Relative Income Deprivation and All-Cause Mortality in Japan: Do Life Priorities Matter?

Abstract

Background/Purpose

Relative deprivation (RD) is proposed to affect health through psychosocial stress stemming from upward social comparisons. This study hypothesized that prioritizing values, such as social engagement and personal growth (as opposed to prioritizing work), would inoculate against the toxic effects of upward social comparisons.

Methods

Prospective data of 9,533 subjects (4,475 men and 5,058 women) participating in the Komo-Ise study answering a baseline questionnaire in 1993 and a follow-up survey in 2000 were analyzed. Associations between RD—using Yitzhaki Index (YI) and Income Rank (IR)—and mortality were evaluated using Cox proportional-hazard regression models. At follow-up, people were also asked about what they prioritized in life: work, social engagement, or personal growth.

Results

1,168 deaths (761 men and 407 women) occurred during follow-up (to the end of 2011). Controlling for sociodemographic factors, the hazard ratio (HR) for mortality was 1.22 (95% confidence interval [CI] = 1.08–1.38) per a standard deviation (SD) increase in YI and 1.18 (95% CI = 1.03–1.35) per an SD decrease in IR. Life priorities (LP) were not statistically significantly associated with mortality. In women, the interaction between LP and YI was statistically significant. In fully adjusted models, women who endorsed only work as very important in their lives had a 2.66 (95% CI = 1.23–5.77) times higher HR for mortality per SD increase in YI compared to women who valued social engagement/personal growth. An interaction between LP and RD was not found in men.

Conclusions

Increased relative income deprivation was associated with a higher risk of all-cause mortality independently of absolute income. Focusing on social engagement/personal growth (as opposed to work) appears to inoculate women against the toxic effects of relative deprivation.

Relative income deprivation is hypothesized to be deleterious to health. In this cohort of older Japanese adults, we found that this association among women who said that work was their only priority in life (as opposed to social relationships or personal growth) was a stronger predictor for elevated mortality risk.

Introduction

People’s income is strongly linked to their health in a graded way, where higher levels of income are correlated with better health. One reason is that money can buy healthy goods and services that promote better health while also providing a protective buffer during stressful life events, such as job loss. By contrast, income poverty leads to higher risks of chronic diseases and mortality through stress and maladaptive “coping mechanisms,” such as smoking, heavy drinking, and other substance abuse [1–6]. However, once the basic necessities (food, housing, and medical care) are satisfied, money can be used to engage in positional competition, that is, consumption of items that signal a person’s relative position in the social hierarchy [7]. Positional consumption includes items such as larger houses and faster cars.

The concept of relative deprivation was often used to define poverty, where “People are relatively deprived if they cannot obtain, at all or sufficiently, the conditions of life - that is, the diets, amenities, standards and services - which allow them to play the roles, participate in the relationships and follow the customary behaviour which is expected of them by virtue of their membership of society” [6]. However, according to a broader definition, relative deprivation is “the extent of the difference between the desired situation and that of the person desiring it”, which holds true for those who do not live in absolute poverty as well [8]. A person who fails to attain their aspirational level of consumption can feel relatively deprived even though they may not be deprived in any absolute sense. In turn, relative deprivation is hypothesized to lead to frustration, anxiety, and resentment [9–11]. One plausible mechanism through which positional competition leads to poor health is via overwork and indebtedness [7, 12, 13]. The logic of positional competition proceeds as follows: in order to achieve higher social status, a person needs to work longer and harder to earn more income—even if they already earn enough to live comfortably. As people spend more hours at work, they are forced to spend less time with family and friends, which is an essential source of preserving good health. However, through spending money on positional goods (e.g., bigger house and faster car), positional competition can eventually lead to indebtedness (which is a significant source of psychological stress).

While income is only one of the numerous factors based on which one can feel relatively deprived, in empirical research, relative deprivation has been operationalized by indicators, such as the Yitzhaki Index (YI), which calculates the average of the sum of the gap between an individual’s income and the incomes of everyone else in their social comparison group. Another measure of relative deprivation is the Income Rank (IR), referring to the ordinal position of an individual’s earnings in the income hierarchy [14, 15]. Relative income deprivation measured by the YI or IR has been shown to be correlated with several health and behavioral outcomes, including incident functional disability, poor self-rated health, mental distress, depression, alcohol consumption, smoking, and obesity, among others [16–25]. However, in various studies, higher relative income deprivation was shown to have a stronger association with adverse health outcomes among men compared to women [23, 25–27].

One proposed way to mitigate the toxic health effects of relative income deprivation is to emphasize the importance of social connections or to foster other means of defining self-worth instead of promoting consumerism [13, 20, 28]. This was exemplified by social movements advocating a simpler way of living where self-reflection, social justice, and time spent for family, friends, and community service were considered more valuable than spending money engaging in conspicuous consumption [13, 28, 29]. There has been an increasing number of people leaving their stressful—and often well-paid—jobs opting for lower income and consumption in exchange for shorter working hours and an increased amount of free time to care for others and to pursue personal development and happiness [28]. Focusing on intrinsic goals (e.g., being with family/friends and pursuing self-improvement) instead of extrinsic ones (e.g., financial success, social recognition/status, or physical appearance—engaging in positional competition) has been linked with higher levels of psychological well-being, self-actualization, and vitality and also with fewer physical symptoms and depression [30]. It was shown that, while people with personal strivings linked to extrinsic goals were prone to engage in distracting activities, including smoking, consuming alcohol, or watching TV, intrinsic goal-oriented individuals were more likely to participate in meaningful activities, such as helping others [31]. Extrinsic goals were also reported to be relatively incompatible with values related to family/friends, community, and self-acceptance, leading to more interpersonal problems, loneliness, and lower marital quality [32]. In short, the more people focus on working longer hours to earn money and status, the less time they might be able to devote to spending time with family and friends or caring for children and older relatives, etc. [7, 12, 13, 33]. It is always possible to earn more money, while the satisfaction provided by owning new things might be short lived [13, 34, 35]. Competing and striving for financial success can be detrimental to personal growth or having healthy committed relationships [36]. “If desires, expectations, and standards of comparison increase as rapidly as ‘achievements’, no increase in income, no matter how large, will increase subjective well-being” [35]. While prioritizing social engagement and personal growth (as opposed to prioritizing work) is theorized to inoculate against the toxic effects of invidious social comparisons, there was no attempt so far to evaluate how these factors might affect the relationship between relative income and health.

Building on previous findings where higher relative income deprivation was shown to be associated with cardiovascular diseases (CVD) mortality among men—but not among women [27], the current study had the following aims: first, to reexamine the associations between relative income deprivation and all-cause mortality in a new sample of Japanese adults utilizing an added measure—IR—as well as the well-established YI; second, to explore potential sex differences in the effects of relative income deprivation on mortality; and third, to assess how the perceived importance of social engagement, personal growth, or work might modify these associations, testing the theory that focusing on nonmaterial values might attenuate the negative association between relative income deprivation and health [13, 20, 28].

Individuals with higher YI and lower IR were expected to have a higher risk of all-cause mortality, especially among men. It was further hypothesized that these associations between relative income and mortality would be attenuated among those who greatly value their social connections or personal growth compared to those who focus more on their work.

Methods

Study Population

Longitudinal data were obtained from the Komo-Ise Study cohort collected between 1993 and 2011 [37–39]. In 1993, self-administered questionnaires inquiring about sociodemographics, mental and physical health (chronic diseases, daily activities, etc.), lifestyle, and social networks were distributed through the residents’ association (Jichikai) in partnership with the local municipal government offices to all residents aged 40–69 years living in the village of Komochi (n = 4,875) and the downtown area of the city of Isesaki (n = 7,755) in Gunma Prefecture, Japan. Participants were identified based on the municipal resident registration file (Jumin Kihon Daicho). The response rate was 91.6%; thus, the baseline sample consisted of 11,565 individuals (5,630 men and 5,935 women). In 2000, a Japanese version of the Alameda County Study 1999 questionnaire was administered as a follow-up survey to update lifestyle and health status information [40]. The second wave of questionnaires was also mailed to individuals who had moved away from the study area, and the total response rate was 88.5% with 9,650 participants.

The Alameda County Study was established in 1965, preceded by a pilot study in 1961 with a follow-up in 1964 to examine the validity and reliability of the questionnaire. Also, in 1968, a shorter version of the baseline survey was filled out by a random sample of 1,530 adult participants residing in Alameda County who, then, completed fundamentally the same questionnaire 1 to 2 weeks later. The index of reliability was 79 for all factors (60 items), 82 for questions related to physical health (35 items, e.g., disabilities, chronic conditions, and symptoms), and 76 for psychological factors (19 items, e.g., anomy, isolation, depression, and happiness) [40]. In all subsequent waves of the study, the wording and format of the questions and responses, as well as the style and length of the questionnaire were consistent [40, 41]. The Japanese version of the survey was created with the cooperation of researchers from the Human Population Laboratory at the University of California, Berkeley, and was pretested enlisting 30 participants aged 50–69 years from the Komo-Ise Study area.

The present study included 9,533 respondents (4,475 men and 5,058 women) who answered both the baseline and the follow-up surveys. We excluded 112 participants who no longer lived in the study area in 2000 and 5 participants whose resident registration file record could not be obtained. Written consent was obtained from all participants at the beginning of the study. The Komo-Ise cohort study was approved by the Epidemiologic Research Ethic Committee of Gunma University Faculty of Medicine, Maebashi, Japan.

Follow-Up Procedures

Information on deaths and changes of residence was obtained from the local municipal resident registration files (Jumin Kihon Daicho). Between 2000 and 2011, 1,168 deaths were observed (761 among men and 407 among women) and 45 participants (22 men and 23 women) moved out of the study area.

Indices of Relative Deprivation

Data on gross annual household income—including earnings from employment, benefits, and transfer payments of all household members—were collected during the second wave of questionnaires in 2000. The mid-point of each of the 16 income categories was used to define the household income of the study subjects. For the open-ended highest income bracket, a median-based Pareto curve estimate was utilized to calculate the mid-point [42]. To determine individual earnings, equivalized income was calculated for each participant dividing their gross annual household income by the square root of their household size.

To evaluate relative income deprivation, the following two measures were used:

• (1) The YI, which can be calculated with the following formula [14]:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{Y}{\mathrm{I}}_i=\frac{1}{N}{\displaystyle \sum _j}(y_j-y_i)I_{ij}{I}_{ij}=\{\begin{array}{l}\begin{array}{lll}1,& {\displaystyle if}& {\displaystyle y_i<y_j}\end{array}\\ \begin{array}{lll}0,& {\displaystyle if}& {\displaystyle y_i\ge y_j}\end{array}\end{array}}\end{array}}$$

where y_i = income of individual i, y_j = income of individual j, and N = total number of individuals in i’s reference group. Therefore, the YI estimates the average magnitude of difference in income between an individual and all other individuals with higher earnings in the same reference group.

• (2) IR, estimated by [15]:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{I}{\mathrm{R}}_i=\frac{i-1}{n-1}}\end{array}}$$

where i – 1 = number of individuals within individual i’s reference group with lower income than i’s and n = total number of individuals within i’s reference group. Thus, IR calculates the ordinal position of one’s earnings in the income hierarchy within a reference group. Calculated values range between 0 and 1, representing the lowest and the highest rank within the group, respectively.

The Komo-Ise Study survey does not contain questions inquiring on whom the study participants compare themselves to in the space of income or wealth. Therefore, based on previously published literature, the similarity between individuals in factors, such as age, sex, occupation, education, and the municipality of residence were assumed to be relevant for income-based social comparisons [19, 27, 43]. The created reference groups included the following combinations of the factors mentioned above: (a) sex and age group, (b) sex and municipality of residence, (c) sex, age group, and municipality of residence, (d) sex, age group, and occupation, (e) sex, age group, and education, (f) sex, age group, occupation, and municipality of residence, and (g) sex, age group, education, and municipality of residence. For example, it was assumed that a 45 year old man living in Isesaki city working as a company employee would compare his income to other men with the same attributes instead of comparing himself to a retired 65 year old woman living in Komochi village.

Life Priorities

In the second wave of surveys, the following question was asked: “As you grow older, how important do you think these factors are in your life?” The list included interaction with family, focusing on work, being active in a group or organization, socializing with friends, as well as personal growth. The possible responses ranged from “not important at all”, “a bit important”, “quite important”, to “very important”. Based on which of these factors the study participants labeled as “very important,” a “life priority” variable with three categories was created. In the first category, at least one of the social engagement variables (family, friends, and group/organization) or personal growth was found very important but not work. The second category included those who found work very important as well (next to social engagement or personal growth), while participants in the third category chose work only.

In the Japanese version of the questionnaire, personal growth replaced the original spirituality factor (“growing spiritually”) to reflect the cultural context of the participants. The wording of the question (“人としてあるいは人格的に成長すること”) implies eudaimonic growth, which is related to meaningful activities/relationships, virtuous values, and wisdom instead of materialistic gain, social status, or egoistic self-image [44]. The social engagement factors (family, friends, and group/organization), as well as personal growth, were chosen based on the social network index, which was tied to mortality in both the Alameda County and the Komo-Ise studies [37, 40, 45, 46].

Covariates

To adjust for potential confounding, age (continuous variable—in years), sex (male or female), equivalized household income (continuous variable—10,000 yen [approximately 100$] unit), education level (compulsory vs. high school/vocational school/college, university, and above), marital status (married vs. separated/divorced/widowed), occupation (agriculture/forestry, company employee, mercantile, self-employed, retired, homemaker, or other), number of illnesses (score from 0 to 13), body mass index (BMI; <18.5, 18.5 to <25, or ≥25), hours of sleep (≤6, 7, 8, or ≥9 hr per day), history of medical health checkup participation (<1 year ago, 1–2 years ago, >2 years ago, or never), vegetable consumption (continuous variable—number of servings per day), walking (taking a long walk often vs. never/sometimes), physical exercise (often vs. never/sometimes), alcohol—wine, beer, and liquor—consumption (never/less than once per week/once or twice per week vs. three or four times per week/nearly every day/every day), smoking status (current smoker, ex-smoker, or never), and perceived stress at work (a great deal vs. some/hardly any) were included to the multivariate models. History of illnesses was self-reported by answering the question “Have you ever had the following diseases?” The list of illnesses consisted of heart disease, stroke, cancer, hypertension, hyperlipidemia, diabetes, ulcer, asthma, pollinosis, depression/neurosis, osteoporosis, low-back pain, and “others.” All variables were assessed using the second wave of questionnaires conducted in 2000.

Statistical Analyses

To account for missing values, sequential regression multiple imputation (SRMI, aka “imputation using chained equations” or “imputation by fully conditional specification”) was conducted on the full data set. Based on a set of 20 imputations, Cox proportional hazards regression models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) for all-cause mortality. Interaction terms between the YI/IR and sex were not statistically significant; therefore, data from men and women were combined for analyzing the associations between relative income deprivation and mortality. Four sets of models were defined for both the YI and the IR: Model 1 was adjusted for age, sex, equivalized household income, marital status, education, and occupation. Model 2 also controlled for the number of illnesses, BMI, vegetable consumption, hours of sleep, walking, physical exercise, health checkup participation, alcohol consumption, smoking status, and perceived stress at work. Model 3 was further adjusted for the “life priority” measure, while Model 4 contained three-way interaction terms between the YI/IR, the “life priority” variable, and sex as well. All models were adjusted for equivalized household income to assess associations between relative income and health independently of absolute income. Lifestyle- and behavior-related covariates added to the models (Models 2 to 4) were considered to be potential mediators linking relative deprivation to all-cause mortality. IR was reverse coded in all analyses to allow for comparability with the YI.

Interaction terms between YI/IR, the “life priority” variable, and sex were included to test for effect modification (Model 4) after confirming that there were statistically significant associations between relative income deprivation and all-cause mortality among the Komo-Ise study participants (Models 1 and 2). HRs based on simple slopes were estimated for all-cause mortality in relation to relative income deprivation within each life priority category, then the differences between the slopes were calculated—expressed as ratios of hazard ratios (RHRs)—and tested for statistical significance. For this study, SAS version 9.4 (SAS Institute Inc., Cary, NC) and SAS-callable IVEware (imputation and Variance Estimation Software) version 0.2 were used to conduct all analyses.

Results

Table 1 shows sex-specific age-adjusted characteristics of the study participants according to the three life priority categories. There were 2,371 men and 2,610 women in the Social engagement/self-growth group, 2,044 men and 2,404 women in the Work and social engagement/self-growth group, while the corresponding number of participants in the Work-only group was 60 and 45, respectively. Among both men and women, focusing only on work—but not on social engagement or personal growth—was positively associated with lower educational attainment, longer sleeping hours (≥9 hr/day), and frequent alcohol consumption (three or more times/week), as well as current smoker status and lower vegetable consumption. These participants were also less likely to have low BMI or to have a recent history of health checkup participation. Among men, those who exclusively prioritized work were less likely to be married. Among women, the prevalence of taking long walks was lowest in the group that exclusively prioritized work. Compared to men, women were less likely to report high work-related stress, to be married, to have longer sleeping hours, to frequently consume alcohol, to be current smokers, or to walk and do sports often and were more likely to have a recent history of health checkup participation.

Table 1.

Age-adjusted characteristics of participants according to life priorities at baseline

	Men			Women
Life priorities	Social engagement/ self-growth	Work and social engagement/ self-growth	Work only	Social engagement/ self-growth	Work and social engagement/ self-growth	Work only
Number of participants, n	2,371	2,044	60	2,610	2,404	45
Age, yearsa	61.18 (8.44)	62.29 (8.17)*	59.88 (8.15)*	61.33 (8.51)	63.10 (7.85)**	63.82 (8.04)**
Equivalized household income, 10,000 yen	359.0	370.2*	288.4*	355.0	336.6**	315.2**
Low educational attainment, %	39	47*	50*	40	49**	52*
Retired, %	32	27*	18*	15	13**	9**
Married, %	90	93*	78*	78	79**	82**
Body mass index <18.5 kg/m2, %	6	4*	1*	6	6**	2*
Number of illnesses	1.89	1.72*	1.65*	1.90	1.69**	2.24**
Hours of sleep ≥9 hr/day, %	8	7*	10*	4	3**	7**
Time since last health checkup >2 years, %	6	6*	10*	5	5**	7**
Vegetable consumption, servings/day	7.05	7.63*	5.26*	8.13	8.40**	7.25**
Taking a long walk often, %	17	18*	16	13	15**	9**
Engaging in physical exercise often, %	17	18*	18	14	16**	13†
Consuming alcohol at least three times a week, %	56	61*	69*	15	15†	19**
Current smoker, %	46	48*	59*	10	13**	16**
High perceived stress at work, %	13	16*	22*	8	7**	8†

Values are least squares means or percentages and are adjusted for the age distribution of the study population.

^aValues are unadjusted means (standard deviation).

*p < .05 for differences between means when compared to the “Social engagement/self-growth” group within the same sex category

**p < .05 for differences between means when compared to the “Social engagement/self-growth” group within the same sex category and p < .05 when women are compared to men within the same life priority category

^† p < .05 for differences between means when compared to men within the same life priority category but p > .05 when compared to the “Social engagement/self-growth” group within the same sex category.

Table 2 provides HRs with 95% CIs for all-cause mortality in relation to relative income deprivation. After adjustment for age, sex, equivalized household income, marital status, education, and occupation (Model 1), both higher YI and lower IR were associated with a higher risk of all-cause mortality in all defined reference groups. Each standard deviation (SD) increase in YI (approximately JPY 850,000 or USD 8,500) was associated with an HR of 1.20 (95% CI = 1.06, 1.36) to 1.25 (95% CI = 1.10, 1.41) for all-cause mortality. The corresponding HRs for an SD decrease in IR (0.29 decrease in rank between 1 and 0) were 1.17 (95% CI = 1.02, 1.35) to 1.22 (95% CI = 1.03, 1.46). After controlling for behavioral and dietary variables (Model 2), the HR estimates for YI were attenuated by 5%–8% but remained statistically significant in all but one reference group—the reference group based on gender and age group. The corresponding HR estimates (Model 2) for IR did not reach statistical significance.

Table 2.

Multivariate adjusted hazard ratios (HRs; 95% confidence intervals [CIs]) of all-cause mortality associated with relative income deprivation for seven reference groups

Person-years	162,522	Yitzhaki Index (1 SD increase)				Income Rank (1 SD decrease)
N	9,533	Model 1 HR (95% CI)	Model 2 HR (95% CI)	Model 3 HR (95% CI)	Model 4 HR (95% CI)	Model 1 HR (95% CI)	Model 2 HR (95% CI)	Model 3 HR (95% CI)	Model 4 HR (95% CI)
Total death, n	1,168
RD reference groups
Gender and age group
Relative income deprivation		1.21 (1.07, 1.38)	1.13 (0.99, 1.28)	1.13 (0.99, 1.28)	1.19 (1.02, 1.39)	1.20 (1.04, 1.39)	1.12 (0.97, 1.29)	1.12 (0.97, 1.29)	1.14 (0.97, 1.35)
Life priority—both social engagement and worka				0.89 (0.78, 1.00)	0.91 (0.79, 1.06)			0.88 (0.78, 1.00)	0.91 (0.79, 1.06)
Life priority—work onlya				1.26 (0.74, 2.15)	1.09 (0.50, 2.38)			1.26 (0.74, 2.16)	1.12 (0.51, 2.46)
RD × Life priority—both social engagement and worka × Sexb					1.21 (0.89, 1.65)				1.10 (0.82, 1.48)
RD × Life priority—work onlya × Sexb					3.10 (1.13, 8.55)				3.63 (0.58, 22.71)
Gender and municipality of residence
Relative income deprivation		1.23 (1.07, 1.40)	1.17 (1.02, 1.33)	1.17 (1.02, 1.34)	1.18 (1.01, 1.38)	1.22 (1.03, 1.46)	1.15 (0.97, 1.37)	1.15 (0.97, 1.37)	1.15 (0.94, 1.39)
Life priority—both social engagement and worka				0.89 (0.78, 1.00)	0.91 (0.78, 1.06)			0.89 (0.78, 1.00)	0.91 (0.77, 1.06)
Life priority—work onlya				1.25 (0.73, 2.14)	1.04 (0.44, 2.45)			1.26 (0.74, 2.16)	1.10 (0.48, 2.51)
RD × Life priority—both social engagement and worka × Sexb					1.03 (0.78, 1.36)				1.00 (0.74, 1.35)
RD × Life priority—work onlya × Sexb					2.57 (0.72, 9.16)				4.00 (0.43, 36.92)
Gender, age group, and municipality of residence
Relative income deprivation		1.25 (1.11, 1.41)	1.18 (1.04, 1.33)	1.18 (1.04, 1.33)	1.22 (1.05, 1.42)	1.22 (1.06, 1.41)	1.14 (0.99, 1.31)	1.14 (0.99, 1.32)	1.16 (0.99, 1.37)
Life priority—both social engagement and worka				0.89 (0.78, 1.00)	0.91 (0.79, 1.06)			0.89 (0.78, 1.00)	0.91 (0.78, 1.06)
Life priority—work onlya				1.26 (0.73, 2.15)	1.05 (0.47, 2.36)			1.26 (0.73, 2.16)	1.11 (0.51, 2.45)
RD × Life priority—both social engagement and worka × Sexb					1.18 (0.87, 1.60)				1.09 (0.82, 1.46)
RD × Life priority—work onlya × Sexb					2.71 (1.02, 7.18)				3.63 (0.56, 23.44)
Gender, age group, and occupation
Relative income deprivation		1.22 (1.07, 1.40)	1.15 (1.01, 1.32)	1.15 (1.01, 1.32)	1.18 (1.00, 1.39)	1.19 (1.03, 1.37)	1.11 (0.97, 1.28)	1.11 (0.97, 1.28)	1.13 (0.96, 1.33)
Life priority—both social engagement and worka				0.89 (0.78, 1.00)	0.91 (0.78, 1.06)			0.89 (0.78, 1.00)	0.91 (0.78, 1.06)
Life priority—work onlya				1.26 (0.73, 2.15)	0.89 (0.35, 2.28)			1.26 (0.74, 2.16)	1.07 (0.47, 2.46)
RD × Life priority—both social engagement and worka × Sexb					1.15 (0.85, 1.55)				1.09 (0.82, 1.46)
RD × Life priority—work onlya × Sexb					2.00 (0.71, 5.60)				3.39 (0.50, 22.86)
Gender, age group, occupation, and municipality of residence
Relative income deprivation		1.25 (1.10, 1.41)	1.19 (1.05, 1.34)	1.19 (1.05, 1.34)	1.20 (1.03, 1.41)	1.20 (1.05, 1.37)	1.13 (0.99, 1.29)	1.13 (0.99, 1.29)	1.14 (0.98, 1.33)
Life priority—both social engagement and worka				0.89 (0.78, 1.00)	0.91 (0.78, 1.06)			0.89 (0.78, 1.00)	0.91 (0.78, 1.06)
Life priority—work onlya				1.25 (0.73, 2.14)	0.85 (0.31, 2.33)			1.26 (0.74, 2.15)	1.07 (0.46, 2.48)
RD × Life priority—both social engagement and worka × Sexb					1.15 (0.86, 1.53)				1.10 (0.82, 1.46)
RD × Life priority—work onlya × Sexb					1.90 (0.68, 5.27)				3.57 (0.49, 25.75)
Gender, age group, and education
Relative income deprivation		1.20 (1.06, 1.36)	1.14 (1.00, 1.29)*	1.14 (1.00, 1.29)*	1.20 (1.03, 1.39)	1.17 (1.02, 1.35)	1.10 (0.96, 1.27)	1.10 (0.96, 1.27)	1.13 (0.96, 1.33)
Life priority—both social engagement and worka				0.89 (0.79, 1.01)	0.91 (0.79, 1.06)			0.89 (0.78, 1.00)	0.91 (0.79, 1.06)
Life priority—work onlya				1.26 (0.74, 2.15)	1.08 (0.50, 2.37)			1.27 (0.74, 2.17)	1.13 (0.53, 2.44)
RD × Life priority—both social engagement and worka × Sexb					1.14 (0.84, 1.56)				1.07 (0.80, 1.44)
RD × Life priority—work onlya × Sexb					3.08 (1.19, 7.97)				5.00 (0.62, 40.06)
Gender, age group, education, and municipality of residence
Relative income deprivation		1.22 (1.08, 1.38)	1.17 (1.03, 1.32)	1.16 (1.03, 1.32)	1.22 (1.05, 1.41)	1.18 (1.03, 1.35)	1.11 (0.97, 1.27)	1.11 (0.97, 1.27)	1.14 (0.97, 1.34)
Life priority—both social engagement and worka				0.89 (0.79, 1.01)	0.92 (0.79, 1.06)			0.89 (0.78, 1.00)	0.91 (0.79, 1.06)
Life priority—work onlya				1.26 (0.74, 2.15)	1.07 (0.49, 2.36)			1.27 (0.74, 2.17)	1.13 (0.52, 2.43)
RD × Life priority—both social engagement and worka × Sexb					1.12 (0.83, 1.53)				1.07 (0.80, 1.44)
RD × Life priority—work onlya × Sexb					2.90 (1.14, 7.35)				5.05 (0.60, 42.43)

Model 1 is adjusted for age, sex, equivalized household income, education, marital status, and occupation. Model 2 is adjusted for age, sex, equivalized household income, education, marital status, occupation, number of diagnosed illnesses, body mass index, vegetable consumption, hours of sleep, walking, physical exercise, health checkup participation, alcohol consumption, smoking status, and perceived stress at work. Model 3 is further adjusted for life priorities. Model 4 also contains three-way interaction terms between relative income deprivation, life priorities, and sex.

RD, relative deprivation; SD, standard deviation.

^aReference = Life priority—social engagement but not work.

^bReference = men.

*p < .05.

The three-way interaction terms between YI, sex, and life priority were statistically significant (Model 4). Table 3 provides sex-specific HRs for all-cause mortality in relation to YI according to the three life priority categories and RHRs with 95 % confidence intervals to test for difference in HR estimates between subgroups. Among women, the YI was associated with an increased risk of all-cause mortality. However, this association was only statistically significant among those who exclusively focused on work. After controlling for all covariates, the HR for mortality per SD increase in YI in the work-only group was 2.56 (95% CI = 0.86, 7.62) to 2.66 (95% CI = 1.08, 6.53), which was 2.49 (95% CI = 0.83, 7.45) to 2.77 (95% CI = 1.20, 6.41) times higher than the HR among those who did not find work very important and 2.29 (95% CI = 0.92, 5.72) to 2.54 (95% CI = 1.15, 5.62) times higher than the HR among those who found both work and social engagement or personal growth essential. Among men, there were no statistically significant differences in HR estimates between subgroups based on life priority category.

Table 3.

Multivariate adjusted hazard ratios (HRs; 95% confidence intervals [CIs]) and ratios of hazard ratios (RHRs) of all-cause mortality associated with relative income deprivation by life priority category

	Yitzhaki Index (1 SD increase)
	Social engagement /self-growth HR (95% CI)	Work and social engagement HR (95% CI)	Work only HR (95% CI)	Work only vs. social engagement/ self-growth (95% CI)	Work only vs. work and social engagement (95% CI)	Work and social engagement vs. social engagement (95% CI)
Women
Person-years	44,969	41,584	754
N	2,610	2,404	45
All-cause mortality, n	221	178	8
RD reference groups
Gender and age group	0.96 (0.74, 1.24)	1.07 (0.82, 1.39)	2.65 (1.16, 6.08)	2.77 (1.20, 6.41)	2.49 (1.07, 5.79)	1.11 (0.87, 1.43)
Gender and municipality of residence	1.03 (0.80, 1.32)	1.06 (0.81, 1.38)	2.56 (0.86, 7.62)	2.49 (0.83, 7.45)	2.41 (0.80, 7.28)	1.03 (0.82, 1.29)
Gender, age group, and municipality of residence	0.99 (0.78, 1.27)	1.12 (0.87, 1.44)	2.61 (1.20, 5.69)	2.62 (1.19, 5.79)	2.34 (1.05, 5.20)	1.12 (0.87, 1.44)
Gender, age group, and occupation	1.00 (0.77, 1.29)	1.11 (0.85, 1.44)	2.58 (1.03, 6.47)	2.58 (1.03, 6.50)	2.32 (0.90, 5.99)	1.11 (0.87, 1.42)
Gender, age group, occupation, and municipality of residence	1.02 (0.81, 1.29)	1.16 (0.90, 1.49)	2.66 (1.08, 6.53)	2.60 (1.07, 6.31)	2.29 (0.92, 5.72)	1.13 (0.89, 1.45)
Gender, age group, and education	0.98 (0.75, 1.26)	1.04 (0.80, 1.36)	2.65 (1.22, 5.77)	2.72 (1.22, 6.05)	2.54 (1.15, 5.62)	1.07 (0.83, 1.38)
Gender, age group, education, and municipality of residence	0.99 (0.77, 1.27)	1.06 (0.81, 1.38)	2.62 (1.23, 5.58)	2.66 (1.23, 5.77)	2.48 (1.15, 5.37)	1.07 (0.83, 1.39)
Men
Person-years	39,835	34,375	1,008
N	2,371	2,044	60
All-cause mortality, n	412	339	10
RD reference groups
Gender and age group	1.23 (1.03, 1.46)	1.12 (0.92, 1.37)	1.02 (0.52, 2.02)	0.83 (0.43, 1.62)	0.91 (0.47, 1.79)	0.92 (0.77, 1.09)
Gender and municipality of residence	1.22 (1.02, 1.47)	1.22 (1.00, 1.48)*	1.18 (0.61, 2.26)	0.96 (0.51, 1.82)	0.97 (0.51, 1.84)	0.99 (0.85, 1.17)
Gender, age group, and municipality of residence	1.26 (1.06, 1.49)	1.19 (0.99, 1.43)	1.14 (0.58, 2.24)	0.91 (0.47, 1.76)	0.97 (0.49, 1.89)	0.94 (0.79, 1.12)
Gender, age group, and occupation	1.19 (0.98, 1.46)	1.16 (0.94, 1.43)	1.41 (0.72, 2.78)	1.18 (0.60, 2.32)	1.22 (0.62, 2.38)	0.97 (0.82, 1.15)
Gender, age group, occupation, and municipality of residence	1.22 (1.02, 1.47)	1.20 (0.99, 1.45)	1.51 (0.75, 3.05)	1.24 (0.61, 2.49)	1.26 (0.63, 2.54)	0.98 (0.83, 1.16)
Gender, age group, and education	1.23 (1.03, 1.46)	1.15 (0.95, 1.39)	1.00 (0.52, 1.96)	0.82 (0.42, 1.58)	0.87 (0.45, 1.70)	0.94 (0.79, 1.11)
Gender, age group, education, and municipality of residence	1.25 (1.06, 1.48)	1.19 (0.98, 1.43)	1.06 (0.55, 2.04)	0.84 (0.44, 1.62)	0.89 (0.46, 1.72)	0.95 (0.80, 1.12)

Models are adjusted for age, equivalized household income, education, marital status, occupation, number of diagnosed illnesses, body mass index, vegetable consumption, hours of sleep, walking, physical exercise, health checkup participation, alcohol consumption, smoking status, and perceived stress at work.

RD, relative deprivation; SD, standard deviation.

*p < .05.

Discussion

Based on longitudinal data from a Japanese population with a mortality follow-up between 1993 and 2011, the findings indicated that increased relative income deprivation was associated with a higher risk of all-cause mortality, independently of absolute income. Furthermore, the associations between relative income—when measured by the YI—and mortality appeared to be stronger among women who exclusively prioritized work in their lives.

The findings are in line with previous studies reporting a significant association between higher relative income deprivation and all-cause mortality [18, 26]. However, while there were no statistically significant differences between men and women in the effects of YI or IR on mortality in the present study, evidence from the published literature showed that men tend to be more susceptible to the adverse health effects of higher relative income deprivation than women [23, 25–27]. The data set for this study contained a relatively small sample size (n = 9,533) compared to the abovementioned studies and was collected only at two locations in Gunma Prefecture, Japan; therefore, the findings may not be generalizable. Also, the age of the participants at baseline was 40–69 years, which is a younger population perhaps with a higher percentage of women actively involved in work—43% in 2000—compared to previous studies conducted in Japan focusing on elderly populations. In addition, women who reported “retired” status—14% in 2000—might have previous working experience; therefore, they may be psychologically primed to make income-based comparisons.

Previously, it has been suggested that IR might be a stronger and more consistent predictor of self-rated and objective health outcomes than the YI [22]. Findings based on the Komo-Ise study showed that HR estimates were similar for both measures of relative income after standardization. After adjustment for behavioral and dietary factors, the YI remained a statistically significant predictor for elevated mortality risk utilizing different reference groups but not IR. There were no significant differences in the magnitude of HRs based on the choice of reference groups.

According to the psychosocial explanation, relative deprivation affects health through psychosocial stress arising from positional competition created by upward social comparisons [47–49]. Both YI and IR measure relative deprivation based on income, assuming that one of the most important dimensions of social comparisons is income and wealth, especially, in capitalist/consumer societies. In this study, not only YI but also IR was statistically significantly associated with mortality. This finding suggests that, while decreasing the magnitude of differences in wages before tax and implementing redistributive tax policies are essential steps toward tackling income inequalities [50, 51], it would not completely solve the issue at hand since, normally, these measures would not change the ordinal position of individuals’ income. To address this problem, it was suggested that basing self-respect/self-worth on other merits and shifting the focus from materialistic aspirations/consumerism toward other values, such as social connections and social engagement, might attenuate the associations between relative income and health. An emphasis on personal growth has been linked to higher levels of psychological well-being and survival [44, 52, 53]. Also, greater social engagement was shown to be associated with better effective mental health and cognitive functioning, lower diabetes incidence, increased health-related quality of life, lower depressive symptoms, and reduced risk of developing a functional disability, as well as lower mortality risk [54–57].

The current study explored how differences in the perceived importance of focusing on work, social engagement, or personal growth might modify the associations between relative income deprivation and mortality. While considering career “very important” in itself might not be harmful, the findings suggested that compared to other groups, among women who focused exclusively on work (but did not give the same consideration to social engagement or personal growth), the YI was a stronger predictor for elevated mortality risk. Comparable effect modification was not found in men. There are several possible explanations for this difference between men and women: first, “focusing on work” does not necessarily equate with “seeking status through work”; for example, it can be viewed as a necessity for survival in the society or as a source of enjoyment as well. However, compared to men, Japanese women’s options in the job market and attitude toward work underwent a rapid change in the past decades. From the 1960s, with the increasing number of service-oriented jobs, the participation of women in the labor market started to grow [58]. Since the 1980s, new opportunities have opened up in large companies, increasing the number of ‘career-track’ women in the Japanese society who were seeking independence and self-actualization while engaging in more lavish consumer spending [58], which perhaps made them more prone to income-based social comparisons. On the other hand, traditionally, men had to ensure financial security for their family, possibly making married men who believe family is very important even more preoccupied with their economic and social status compared to men focusing only on their career due to the additional responsibility possibly leading to more stress [58, 59]. Second, a previous study reported a stronger link between materialism and lower well-being when the proportion of women was higher among respondents [34]. Women might be more concerned about keeping up physical appearances (image), which is related to their sense of self-worth and identity [60]. Also, traditionally, it is more socially acceptable for men to focus on extrinsic goals than for women, which might negatively affect career-oriented women’s psychological well-being [33, 34]. Third, while social engagement was reported to be beneficial for both sexes, friends and other social groups can represent important sources of social comparisons, which might affect men and women differently [55, 61].

Limitations

The following limitations should be taken into consideration when interpreting the described results. First, while defining comparison groups is essential for studying relative income deprivation, the Komo-Ise study did not contain information on the preferred reference groups of the participants. Therefore, following the example of previous publications, multiple reference groups were constructed based on demographic and socioeconomic factors to address this problem. However, we cannot exclude the possibility that there were additional important characteristics based on which participants chose a group of people as their reference. Second, those who have lower income compared to others might not perceive themselves as relatively deprived. Since a subjective measure for relative deprivation was not available for this study, two objective measures were utilized and compared; the widely accepted YI and the more novel IR. Third, life priorities were only assessed at baseline. However, it is possible that preferences have changed over time. Fourth, group participation might include organizations that are focused on career development, which could potentially lead to the misclassification of some participants to the wrong life priority group. Also, some women who believe work is very important might instead choose friends and family in their response to the questionnaire due to social desirability. However, these issues would lead to more conservative estimates for the work-only group without invalidating the findings. Fifth, the number of participants prioritizing solely work was limited (60 men and 45 women). Also, the small number of mortality cases in this group might have resulted in inflated HR estimates; therefore, the results have to be interpreted with caution; it is plausible that point estimates closer to the lower limit of the 95% CI would be more appropriate. Sixth, the data for this study were collected from two locations within the same prefecture in Japan; therefore, it might not be generalizable to the rest of the country. The study results should be confirmed based on a larger number of participants recruited from all over Japan. Also, the finding should be explored in other countries with different sociocultural context as well.

Conclusion

The study findings indicate that both lower IR and higher YI were associated with an increased risk of all-cause mortality, independently of absolute income. Also, among women, focusing on work instead of social engagement or personal growth might be related to stronger associations between relative income deprivation and mortality. These results are consistent with reports showing negative associations between extrinsic goals or materialism and health [30, 32, 34]. However, while focusing solely on work and material gains might divert the attention from meaningful social connections, altruistic behavior, and self-improvement, the opposite is also true: focusing on intrinsic values (e.g., caring for others, community goals, personal growth, and social engagement) might prevent materialistic goals from becoming overprioritized [32, 33]. Limiting the exposure to social messages emphasizing the importance of consumer culture, money, status, and power might also be an important step toward encouraging intrinsic goals [32].

Compliance with Ethical Standards

Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards The authors report no conflicts of interest.

Funding The present study received support from the International Eco Health Research Group [Komo‐Ise research grant], the Japan Society for the Promotion of Science (Overseas Challenge Program for Young Researchers and Grant-in-Aid for JSPS Fellows [Grant no. 18J13078]), and by Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Advancing Translational Sciences, National Institutes of Health Award UL 1TR002541) and financial contributions from Harvard University and its affiliated academic healthcare centers. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health.

Authors’ Contributions Krisztina Gero conceived the initial study plan, analyzed the data, and wrote the first draft of the manuscript. Atsushi Miyawaki and Ichiro Kawachi contributed to the study design, and provided critical feedback on the data analyses as well as the manuscript. Ichiro Kawachi supervised the study.

Ethical Approval The Komo-Ise cohort study was approved by the Epidemiologic Research Ethic Committee of Gunma University Faculty of Medicine, Maebashi, Japan. The Institutional Review Board of the Harvard T.H. Chan School of Public Health determined that the current study is not human subjects research as defined by DHHS or FDA regulations.

Informed Consent Written consent was obtained from all participants at the beginning of the Komo-Ise cohort study.