Association of housing tenure and unaffordable housing with preterm birth and other adverse birth outcomes in Canada: a population-based study

Abstract

Background

Socioeconomic risk factors are known drivers of adverse birth outcomes. Housing is a key target for policy interventions.

Objective

To estimate the associations of housing tenure (renting vs owning) and unaffordable housing with preterm birth and other adverse birth outcomes.

Methods

We used 2014–2016 Canadian birth registration data linked with the 2016 long-form census and included singleton births among homeowners and renters. Unaffordable housing was defined at the family level as the proportion of pre-tax income spent on shelter, using a 30% cut-off. The primary outcome was preterm birth. Secondary outcomes were stillbirth and infant death. Log-binomial regression estimated the association of housing tenure and unaffordability with outcomes adjusting for sociodemographic risk factors and parity.

Results

Among 162 700 live births and stillbirths (52 740 renters, 109 960 owners), 31% of renters and 17% of owners experienced unaffordable housing. Renting was associated with an increased risk of preterm birth (7.5% vs 6.1%; adjusted risk ratio (aRR) 1.13; 95% CI 1.08 to 1.17), stillbirth (9.5 vs 6.6 per 1000; aRR 1.33, 95% CI 1.14 to 1.56) and infant death (4.2 vs 3.0 per 1000; aRR 1.52, 95% CI 1.26 to 1.82). There was no association of housing unaffordability with preterm birth or other adverse birth outcomes among owners or renters.

Conclusions

This nationally representative study in Canada found associations between renting versus owning and preterm birth, stillbirth and infant death, as well as a high burden of unaffordable housing, particularly among renters. This study suggests that home tenure itself is a social determinant of adverse birth outcomes.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Socioeconomic risk factors are known drivers of disparities in preterm birth and other adverse birth outcomes and housing is a key social determinant of health. We searched MEDLINE on 21 February 2024 for “housing” AND (“preterm delivery” OR “preterm birth” OR “gestational age” OR “stillbirth” OR “perinatal death*” OR “neonatal death*” OR “fetal death*” OR “fetal demise”). We found a previous systematic review including literature up to November 2020 that reported that homelessness and unstable housing were associated with preterm birth, with adjusted ORs ranging from 1.20 to 1.26 in three studies. A second more recent systematic review, which included literature up to September 2022, reported that rental evictions during pregnancy, measured at the census block level for five studies and at the individual level in one study, were associated with preterm birth. One ecological study reported that census tracts with higher rental eviction rates had higher infant mortality rates. Current evidence suggests that homelessness, precarious housing and rental evictions are associated with preterm birth. Stillbirth and infant death were not assessed in such studies at the individual level. We found no studies that investigated the association of either housing tenure or unaffordable housing with preterm birth or any other adverse birth outcomes.

WHAT THIS STUDY ADDS

• To our knowledge, this is the first population-based study assessing the associations of housing tenure and unaffordable housing with preterm birth and other adverse birth outcomes. This nationally representative cohort in Canada which included 52 740 home renters and 109 960 homeowners found that renting versus owning was associated with preterm birth, stillbirth and infant death, after adjusting for parity, maternal age, education, visible minority status, maternal birthplace and after-tax family income. It further found a high burden of unaffordable housing among renters (31%) vs owners (17%). Unaffordable housing itself was not associated with any adverse birth outcomes.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY.

This study builds on previous research indicating that rental evictions are associated with preterm birth and suggests that renting versus owning is associated with preterm birth, stillbirth and infant death in a representative sample of the Canadian population. More broadly, this study suggests that home tenure itself is a social determinant of adverse birth outcomes. There is a need to further investigate what aspects of being a renter, besides facing evictions, can contribute to adverse birth outcomes.

Introduction

Preterm birth is the leading cause of child morbidity and mortality globally,¹ and can lead to chronic health conditions, learning and behavioural disorders, and other adverse outcomes in childhood and throughout the life course.² Rates of preterm birth have stagnated according to the WHO and there is a need for more comprehensive preventive strategies.¹ However, a focus on individual risk factors in association with preterm birth and other adverse birth outcomes can pathologise persons receiving prenatal care.³ Such an emphasis on each individual’s behaviour (eg, diet, exercise, alcohol and other substance use) and pre-existing health conditions (eg, diabetes) is potentially ‘victim-blaming’ and does not address the life circumstances that drive behaviours and underlie chronic health conditions.³ An upstream focus can inspire broader political and economic change, and therefore, have wider public health impacts.^{4 5}

Socioeconomic risk factors are known drivers of disparities in preterm birth and other adverse birth outcomes.^{6 7} The ability to own a home, the conditions associated with renting one’s home and unaffordable housing, in particular, are modifiable structural determinants of health that are amenable to policy interventions.⁸ While the majority of studies have investigated the effects of homelessness on preterm birth,⁹ and more recent studies have focused on rental evictions,¹⁰ the effects of housing tenure (ie, renting vs owning one’s housing) and unaffordable housing on birth outcomes are less understood. Renting one’s home is associated with various stressors as compared with owning property, including the fear and stress associated with potential evictions,¹¹ fear of rental cost increases, lack of control over the property and a loss of income without the accompanying wealth accumulation of home ownership.¹² Unaffordable housing is a human rights violation in itself and is a negative indicator of child and family well-being.^{13 14} In Canada, there was a marked growth of renter versus owner households between 2011 and 2021,¹⁵ and renting is reported to have become less affordable for low-wage workers between 2018 and 2022.¹⁶

A mechanism of psychosocial stress linking renting versus owning and unaffordability with preterm birth and other adverse birth outcomes is possible⁷; however, there is a paucity of research on this topic. The purpose of this study was to estimate the burden of housing unaffordability among those giving birth and to estimate the association between renting versus owning and unaffordable housing with preterm birth and other adverse birth outcomes.

Methods

Study design, data sources and study population

We conducted a population-based study using data from the 2016 Canadian Birth-Census Cohort, which consists of 2014–2016 vital statistics live birth and stillbirth records linked with the 2016 long-form census.¹⁷ The Canadian long-form census is administered to a subset of Canadian households and must be completed by law; it includes survey questions on housing, education, ethnicity and income.¹⁸ This study was restricted to all singleton live births and stillbirths among home renters and owners in a private dwelling. Those who lived in farm dwellings and collective housing (eg, hospital, correctional facility and shelter) or band housing (situated on Indigenous reserves or settlement) were excluded; this definition follows national standards on the reporting of housing unaffordability.¹⁹ Those households reporting zero or negative incomes or expenditures were excluded, as such circumstances make interpretations of unaffordability difficult.^{20 21}

Exposure: home tenure and housing unaffordability

Home tenure refers to whether a household owns or rents its private dwelling.¹⁹ Being a renter or owner was defined based on whether some household members rented or owned the dwelling as assessed by the census questionnaire.²² We used the housing unaffordability definition of the federally funded Canada Mortgage and Housing Corporation, which agreed with the provinces in 1986 to assess housing affordability using the threshold of whether the household spent 30% or more of its average total annual income on shelter costs.¹⁹ This definition of unaffordable housing is accepted internationally.23,25

To calculate unaffordability, the total annual household income was defined as all income (before taxes) reported from earnings, investments (eg, net rental income, interest), government transfers (eg, employment insurance benefits, social assistance) and other sources (eg, child support, scholarships or income from outside Canada). This income was based on data from income tax records, reference year 2015, that were integrated into the 2016 census. Shelter costs included rent, regular mortgage payments (principal and interest), property taxes, condominium fees and utilities (eg, water, fuel and electricity). The calculation is described in further detail in online supplemental appendices.

Outcome

The primary outcome was preterm birth, defined as gestational age at birth of less than 37 weeks. Secondary outcomes were stillbirth and infant death, where infant death was defined as death in the first year of life (0–364 days).²⁶

Covariates

Covariates of interest included maternal age in years, parity, maternal education level, maternal ethnicity, maternal marital status and maternal birthplace. After-tax family income was adjusted to each family’s community and family size and reported in tertiles and included as a continuous spline in all models (details of the calculation are provided in online supplemental appendices).

Statistical analysis

The long-form census data were available for approximately 25% of the birth cohort and weights were applied to generalise the findings to all births.¹⁷ Bootstrap weights that considered census sampling design, non-responses and missing birth linkages were used to calculate the variance of analytic parameters. Maternal and infant characteristics, adverse birth outcomes, and unaffordable housing (shelter-to-income ratio ≥30%) were compared by renter versus owner status. Frequency tables were generated and risk ratios and risk differences were calculated. Log-binomial regression was used to estimate the associations of housing tenure and unaffordability with outcomes after adjusting for maternal age, education, visible minority status, marital status, parity and maternal birthplace. Generalised linear models were used with a binomial outcome distribution and a log link function to estimate risk ratios and an identity link function to estimate risk differences.²⁷ The models for unaffordable housing were stratified by tenure a priori based on our assumption that unaffordability affects renters and owners differently.²⁸ The models for housing tenure were additionally adjusted for after-tax family income tertile, while the models for housing unaffordability were not, as income is a component of the exposure measure.

To assess the robustness of the findings, several sensitivity analyses were conducted. First, we also assessed the associations using a discrete housing unaffordability measure: <15%, 15% to <30%, 30% to <50%, 50% to <70% and ≥70% for the preterm birth outcome; other outcomes were too infrequent within strata for such analyses. Second, analyses of unaffordability were assessed by strata of after-tax family income tertiles. This is due to the fact that for a high-income family, the cut-off represents a smaller overall burden as they still have a substantial amount of income left over for other expenses. Third, we restricted the analyses to births for which the birthing parent did not move between the birth and the census date, as those who moved may have experienced changes in their housing status. Fourth, we restricted the analyses to births in the second year of study, which was closer to the census date and may, therefore, more accurately capture family incomes during pregnancy. Fifth, family income was calculated without adjustment for geography and family size to determine whether this affected the findings. Data manipulation was performed with SAS V.9.4 software, while regression analyses were performed with Stata V.18 software, using bootstrap weights to account for the complex sampling design of the cohort.

Results

Among 162 700 live births and stillbirths (52 740 renters; 109 960 owners), housing unaffordability was experienced by 31% of home renters and 17% of homeowners. Figure 1 shows the flow diagram for cohort creation. Renters were more likely to be in lower income tertiles than homeowners. Among renters, 46.4% in the lowest income tertile vs 0.6% in the highest income tertile experienced unaffordable housing, while among homeowners, 54.6% in the lowest income tertile and 2.6% in the highest tertile experienced unaffordable housing (table 1).

Figure 1. Flow diagram for cohort creation, 2016 Canadian Birth-Census Cohort.

Table 1. Maternal and infant characteristics, adverse birth outcomes, and shelter-to-income ratio ≥30% for renter versus owner, singleton births, 2016 Canadian Birth-Census Cohort.

	Renter	Shelter-to-income ratio	Owner	Shelter-to-income ratio
Characteristic	N (per cent)*	per cent ≥30%†	N (per cent)*	per cent ≥30% †
N	52 740	31.3	109 960	16.8
Gestational age (weeks)
13–27	440 (1.2)	30.4	620 (0.8)	19
28–31	320 (0.7)	32	500 (0.5)	18.8
32–33	390 (0.8)	36.7	620 (0.6)	16.9
34–36	2660 (4.9)	32.9	4720 (4.2)	17.4
37–41	48 630 (92.0)	30.9	103 000 (93.5)	16.8
≥42	250 (0.5)	31.1	430 (0.4)	17.7
Missing	50 (0.0)		50 (0.0)
Maternal age (years)
<20	2340 (3.5)	48.4	1030 (0.8)	16.5
20–24	10 690 (19.2)	37.5	7940 (6.8)	17.9
25–29	15 750 (30.0)	29.8	31 260 (28.0)	16.6
30–34	14 770 (29.1)	27.7	44 040 (40.7)	16
35–39	7 410 (14.6)	28.2	21 450 (19.9)	17.7
≥40	1790 (3.5)	28.7	4230 (3.9)	21.4
Parity
1	22 560 (42.9)	29.4	46 840 (42.7)	15.8
2	15 840 (30.4)	31.7	40 420 (36.9)	16.6
3	7500 (14.4)	33.5	13 960 (12.7)	19.3
≥4	5240 (9.8)	33.2	5730 (5.1)	23
Missing	1580 (2.6)		3020 (2.6)
Maternal education level
Less than high school	10 330 (18.3)	39.1	5190 (4.5)	28.7
High school	13 780 (25.9)	35.3	17 020 (15.2)	22.4
Postsecondary diploma or certificate	15 790 (30.4)	28.7	37 120 (33.5)	16.7
University degree	12 860 (25.5)	23.8	50 630 (46.8)	13.9
Maternal marital status
Married or cohabiting	40 440 (77.3)	24.5	103 610 (94.4)	16
Not married	12 310 (22.8)	53.5	6340 (5.6)	30.5
Maternal ethnicity
Not visible minority	27 940 (53.3)	29.7	79 160 (71.3)	11.9
Visible minority, not Indigenous	18 550 (37.0)	32.2	26 980 (25.7)	31.1
Indigenous	6250 (9.7)	34.5	3810 (3.1)	11.9
Maternal birthplace
Canada	33 190 (61.1)	30.8	81 960 (73.6)	11.6
Outside Canada	19 550 (38.9)	31.4	28 000 (26.5)	31.4
Shelter-to-income ratio
<15%	13 480 (24.3)	–	33 390 (29.9)	–
15% to <30%	23 130 (44.7)	–	58 450 (53.3)	–
30% to <50%	9840 (19.0)	–	12 710 (11.8)	–
50% to <70%	3150 (6.1)	–	2740 (2.6)	–
≥70%	3130 (6.0)	–	2650 (2.5)	–
After-tax economic family income tertile
1 (lowest)	33 710 (64.1)	46.4	20 760 (18.9)	54.6
2	13 890 (26.2)	4.8	40 360 (36.7)	14.5
3 (highest)	5150 (9.7)	0.6	48 850 (44.4)	2.6
Adverse birth outcome
Preterm birth	3810 (7.5)	32.8	6470 (6.1)	17.7
Stillbirth	340 (9.5 per 1000)	33.5	500 (6.6 per 1000)	17.5
Infant death	200 (4.2 per 1000)	32.4	280 (3.0 per 1000)	17.7

^*Event numbers are unweighted and weighted per cent or per thousand for stillbirth and infant death. Event numbers are randomly rounded up or down to the nearest 10 in accordance with Statistics Canada disclosure regulations.

^†Weighted percentage having shelter-to-income ratio ≥30% within each stratum (row).

Housing unaffordability was more common among single versus married mothers/parents (table 1). In addition, it decreased with higher education and was higher among Indigenous and visible minority renters versus non-visible minority and non-Indigenous renters, and among visible minority versus non-visible minority owners. Unaffordable housing was more common among owners whose maternal birthplace was outside of Canada versus within Canada.

Preterm birth incidence was higher among renters versus owners at all levels of the shelter-to-income ratio (figure 2). Table 2 summarises the association of housing tenure with birth outcomes. Renting was associated with an increased risk of preterm birth (7.5% vs 6.1%; adjusted risk ratio (aRR) 1.13; 95% CI 1.08 to 1.17), stillbirth (9.5 vs 6.6 per 1000; aRR 1.33, 95% CI 1.14 to 1.56) and infant death (4.2 vs 3.0 per 1000; aRR 1.52, 95% CI 1.26 to 1.82).

Figure 2. Preterm birth in Canada among renters versus owners by categories of shelter-to-income ratio, 2016 Canadian Birth-Census Cohort. Vertical bars are 95% CIs accounting for survey design using bootstrap weights.

Table 2. Associations between housing tenure (renting vs owning) and adverse birth outcomes, 2016 Canadian Birth-Census Cohort.

Outcome	Event number (incidence risk)*		Risk difference (95% CI)		Risk ratio (95% CI)
	Renter	Owner	Crude	Adjusted†	Crude	Adjusted†
Preterm birth	3820 (7.5 per 100)	6480 (6.1 per 100)	1.40 (1.19, 1.61)	0.75 (0.50, 1.00)	1.25 (1.21, 1.29)	1.13 (1.08, 1.17)
Stillbirth	340 (9.5 per 1000)	500 (6.6 per 1000)	2.84 (2.08, 3.61)	2.13 (0.92, 3.34)	1.43 (1.31, 1.57)	1.33 (1.14, 1.56)
Infant death	200 (4.2 per 1000)	270 (3.0 per 1000)	1.19 (0.69, 1.68)	1.47 (0.79, 2.16)	1.40 (1.23, 1.60)	1.52 (1.26, 1.82)

^*Event numbers are unweighted and incidence risks are weighted. Event numbers are randomly rounded up or down to the nearest 10 in accordance with Statistics Canada disclosure regulations.

^†Adjusted models include maternal age, education, visible minority status, marital status, parity, maternal birthplace, and after-tax economic family income. All models account for bootstrap weighting.

CIconfidence interval

Among renters, no meaningful associations were observed between housing unaffordability and preterm birth (7.9% vs 7.3%; aRR 1.03; 95% CI 0.95 to 1.10), stillbirth (10.2 vs 9.1 per 1000; aRR 1.08; 95% CI 0.91 to 1.27) or infant death (4.3 vs 4.1 per 1000; aRR 1.02; 95% CI 0.78 to 1.33). Similarly, among owners, there were no meaningful associations between housing unaffordability and preterm birth (6.4% vs 6.1%; aRR 0.99; 95% CI 0.93 to 1.04), stillbirth (6.9 vs 6.6 per 1000; aRR 0.87; 95% CI 0.65 to 1.15) or infant death (3.1 vs 2.9 per 1000; aRR 0.88; 95% CI 0.67 to 1.16, table 3). Crude and adjusted risk differences describing the associations of housing tenure and unaffordable housing with birth outcomes are presented in tables2 3.

Table 3. Associations between unaffordable housing (≥30% vs <30% shelter-to-income ratio) and adverse birth outcomes among owners and renters, 2016 Canadian Birth-Census Cohort.

	Event number (incidence risk)*		Risk difference (95% CI)		Risk ratio (95% CI)
	Shelter-to-income ratio <30%	Shelter-to-income ratio ≥30%	Crude	Adjusted†	Crude	Adjusted†
Renters
Preterm birth	2600 (7.3 per 100)	1210 (7.9 per 100)	0.61 (0.05, 1.17)	0.18 (−0.34, 0.69)	1.09 (1.01, 1.18)	1.03 (0.95, 1.10)
Stillbirth	240 (9.1 per 1000)	100 (10.2 per 1000)	1.08 (−0.65, 2.81)	0.66 (−0.84, 2.17)	1.12 (0.94, 1.34)	1.08 (0.91, 1.27)
Infant death	140 (4.1 per 1000)	60 (4.3 per 1000)	0.27 (−1.05, 1.58)	0.08 (−1.00, 1.15)	1.07 (0.78, 1.45)	1.02 (0.78, 1.33)
Owners
Preterm birth	5360 (6.1 per 100)	1110 (6.4 per 100)	0.37 (0.07, 0.67)	−0.08 (−0.39, 0.23)	1.07 (1.01, 1.12)	0.99 (0.93, 1.04)
Stillbirth	420 (6.6 per 1000)	70 (6.9 per 1000)	0.31 (-1.06, 1.68)	−0.85 (−2.49, 0.78)	1.05 (0.86, 1.28)	0.87 (0.65, 1.15)
Infant death	230 (2.9 per 1000)	40 (3.1 per 1000)	0.18 (-0.60, 0.95)	−0.35 (−1.10, 0.40)	1.06 (0.82, 1.36)	0.88 (0.67, 1.16)

^*Event numbers are unweighted and incidence risks are weighted. Event numbers are randomly rounded up or down to the nearest 10 in accordance with Statistics Canada disclosure regulations.

^†Adjusted models include maternal age, education, visible minority status, marital status, parity, and maternal birthplace. All models account for bootstrap weighting.

Sensitivity analyses revealed no clear relationship between unaffordability as a discrete measure and preterm birth among renters or owners (online supplemental etable 1). The association of unaffordability with preterm birth did not differ markedly when stratified by income tertiles among owners, although there was a larger effect size at the highest income tertile (aRR 1.19, 95% CI 0.98 to 1.45) (online supplemental etable 2); results were not stratified by income tertile among renters due to small numbers of preterm births in the highest income tertile. Restriction to mothers who did not move between the birth and census date (online supplemental etables 3,4), and to those only in the second year of the cohort (online supplemental etables 5,6) did not materially affect the findings. In addition, analysis of family income without adjustment for geography or family size yielded similar results.

Discussion

This nationally representative study in Canada, using the 2016 long-form Census, found a high burden of unaffordable housing, with 31% among renters and 17% among owners, as well as associations between renting versus owning and preterm birth, stillbirth and infant death. Based on the traditional definition of 30% or more of income spent on housing, as well as the use of a discrete shelter-to-income ratio measure sensitivity analyses, no salient associations were found between unaffordable housing and preterm birth or other adverse birth outcomes among renters or owners. The findings suggest that home tenure itself is one aspect of housing that is associated with adverse birth outcomes.

Housing is a fundamental social determinant of health. Despite this, to our knowledge, this study is the first population-based study that investigated preterm birth and other adverse birth outcomes by home tenure and unaffordability in the general population. Previous studies on birth outcomes have focused on the deleterious effects of homelessness and unstable housing on preterm birth or have otherwise restricted analyses to specific subpopulations.⁹ While homelessness is an important and extreme example of inadequate housing, other negative aspects of housing may have wider reaching impacts, and unaffordable housing predisposes families, particularly renters, to homelessness. A focus on rental evictions in recent years highlights that evictions are one mechanism through which renting may be associated with adverse birth outcomes.¹⁰ A recent systematic review reported that rental evictions during pregnancy, measured at the census block level for five studies and at the individual level in one study, were associated with preterm birth.¹⁰ In addition, one ecological study reported that census tracts with higher rental eviction rates had higher infant mortality rates.²⁹ Evictions are associated with prenatal stress, economic hardship which can compromise access to nutritious food and healthcare, and force families to live in poor quality and unsafe homes.¹⁰

Beyond rental evictions, a mechanism for the association between renting versus owning with adverse birth outcomes may be the deleterious mental health effects associated with renting.^{12 24} Renting has been associated with lower mental health scores,¹² self-reported well-being²³ and life satisfaction. Some have hypothesised that home ownership may be viewed as a rite of passage, leading to higher self-esteem.¹² In addition, home ownership is a forced investment in long-term savings, while the burden of rental payments puts home ownership out of reach for many, especially low-income renters.¹² Renting can lead to a feeling of social exclusion, which can be socially and psychologically damaging.^{8 24} There may be neighbourhood effects; for instance, well-being is rated higher in neighbourhoods with more homeowners.²³ Other direct sources of stress for renters include a lack of control over the property, challenges in dealing with landlords and the fear of future eviction, all contributing to an inability to feel secure and stable in their homes.³⁰

The lack of an association between overall unaffordable housing and preterm birth and other adverse birth outcomes may be due to the fact that high housing costs can ‘purchase’ various health advantages, including access to neighbourhoods with preferred characteristics, such as proximity to parks and community centres, nearby grocery stores and day cares, a lower crime rate, lower commute times and better conditions within the household. A close proximity of the home to work, community support and recreation is particularly important for single parents, who may not otherwise be able to manage the household. On the other hand, research supports that a high shelter-to-income ratio decreases spending on food and leisure activities and makes it difficult to deal with unexpected stressors such as a job loss or home payment increases. Studies have reported that high housing costs relative to income lead individuals to forego expenditures in critical areas such as taking their medications and accessing healthcare.³¹ One study reported that unaffordable housing was associated with poor mental health among home renters but not owners in the USA,¹² while another study reported that unaffordable housing was associated with depressive symptoms among both older adult renters and owners in Korea.³² In contrast, our study found no notable association of unaffordability with birth outcomes among renters or owners.

Strengths of this study include that it was population-based and the sampling strategy helped ensure that the findings are generalisable to all Canadian households.²⁶ In addition, the census collected many details about housing expenditures including utilities. The following are some limitations. First, we cannot discount the possibility that differences in birth outcomes observed by tenure may be due to residual confounding by characteristics of the renters, including unmeasured factors or residual imbalances. Renters were more likely to be younger, visible minority or Indigenous, single and have lower income levels. We have accounted for these and other demographic factors in our study. We cannot, however, discount the possibility that other unmeasured characteristics, such as disability, smoking, alcohol and drug use, chronic health conditions or adverse personal circumstances, were more prevalent among renters, thereby potentially overestimating associations in our study. Other unmeasured factors include family wealth and savings that could facilitate the down payment required for a mortgage; this may have resulted in residual confounding and a possible overestimation of associations.

Second, this study provided a snapshot of the household housing situation, and we were unable to follow households longitudinally. As a result, some households may have both rented and owned in 1 year, and renting and unaffordability may have been a brief, transient experience for some, which may have diluted the observed effects. To address this concern, in sensitivity analyses, we restricted the cohort to individuals who did not move between the birth and census date and found no marked difference in the findings. Third, we did not have detailed information about the type of employment, nor the funds held in savings accounts and self-employed earners with earnings that fluctuate may use savings or borrowed funds. To account for this, we excluded those with zero or negative incomes but cannot discount the possibility that some low-income families used their savings income. Fourth, we did not capture housing conditions such as crowding and the need for household repairs, nor neighbourhood characteristics and exposure to environmental toxins, both of which are important public health aspects of housing.³³ Future research should investigate the interactions between renting versus owning, unaffordability, housing conditions and neighbourhood characteristics and assess them longitudinally. Given that acute versus chronic forms of stress may have different influences on preterm birth and other adverse birth outcomes, there is a need to investigate the effect of housing on birth outcomes longitudinally. Fifth, housing expenditures and some of the covariates (eg, education, marital status, ethnicity and birthplace) were self-reported and may be subject to misclassification. Sixth, although the cohort weights were designed to help mitigate potential inclusion bias associated with missed linkages, unknown bias might exist if the in-scope stillbirths and infant deaths missing from the cohort differed systematically from those that were included.¹⁷

Conclusions

This nationally representative population-based study in Canada found associations between renting versus owning one’s home and preterm birth, stillbirth and infant death, as well as a high burden of unaffordable housing, particularly among renters. This study suggests that home tenure itself is a social determinant of adverse birth outcomes. There is a need to further investigate what aspects of renting, besides evictions, can contribute to adverse birth outcomes. Physicians and allied health workers can serve as important advocates for patients and communities in addressing the social determinants of health, including housing.³⁴ This can include promoting an agenda to eliminate unaffordable housing and addressing health disparities among renters versus owners. There remains a need for more population-based research on the effects of housing on the health of families, pregnant individuals and children. Such research should focus on factors amenable to policy interventions that have the potential to be widely implemented, such as rent controls, tenant protection policies, increases in affordable rental housing and measures to make home ownership accessible to more families.

Data availability statement

Data are available on reasonable request.