Introduction

Social determinants that contribute to health inequality, such as educational background, economic status, and occupational class, have received a lot of attention. Housing tenure (HT) is also considered one of the important factors in determining health [1–3].

The association between HT and health has been studied in many countries, and many previous studies have reported that owner-occupiers have better health than people who live in privately rented or publically subsidized housing [4–12]. Although HT can be taken as a surrogate indicator of social class and wealth [6], whether or not the association between HT and health is independent of socioeconomic indicators may vary across countries [5,10,11].

First, HT-related factors, such as home ownership rate, household preferences, and housing administration policies, vary depending on the country [10,11,13]. According to a report by the OECD [14], in 2016 on average 69% of households across the OECD owned a dwelling, compared to 26% of households who rented a dwelling. Although owning a home is the most common form of housing, Switzerland and Germany have a majority of renters (60% in Switzerland and 55% in Germany). Regarding housing policies, most OECD countries have a system of social rental housing, but the size of the social housing stock differs from country to country: The Netherlands, Austria, Denmark, France, and the United Kingdom have a high rate of social housing stock, accounting for 15% or more of the total housing stock. In Japan, 80% of households own their dwelling, placing Japan fifth highest in the OECD ranking of ownership rates. However, Japan has a low rate of social housing stock, comprising about 5% of the total housing stock. Inadequate housing policies may increase the proportion of economically vulnerable people living in low-rent and poorly-maintained accommodation in the private rental sector. This could have adverse health effects due to financial stress and difficulties affording health care [3].

Second, Japan has among the highest proportion of older adults aged 65 and older in the world [15]. Because older people tend to spend most of their time at home and are vulnerable to barriers and problems of the home environment [16], Japanese people may be more exposed to health risks associated with their own home. Furthermore, previous studies suggest that the housing-health association can be partly explained by neighborhood environments [17,18]. A cross-national research study in China, Japan, and South Korea has reported that the association between neighborhood social environment and self-rated health is strongest in Japan [17]. Japanese culture is greatly influenced by Confucian ideals, where harmony among people is to be valued. Japanese people may tend to place emphasis on neighborly ties and therefore their health may be more affected by their neighborhood social environment. Based on these assumptions, the HT-health association may be stronger in Japan than in other countries. Therefore, we hypothesize that HT is significantly related to the health status of Japanese people, independent of other socioeconomic factors, such as education, expenditure, and occupation.

Self-rated health (SRH) is a subjective indicator of current general health, which contains diverse aspects of health including physical and mental health, well-being, and life satisfaction [19]. Additionally, SRH, which is frequently used in international comparative statistics [20] and epidemiological studies, has been established as an independent predictor not only of mortality in general populations [21] and young adulthood [22], but also of functional decline in community-dwelling older adults [23]. Therefore, SRH is a crucial health outcome in nationally representative health surveys.

Previous studies on HT and SRH mostly focus on two categories of HT, such as home ownership and rental [4,7,10], or owner-occupiers and social renters [5]. To the best of our knowledge, no studies have examined the association between detailed types of HT and SRH in Japan. Therefore, in this study, we used anonymized data from a nationwide cross-sectional survey targeted at households and household members throughout the country and investigated the following study questions: 1) whether HT classified into five small groups is associated with SRH among the general Japanese population; 2) whether this association is independent of demographic characteristics and socioeconomic status (SES) factors; and 3) whether the relationship with HT and SRH varies depending on social demographic attributes.

Materials and methods

Study participants

Fig 1 displays a flowchart of study participants. Our study’s aim was to examine the cross-sectional association between HT and SRH, independent of demographic factors and SES factors. People with ADL disability were excluded from our analyses because they have high potential for reverse causation; ADL disability may be a cause of some renters requiring publically subsidized housing [5]. Additionally, people in a hospital/facility and minors under twenty years of age were not required respond to questions about health status and lifestyle factors such as smoking and drinking. Therefore, we excluded 33,946 persons from our analyses because of being aged <20 (n = 16,951), being in a hospital/facility (n = 998), having received long-term care needing certification (n = 320), having a self-reported ADL disability (n = 1,611), and missing data on age, hospital admission, ADL, HT, household crowding, and/or SRH (n = 13,956). Eventually, we restricted our analyses to the data of 59,784 persons of 20 years or older (28,641 men and 31,143 women) without ADL disability whose HT, household crowding, and SRH (shown below) were available. Individuals excluded from this study due to missing information were older than those included in our analyses, but there was no gender difference between the two groups (data is provided in S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Selection of study participants.

ADL, activities of daily living; HT, housing tenure; SRH, self-rated health.

https://doi.org/10.1371/journal.pone.0224821.g001

Results

Participant characteristics

Of the 59,784 participants, 24.1% were older adults aged 65 years or older, 47.9% were men, 75.9% were owner occupiers, and 14.6% reported poor SRH. Table 1 shows the characteristics of study participants by HT status. Owner-occupiers were more likely to be aged 65 years and older and have chronic medical conditions, and less likely to be current smokers and live in crowded houses; and private renters had the second highest tobacco use; because a large number of people living in provided housing live apart from their family at the new workplace, they had the highest percentage of the married, males, and upper non-manual workers, the second highest percentage of persons living alone, and the lowest percentage of persons aged 65 or older and those who had a low level of education. Publically subsidized renters tended to have junior high school education, live in crowded homes, and report poor SRH; and those living in rented rooms were more likely to be current smokers and have a low EHE.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of study participants by housing tenure status.

https://doi.org/10.1371/journal.pone.0224821.t001

Cross-sectional association between HT and SRH

The ORs for poor SRH associated with HT are presented in Table 2. Publically subsidized renters had a significantly higher prevalence of poor SRH. This significant relationship persisted after adjustment for demographic factors and SES factors (adjusted OR 1.33; 95% CI, 1.19–1.48 in Model 3). Persons living in provided housing were less likely to have poor SHR than owner-occupiers. However, after adjustment for age and gender, this association did not remain significant: adjusted OR (95% CI) was 1.11 (0.94–1.32) in model 1 and 1.12 (0.94–1.34) in model 3. In contrast, in the crude model, private renters and persons inhabiting rented rooms showed no association with poor SRH relative to owner-occupiers. After adjustment for age and gender, private renters and residents in rented rooms were more likely to have poor SRH than owner-occupiers. These significant associations were unchanged after adjustment for SES factors as well as demographic factors (Model 3): adjusted OR (95% CI) was 1.36 (1.26–1.47) in private renters and 1.41 (1.22–1.62) in residents in rented rooms. Given that nine out of every ten people in this study had more than one member in their household, there was a possibility that we had the violation of the independence assumption in the association between HT and SRH. To address this concern, we conducted subset analyses limited to the head of the household (n = 27,849). The results did not change significantly with the subset analyses: in Model 3, adjusted OR (95% CI) was 1.46 (1.32–1.62) in private renters, 1.43 (1.24–1.65) in publically subsidized renters, and 1.35 (1.12–1.61) in residents in rented rooms, compared to owner-occupiers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Odd ratios for poor self-rated health associated with housing tenure.

https://doi.org/10.1371/journal.pone.0224821.t002

Consideration of household crowding

The association between household crowding and SRH by HT is shown in Table 3. In the crude model, a significantly lower OR for household crowding of having poor SRH compared with non-crowded homes was found in owner-occupiers, private renters, and publically subsidized renters. After adjustment for age and gender (Model 1), significant associations in private renters and publically subsidized renters disappeared. After adjustment for all demographic factors (Model 2), owner-occupiers had no association between household crowding and SRH. In contrast, for those living in rented rooms, household crowding was not associated with SRH in crude model. After adjustment for demographic factors (Model 2), persons living in crowded homes were more likely to have poor SRH than persons without crowded homes. These significant associations remained after additional adjustment for SES factors (Model 3): adjusted OR (95% CI) was 1.43 (1.05–1.95) in persons inhabiting crowded rented rooms, compared to those living in non-crowded rented rooms.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association between household crowding and self-rated health by housing tenure.

https://doi.org/10.1371/journal.pone.0224821.t003

Discussion

Our study found that publically subsidized renters tended to have significantly poorer SRH than owner-occupiers, independent of demographic and SES factors. This result supports our hypothesis as well as agrees with the results of previous studies on HT and health in non-Japanese countries [4–7,10,12]. Additionally, we have revealed that not only residents in publically subsidized housing but also those in other types of HT had a significantly worse SRH than owner-occupiers. Our results based on stratified analyses showed that HT status had a greater association with poor SRH among people with weaker social positions than those with a higher level of socio-demographic status. Our findings are consistent with those of previous studies [1,11,35] which suggest that housing environment exerts a greater impact on the health of vulnerable groups such as ethnic minorities, older adults, and the unemployed than the socially advantaged.

The association between HT and SRH may be explained by three mechanisms: housing circumstances, health behaviors, and the neighborhood. First, previous studies suggest that the housing environment of vulnerable groups tends to be crowded due to a large number of people living in a limited area, unsanitary living conditions due to inadequate garbage and sewage treatment, bad air with harmful chemical substances and poor ventilation, and poor hygrothermal conditions such as warmth and humidity [3,36]. Crowding and poor air quality can cause respiratory disease and communicable diseases, a polluted water supply can spread waterborne diseases, and cold homes can elevate the risk for cardiovascular diseases and poor mental health [36,37]. In Japan, because universal access to a clean water supply and an effective sanitation system has improved the level of public hygiene, unsanitary living conditions may not be an issue nowadays. However, because Japan has a high death toll during the winter months [38], excess cold in the home is an ongoing problem in Japan: People with financial difficulties may not be able to afford adequate heating, with consequent negative impact on their health. In this study, among persons living in rented rooms, household crowding was associated with SRH. Our results suggest that people living in rented rooms may be exposed to more hazards in the home environment than people living in other types of HT. For example, persons inhabiting crowded rented rooms seem likely to face the risk of intruders and excess noise. These conditions together with crowding in housing are considered as psychological hazards relating to housing, and could lead to increased risk of illness [37].

Second, previous studies have reported that the social housing group tend to have a lower level of physical activity and a higher prevalence of obesity than those in other housing sectors [39] and that house owners are more likely to have regular medical checkups [40] and to be non-smokers [11] than persons without house ownership; they indicate that the positive health effect of house ownership may be the result of healthy lifestyle and active healthy behaviors among owner-occupiers. These findings from previous studies agree with our results showing that, among the five types of HT, house owners have the lowest rate of current smokers.

Third, the neighborhood environment of dwellings in the publically rented sector is often characterized by poor access to recreational facilities, shops, public transport, social support networks, and health services [6,39]. A neighborhood environment that discourages physical activity and health care may affect health behaviors and the health of residents [3]. On the other hand, research has shown that people living in a neighborhood with more sports/recreational facilities, more walkable green spaces, and more people who practice healthy lifestyles such as choosing healthy foods and doing active physical exercise are more likely to report being in good health [41–44]. The communities with a high homeownership rate are considered to have a healthy neighborhood [11], which promotes the health of house owners.

Our study has several strengths. First, this study is based on nationally representative data. Therefore, we have the advantage of generalization and have been able to examine the SRH-association based on five categories of HT. Second, the CSLC collects a large number of socio-demographic indicators such as marital status, chronic medical conditions, education, and occupation. The use of many covariates in this study makes the validity of the results a strong point.

Our study has several limitations. First, we cannot omit the effect of reverse causation due to cross-sectional design. In particular, we should consider the fact that health problems can trigger unemployment and loss of stable employment [5,45]. As a result, people may end up in poverty and need to sell their houses. In this case, poor health status is not attributed to housing environment. However, it is highly likely that residents in privately rented residences, publically subsidized housing, and rented rooms who relocate due to health issues will have poor SRH. Second, our results are based on self-assessment. Therefore, we have the possibility of problems with common method variance and an overestimation between HT and SRH [46]. Future studies are needed to confirm the association between HT and health using an objective assessment of health based on a prospective study design. Third, anonymized data eliminated the rare households in order to prevent identification of individuals, and our study failed to include persons with missing data. The former ruled out social minorities and the latter resulted in selective loss of older people. That is, persons omitted from our study are vulnerable groups, and potentially have poor health and poor housing. This bias might lead to an underestimation of the association between HT and SRH in this study.

Conclusions

This study has revealed a significant association between HT and SRH in Japan, independent of SES factors and demographic characteristics. The Japanese government has adopted reduction of health disparities as one of the main target objectives for the national health promotion plan, Health Japan 21 (the second term) [47]. For a better understanding of disparities in health status among the Japanese, our findings suggest that HT is an important factor that deserves notice. From a policy perspective, policymakers need to pay more attention to HT as a social determinant that contributes to health inequality in Japan.

Supporting information

Acknowledgments

We wish to thank the participants of this study and support staff who make the study possible. We also thank Dr. Heather Hill for her English language editing.