3.27	7.867	5.35	10.921
Number of jobs	2.06	1.466	1.85	1.395	2.36	1.512	2.09	1.469	1.89	1.400	2.38	1.513
N	174,235		101,988		72,247		155,849		90,974		64,875
n	30,789		17,527		13,262		27,246		15,462		11,784

The Cox model is advantageous because it does not make potentially untenable distributional assumptions about the hazard rate. A positive Cox regression coefficient shows a higher hazard probability for an independent variable. The dependent variable of the study in such a survival analysis is the risk of an increase in all-cause mortality in a given year. The primary independent variable is binary, with “1” indicating that the participants work more than 49 hours per week. 1

Moreover, to check the robustness of the results, we used the following regression form for the probit: $D_{M} = α + β LWH + γ C + ɛ$ (2)

where D_M equals 1 if the person died during the sample years. LWH is a binary variable that equals 1 if the person works 50 hours per week or more. C is a matrix that includes the control variables that are given in Table A1 in the Appendix.

In this study, the dependent variable in the survival analysis is the risk of a subject dying in a given year, and it is analyzed regardless of specific cause of death. We used the respondents’ death information from HRS RAND. Our primary independent variable was working 50 hours per week or more. 2 The control variables included demographic factors (e.g., age, level of education, gender, race), socioeconomic factors (e.g., total household real income, marital status), health-related factors (e.g., obesity, smoking, alcohol use, self-reported health), and work-related factors (e.g., tenure, number of jobs, stress) across five industries. 3 Table A1 in the Appendix provides detailed information on the creation of these variables. As we did not include members of the armed forces in the analysis, the sample involves only the civilian workforce. 4

Table 1 presents summary statistics for the full sample of older participants analyzed in this study. From 1992 to 2016, after excluding observations with missing data, the overall sample includes a total of 17,527 and 13,262 person-years for women and men, respectively. The mean age of sample participants was approximately 69 years for both genders. Approximately 33% of women and 67% of men in the sample worked 50 hours per week or more. 89% of the overall sample had a migrant background, with a mean age of 69 years as well. The immigrant sample comprised 15,462 female and 11,784 male person-years, in which almost 33% of female immigrants and 68% of male immigrants worked 50 hours per week or more. Finally, 1.35% of the overall sample and 1.39% of the immigrant sample died during the survey period. According to the Centers for Disease Control and Prevention (CDC), the death rate in the United States is 863.8 deaths per 100,000 population. Higher death rates compared with the overall population are expected in this age cohort.

Table 2

The results of Cox regression analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (full sample)

	Full sample			Women			Men
Explanatory variables at baseline	HRR	SE	95% CI	HRR	SE	95% CI	HRR	SE	95% CI
Long working hours (50+)^a	1.415	0.365	[0.853 –2.345]	0.456	0.485	[0.057 –3.659]	1.391	0.389	[0.804 –2.408]
Age	1.073^***	0.003	[1.067 –1.078]	1.075^***	0.004	[1.067 –1.083]	1.066^***	0.004	[1.058 –1.074]
White	1.019	0.055	[0.916 –1.134]	1.011	0.076	[0.872 –1.172]	1.029	0.081	[0.881 –1.201]
Income	1.023	0.021	[0.982 –1.066]	1.018	0.028	[0.964 –1.074]	1.004	0.028	[0.949 –1.061]
Married	0.966	0.048	[0.877 –1.064]	0.895	0.069	[0.770 –1.041]	0.799^***	0.054	[0.700 –0.912]
Number of cohabitants	1.049^***	0.018	[1.015 –1.085]	1.064^***	0.023	[1.019 –1.110]	1.024	0.028	[0.972 –1.080]
Number of children	0.994	0.009	[0.976 –1.012]	0.991	0.013	[0.966 –1.017]	0.993	0.013	[0.969 –1.018]
Based_self-reported health	1.095^***	0.026	[1.046 –1.146]	1.123^***	0.037	[1.053 –1.197]	1.065^*	0.035	[0.998 –1.136]
Self-reported health	1.539^***	0.047	[1.449 –1.634]	1.517^***	0.064	[1.396 –1.647]	1.534^***	0.068	[1.407 –1.673]
Change in health status perceived by the individual	1.409^***	0.068	[1.282 –1.547]	1.395^***	0.093	[1.224 –1.590]	1.470^***	0.101	[1.284 –1.682]
Obesity	0.533^***	0.037	[0.464 –0.612]	0.599^***	0.058	[0.496 –0.723]	0.538^***	0.055	[0.441 –0.656]
Based_body mass index	1.018^***	0.006	[1.006 –1.031]	1.004	0.008	[0.988 –1.021]	1.024^**	0.010	[1.005 –1.044]
Hospitalization	1.708^***	0.076	[1.566 –1.864]	1.694^***	0.104	[1.502 –1.910]	1.664^***	0.107	[1.467 –1.887]
Number of diseases	1.169^***	0.018	[1.134 –1.205]	1.197^***	0.027	[1.146 –1.250]	1.143^***	0.025	[1.096 –1.192]
Health limits work	1.249^***	0.059	[1.138 –1.371]	1.422^***	0.095	[1.248 –1.622]	1.081	0.073	[0.947 –1.234]
Alcohol consumption	0.783^***	0.039	[0.711 –0.863]	0.722^***	0.056	[0.621 –0.840]	0.748^***	0.049	[0.658 –0.850]
Cigarette consumption	1.432^***	0.096	[1.256 –1.632]	1.500^***	0.148	[1.236 –1.820]	1.277^***	0.117	[1.068 –1.527]
Stress	0.686^*	0.145	[0.454 –1.037]	0.424^**	0.184	[0.181 –0.994]	0.870	0.214	[0.537 –1.410]
Physical effort	1.101	0.233	[0.727 –1.668]	0.830	0.369	[0.347 –1.986]	1.291	0.325	[0.788 –2.116]
Tenure	1.006	0.008	[0.991 –1.021]	1.003	0.019	[0.967 –1.041]	1.004	0.008	[0.988 –1.021]
Number of jobs	1.036^**	0.017	[1.004 –1.070]	0.972	0.027	[0.921 –1.026]	1.027	0.022	[0.985 –1.072]
Education level dummy variables	✓			✓			✓
Industry dummy variables	✓			✓			✓
N	174,235			101,988			72,247

Abbreviations: HRR = Hazard Rate Ratio; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals 1; the individual worked more than 49 hours a week. ^*p < 0.10,^**p < 0.05,^***p < 0.01.

4 Results

4.1 Cox regression analysis

In this section, we present the estimates of the association of working 50 hours per week or more with all-cause mortality for people aged 50 years and older in the United States. Hazard Rate ratios (HRs) for all-cause mortality were estimated as a function of weekly working hours while controlling for demographic, socioeconomic, health-related, and work-related factors.

The Cox regression provided in Equation 1 was conducted for all participants, including those who were part-time, full-time, retired, and unemployed workers. The findings are reported in Table 2. The results indicate that there is no significant association between working 50 hours per week or more and all-cause mortality for older workers in the United States (HR: 1.41, 95% CI: 0.85, 2.34). The effects are insignificant, but the coefficient is positive, which matched our expectations for this model. Regarding the control variables, as reported in Table 2, older adults are more vulnerable to mortality; a rise in the age of the respondent is associated with a 7% (95% CI: 1.06, 1.07) higher HR for risk of all-cause mortality. Moreover, most health-related control variables are significant in the full sample model. Especially, deterioration in respondents’ self-reported health results is statistically significant 54% (95% CI: 1.45, 1.63) higher HR for risk of all-cause mortality. A rise in the respondent’s number of diseases caused a 17% (95% CI: 1.13, 1.21) higher HR for risk of all-cause mortality. People who reported an overnight hospital stay had a 71% (95% CI: 1.56, 1.86) higher HR for risk of all-cause mortality. Having an impairment or health problem that limited the kind or amount of paid work led to a 25% (95% CI: 1.14, 1.37) higher HR for risk of all-cause mortality. On the other hand, job-related control variables such as working jobs requiring great physical effort all or most of the time were not associated with risk of all-cause mortality.

Table 3
The results of Cox regression analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (workers in laborforce)

Explanatory variables at baseline Full sample Women Men

HRR SE 95% CI HRR SE 95% CI HRR SE 95% CI

Long working hours (50+)^a 1.451 0.389 [0.858 –2.454] 0.511 0.555 [0.061 –4.283] 1.450 0.435 [0.805 –2.611]

Age 1.078^* 0.016 [1.046 –1.110] 1.047 0.031 [0.989 –1.110] 1.078^* 0.020 [1.039 –1.118]

White 0.991 0.269 [0.583 –1.687] 0.395^ 0.175 [0.166 –0.939] 1.619 0.604 [0.780 –3.362]

Income 1.056 0.080 [0.910 –1.226] 0.854^ 0.061 [0.743 –0.982] 1.133 0.109 [0.939 –1.367]

Married 0.889 0.202 [0.569 –1.388] 1.146 0.465 [0.517 –2.541] 0.581^ 0.153 [0.347 –0.973]

Number of cohabitants 0.930 0.084 [0.779 –1.111] 0.898 0.162 [0.631 –1.279] 0.952 0.096 [0.782 –1.160]

Number of children 1.054 0.054 [0.953 –1.166] 1.001 0.104 [0.816 –1.228] 1.081 0.061 [0.969 –1.207]

Based_self-reported health 0.968 0.136 [0.735 –1.275] 0.919 0.251 [0.538 –1.568] 0.981 0.162 [0.710 –1.356]

Self-reported health 1.376^ 0.215 [1.013 –1.870] 1.277 0.348 [0.748 –2.179] 1.359^* 0.247 [0.952 –1.939]

Change in health status perceived by the individual 1.315 0.310 [0.828 –2.088] 3.908^* 1.652 [1.707 –8.947] 0.906 0.254 [0.523 –1.570]

Obesity 0.561^ 0.152 [0.330 –0.956] 0.636 0.314 [0.242 –1.674] 0.629 0.198 [0.339 –1.167]

Based_body mass index 1.050^ 0.026 [1.001 –1.101] 1.027 0.042 [0.947 –1.113] 1.027 0.035 [0.961 –1.097]

Hospitalization 1.732^ 0.393 [1.110 –2.701] 2.111 1.040 [0.804 –5.542] 1.532 0.406 [0.911 –2.576]

Number of diseases 1.345^* 0.121 [1.128 –1.603] 1.416^ 0.235 [1.023 –1.960] 1.340^*** 0.141 [1.090 –1.646]

Health limits work 1.503 0.375 [0.922 –2.450] 1.478 0.714 [0.573 –3.811] 1.603^* 0.458 [0.916 –2.806]

Alcohol consumption 0.876 0.184 [0.580 –1.323] 0.949 0.376 [0.436 –2.063] 0.781 0.193 [0.482 –1.268]

Cigarette consumption 1.582^* 0.430 [0.929 –2.697] 1.598 0.899 [0.531 –4.815] 1.458 0.465 [0.781 –2.722]

Stress 0.735 0.164 [0.474 –1.138] 0.421^* 0.195 [0.169 –1.045] 0.924 0.240 [0.556 –1.537]

Physical effort 1.251 0.292 [0.791 –1.977] 0.874 0.431 [0.332 –2.298] 1.536 0.430 [0.887 –2.659]

Tenure 1.007 0.009 [0.990 –1.024] 1.014 0.025 [0.967 –1.064] 1.001 0.009 [0.983 –1.019]

Number of jobs 1.069 0.076 [0.931 –1.228] 1.032 0.187 [0.724 –1.471] 1.033 0.081 [0.886 –1.205]

Education level dummy variables ✓ ✓ ✓

Industry dummy variables ✓ ✓ ✓

N 53,839 27,601 26,238

Explanatory variables at baseline	Full sample	Women	Men
Long working hours (50+)^a	1.451	0.389	[0.858 –2.454]	0.511	0.555	[0.061 –4.283]	1.450	0.435	[0.805 –2.611]
Age	1.078^***	0.016	[1.046 –1.110]	1.047	0.031	[0.989 –1.110]	1.078^***	0.020	[1.039 –1.118]
White	0.991	0.269	[0.583 –1.687]	0.395^**	0.175	[0.166 –0.939]	1.619	0.604	[0.780 –3.362]
Income	1.056	0.080	[0.910 –1.226]	0.854^**	0.061	[0.743 –0.982]	1.133	0.109	[0.939 –1.367]
Married	0.889	0.202	[0.569 –1.388]	1.146	0.465	[0.517 –2.541]	0.581^**	0.153	[0.347 –0.973]
Number of cohabitants	0.930	0.084	[0.779 –1.111]	0.898	0.162	[0.631 –1.279]	0.952	0.096	[0.782 –1.160]
Number of children	1.054	0.054	[0.953 –1.166]	1.001	0.104	[0.816 –1.228]	1.081	0.061	[0.969 –1.207]
Based_self-reported health	0.968	0.136	[0.735 –1.275]	0.919	0.251	[0.538 –1.568]	0.981	0.162	[0.710 –1.356]
Self-reported health	1.376^**	0.215	[1.013 –1.870]	1.277	0.348	[0.748 –2.179]	1.359^*	0.247	[0.952 –1.939]
Change in health status perceived by the individual	1.315	0.310	[0.828 –2.088]	3.908^***	1.652	[1.707 –8.947]	0.906	0.254	[0.523 –1.570]
Obesity	0.561^**	0.152	[0.330 –0.956]	0.636	0.314	[0.242 –1.674]	0.629	0.198	[0.339 –1.167]
Based_body mass index	1.050^**	0.026	[1.001 –1.101]	1.027	0.042	[0.947 –1.113]	1.027	0.035	[0.961 –1.097]
Hospitalization	1.732^**	0.393	[1.110 –2.701]	2.111	1.040	[0.804 –5.542]	1.532	0.406	[0.911 –2.576]
Number of diseases	1.345^***	0.121	[1.128 –1.603]	1.416^**	0.235	[1.023 –1.960]	1.340^***	0.141	[1.090 –1.646]
Health limits work	1.503	0.375	[0.922 –2.450]	1.478	0.714	[0.573 –3.811]	1.603^*	0.458	[0.916 –2.806]
Alcohol consumption	0.876	0.184	[0.580 –1.323]	0.949	0.376	[0.436 –2.063]	0.781	0.193	[0.482 –1.268]
Cigarette consumption	1.582^*	0.430	[0.929 –2.697]	1.598	0.899	[0.531 –4.815]	1.458	0.465	[0.781 –2.722]
Stress	0.735	0.164	[0.474 –1.138]	0.421^*	0.195	[0.169 –1.045]	0.924	0.240	[0.556 –1.537]
Physical effort	1.251	0.292	[0.791 –1.977]	0.874	0.431	[0.332 –2.298]	1.536	0.430	[0.887 –2.659]
Tenure	1.007	0.009	[0.990 –1.024]	1.014	0.025	[0.967 –1.064]	1.001	0.009	[0.983 –1.019]
Number of jobs	1.069	0.076	[0.931 –1.228]	1.032	0.187	[0.724 –1.471]	1.033	0.081	[0.886 –1.205]
Education level dummy variables	✓	✓	✓
Industry dummy variables	✓	✓	✓
N	53,839	27,601	26,238

Abbreviations: HRR = Hazard Rate Ratio; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals 1; the individual worked more than 49 hours a week. ^*p < 0.10,^**p < 0.05,^***p < 0.01.

As mortality generally varies by sex [70, 71], to ascertain sex-specific outcomes for our study, we also re-examined the association LWH and all-cause mortality separately for women and men. The findings presented in Table 2 indicate that for both women (HR: 0.46, 95% CI: 0.06, 3.65) and men (HR: 1.39, 95% CI: 0.80, 2.41), there is no significant association between working 50 hours per week or more and all-cause mortality, compared to their counterparts working fewer than 50 hours per week. Most health-related control variables reveal similar results in each of the three models. A deterioration in respondents’ self-reported health for women and men is associated with 52% (95% CI: 1.40, 1.65) and 53% (95% CI: 1.41, 1.67) higher HR for risk of all-cause mortality, respectively. Women and men who reported an overnight hospital stay had a 69% (95% CI: 1.50, 1.91) and 66% (95% CI: 1.47, 1.89) higher HR for risk of all-cause mortality. Having an impairment or health problem that limited the kind or amount of paid work was associated with a 42% (95% CI: 1.24, 1.62) higher HR for risk of all-cause mortality for women, while it did not have an effect on men’s risk of all-cause mortality. On the other hand, most of the job-related control variables were not associated with higher HR for the risk of all-cause mortality for both men and women.

Moreover, the Cox regression provided in Equation 1 was also estimated for workers in the labor force, and the findings of the impact of LWH on mortality are presented in Table 3. The results suggest that there is no significant association between working 50 hours per week or more and all-cause mortality, referencing older working respondents working fewer than 50 hours per week in the United States both in the full sample and subsamples. Compared with other employees of the same sex, all-cause mortality HRs for individuals working 50 hours per week or more was estimated at 0.51 (95% CI: 0.06, 4.28) among women and 1.45 (95% CI: 0.81, 2.61) among men. Regarding the control variables, for instance, while an increment in individuals’ body mass index had a 5% (95% CI: 1.00, 1.10) higher HR for the risk of all-cause mortality in the full sample, there was no association across the subgroups. People who reported overnight hospital stays had 73% (95% CI: 1.11, 1.70) higher HR for the risk of all-cause mortality, with no difference between women and men employees. Furthermore, a deterioration in working individuals’ self-reported health was statistically associated with a 38% (95% CI: 1.01, 1.87) increase in all-cause mortality for the whole sample, as well as a 36% (95% CI: 0.95, 1.94) higher risk for all-cause mortality among men, but no significant enhanced risk for women (HR 1.27, 95% CI: 0.75, 2.18). Moreover, a rise in the respondent’s number of diseases was statistically significantly associated with increased all-cause mortality rates among the full sample (HR 1.35, 95% CI: 1.13, 1.60), men workers (HR 1.34, 95% CI: 1.09, 1.65), as well as for women workers (HR 1.42, 95% CI: 1.02, 1.9).

Table 4

The results of Cox regression analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (immigrant sample)

Explanatory variables at baseline	Full sample			Women			Men
	HRR	SE	95% CI	HRR	SE	95% CI	HRR	SE	95% CI
Long working hours (50+)^a	1.426	0.377	[0.849 –2.393]	0.495	0.527	[0.061 –3.992]	1.380	0.398	[0.784 –2.427]
Age	1.071^***	0.003	[1.065 –1.077]	1.074^***	0.004	[1.065 –1.082]	1.064^***	0.004	[1.056 –1.072]
White	1.032	0.059	[0.922 –1.154]	1.006	0.080	[0.860 –1.176]	1.055	0.088	[0.896 –1.242]
Income	1.010	0.022	[0.968 –1.055]	1.003	0.029	[0.948 –1.061]	0.991	0.029	[0.934 –1.050]
Married	0.964	0.049	[0.872 –1.066]	0.895	0.072	[0.765 –1.047]	0.795^***	0.055	[0.693 –0.910]
Number of cohabitants	1.079^***	0.019	[1.042 –1.117]	1.086^***	0.025	[1.038 –1.136]	1.067^**	0.030	[1.010 –1.127]
Number of children	1.001	0.009	[0.983 –1.020]	1.000	0.014	[0.973 –1.027]	1.001	0.013	[0.976 –1.026]
Based_self-reported health	1.090^***	0.027	[1.039 –1.144]	1.125^***	0.039	[1.052 –1.204]	1.052	0.036	[0.984 –1.125]
Self-reported health	1.569^***	0.050	[1.474 –1.670]	1.546^***	0.068	[1.417 –1.686]	1.561^***	0.071	[1.429 –1.707]
Change in health status perceived by the individual	1.424^***	0.071	[1.291 –1.570]	1.394^***	0.097	[1.215 –1.598]	1.503^***	0.108	[1.306 –1.730]
Obesity	0.514^***	0.038	[0.445 –0.595]	0.585^***	0.060	[0.479 –0.714]	0.512^***	0.055	[0.415 –0.631]
Based_body mass index	1.020^***	0.006	[1.007 –1.032]	1.006	0.009	[0.989 –1.023]	1.025^**	0.010	[1.005 –1.046]
Hospitalization	1.657^***	0.076	[1.514 –1.814]	1.613^***	0.103	[1.424 –1.828]	1.649^***	0.110	[1.448 –1.878]
Number of diseases	1.156^***	0.019	[1.120 –1.194]	1.177^***	0.027	[1.125 –1.232]	1.137^***	0.026	[1.088 –1.188]
Health limits work	1.238^***	0.062	[1.122 –1.365]	1.434^***	0.101	[1.249 –1.646]	1.057	0.075	[0.920 –1.214]
Alcohol consumption	0.795^***	0.041	[0.718 –0.879]	0.728^***	0.059	[0.621 –0.854]	0.763^***	0.051	[0.669 –0.871]
Cigarette consumption	1.424^***	0.098	[1.244 –1.630]	1.506^***	0.154	[1.232 –1.841]	1.265^**	0.118	[1.053 –1.520]
Stress	0.629^**	0.137	[0.411 –0.963]	0.376^**	0.172	[0.154 –0.920]	0.807	0.204	[0.492 –1.326]
Physical effort	1.217	0.261	[0.799 –1.853]	0.918	0.406	[0.385 –2.185]	1.438	0.369	[0.870 –2.378]
Tenure	1.004	0.008	[0.989 –1.020]	1.000	0.020	[0.961 –1.041]	1.002	0.008	[0.986 –1.019]
Number of jobs	1.023	0.018	[0.989 –1.058]	0.955	0.028	[0.902 –1.011]	1.017	0.023	[0.972 –1.063]
Education level dummy variables	✓			✓			✓
Industry dummy variables	✓			✓			✓
N	155,849			90,974			64,875

Abbreviations: Hazard Rate Ratio; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals 1; the individual worked more than 49 hours a week. ^*p < 0.10, ^**p < 0.05, ^***p < 0.01.

Considering immigrants, the general point of view is that foreign-born individuals tend to have a lower risk for mortality when compared to their United States-born counterparts [46, 72]. If this situation is valid for our study, it could provide another explanation for the established findings between LWH and mortality, since about 89% of the sample consists of immigrants. To examine this connection, we reconducted the survival analysis with only immigrants’ data. Moreover, we carried out the analysis for male and female immigrants separately for two reasons. First, gender is a crucial selection factor in migration [73], and second, it is an important indicator of the healthy immigrant effect [74].

Table 5

The results of Cox regression analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (immigrant workers in labor force)

Explanatory variables at baseline	Full sample			Women			Men
	HRR	SE	95% CI	HRR	SE	95% CI	HRR	SE	95% CI
Long working hours (50+)^a	1.463	0.401	[0.855 –2.503]	0.550	0.604	[0.064 –4.741]	1.438	0.440	[0.789 –2.620]
Age	1.073^***	0.017	[1.041 –1.106]	1.042	0.031	[0.983 –1.104]	1.074^***	0.020	[1.035 –1.114]
White	0.907	0.249	[0.530 –1.555]	0.343^**	0.157	[0.140 –0.842]	1.476	0.553	[0.708 –3.075]
Income	1.050	0.086	[0.895 –1.232]	0.846^**	0.068	[0.722 –0.991]	1.123	0.115	[0.918 –1.372]
Married	0.829	0.192	[0.526 –1.307]	1.060	0.441	[0.469 –2.394]	0.527^**	0.141	[0.313 –0.889]
Number of cohabitants	0.952	0.089	[0.794 –1.143]	0.872	0.181	[0.581 –1.310]	1.002	0.101	[0.822 –1.221]
Number of children	1.053	0.056	[0.948 –1.169]	0.991	0.105	[0.804 –1.221]	1.084	0.062	[0.968 –1.213]
Based_self-reported health	0.982	0.141	[0.741 –1.301]	0.901	0.266	[0.505 –1.609]	1.022	0.167	[0.741 –1.408]
Self-reported health	1.422^**	0.227	[1.041 –1.944]	1.331	0.381	[0.759 –2.334]	1.384^*	0.255	[0.965 –1.986]
Change in health status perceived by the individual	1.231	0.300	[0.764 –1.984]	3.710^***	1.602	[1.591 –8.650]	0.848	0.245	[0.481 –1.493]
Obesity	0.546^**	0.152	[0.316 –0.943]	0.560	0.289	[0.204 –1.542]	0.631	0.203	[0.336 –1.186]
Based_body mass index	1.050^**	0.026	[1.001 –1.102]	1.023	0.044	[0.940 –1.112]	1.033	0.034	[0.968 –1.103]
Hospitalization	1.819^***	0.415	[1.163 –2.846]	2.376^*	1.174	[0.902 –6.259]	1.592^*	0.425	[0.944 –2.687]
Number of diseases	1.325^***	0.125	[1.102 –1.594]	1.411^*	0.249	[0.999 –1.994]	1.309^**	0.144	[1.055 –1.624]
Health limits work	1.398	0.362	[0.841 –2.323]	1.287	0.681	[0.456 –3.632]	1.545	0.450	[0.873 –2.733]
Alcohol consumption	0.940	0.202	[0.617 –1.432]	0.970	0.399	[0.433 –2.174]	0.845	0.212	[0.516 –1.382]
Cigarette consumption	1.549	0.428	[0.902 –2.661]	1.530	0.908	[0.478 –4.896]	1.436	0.462	[0.764 –2.698]
Stress	0.664^*	0.152	[0.424 –1.040]	0.363^**	0.174	[0.142 –0.928]	0.843	0.223	[0.502 –1.415]
Physical effort	1.352	0.318	[0.852 –2.144]	1.030	0.495	[0.402 –2.644]	1.643^*	0.472	[0.936 –2.884]
Tenure	1.006	0.009	[0.988 –1.024]	1.009	0.026	[0.960 –1.062]	1.002	0.009	[0.984 –1.020]
Number of jobs	1.064	0.079	[0.920 –1.231]	0.997	0.195	[0.680 –1.462]	1.040	0.084	[0.888 –1.220]
Education level dummy variables	✓			✓			✓
Industry dummy variables	✓			✓			✓
N	48,227			24,826			23,401

Abbreviations: Hazard Rate Ratio; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals 1; the individual worked more than 49 hours a week. ^*p < 0.10, ^**p < 0.05, ^***p < 0.01.

Table 6

The results of probit model analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (full sample)

Explanatory variables at baseline	Full sample			Women			Men
	COEF.	SE	95% CI	COEF.	SE	95% CI	COEF.	SE	95% CI
Long working hours (50+)^a	0.164	0.121	[–0.074 –0.402]	–0.313	0.426	[–1.147 –0.522]	0.164	0.131	[–0.093 –0.420]
Age	0.050^***	0.002	[0.045 –0.054]	0.051^***	0.004	[0.044 –0.058]	0.047^***	0.003	[0.040 –0.053]
White	0.034	0.035	[–0.035 –0.104]	0.026	0.051	[–0.073 –0.126]	0.047	0.050	[–0.052 –0.146]
Income	0.026^**	0.012	[0.004 –0.049]	0.021	0.015	[–0.008 –0.051]	0.018	0.016	[–0.013 –0.049]
Married	–0.016	0.029	[–0.074 –0.042]	–0.072	0.045	[–0.160 –0.016]	–0.148^***	0.042	[–0.230 ––0.067]
Number of cohabitants	0.032^***	0.011	[0.011 –0.054]	0.039^***	0.015	[0.009 –0.068]	0.024	0.016	[–0.008 –0.055]
Number of children	–0.012^**	0.006	[–0.024 ––0.000]	–0.014	0.009	[–0.031 –0.004]	–0.011	0.008	[–0.028 –0.005]
Based_self-reported health	0.211^***	0.017	[0.178 –0.244]	0.221^***	0.024	[0.174 –0.268]	0.195^***	0.023	[0.149 –0.240]
Self-reported health	0.207^***	0.018	[0.173 –0.242]	0.209^***	0.025	[0.160 –0.259]	0.197^***	0.025	[0.149 –0.246]
Change in health status perceived by the individual	0.225^***	0.026	[0.175 –0.276]	0.216^***	0.036	[0.145 –0.287]	0.257^***	0.037	[0.184 –0.331]
Obesity	–0.433^***	0.041	[–0.514 ––0.353]	–0.362^***	0.057	[–0.474 ––0.250]	–0.417^***	0.059	[–0.533 ––0.300]
Based_body mass index	0.013^***	0.004	[0.006 –0.020]	0.005	0.005	[–0.005 –0.014]	0.014^**	0.006	[0.003 –0.025]
Hospitalization	0.300^***	0.024	[0.254 –0.347]	0.300^***	0.033	[0.234 –0.365]	0.285^***	0.034	[0.218 –0.351]
Number of diseases	0.087^***	0.011	[0.065 –0.108]	0.107^***	0.016	[0.075 –0.139]	0.073^***	0.015	[0.044 –0.103]
Health limits work	–0.003	0.025	[–0.053 –0.047]	0.046	0.036	[–0.025 –0.117]	–0.066^*	0.036	[–0.137 –0.005]
Alcohol consumption	–0.177^***	0.029	[–0.233 ––0.121]	–0.229^***	0.044	[–0.316 ––0.142]	–0.210^***	0.038	[–0.284 ––0.136]
Cigarette consumption	0.197^***	0.040	[0.119 –0.276]	0.193^***	0.061	[0.073 –0.312]	0.146^***	0.054	[0.040 –0.252]
Stress	–0.174^*	0.093	[–0.356 –0.009]	–0.317^*	0.176	[–0.663 –0.028]	–0.096	0.111	[–0.313 –0.121]
Physical effort	0.009	0.096	[–0.179 –0.198]	–0.202	0.194	[–0.582 –0.178]	0.106	0.113	[–0.115 –0.327]
Tenure	–0.003	0.004	[–0.010 –0.005]	–0.005	0.008	[–0.022 –0.011]	–0.004	0.004	[–0.012 –0.004]
Number of jobs	–0.025^**	0.010	[–0.046 ––0.005]	–0.067^***	0.018	[–0.102 ––0.033]	–0.038^***	0.014	[–0.065 ––0.010]
Education level dummy variables	✓			✓			✓
Industry dummy variables	✓			✓			✓
N	203,636			118,093			85,542

Abbreviations: COEF. = Coefficient; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals to 1; the individual worked more than 49 hours a week. ^*p < 0.10,^**p < 0.05,^***p < 0.01.

Table 7

The results of probit model analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (workers in laborforce)

Explanatory variables at baseline	Full sample			Women			Men
	COEF.	SE	95% CI	COEF.	SE	95% CI	COEF.	SE	95% CI
Long working hours (50+)	0.105	0.131	[–0.152 –0.361]	–0.611	0.812	[–2.202 –0.981]	0.107	0.156	[–0.199 –0.414]
Age	0.040^***	0.010	[0.021 –0.059]	0.052	0.041	[–0.029 –0.133]	0.042^***	0.012	[0.019 –0.066]
White	0.035	0.131	[–0.222 –0.292]	–0.592	0.477	[–1.527 –0.344]	0.303	0.210	[–0.109 –0.715]
Income	0.084^*	0.048	[–0.010 –0.178]	–0.041	0.075	[–0.187 –0.106]	0.124^**	0.060	[0.006 –0.242]
Married	–0.054	0.112	[–0.274 –0.167]	0.271	0.295	[–0.308 –0.849]	–0.341^**	0.165	[–0.664 ––0.018]
Number of cohabitants	–0.014	0.042	[–0.097 –0.068]	–0.074	0.100	[–0.271 –0.123]	0.003	0.053	[–0.100 –0.107]
Number of children	0.026	0.026	[–0.025 –0.077]	–0.035	0.061	[–0.154 –0.084]	0.050	0.034	[–0.015 –0.116]
Based_self-reported health	0.086	0.076	[–0.063 –0.235]	0.222	0.257	[–0.283 –0.726]	0.063	0.095	[–0.124 –0.250]
Self-reported health	0.155^*	0.081	[–0.003 –0.313]	0.217	0.175	[–0.127 –0.560]	0.159	0.103	[–0.044 –0.361]
Change in health status perceived by the individual	0.123	0.111	[–0.094 –0.341]	0.764^*	0.397	[–0.013 –1.541]	–0.087	0.155	[–0.390 –0.217]
Obesity	–0.363^**	0.149	[–0.655 ––0.071]	–0.432	0.370	[–1.157 –0.294]	–0.320^*	0.184	[–0.680 –0.041]
Based_body mass index	0.023^*	0.012	[–0.000 –0.047]	0.018	0.028	[–0.036 –0.072]	0.017	0.018	[–0.018 –0.052]
Hospitalization	0.323^***	0.105	[0.118 –0.528]	0.381	0.251	[–0.112 –0.873]	0.315^**	0.132	[0.057 –0.572]
Number of diseases	0.144^***	0.055	[0.036 –0.252]	0.201	0.165	[–0.122 –0.525]	0.175^**	0.072	[0.033 –0.317]
Health limits work	0.198	0.122	[–0.042 –0.437]	0.211	0.331	[–0.438 –0.860]	0.247	0.154	[–0.054 –0.549]
Alcohol consumption	–0.119	0.100	[–0.315 –0.077]	–0.172	0.236	[–0.634 –0.291]	–0.158	0.129	[–0.412 –0.096]
Cigarette consumption	0.183	0.128	[–0.067 –0.433]	0.347	0.324	[–0.288 –0.983]	0.106	0.174	[–0.235 –0.447]
Stress	–0.189^*	0.109	[–0.403 –0.025]	–0.503^*	0.274	[–1.040 –0.035]	–0.082	0.136	[–0.348 –0.184]
Physical effort	0.057	0.110	[–0.158 –0.273]	–0.309	0.371	[–1.037 –0.419]	0.191	0.140	[–0.082 –0.465]
Tenure	–0.000	0.005	[–0.009 –0.009]	0.003	0.014	[–0.025 –0.030]	–0.002	0.005	[–0.012 –0.009]
Number of jobs	0.023	0.035	[–0.046 –0.093]	0.021	0.104	[–0.183 –0.226]	0.014	0.044	[–0.072 –0.101]
Education level dummy variables	✓			✓			✓
Industry dummy variables	✓			✓			✓
N	66,684			33,525			33,159

Abbreviations: COEF. = Coefficient; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals to 1; the individual worked more than 49 hours a week. ^*p < 0.10,^**p < 0.05,^***p < 0.01.

Table 8

The results of probit model analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (immigrant sample)

Explanatory variables at baseline	Full sample			Women			Men
	COEF	SE	95% CI	COEF	SE	95% CI	COEF	SE	95% CI
Long working hours (50+)^a	0.160	0.123	[–0.081 –0.402]	–0.282	0.421	[–1.107 –0.542]	0.154	0.135	[–0.110 –0.418]
Age	0.047^***	0.002	[0.042 –0.052]	0.048^***	0.004	[0.041 –0.055]	0.044^***	0.004	[0.037 –0.051]
White	0.033	0.037	[–0.040 –0.105]	0.016	0.052	[–0.087 –0.119]	0.049	0.053	[–0.056 –0.154]
Income	0.015	0.012	[–0.009 –0.039]	0.009	0.016	[–0.023 –0.041]	0.006	0.016	[–0.026 –0.039]
Married	–0.026	0.030	[–0.086 –0.033]	–0.083^*	0.047	[–0.174 –0.008]	–0.160^***	0.043	[–0.245 ––0.075]
Number of cohabitants	0.048^***	0.012	[0.026 –0.071]	0.051^***	0.016	[0.020 –0.082]	0.046^***	0.017	[0.013 –0.080]
Number of children	–0.008	0.006	[–0.020 –0.004]	–0.008	0.009	[–0.025 –0.010]	–0.008	0.009	[–0.025 –0.009]
Based_self-reported health	0.201^***	0.017	[0.168 –0.235]	0.213^***	0.025	[0.165 –0.261]	0.184^***	0.024	[0.137 –0.230]
Self-reported health	0.220^***	0.018	[0.184 –0.255]	0.220^***	0.026	[0.169 –0.271]	0.212^***	0.026	[0.162 –0.262]
Change in health status perceived by the individual	0.224^***	0.027	[0.171 –0.276]	0.208^***	0.037	[0.134 –0.281]	0.264^***	0.039	[0.188 –0.340]
Obesity	–0.447^***	0.043	[–0.532 ––0.362]	–0.372^***	0.060	[–0.490 ––0.254]	–0.437^***	0.063	[–0.559 ––0.314]
Based_body mass index	0.014^***	0.004	[0.007 –0.021]	0.005	0.005	[–0.005 –0.016]	0.016^***	0.006	[0.005 –0.027]
Hospitalization	0.280^***	0.024	[0.232 –0.328]	0.272^***	0.034	[0.205 –0.339]	0.275^***	0.035	[0.206 –0.344]
Number of diseases	0.080^***	0.011	[0.057 –0.102]	0.097^***	0.017	[0.065 –0.130]	0.069^***	0.016	[0.038 –0.101]
Health limits work	–0.003	0.026	[–0.055 –0.048]	0.046	0.038	[–0.028 –0.120]	–0.066^*	0.038	[–0.140 –0.008]
Alcohol consumption	–0.157^***	0.029	[–0.215 ––0.100]	–0.213^***	0.046	[–0.303 ––0.123]	–0.184^***	0.039	[–0.260 ––0.108]
Cigarette consumption	0.185^***	0.041	[0.105 –0.265]	0.184^***	0.062	[0.063 –0.305]	0.134^**	0.056	[0.025 –0.243]
Stress	–0.200^**	0.096	[–0.388 ––0.013]	–0.350^*	0.182	[–0.706 –0.006]	–0.120	0.114	[–0.344 –0.103]
Physical effort	0.047	0.097	[–0.143 –0.238]	–0.164	0.193	[–0.542 –0.214]	0.146	0.115	[–0.079 –0.371]
Tenure	–0.003	0.004	[–0.010 –0.004]	–0.006	0.009	[–0.023 –0.011]	–0.004	0.004	[–0.012 –0.004]
Number of jobs	–0.031^***	0.011	[–0.052 ––0.010]	–0.076^***	0.018	[–0.112 ––0.040]	–0.041^***	0.015	[–0.070 ––0.013]
Education level dummy variables	✓			✓			✓
Industry dummy variables	✓			✓			✓
N	180,951			104,633			76,317

Abbreviations: COEF = Coefficient; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals to 1, the individual worked more than 49 hours in a week. ^*p < 0.10, ^**p < 0.05, ^***p < 0.01.

The results obtained from the fundamental analysis on the impact of LWH on all-cause mortality are provided in Table 4, and findings for immigrant workers in the labor force are presented in Table 5. According to both tables, there was no significant association between all-cause mortality and working 50 hours per week or more across all sub-samples. More specifically, in Table 4, compared with other older adult immigrants of the same sex, all-cause mortality HRs for individuals working 50 hours per week or more was estimated at 0.49 (95% CI: 0.06, 4.00) among the female immigrants and 1.38 (95% CI: 0.78, 2.43) among the male immigrants. As for the control variables, an increase in older immigrants’ age significantly raises the mortality risk about 7% for women (95% CI: 1.06, 1.08) and 6% for men (95% CI: 1.05, 1.07). Additionally, most of the prominent health-related factors potentially contributed to the higher mortality rates in all sub-samples. For instance, increased mortality risk was observed among those reported an overnight hospital stay for the full sample at 65% (95% CI: 1.51, 1.81), for women at 61% (95% CI: 1.42, 1.83), and for men at 64% (95% CI: 1.45, 1.88). A rise in an immigrant’s number of diseases created a higher mortality risk for the full sample (HRR: 1.15, 95% CI: 1.12, 1.19), for female immigrants (HRR: 1.17 95% CI: 1.12, 1.23), and for male immigrants (HRR: 1.13, 95% CI: 1.09, 1.19). A deterioration in immigrants’ self-reported health remains a life-threatening condition with an increased mortality rates by about 57% (95% CI: 1.47, 1.67) for the full sample, 55% (95% CI: 1.42, 1.69) for women, and 56% (95% CI: 1.43, 1.71) for men. On the other hand, as Table 5 shows, all-cause mortality HRs for individuals working 50 hours per week or more was estimated at 0.55 (95% CI: 0.06, 4.75) among the female immigrants and 1.44 (95% CI: 0.79, 2.62) among the male immigrants in the labor force. An overnight hospital stay contributed to the significantly higher mortality rates for all immigrants (HRR:1.82, 95% CI: 1.16, 2.84), as well as women (HRR: 2.37, 95% CI: 0.90, 6.26) and men (HRR: 1.59, 95% CI: 0.94, 2.69) separately. An increase in an immigrant’s number of diseases significantly raised the mortality risk for the full sample by about 32% (95% CI: 1.10, 1.59), for women by 41% (95% CI: 1.00, 2.00), and for men by 31% (95% CI: 1.05, 1.62). Aging leads to a 7.3% (95% CI: 0.85, 2.50) higher HR for the risk of all-cause mortality for the all immigrants and male immigrants. A deterioration of immigrants’ self-reported health created statistically significant mortality risk for all immigrant workers (HRR. 1.42, 95% CI: 1.04, 1.94) and for female immigrants (HRR: 1.33, 95% CI: 0.76, 2.33).

4.2 Logistic regression analysis estimates

Although results estimated through the Cox regression analysis strongly suggest that working longer hours is not associated with all-cause mortality, we also performed a probit regression model using the panel data as a robustness check for all the aforementioned models. As shown in Tables 6 and 7, the coefficient of long working hours (50+) estimated from the probit models aligns with the results of the survival analysis, indicating that we would fail to reject the null hypothesis and conclude that the regression coefficient for working 50 hours per week or more is not statistically different from zero in all models. Moreover, the sign of the coefficients matched the findings reported in Tables 2 and 3. In Table 6, the coefficient of the 50-hour dummy is 0.164 (Marginal effect (ME): 0.0042, Standard Error (SE): 0.0031) for the full sample, –0.313 (ME: –0.0071, SE: 0.0097) for women, and 0.164 (ME: 0.0048, SE: 0.0038) for men. In Table 7, the coefficient of the 50-hour dummy is 0.105 (ME: 0.00039, SE: 0.00049) for the full sample, –0.611 (ME: –0.0010, SE: 0.0011) for women, and 0.107 (ME: 0.00057, SE: 0.00083) for men. In both tables, although not significant, this means that people who work 50 hours per week or more are more likely to face increased risk of death when considering the full sample and the male subsample, while the opposite is true for women.

As concerns the control variables, the estimated regression coefficients of most of the health and job-related factors presented in Table 6 are significant and reveal similar results in each of the three models, whereas some of the established coefficients regarding health and job-related factors presented in Table 7, such as hospitalization and number of diseases, are significant and vary across subgroups.

As can be seen in Table 8, the coefficient of the 50-hour dummy was 0.160 (ME: 0.0043, SE: 0.0033) for all immigrants in the sample, –0.282 (ME: –0.0066, SE: 0.0099) for female immigrants, and 0.154 (ME: 0.0047, SE: 0.0041) for male immigrants. Regarding the health-related factors, an increase in the number of diseases, a deterioration in self-reported health, and having an impairment or health problem was associated with a higher mortality risk across all sub-samples. Additionally, some of the demographic factors, such as number of children and increased age, also led to an increase in immigrants’ mortality risk.

Probit regression analyses were also conducted for immigrant employees in the labor force who worked 50 hours per week or more, and the results are presented in Table 9. Although not significant, the coefficient of the 50-hour dummy is 0.112 (ME: 0.00044, SE: 0.00054) for the full sample, –0.607 (ME: –0.0011, SE: 0.0014) for female immigrants, and 0.117 (ME: 0.00066, SE: 0.00092) for male immigrants. For the entire immigrant sample and the male immigrant subsample, an overnight hospital stay, an increase in the number of diseases, a decrease in self-reported health, and advanced age all contributed to the risk of mortality.

Table 9
The results of probit model analysis; Effect of long working hours on the risk of mortality in the Health and Retirement Study, 1992–2016 (immigrant workers in labor force)

Explanatory variables at baseline Full sample Women Men

COEF SE 95% CI COEF SE 95% CI COEF SE 95% CI

Long working hours (50+)^a 0.112 0.136 [–0.155 –0.379] –0.607 0.885 [–2.342 –1.127] 0.117 0.165 [–0.206 –0.440]

Age 0.040^* 0.010 [0.020 –0.061] 0.050 0.044 [–0.036 –0.136] 0.043^* 0.013 [0.017 –0.070]

White 0.015 0.138 [–0.257 –0.286] –0.675 0.526 [–1.706 –0.356] 0.330 0.238 [–0.136 –0.796]

Income 0.084 0.054 [–0.023 –0.190] –0.043 0.088 [–0.215 –0.129] 0.124^* 0.069 [–0.011 –0.259]

Married –0.089 0.117 [–0.319 –0.141] 0.232 0.288 [–0.332 –0.797] –0.405^ 0.182 [–0.762 ––0.048]

Number of cohabitants –0.020 0.044 [–0.107 –0.067] –0.080 0.107 [–0.289 –0.129] 0.004 0.057 [–0.107 –0.115]

Number of children 0.027 0.028 [–0.027 –0.082] –0.042 0.064 [–0.168 –0.083] 0.057 0.037 [–0.015 –0.128]

Based_self-reported health 0.103 0.080 [–0.054 –0.260] 0.221 0.276 [–0.319 –0.762] 0.090 0.099 [–0.103 –0.284]

Self-reported health 0.188^ 0.086 [0.018 –0.357] 0.242 0.181 [–0.112 –0.596] 0.194^* 0.114 [–0.030 –0.419]

Change in health status perceived by the individual 0.089 0.116 [–0.139 –0.316] 0.724^* 0.406 [–0.071 –1.520] –0.132 0.165 [–0.456 –0.191]

Obesity –0.362^** 0.156 [–0.667 ––0.057] –0.512 0.399 [–1.294 –0.270] –0.292 0.191 [–0.666 –0.081]

Based_body mass index 0.022^* 0.012 [–0.002 –0.047] 0.015 0.028 [–0.039 –0.069] 0.018 0.018 [–0.017 –0.054]

Hospitalization 0.324^*** 0.108 [0.112 –0.536] 0.419^* 0.253 [–0.076 –0.915] 0.306^ 0.139 [0.034 –0.578]

Number of diseases 0.138^ 0.058 [0.025 –0.251] 0.206 0.181 [–0.149 –0.561] 0.166^ 0.076 [0.016 –0.315]

Health limits work 0.161 0.126 [–0.086 –0.409] 0.076 0.287 [–0.486 –0.639] 0.241 0.162 [–0.076 –0.559]

Alcohol consumption –0.093 0.104 [–0.297 –0.110] –0.147 0.239 [–0.615 –0.321] –0.129 0.135 [–0.394 –0.136]

Cigarette consumption 0.179 0.133 [–0.081 –0.440] 0.329 0.338 [–0.334 –0.991] 0.105 0.184 [–0.256 –0.466]

Stress –0.234^ 0.117 [–0.462 ––0.005] –0.562^* 0.297 [–1.143 –0.019] –0.129 0.147 [–0.416 –0.158]

Physical effort 0.101 0.112 [–0.119 –0.320] –0.221 0.361 [–0.929 –0.487] 0.241^* 0.145 [–0.043 –0.525]

Tenure –0.001 0.005 [–0.010 –0.009] –0.001 0.014 [–0.029 –0.027] –0.001 0.006 [–0.012 –0.010]

Number of jobs 0.023 0.038 [–0.051 –0.096] –0.001 0.105 [–0.206 –0.205] 0.020 0.048 [–0.074 –0.114]

Education level dummy variables ✓ ✓ ✓

Industry dummy variables ✓ ✓ ✓

N 59,375 29,968 29,407

Explanatory variables at baseline	Full sample	Women	Men
Long working hours (50+)^a	0.112	0.136	[–0.155 –0.379]	–0.607	0.885	[–2.342 –1.127]	0.117	0.165	[–0.206 –0.440]
Age	0.040^***	0.010	[0.020 –0.061]	0.050	0.044	[–0.036 –0.136]	0.043^***	0.013	[0.017 –0.070]
White	0.015	0.138	[–0.257 –0.286]	–0.675	0.526	[–1.706 –0.356]	0.330	0.238	[–0.136 –0.796]
Income	0.084	0.054	[–0.023 –0.190]	–0.043	0.088	[–0.215 –0.129]	0.124^*	0.069	[–0.011 –0.259]
Married	–0.089	0.117	[–0.319 –0.141]	0.232	0.288	[–0.332 –0.797]	–0.405^**	0.182	[–0.762 ––0.048]
Number of cohabitants	–0.020	0.044	[–0.107 –0.067]	–0.080	0.107	[–0.289 –0.129]	0.004	0.057	[–0.107 –0.115]
Number of children	0.027	0.028	[–0.027 –0.082]	–0.042	0.064	[–0.168 –0.083]	0.057	0.037	[–0.015 –0.128]
Based_self-reported health	0.103	0.080	[–0.054 –0.260]	0.221	0.276	[–0.319 –0.762]	0.090	0.099	[–0.103 –0.284]
Self-reported health	0.188^**	0.086	[0.018 –0.357]	0.242	0.181	[–0.112 –0.596]	0.194^*	0.114	[–0.030 –0.419]
Change in health status perceived by the individual	0.089	0.116	[–0.139 –0.316]	0.724^*	0.406	[–0.071 –1.520]	–0.132	0.165	[–0.456 –0.191]
Obesity	–0.362^**	0.156	[–0.667 ––0.057]	–0.512	0.399	[–1.294 –0.270]	–0.292	0.191	[–0.666 –0.081]
Based_body mass index	0.022^*	0.012	[–0.002 –0.047]	0.015	0.028	[–0.039 –0.069]	0.018	0.018	[–0.017 –0.054]
Hospitalization	0.324^***	0.108	[0.112 –0.536]	0.419^*	0.253	[–0.076 –0.915]	0.306^**	0.139	[0.034 –0.578]
Number of diseases	0.138^**	0.058	[0.025 –0.251]	0.206	0.181	[–0.149 –0.561]	0.166^**	0.076	[0.016 –0.315]
Health limits work	0.161	0.126	[–0.086 –0.409]	0.076	0.287	[–0.486 –0.639]	0.241	0.162	[–0.076 –0.559]
Alcohol consumption	–0.093	0.104	[–0.297 –0.110]	–0.147	0.239	[–0.615 –0.321]	–0.129	0.135	[–0.394 –0.136]
Cigarette consumption	0.179	0.133	[–0.081 –0.440]	0.329	0.338	[–0.334 –0.991]	0.105	0.184	[–0.256 –0.466]
Stress	–0.234^**	0.117	[–0.462 ––0.005]	–0.562^*	0.297	[–1.143 –0.019]	–0.129	0.147	[–0.416 –0.158]
Physical effort	0.101	0.112	[–0.119 –0.320]	–0.221	0.361	[–0.929 –0.487]	0.241^*	0.145	[–0.043 –0.525]
Tenure	–0.001	0.005	[–0.010 –0.009]	–0.001	0.014	[–0.029 –0.027]	–0.001	0.006	[–0.012 –0.010]
Number of jobs	0.023	0.038	[–0.051 –0.096]	–0.001	0.105	[–0.206 –0.205]	0.020	0.048	[–0.074 –0.114]
Education level dummy variables	✓	✓	✓
Industry dummy variables	✓	✓	✓
N	59,375	29,968	29,407

Abbreviations: COEF = Coefficient; SE = Standard Error; CI = Confidence Interval. Note: ^aIt equals to 1, the individual worked more than 49 hours in a week. ^*p < 0.10, ^**p < 0.05, ^***p < 0.01.

In conclusion, the estimation results from the probit models for the immigrant population affirm those in our base specification reported in Tables 4 and 5. We failed to find any statistically significant evidence of an association between working 50 hours per week or more and all-cause mortality, and the signs of the probit model estimates coincide with the results estimated from the Cox model across all specifications.

5 Discussion and conclusion

This study used the HRS data set, which follows individuals biennially from 1992 to 2016, and constituted a first attempt to uncover the relationship between LWH and mortality among older adults in the United States. Despite the fatal potential of some health outcomes of LWH, such as stroke [12], cardiovascular disease [18], and diabetes [15], the findings of this study show that LWH is not associated with a statistically significant higher or lower probability of having all-cause mortality for all participants, including part-time, full-time, retired and unemployed workers. The main findings do not change with several control methods, including regarding gender and immigrant background. The estimation results affirm those from our base specification.

The results obtained from the Cox regression and Probit model analyses confirm and extend the findings of earlier studies by taking into consideration the potential impact of many demographic, socioeconomic, work-related, and health-related factors. Any comparison with earlier studies should be made carefully due to different age intervals, definitions of long working hours, and time frames. Nonetheless, our findings largely confirm those of Holtermann et al. [37] and Virtanen et al. [36], which did not find an increased risk of mortality resulting from working overtime It also lends support to the findings of O’reilly and Rosato [26] and Hannerz and Soll-Johanning [38] on the basis of nationally representative longitudinal data. On the other hand, it contradicts the results of Goh, Pfeffer, and Zenios [40], at least for the older population. Our study also did not find gendered differences as Nylen, Voss, and Floderus’ [26], which evidenced elevated mortality for women working five or more overtime hours per week.

Thanks to the survey structure of HRS, the present study could focus on older immigrants. The general point of view is that foreign-born individuals tend to have a lower mortality risk than their peers born in the United States [47, 73]. If this situation was valid for our study, it could provide another explanation for the established findings between LWH and mortality. Therefore, we reconducted the same analyses for immigrant workers in the labor force. The results were similar to the estimates obtained from our baseline model.

This study has important strengths. Its main strength is that the entirety of the working older participants was randomly selected from the target inhabitants in the United States. The second significant advantage is germane to a healthy worker effect, which proposes that individuals who has poor health may cut back on working hours. This may be the cause of the non-significant association. To circumvent this problem, we included several variables related to personal health (e.g., self-reported general health status, obesity, health problems limiting the kind or amount of paid work, number of diseases, overnight hospital stay, and alcohol and cigarette consumption). Furthermore, in line with the literature [75], because our study used twelve waves of flow-up data, the healthy worker effect was substantially reduced. The third strength is that contrary to other studies [39], this study drew on information on working hours for both main jobs and subsidiary occupations from 1992 to 2016, which prevented the underestimation of results.

While we did not find a relationship between LWH and mortality, our results should be carefully considered, as the sample only included older adults. The need for work time directives to reduce the risk of health problems arising from long work hours should not be dismissed. Moreover, policies and interventions should take into account our auxiliary findings. In this regard, older adults’ self-assessed health, number of diseases, and record of overnight hospital stay should be closely monitored by physicians and occupational health service providers so that precautionary measures can be taken.

This study has some important limitations, mostly owning to the nature of the HRS. For example, all of the study variables (e.g., height, weight, etc.) were self-reported by survey respondents. Additionally, the survey data also excludes institutionalized adults. Finally, the dataset does not include information on youth, so our analysis could not extend to the younger generations.

Further research should be conducted covering a wider age span and incorporating data from different countries. In addition, studies could be undertaken that consider the association between LWH and mortality for specific sectors and occupations, to assess if there are higher risk occupations for health hazards related to overtime work.

Footnotes

Appendix

Table A1

Table A1 Definition of variables using study¹

Demographic factors
Age	Age of the respondent
White	Race of the respondent (Baseline = White, Reference = Black/African American and Other)
Education level dummy variables	The years of education variable is assigned by looking at reports all waves of data (years of school completed; 0–11, 12, 13–15, 16+)
Socioeconomic factors
Household total real income	Total income respondent and spouse (deflated with CPI index and got logarithm)
Married	Marital status reported for each wave (Baseline = married, Reference = married, spouse absent, partnered, separated, divorced, separated/divorced, widowed, never married)
Number of cohabitants	Number of people living in the household, including the respondent and spouse
Number of children	The number of living children of the respondent and spouse or partner
Health-related factors
Mortality risk	If the person dies during the sample years, it will equal to one.
Based_self-reported health	Respondent’s self-reported general health status based on interviewed year (excellent, very good, good, fair, poor)
Obesity	Body Mass Index> = 30
Based_body mass index	Respondent’s body mass index, height, and weight. BMI is weight divided by the square height (in the first year of the interview)
Hospitalization	It indicates whether the respondent reports any overnight hospital stay in the Reference period (Baseline = yes, reence = no)
Self-reported health	The Respondent’s self-reported general health status in the interviewed year (1 means excellent and 5 means poor)
Number of diseases	The number of diseases the respondent reports ever having
Health limits work	It indicates whether an impairment or health problem limits the kind or amount of paid work for the respondent (Baseline = yes, Reference = no)
Change in health status	Respondent’s self-reported change in health since the last interview or in the last two years (Baseline = Somewhat worse, Much worse, Reference = Much better, Somewhat better, Same)
Alcohol consumption	It indicates whether the Respondent ever drinks alcoholic beverages.
Cigarette consumption	It indicates whether the respondent smokes cigarettes now
Work-related factors
Long working hours (50+)	It would equal 1 if the individual worked 50 h/week or more and 0 in other cases. Working hours per week include both for main and second jobs.
Physical effort	Respondent says her/his job requires lots of physical effort (Baseline = all/almost all the time, most of the time, Reference = some of the time, none/almost none of the time, does not apply)
Stress	Respondent agrees with the statement that her/his job involves lots of stress (Baseline = strongly agree, agree, Reference = strongly disagree, does not apply)
Tenure	Respondent’s years of tenure on the current job
Number of jobs	It is the number of jobs the Respondent reports having through job history.
Industry dummy variables	1. Agric/Forest/Fish, Mining and Constr
	2. Mnfg: Non-durable,Mnfg: Durable,Transportation
	3. Wholesale, Retail,Finan/Ins/RealEs
	4. Burns/Repair Svc, Personal Service, Entertain/Recreation, Prof/Related Svcs
	5. Public Administration

¹Definitions are written from the RAND HRS Longitudinal File 2016 (V1) Documentation.

Acknowledgments

The authors would like to thank the editor Karen Jacobs, the editor’s assistants Amanda Nardone and Lindsey Sousa and associate publisher Axana Scherbeijn for their assistance in the evaluation process and anonymous referees for their insightful comments, which helped us to significantly improve our paper.

Conflict of interest

The authors declare that there are no conflicts of interest.

Funding

This research received no external funding.

Working hours per week include both main and second jobs.

In previous studies, very long hours, or extreme work hours, or extreme work are used as synonyms and are conceptualized as individuals’ weekly work hours of 50 or more [].

We re-categorized 13 industries into 5 industrial dummy variables.

When we include military officers, the results do not change.

References

Ferting

, Schmidt

, Kösters

, Board

, Frommknecht

, Heinke

, et al. Gerontocracy in motion?: European cross-country evidence on the labor market consequences of population ageing. Essen: Rheinisch-Westfälisches Institut für Wirtschaftsforschung; 2003.

Rada

. Population Ageing and its Global Challenges. In: McDaniel SA, Zimmer Z, editors. Global Ageing in the Twenty-First Century: Challenges, Opportunities and Implications. Surrey: Ashgate; 2013;159-74.

Jones

, Latreille

, Sloane

, Staneva

. Work-related health risks in Europe: Are older workers more vulnerable? Social Science & Medicine. 2013;88:18–29.

Henseke

. Good jobs, good pay, better health? The effects of job quality on health among older European workers. The European Journal of Health Economics. 2018;19(1):59–73.

McMahan

, Sturz

. Implications for an aging workforce. Journal of Education for Business. 2006;82(1):50–5.

Dembe

. Long working hours: The scientific basis for concern. Members-only Library. 2016;8(2).

Genin

, Haines

III VY

, Pelletier

, Rousseau

, Marchand

. Why the long hours? Job demands and social exchange dynamics. Work. 2016;55(3):539–48.

ILO. Guide to developing balanced working time arrangements. Geneva: ILO; 2019.

OECD. Percentage point change in the proportion of the employed working 40 hours or more, by gender, 1995 to 2017. OECD Family Database. Paris: OECD; 2018.

10.

ILO. Employment by sex and weekly hours actuallyworked. In: ILO, editor. ILO Stat Explorer: ILO; 2020.

11.

Labor USDo. Household Data Annual Averages: 19. Persons at work in agriculture and nonagricultural industries by hours of work (2017). In: Statistics BoL, editor. Labor Force Statistics from the Current Population Survey 2020.

12.

Kivimäki

, Jokela

, Nyberg

, Singh-Manoux

, Fransson

, Alfredsson

, et al. Long working hours and risk of coronary heart disease and stroke: a systematic review and meta-analysis of published and unpublished data for 603 838 individuals. The Lancet. 2015;386(10005):1739–46.

13.

Lee

J-G

, Kim

, Jung

, Kim

, Lee

J-H

, Lee

K-J

. The association between long working hours and work-related musculoskeletal symptoms of Korean wage workers: data from the fourth Korean working conditions survey (a cross-sectional study). Annals of Occupational and Environmental Medicine. 2018;30(1):67.

14.

Barlin

, Mercan

. Occupation, Working Hours and Arthritis: Evidence from a Nationally Representative Sample of Older Age Adults. Advances in Management & Applied Economics. 2020;10(2):16.

15.

Kivimäki

, Virtanen

, Kawachi

, Nyberg

, Alfredsson

, Batty

, et al. Long working hours, socioeconomic status, and the risk of incident type 2 diabetes: a meta-analysis of published and unpublished data from 222 120 individuals. The lancet Diabetes & Endocrinology. 2015;3(1):27–34.

16.

Guner

. Is there a relationship between working hours and hypertension? Evidence from the United States. Work. 2019;62(2):279–85.

17.

Kobayashi

, Suzuki

, Takao

, Doi

. Long working hours and metabolic syndrome among Japanese men: a cross-sectional study. BMC Public Health. 2012;12(1):395.

18.

Virtanen

, Jokela

, Madsen

, Magnusson Hanson

, Lallukka

, Nyberg

, et al. Long working hours and depressive symptoms: systematic review and meta-analysis of published studies and unpublished individual participant data. Scandinavian journal of work, Environment & Health. 2018(3):239–50.

19.

Lee

, Suh

, Kim

J-E

, Park

. The impact of long working hours on psychosocial stress response among white-collar workers. Industrial Health. 2017;55(1):46–53.

20.

Virtanen

, Ferrie

, Gimeno

, Vahtera

, Elovainio

, Singh-Manoux

, et al. Long working hours and sleep disturbances: the Whitehall II prospective cohort study. Sleep. 2009;32(6):737–45.

21.

Kim

K-U

, Park

S-G

, Kim

H-C

, Lim

J-H

, Lee

S-J

, Jeon

S-H

, et al. Association between Long Working Hours and Suicidal Ideation. Korean J Occup Environ Med.. 2012;24(4):339–46.

22.

Nagashima

, Suwazono

, Okubo

, Uetani

, Kobayashi

, Kido

, et al. Working hours and mental and physical fatigue in Japanese workers. Occupational Medicine. . 2007;57(6):449–52.

23.

Castro

, Carvalhais

, Teles

. Irregular working hours and fatigue of cabin crew. Work. 2015;51(3):505–11.

24.

Ettner

, Grzywacz

. Workers’ perceptions of how jobs affect health: A social ecological perspective. Journal of Occupational Health Psychology. 2001;6(2):101.

25.

Nylen

, Voss

, Floderus

. Mortality among women and men relative to unemployment, part time work, overtime work, and extra work: a study based on data from the Swedish twin registry. Occupational and Environmental Medicine. 2001;58(1):52–7.

26.

O’reilly

, Rosato

. Worked to death? A census-based longitudinal study of the relationship between the numbers of hours spent working and mortality risk. International Journal of Epidemiology. 2013;42(6):1820–30.

27.

Lallukka

, Lahelma

, Rahkonen

, Roos

, Laaksonen

, Martikainen

, et al. Associations of job strain and working overtime with adverse health behaviors and obesity: evidence from the Whitehall II Study, Helsinki Health Study, and the Japanese Civil Servants Study. Social Science & Medicine. 2008;66(8):1681–98.

28.

Artazcoz

, Cortès

, Escribà-Agüir

, Cascant

, Villegas

. Understanding the relationship of long working hours with health status and health-related behaviours. Journal of Epidemiology & Community Health. 2009;63(7):521–7.

29.

Wong

, Chan

, Ngan

. The effect of long working hours and overtime on occupational health: a meta-analysis of evidence from 1998 to 2018. International Journal of Environmental Research and Public Health. 2019;16(12):2102.

30.

Wang

P-C

, Delp

. Health status, job stress and work-related injury among Los Angeles taxi drivers. Work. . 2014;49(4):705–12.

31.

WHO. Causes of death (2016). World Health Observatory. Geneva 2020.

32.

Barnay

. Health, work and working conditions: a review of the European economic literature. The European Journal of Health Economics. 2016;17(6):693–709.

33.

Harrington

. Health effects of shift work and extended hours of work. Occupational and Environmental Medicine. 2001;58(1):68–72.

34.

Seth

, Singh

. Adverse health consequences-a result of long work hours: a review of the current evidence. Indian J Nutr. 2016;3:125.

35.

Van der Hulst

. Long workhours and health. Scandinavian journal of work, Environment & Health. 2003;171-88.

36.

Virtanen

, Ferrie

, Singh-Manoux

, Shipley

, Vahtera

, Marmot

, et al. Overtime work and incident coronary heart disease: the Whitehall II prospective cohort study. European Heart Journal. 2010;31(14):1737–44.

37.

Holtermann

, Mortensen

, Burr

, Søgaard

, Gyntelberg

, Suadicani

. Long work hours and physical fitness: 30-year risk of ischaemic heart disease and all-cause mortality among middle-aged Caucasian men. Heart. 2010;96(20):1638–44.

38.

Hannerz

, Soll-Johanning

. Working hours and all-cause mortality in relation to the EU Working Time Directive: a Danish cohort study. European Journal of Public Health. 2018;28(5):810–4.

39.

Goh

, Pfeffer

, Zenios

, Rajpal

. Workplace stressors & health outcomes: Health policy for the workplace. Behavioral Science & Policy. 2015;1(1):43–52.

40.

Goh

, Pfeffer

, Zenios

. The relationship between workplace stressors and mortality and health costs in the United States. Management Science. 2016.

41.

Migration Policy Institute. 2018.

42.

Treas

, Batalova

. Immigrants and aging. International handbook of population aging: Springer; 2009;365–94.

43.

LaPierre

, Hughes

. Population Aging in Canada and the United States. In: Uhlenberg P, editor. International Handbook of Population Aging. Dordrecht: Springer Netherlands; 2009;191-230.

44.

Cho

, Hummer

. Disability status differentials across fifteen Asian and Pacific Islander groups and the effect of nativity and duration of residence in the US. Social Biology. 2001;48(3-4):171–95.

45.

Markides

, Eschbach

. Hispanic paradox in adult mortality in the United States. International handbook of adult mortality: Springer; 2011;227-40.

46.

Singh

, Siahpush

. Ethnic-immigrant differentials in health behaviors, morbidity, and cause-specific mortality in the United States: an analysis of two national data bases. Human Biology. 2002:83-109.

47.

Abraído-Lanza

, Armbrister

, Flórez

, Aguirre

. Toward a theory-driven model of acculturation in public health research. American Journal of Public Health. 2006;96(8):1342–6.

48.

Antecol

, Bedard

. Unhealthy assimilation: why do immigrants converge to American health status levels? Demography. 2006;43(2):337–60.

49.

Uretsky

, Mathiesen

. The effects of years lived in the United States on the general health status of California’s foreign-born populations. Journal of Immigrant and Minority Health. 2007;9(2):125–36.

50.

Hummer

, Rogers

, Nam

, LeClere

. Race/ethnicity, nativity, and US adult mortality. Social Science Quarterly. 1999:136-53.

51.

Singh

, Siahpush

. All-cause and cause-specific mortality of immigrants and native born in the United States. American Journal of Public Health. 2001;91(3):392.

52.

Dupre

, Gu

, Vaupel

. Survival differences among native-born and foreign-born older adults in the United States. PLoS One. 2012;7(5).

53.

Arenas

, Goldman

, Pebley

, Teruel

. Return migration to Mexico: Does health matter? Demography. 2015;52(6):1853–68.

54.

Markides

, Eschbach

. Aging, migration, and mortality: current status of research on the Hispanic paradox. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 2005;60(Special_Issue_2):S68–S75.

55.

Palloni

, Arias

. Paradox lost: explaining the Hispanic adult mortality advantage. Demography.. 2004;41(3):385–415.

56.

Turra

, Elo

. The impact of salmon bias on the Hispanic mortality advantage: New evidence from social security data. Population Research and Policy Review. 2008;27(5):515.

57.

Myers

, Gao

, Emeka

. The gradient of immigrant age-at-arrival effects on socioeconomic outcomes in the US. International Migration Review. 2009;43(1):205–29.

58.

Choi

. Insurance status and health service utilization among newly-arrived older immigrants. Journal of Immigrant and Minority Health. 2006;8(2):149–61.

59.

Morales

, Lara

, Kington

, Valdez

, Escarce

. Socioeconomic, cultural, and behavioral factors affecting Hispanic health outcomes. Journal of Health Care for the Poor and Underserved. 2002;13(4):477.

60.

Ribble

, PhD

, Keddie

. Understanding the Hispanic paradox. Ethn Dis. 2001;11(3):496–518.

61.

Abraido-Lanza

, Chao

, Flórez

. Do healthy behaviors decline with greater acculturation?: Implications for the Latino mortality paradox. Social Science & Medicine. 2005;61(6):1243–55.

62.

Finch

, Vega

. Acculturation stress, social support, and self-rated health among Latinos in California. Journal of Immigrant Health. 2003;5(3):109–17.

63.

Parent-Lamarche

, Marchand

. Work stress, personality traits, and cortisol secretion: Testing a model for job burnout. Work. 2018;60(3):485–97.

64.

Carrasquillo

, Carrasquillo

, Shea

. Health insurance coverage of immigrants living in the United States: differences by citizenship status and country of origin. American Journal of Public Health. 2000;90(6):917.

65.

Leclere

, Jensen

, Biddlecom

. Health care utilization, family context, and adaptation among immigrants to the United States. Journal of Health and Social Behavior. 1994:370-84.

66.

Hunter

. The spatial association between US immigrant residential concentration and environmental hazards. International Migration Review. 2000;34(2):460–88.

67.

Loh

, Richardson

. Foreign-born workers: trends in fatal occupational injuries, 1996–2001. Monthly Lab Rev. 2004;127:42.

68.

Conway

, Cayuela

, Delclos

, Pompeii

, Ronda

. Association between long work hours and poor self-reported general health among Latin American immigrant and native workers in the United States and Spain. American Journal of Industrial Medicine. 2016;59(12):1105–11.

69.

Burger

. Extreme work hours in Western Europe and North America: diverging trends since the s. Socio-Economic Review. 2020;18(4):1065–87.

70.

Kruger

, Nesse

. Economic transition, male competition, and sex differences in mortality rates. Evolutionary Psychology. 2007;5(2):147470490700500213.

71.

Rogers

, Everett

, Saint Onge

, Krueger

. Social, behavioral, and biological factors, and sex differences in mortality. Demography. 2010;47(3):555–78.

72.

Mehta

, Elo

, Engelman

, Lauderdale

, Kestenbaum

. Life expectancy among US-born and foreign-born older adults in the United States: Estimates from linked social security and medicare data. Demography. 2016;53(4):1109–34.

73.

Cerrutti

, Massey

. On the auspices of female migration from Mexico to the United States. Demography. . 2001;38(2):187–200.

74.

Kwak

. Age and gender variations in healthy immigrant effect: a population study of immigrant well-being in Canada. Journal of International Migration and Integration. 2018;19(2):413–37.

75.

Shah

. Healthy worker effect phenomenon. Indian Journal of Occupational and Environmental Medicine. . 2009;13(2):77.