Background: This study examined whether living in rented housing is associated with increased all-cause and cause-specific mortality, and whether the association between home ownership and mortality can be explained by household income, occupational class, and educational level.
Methods: A random sample including every seventh Finn aged 40–80 years at the end of 1997 (N = 308 291) was derived from the population register of Finland. The sample was followed up for mortality until the end of 2003 during which time there were 22 721 deaths.
Results: The hazard ratio for all-cause mortality among renters compared with owner-occupiers was 2.06 (95% CI 1.98 to 2.14) in men and 1.73 (1.65 to 1.81) in women. Adjusting for household income, occupational class, and educational level attenuated the excess mortality among renters by 30% in men and 19% in women. The effect of income was larger among the under 65 year olds than those aged 65 years or over. Excess mortality among renters was particularly high for alcohol-related diseases, respiratory diseases, lung cancer, as well as endocrine, metabolic and nutritional diseases, and infections.
Conclusions: Renters had higher mortality than owner-occupiers even after adjusting for household income, occupational class, and educational level. Home ownership may indicate material living standards and cumulative wealth that cannot sufficiently be captured by conventional socioeconomic indicators. Analysing home ownership may thus increase understanding of the factors producing inequalities in health.
Statistics from Altmetric.com
Home ownership has been shown to be an important discriminator of good and bad health. Those who live in rented housing have more symptoms and long-term illnesses, and they report poorer general and mental health than owner-occupiers.1–5 The evidence on mortality is scarce but previous studies also suggest higher mortality among renters than owner-occupiers. The estimates of excess mortality among renters have varied between 30% and 70%.6–9
Home ownership has often been used as an indicator of socioeconomic circumstances.10 11 Information on home ownership has been easily available and it has shown even stronger associations with health than such conventional socioeconomic indicators as income, occupational class, and educational level.6 12 In particular, home ownership has been used as a substitute for income, which often cannot be found in registers and may be unreliable when asked in surveys. Home ownership is a plausible marker for income because housing costs typically account for a large part of the total household expenditure and for many families buying a home is the largest investment in their lifetime. No studies have, however, examined the association of home ownership and health status controlling for income and other socioeconomic indicators.
Previous evidence on the associations between home ownership and mortality comes almost exclusively from the United Kingdom. We examined whether home renting is associated with all-cause and cause-specific mortality in Finland, and whether the association can be explained by conventional socioeconomic indicators. We had access to large nationally representative register data, which provided a unique possibility to examine these issues using detailed information on various socioeconomic indicators and data on subsequent mortality with minimal loss to follow-up. The main analyses were conducted separately among the under 65 year olds and those aged 65 years or over as the associations of socioeconomic indicators with mortality may be different after retirement age.
The specific aims were to examine: (1) whether living in rented housing is associated with excess mortality among men and women across various age groups; (2) whether the association can be explained by household income, occupational social class and educational level; and (3) whether the strength of the association varies by cause of death.
The study sample was drawn at random from the population register of Finland. The sample consisted of every seventh Finn (14.3%) aged 40 years or over at the end of 1997. Among the under 40 year olds mortality is low and clusters to specific causes of death. Those aged over 80 years were excluded from the study as they increasingly live in long-term institutional care. The population register data were linked to death records from the beginning of 1998 until the end of 2003 using personal identification codes. The data consisted of 146 472 men and 162 449 women, and there were 22 721 deaths during the follow-up period.
Home ownership was divided into owner-occupiers and renters. Other housing types (2.6%) were excluded. Altogether 17.8% of men and 18.4% of women lived in rented housing.
The conventional socioeconomic indicators were household income, occupational social class and educational level. Household income included all taxable income received by the household members, such as wages and salaries, capital income and taxable income transfers, but excluded taxes and social security payments. The information on the sources of income came from the registers of the Finnish Tax Administration and the Social Insurance Institution. Household size was taken into account by dividing household disposable income by the square root of the number of household members.13 Household income was then divided into deciles with separately defined cut-off points for men and women. To check the robustness of income’s effect several other classifications were tried and they all provided identical results.
Occupational class was classified into seven categories: upper non-manual employees; lower non-manual employees; skilled manual workers; unskilled manual workers; farmers; other self-employed workers; and others.14 The unemployed and the retired were classified according to their previous occupation, and housewives were categorised according to the occupation of the head of the household. Occupational class was unknown for 0.2% of the respondents and they were excluded from the analyses.
Educational level was based on the highest completed educational degree or certificate. We separated five categories ranging from primary school to university degree.
Causes of death were classified according to the Finnish edition of the International Classification of Diseases version 10 into eight broad categories and their subcategories.
Cox proportional hazards models were used to estimate hazard ratios with 95% confidence intervals (CI) comparing renters with owner-occupiers across five-year age groups. The effect of adjusting for household income, occupational class, and educational level was then examined in the whole study population as well as separately among the under 65 year olds and those aged 65 years or over, following the most common formal retirement age in Finland. Age (five-year age groups), marital status (married, never married, divorced, and widowed), region of residence (the Helsinki metropolitan area and five provinces of Finland), and language (Finnish or Swedish) were adjusted for as confounders. The percentage reduction of the hazard ratio (HR) caused by adjustments was calculated as: (unadjusted HR − adjusted HR)/(unadjusted HR − 1)*100. The association between home ownership and mortality was finally examined by causes of death in the whole study population. All analyses were carried out separately for men and women using STATA 9.2 (STATA Corp, College Station, Texas, USA).
The number of deaths strongly increased whereas the proportion of renters decreased slightly by age (table 1). Renting was strongly associated with all-cause mortality in all age groups although the excess mortality risk decreased by age. The overall age-adjusted hazard ratio for mortality among renters compared with owner-occupiers was 2.06 in men and 1.73 in women.
Table 2 shows the association between home ownership and all-cause mortality adjusting for age, marital status, region of residence, and language in the whole study population and separately among the under 65 year olds and those aged 65 years or over. Mainly because of the adjustment for marital status excess mortality among the renters was somewhat lower than that found in table 1, especially in men. Adjusting for conventional socioeconomic indicators only slightly lowered the excess mortality among renters. The excess mortality was lowered by 30% in men and 19% in women when all three socioeconomic indicators were adjusted for. Among the under 65 year olds adjusting for household income attenuated the excess mortality among renters more than adjusting for occupational class or educational level but this was less clear among those aged 65 years or over. Adjusting simultaneously for all three conventional socioeconomic indicators attenuated the excess mortality among renters more than adjusting for household income only.
Excess mortality among renters was observed for nearly all causes of death (table 3). Exceptions were some cancers, such as prostate cancer in men and breast cancer in women. The excess mortality was particularly large for alcohol-related diseases, respiratory diseases, lung cancer, as well as endocrine, metabolic and nutritional diseases, and infections. Adjusting for conventional socioeconomic indicators attenuated the excess mortality among renters throughout all causes of death but did not substantially alter the findings.
Renters had higher mortality than owner-occupiers among Finns aged 40 to 80 years. This was evident in all age groups although the relative difference in mortality decreased with increasing age. After adjusting for age, renters had 106% excess mortality compared with owner-occupiers in men and 73% excess mortality in women. These excess rates are higher than6 9 or similar to7 8 those previously reported elsewhere. Comparisons with previous studies need to be carried out with caution, however, as the strength of the association is heavily dependent on the age groups studied.
What is already known on this subject
Home ownership has commonly been used as a socioeconomic indicator in health research, especially in the United Kingdom
Home ownership has shown even stronger associations with health than conventional socioeconomic indicators, ie income, occupational class and education, but studies on home ownership and mortality are rare
The excess mortality among renters could only partly be accounted for by household income, occupational class and educational level. The association between home ownership and mortality is thus largely independent of these conventional socioeconomic indicators. Our results therefore suggest that if home ownership reflects socioeconomic circumstances these must be dimensions that are poorly captured by conventional indicators. One such dimension could be cumulative wealth. This is plausible because home ownership is likely to reflect accumulated income and inherited wealth over a prolonged period of time. Home ownership may even reflect material circumstances of the family of origin and even earlier generations.15
It has also been suggested that the association between home renting and poor health might reflect direct influences of housing-related health hazards.16–18 In previous studies the association between home renting and poor health has been partly explained by housing conditions, such as damp, crowded, poorly equipped or cold housing, and neighbourhood environment.1 19 Although such direct effects are possible, however, housing conditions and neighbourhood environment depend on material living standards as the wealthier can afford better housing and live in better neighbourhoods. Therefore, housing conditions may be one mechanism contributing to socioeconomic inequalities in mortality but even this mechanism reflects differences in material circumstances. More equal distribution of income and wealth would probably also lead to narrowing mortality differences between home ownership groups.
What this study adds
Using a nationally representative register-based dataset including 308 291 individuals we found that Finnish male renters had 106% higher mortality and female renters 73% higher mortality than owner-occupiers
The excess mortality among renters was largely independent of conventional socioeconomic indicators, although among those of working age household income explained a third of the excess mortality
The association between home ownership and mortality may indicate material living standards and cumulative wealth, which are poorly captured by conventional socioeconomic indicators
Household income explained more of the excess mortality among renters who were under 65 year olds than among those aged 65 years or over. This may be caused by weaker effects of income on mortality among those who have passed retirement age. After retirement absolute income differences are smaller than among those who are active in working life. A previous study showed that the associations of income and occupational class with all-cause mortality weakened by increasing age, and after the retirement age mortality differentials between income groups narrowed more rapidly than did those between occupational classes.20 Many socioeconomic indicators may be problematical among those who have retired.21 22 Home ownership, in contrast, is likely to be relatively stable and therefore potentially also a more appropriate indicator among those who have passed retirement age. Home ownership may, however, also have more independent effects among the retired if the relevance of other socioeconomic indicators has diminished.
Our cause-specific analyses show similarities with previous studies examining socioeconomic inequalities in mortality by other socioeconomic indicators. Excess mortality among renters was particularly large for the same causes of death that have been found to be more common among those with low income20 and social class.23 In contrast, home ownership was not associated with some cancers, such as prostate cancer in men and breast cancer in women, which have higher incidence (but also higher survival) in the higher socioeconomic groups.24 These analyses thus support the idea that home ownership may reflect socioeconomic circumstances. Our findings are also in accordance with UK studies reporting less detailed cause-specific analyses.6 7
The sample of this study was large and representative of the Finnish population, there was loss to follow-up only through emigration, and the data were practically complete for all study variables. We had access to detailed information on various socioeconomic indicators including retrospective data for occupational social class. A limitation was that we could not distinguish private renters from local government and other public sector renters. It is known from other sources that approximately half of all renters in Finland live in local government rented housing.25 Local government rented housing has income limits and the selection criteria include reasons such as having notice or a decision on eviction, overcrowding, and poor conditions of current housing. UK studies have shown that although the excess mortality among private renters does not reach the level of local government renters it is still considerably higher than that among owner-occupiers.6–8 A previous Finnish study conducted among a relatively well-off group of municipal employees showed that renters report poorer self-rated health than owner-occupiers.3 Although further studies are warranted these findings suggest that the excess mortality among renters observed in our study is unlikely to be confined to local government renters only.
In Finland home ownership is an important determinant of mortality independently of household income and other socioeconomic indicators. Among individuals of working age household income explained a third of the excess mortality among renters. Home ownership, reflecting material living standards and long-term cumulative wealth, may indicate socioeconomic conditions not sufficiently captured by conventional indicators. Examining home ownership is thus likely to increase our understanding of the factors producing socioeconomic inequalities in ill-health and mortality. More research is needed on the effects of specific housing conditions, further socioeconomic factors, and characteristics of renters as explanations for the excess mortality among renters.
The authors are grateful to Statistics Finland for permission (TK 53-574-04) to use the data.
Funding: ML (#204894), PM (#70631, #48600, #210752), OR (#210435) and EL (#205588) are supported by the Academy of Finland.
Funding: None declared.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.