Longitudinal trends and associations between housing and health among migrants in Australia: evidence from the Building a New Life in Australia study (2013—2023)

Abstract

Background

Housing insecurity is a fundamental social determinant of migrant health. While the relationship between housing and health outcomes is well-documented, there is a paucity of longitudinal data examining this association among humanitarian migrants. Accordingly, we aimed to assess i) longitudinal trends in housing insecurity, psychological distress, post-traumatic stress disorder (PTSD), and self-rated health and ii) the longitudinal associations between housing insecurity and mental and physical health outcomes among humanitarian migrants over the period 2013/14 to 2023.

Methods

Data were drawn from the Building a New Life in Australia Longitudinal Study (n = 6978) with repeated measures over six data waves from 2013/14 to 2023. Generalised Estimating Equations with logistic regression were used to test the association between housing insecurity and psychological distress, PTSD and self-rated health, adjusting for potential confounders and time-varying covariates. The adjusted odds ratios (AOR) and the 95% confidence interval were reported to determine the relationship between health and housing insecurity. A p-value < 0.05 was determined statistically significant.

Results

We found that the prevalence of housing insecurity decreased significantly, from 73.3% in Wave 1 (2013/14) to 39.7% in Wave 6 (2023). In contrast, psychological distress showed an alternating trend, increasing slightly from 17.2% in Wave 1 to 22.7% in Wave 6. Similarly, PTSD followed a fluctuating trend, decreasing from 34.12% in Wave 1 to 31.3% in Wave 6; self-rated health remained fairly consistent over time. After adjustment, housing insecurity was significantly related to increased odds for psychological distress (AOR = 1.66; 95% CI: 1.43, 1.93), PTSD (AOR = 1.41; 95% CI: 1.24, 1.60), and poor self-rated health (AOR = 1.42; 95% CI: 1.22, 1.66).

Conclusion

While housing insecurity declined over time, our study presents strong evidence of the longitudinal associations between housing insecurity and (psychological distress, PTSD, and self-rated health) among humanitarian migrants. Addressing housing insecurity through targeted public health and housing policies may improve health outcomes for humanitarian migrants. These findings offer essential information for policymakers by demonstrating that housing insecurity is significantly associated with poorer mental and physical health outcomes among humanitarian migrants.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27704-7.

Background

The number of forcibly displaced people worldwide reached 122.6 million by the end of June 2024 [1]. In 2024 alone, more than 2.4 million refugees are projected to require resettlement [2]. Australia is a significant host for refugees and humanitarian1 entrants [3]. For the 2024—25 period, the Australian Government allocated 20,000 places in its Humanitarian Program [3], with the top countries of origin for humanitarian entrants being Afghanistan, Iraq, and Myanmar [4]. This number remains the same compared to 2023/24 but shows a slight increase from 18,750 new entrants in 2018/19 [5]. Resettlement is a good option for different citizens around the world; however, migrants from culturally diverse backgrounds may face unique challenges that can impact their mental health and wellbeing, such as the stress of acculturation, government processes, language barriers, limited employment opportunities, isolation, and family pressures [6, 7]. Additionally, housing instability or affordability challenges and decreased housing satisfaction are linked with poor mental health outcomes, such as depression and psychological distress [8].

Increasing migration and population mobility have both positive and negative public health consequences. Benefits include advancing a range of cultural, societal, and intellectual interests [9] for countries of origin and destination. However, it may also create negative impacts for migrants in terms of their overall well-being [10]. Predeparture factors that may affect health outcomes include social determinants of health (including poverty, income, asylum-seeking, discrimination, social support, experiences of detention, employment, and housing conditions and housing insecurity2 including homelessness), violation of human rights (access to education, health, food and water, discrimination, gender inequality, and being forced to move for safety), ecological determinants (including natural disasters, environmental pollution or climate change), and separation from family [11–14]. During transit, migrants may face threats to their physical safety because of unsafe transportation and the risk of being exploited [15, 16]. In general, migration may be considered one of the catalysts for various health conditions, exacerbating the existing conditions or may increase different vulnerabilities [7]. Resettled migrants and refugees may experience difficulties in accessing safe, affordable, and adequate housing [17, 18]. Housing is a basic human right and is the foundation for an individual’s stability and security [19]. For refugees, housing is an essential first step toward successful settlement and integration. It is the first place where they may begin to feel safe and protected after fleeing their home country [20]. Secure housing not only provides a sense of protection but also enables refugees to focus on essential aspects of their settlement, such as employment, education, and healthcare access, which is crucial for mental health and well-being, especially after traumatic experiences [21, 22].

Humanitarian migrants in Australia experience a greater burden of mental health conditions, such as depression, anxiety, post-traumatic stress disorder (PTSD) and psychological distress [6, 12, 23, 24]. For example, initial data from the Building a New Life in Australia (BNLA) longitudinal study indicated that 33.3% of observations had PTSD at baseline, slightly decreasing to 28.3% over five years. However, the prevalence of psychological distress has persisted, affecting about 17% of the population studied [24–26]. Migrant mental health is influenced by various stressors, including pre-migration trauma, the uncertainty of settlement, and post-migration challenges [27]. Evidence shows that poor social integration, including economic instability, language barriers, low social interactions, loneliness, temporary housing contracts and short-term leases [28, 29], discrimination and financial hardship could be attributed to the associated mental health outcomes in these populations [26, 29, 30]. Furthermore, humanitarian migrants living in unaffordable housing are 60% more likely to have worse mental health compared to the general population [31]. Housing insecurity not only affects mental health but also contributed to the development and exacerbation of chronic physical conditions, poorer self-rated health, and increased risk of infectious diseases [32].

The relationship between housing status and health is well-documented [33, 34]. However, there is a need to better understand the relationship between housing insecurity and health using longitudinal data, considering a broader spectrum of housing insecurity as an exposure variable and its long-term association with both mental and physical health. Australia’s Refugee and Humanitarian Program have two main components: offshore and onshore protection, which offer permanent protection pathways and access to settlement supports [35]. The BNLA study tracks humanitarian migrants who entered through these structured pathways, providing a unique longitudinal cohort for examining housing and health. This is important because it differs from broader refugee samples, which often mix visa categories and migration pathways. The availability of such data can help highlight trends in housing insecurity and health among humanitarian migrants. Although there is ample research using BNLA data to assess the magnitude and associated factors of mental illness [10, 24–26, 30, 36], there is a paucity of analysis focusing on the longitudinal relationship between housing insecurity and health. Therefore, this study aimed to assess longitudinal trends in housing insecurity, psychological distress, PTSD, and self-rated health, and the longitudinal association of housing insecurity and mental and physical health outcomes among humanitarian migrants over a ten-year period, from 2013/14 to 2023, using BNLA cohort. By using six waves of data from the BNLA longitudinal cohort, our study advances existing research in three important ways. First, by examining trends in housing and health across the first decade of resettlement. Second, by estimating longitudinal associations, and third, by focusing specifically on government-resettled humanitarian migrants, a population that remains underrepresented in international housing-health research. This longitudinal data provides better insights by following the same individuals over time, enabling more robust temporal interpretation and stronger estimates of associations. We hypothesised that housing insecurity is a predictor of poor mental and physical health outcomes among humanitarian migrants over time, and that improvements in housing insecurity will be related to better health trajectories. The findings will provide policy-relevant insights to improve the well-being and integration of humanitarian migrants in Australia.

Methods

Study design, setting and observations

This longitudinal study used data from the BNLA study [37], which was conducted from 2013 to 2023 in six waves. The BNLA is the first and continuing national cohort study, funded by the Australian Department of Social Services (DSS) and undertaken by the Australian Institute of Family Studies. Released in 2013, the study tracks the settlement experiences of humanitarian migrants in domains such as mental health, housing, employment, education, and social integration [6, 23, 38]. By tracing individuals and families who settled via humanitarian resettlement programs, the BNLA provides a unique insight into the specific challenges faced by this population [37]. BNLA fieldwork interviews were conducted by independent third-party agencies Colmar Brunton Social Research and Multicultural Marketing and Management.

Participation requirements included being over 18 years old and having been issued a permanent humanitarian visa within the past three to six months. The Australian Government Department of Immigration and Border Protection provided contact information of qualified principal applicants residing in 11 selected regions (five major cities: Adelaide, Brisbane, Sydney, Melbourne, and Perth, with six additional sites) [24] and forwarded them to fieldwork interviewers. During the recruitment process, all eligible principal applicants who were successfully contacted and willing to participate were invited to take part in the study. Secondary applicants, who were aged 15 years and older and part of the same visa application as the eligible principal applicants, were also invited to voluntarily participate in the study [24]. Findings are presented in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist [39].

Inclusion criteria

The BNLA is a longitudinal cohort survey that follows the same individuals over multiple waves. Therefore, 2399 individuals were observed over ten years. At the baseline survey, the cohort consisted of 2,399 unique individuals who contributed a total of 6978 person-wave observations to our analysis across six survey waves. No new participants were added after baseline. The larger number reflects repeated measurements of the same individuals over time. Reflecting the attrition rate, participation declined over time. Specifically, 2,399 participants were interviewed at Wave 1 (baseline), 2,009 at Wave 2, 1,894 at Wave 3, 1,929 at Wave 4, 1,881 at Wave 5, and 1,223 at Wave 6 [37]. Following data consolidation across waves, the sample sizes included 8,109 for housing insecurity, 10,814 for PTSD, 10,909 for psychological distress, and 11,324 for self-rated health outcomes. The final analytic sample was determined after tabulating all covariates and dependent variables. Thus, the final complete-case sample included 6,978 observations across all six waves, with no missing values.

Outcome measures

For this study, the main outcomes were mental health (psychological distress and PTSD) and physical health (self-rated general) outcomes. Accordingly, psychological distress was measured by the Kessler-6 (K6) screening scale [40]. K6 contains six items, on which observations are asked to mark how often during the last four weeks they experienced nervousness, hopelessness, restlessness or fidgeting, effort to do even small things, sadness that nothing could cheer them up, and worthlessness, with a response scale 1 = ‘None of the time’, 2 = ‘A little of the time’, 3 = ‘Some of the time’, 4 = ‘Most of the time’, 5 = ‘All of the time’. Items were then summed to provide a total score out of 6—30, with a score of 19 or higher indicating elevated psychological distress [40]. K6 has been validated across diverse populations and settings and has shown strong psychometric properties [41].

PTSD was measured using the PTSD-8, a screening questionnaire used to assess three subscales: intrusion, avoidance, and hypervigilance (per the DSM-IV criteria for PTSD) [42]. The PTSD-8 is derived from the Harvard Trauma Questionnaire [42], which has demonstrated good performance in screening for PTSD across various cultural backgrounds. A response of either 'sometimes' or 'most of the time' for at least one item in each of the intrusion, avoidance, and hypervigilance clusters is considered a positive screen for PTSD where PTSD is indicated if at least one symptom from each of the three PTSD-8 subscales has a score of 3 ('Sometimes') or 4 ('Most of the time') on the response scale (1 = 'Not at all,' 2 = 'Rarely,' 3 = 'Sometimes,' 4 = 'Most of the time')” [43]. The internal consistency of the K6 and PTSD scales, as estimated with Cronbach's alpha in our study, is 0.93 and 0.96, respectively.

Self-rated health was assessed using a single item: "Overall, how would you rate your health over the last four weeks?". This comes from health-related quality of life (HRQoL) measures or self-reported health status. This measure might be affected by individual responses to how they see themselves based on the context of the community or their cultural expectations [44]. However, it has been found that self-rated health predicts actual health [45]. Original scores were recorded on a six-point scale (1 = Excellent, 2 = Very good, 3 = Good, 4 = Fair, 5 = Poor, 6 = Very poor). During analysis, these were recoded into two categories [1 = Poor (poor and very poor) and 0 = Good (fair, good, very good, and excellent)], to produce a sufficient sample for the housing insecurity variable.

Exposure

The main exposure of the study was housing insecurity. There are no internationally agreed-upon measurements for housing insecurity, though being made to develop a unified measurement of housing insecurity [46]. In our study, housing insecurity was operationalised using three component indicators reflecting residential instability, affordability stress, and insecure housing tenure [47]. An index measure of housing insecurity was created if observations had moved at least twice in the previous 12 months (in response to “In the last 12 months, how many times have you moved homes?”); or were residing in temporary or short-term rentals (in response to “Is your current housing arrangement [Temporary rental (e.g., no lease/contract), Short-term rental (i.e., lease/contract for six months or less), long-term rental (i.e., lease/contract for more than six months), own your home outright or paying off a mortgage, or other]?”); or were unable to pay the rent or mortgage on time. Observations were classified as experiencing housing insecurity if they met at least one of these three conditions (Table 1).

Table 1.

Operationalisation of housing insecurity components

Housing insecurity component	BNLA variable	Survey item	Survey response	Coding used in analysis
Residential instability/housing mobilities	ho03	And in the last 12 months, how many times have you moved homes?	0. None 2. range 1 to 30	0 = None moved, 1 = moved two or more times
Housing affordability stress	em26b	Could not pay rent/mortgage on time because of shortage of money	1.Prefer not to say 2.Don't know 3. Yes 2. No	0 = no; 1 = yes Other excluded
Housing tenure insecurity	ho05	Is your current housing arrangement…?	1. Temporary rental (e.g. no lease/contract) 2. Short term rental (i.e. lease/contract for 6 months or less) 3. Long term rental (i.e. lease/contract for more than 6 months) 4. Own your home outright or paying off a mortgage 5. Other	(Option 1 and 2) = 1 (Option 3 and 4 = 0). We excluded other from the analysis

Covariates

Potential confounders shown in previous research [48, 49] and the information collected in the surveys were considered. These variables were: sociodemographic characteristics (age, gender, marital status [married, separated, divorced, widowed, never married], total number of household members [1—2, 3—4, ≥ 5], country of birth [North Africa, Middle East,Southern Asia, Central Asia, Central and West Africa, Southern and East Africa], education [never attended school, 6 or fewer years, 7—12, trade/technical qualification, university degree], visa subclass (onshore and offshore), language and cultural integration (English language proficiency [very well, well, not well, not at all], sense of belonging a feeling of being part of a larger group or connected to a broader community, specifically measures subjective feeling of being part of the Australian community [very good, good, hard, very hard, never]), settlement experience (overall settlement experience [very good, good, hard, very hard], discrimination [yes, no] and time-related factor (survey wave 1 to wave 6).

Data management and statistical analysis

We extracted, cleaned, and analyzed data from the BNLA data using STATA 18.5 (Stata Corporation, College Station, TX, USA). Survey completion analysis revealed underrepresentation amongst humanitarian migrants who were older, male, residing in Australian state capital cities, non-unauthorized maritime arrival visa holders, or had Myanmar, other Asian, or African as their country of birth. The BNLA data user guide recommended to use the all participant survey weight (awgt) as the most appropriate survey weight to use when analysing variables [37]. Therefore, to adjust for differential response probabilities and attrition over time and to maximize representativeness, survey weights were then applied throughout our analysis.

Descriptive statistics were used to characterise observations sociodemographic and the exposure and dependent variable characteristics across waves. For all variables including the confounders, covariates and the dependent variables, we conducted complete case analysis. Moreover, attrition was assessed by comparing baseline housing insecurity and health outcomes between observations retained and lost to follow-up. Observations were classified as lost to follow-up if they were not observed in later survey waves (supplementary Table S4). Subsequently, univariable and multivariable Generalised Estimating Equation (GEE) analyses were conducted to examine the longitudinal associations between housing insecurity and psychological distress, PTSD, and self-rated overall health. We employed GEE with robust standard errors and applied survey weights to address the sampling design and differential attrition over time [50]. For participant i at wave t, the model was specified as

where is the outcome for individual i at wave t (PTSD, psychological distress, or poor self-rated health), (housing insecurity status),(time-varying covariates), (time-invariant covariates), and denotes survey-wave indicators. The models were specified with a binomial family and a logit link function, and an unstructured correlation matrix to account for within-individual correlation over time [51]. It differs from standard multivariable logistic regression, since the GEE accounts for the correlation between repeated observations within individuals and estimates population-averaged associations [50, 51]. Finally, adjusted odds ratios (AOR) with 95% confidence intervals (CI) and p-values ≤ 0.05 were used in the multivariable analysis to identify and report the associations between housing insecurity and the dependent variables.

Sensitivity analysis

To assess whether the findings were sensitive to the operational definition of housing insecurity, we conducted sensitivity analyses using an ordinal category based on the number of housing insecurity conditions met (0, 1, 2, or 3) (Supplementary Table S4). We also assessed selective attrition among Wave 1 respondents, with non-observation after Wave 1 coded as 1 and observation in at least one later wave coded as 0 (Supplementary Table S3). In addition, we estimated individual fixed-effects logistic regression with wave fixed effects as a further robustness check. This allows us to compare respondents to themselves over time and therefore controls for all time-invariant observed and unobserved individual characteristics (Supplementary Tables S5–S7). Further analysis was conducted to assess the potential confounding effects of covariates that were not included in the final GEE models due to the information collected was from a smaller proportion of the observations. For psychological distress and PTSD outcomes, we further adjusted for variables such as support from the religious community, loneliness, time spent in immigration detention centres and pre-arrival trauma experience (Supplementary table S8—9).

Results

Sociodemographic characteristics of respondents

Of the 6,978 observations across all waves included in the final analysis, the mean age was 40.1 years (standard deviation = 13.5). The majority (n = 4,445, 63.7%) were male. More than half (n = 3,660, 52.5%) originated from Middle Eastern countries. Regarding educational attainment, a higher proportion (n = 3,102, 44.5%) had completed 7—12 years of school (Table 2).

Table 2.

Sociodemographic variables and other related variables (n = 6,978)

Variables	Categories	Frequency	Percent
Age group	15–25	1,010	14.5
	26–35	1,928	27.6
	36–45	1,813	26.0
	46–55	1,235	17.7
	56–80	992	14.2
Marital status	Married	4,341	62.2
	Separated	146	2.1
	Divorced	148	2.1
	Widowed	666	9.5
	Never married	1,677	24.0
Total number of household members	1–2	1,880	26.9
	3–4	2,453	35.2
	≥ 5	2,645	37.9
Gender	Male	4,445	63.7
	Female	2,533	36.3
Country of birth	North Africa	177	2.5
	Middle East	3,660	52.5
	South-East Asia	371	5.3
	Southern Asia	609	8.7
	Central Asia	1,924	27.6
	Central and West Africa	112	1.6
	Southern and East Africa	125	1.8
Education prearrival	Never attended school	1,119	16.0
	6 or less years of schooling	1,498	21.5
	7—12 years of schooling	3,102	44.5
	Trade or technical qualification	478	6.9
	University degree	781	11.2
Discrimination	No	6,384	91.5
	Yes	594	8.5
Housing insecurity	No	3,508	50.3
	Yes	3,470	49.7
Psychological distress	No probable serious mental illness	5,741	82.3
	Probable serious mental illness	1,237	17.7
PTSD	Unlikely to have PTSD	4,881	70.0
	May have PTSD	2,097	30.1
Past 4 weeks overall health	Good	5,724	82.0
	Poor	1,254	18.0
Visa subclass	Offshore	5,786	82.9
	Onshore	1,192	17.1
English language proficiency	Very well	528	7.6
	Well	2,580	37.0
	Not well	2,895	41.5
	Not at all	975	14.0
Overall settlement experience	Very good	1,987	28.5
	Good	3,929	56.3
	Hard	873	12.5
	Very hard	189	2.7
Sense of belonging	Always	3,631	52.0
	Most of the time	1,738	24.9
	Some of the time	1,224	17.5
	Hardly ever	231	3.3
	Never	154	2.2

Attrition analysis

Attrition was 50.2% among housing-insecure observations compared with 46.6% among housing-secure observations. Similarly, attrition was not more likely for those individuals with psychological distress and PTSD than for those without these conditions. Logistic regression models indicated no statistically significant association was observed between baseline housing insecurity and subsequent attrition (OR = 1.23, 95% CI: 0.97–1.55, p = 0.088). Baseline psychological distress (OR = 1.02, p = 0.921), PTSD symptoms (OR = 0.85, p = 0.180), and poor self-rated health (OR = 0.76, p = 0.072). These findings suggest that individuals with housing insecurity or health status (poor self-rated health, psychological distress/PTSD) at baseline were not systematically more likely to have greater attrition over time (Supplementary Table S1-S3). In addition, attrition analyses suggested no clear evidence that baseline housing insecurity, PTSD, psychological distress, or poor self-rated health were associated with non-observation after Wave 1. However, female respondents had higher odds of attrition, whereas offshore entrants had lower odds of attrition (Supplementary Table S3).

Housing insecurity and health trends over time

The overall prevalence of housing insecurity was 49.7% (n = 3,470) (95%CI, 48.5, 50.9) for all waves; prevalence declined from 73.3% (Wave 1) to 39.7% (Wave 6). It fluctuated over the six waves. The overall prevalence of psychological distress was 17.7% (95%CI, 16.9,18.6); it fluctuated over the six waves, with a slight upward trend from 17.2% (Wave 1) to 22.1% (Wave 6). The lowest level of distress was observed in Wave 4 (15.2%). Similarly, the overall prevalence of PTSD was 30.1% (95%CI, 28.0,31.1); it fluctuated across the six waves, decreasing over time, from 34.1% (Wave 1) to 31.3% (Wave 6). There was a significant decline in PTSD from 33.9% to 22.9% in Wave 3 and Wave 4, and the rate again rose to 31.3% in Wave 6. Housing insecurity, PTSD, and psychological distress followed a similar trajectory, with the highest prevalence observed at Wave 1, a decline reaching a low point around Wave 4, and a slight increase by Wave 6 (Fig. 1). The overall prevalence of poor self-rated health was 18.00% (95% CI, 17.1, 18.9). However, poor self-rated health showed a fairly similar trend, with a slight variation from 15.6% in wave 1 to 20.0% in wave 3 (Fig. 1).

Fig. 1.

Trends of housing insecurity, psychological distress, PTSD and poor self-rated health among humanitarian migrants from 2013/2014 to 2023

Association between housing insecurity and health outcomes

After adjusting for confounders, observations who had been experiencing housing insecurity had 66% higher odds of reporting psychological distress compared to those who were not (AOR = 1.66; 95% CI: 1.43—1.1.93). Housing insecurity had also 41% higher odds of PTSD (AOR = 1.41; 95% CI: 1.24–1.60) and 42% higher odds of poorer self-rated health (AOR = 1.42; 95% CI: 1.22–1.66) (Table 3). We subsequently conducted a sensitivity analysis by adjusting for additional variables. The results showed that housing insecurity was associated with psychological distress and PTSD, indicating that the effect estimates were accurate (Supplementary Table S8—9). In addition, in supplementary analyses, we run analyses of housing insecurity as the number of conditions met. A graded association was observed especially for PTSD and psychological distress. Compared with participants with no housing insecurity conditions, those with one condition had higher odds of PTSD (AOR 1.32, 95% CI 1.16–1.51), psychological distress (AOR= 1.50, 95% CI 1.29–1.75), and poor self-rated health (AOR 1.32, 95% CI 1.12–1.56). Participants with two conditions had even greater odds of PTSD (AOR=1.84, 95% CI 1.48–2.29), psychological distress (AOR=2.48, 95% CI 1.92–3.20), and poor self-rated health (AOR= 2.06, 95% CI 1.57–2.70). Those with three conditions had substantially higher odds of PTSD (AOR= 3.57, 95% CI 1.84–6.90) and psychological distress (AOR=5.52, 95% CI 2.96–10.32), while the association with poor self-rated health was elevated but not statistically significant (AOR = 1.52, 95% CI 0.71–3.26)(Supplementary Table S4). Fixed-effects analyses yielded substantively similar results, with housing insecurity remaining positively associated with PTSD, psychological distress, and poor self-rated health. Specifically, the fixed-effects estimates were 1.21 (95% CI 1.02–1.43) for PTSD, 1.52 (95% CI 1.23–1.88) for psychological distress, and 1.45 (95% CI 1.18–1.79) for poor self-rated health (Supplementary Tables S5–S7).

Table 3.

The association between housing insecurity and poor health outcomes (mental and self-rated health) (GEE model)

Health outcomes	COR, (95% CI)	P value	AOR, (95% CI)#	P value
Psychological distress	1.52 (1.34,1.73)	< 0.001	1.66 (1.43, 1.93)	< 0.001
PTSD	1.41(1.26,1.58)	< 0.001	1.41 (1.24, 1.60)	< 0.001
Self-rated health	1.21 (1.10, 1.38)	0.007	1.42 (1.22, 1.66)	< 0.001

^#Adjusted for age, gender, marital status, total household size, English proficiency, country of birth, education, overall settlement experience, sense of belonging, discrimination, visa subclass, and survey wave

Discussion

This longitudinal study examined the complex relationship between housing insecurity and health outcomes among humanitarian migrants in Australia, specifically, mental health and general health outcomes. We observed an association between housing insecurity and increased psychological distress and PTSD, and lower self-rated health. Our results contribute to the growing evidence base linking social determinants of health to refugee wellbeing and highlight the critical need for integrated policy approaches. These findings can inform targeted public health interventions and housing policy reforms specifically designed to address the unique challenges humanitarian migrants face, potentially yielding substantial improvements in their long-term health trajectories and successful integration.

The longitudinal data showed a considerable decline in housing insecurity among humanitarian migrants over time, from around three-quarters in 2013/14 to less than half in 2023. This is likely due to refugees usually being accommodated in temporary housing and relying on service providers for the assistance that they require to secure more stable accommodation once resettlement has commenced [52]. Over time, their reliance on formal housing assistance should then decrease, while their satisfaction with the quality of housing increases [52]. This decline may be linked with increased access to employment, linguistic ability, and social support during the process of integration [53]. However, housing insecurity persisted. The high rate of persistent housing insecurity over time may be attributed to chronic structural barriers such as the high cost of private rentals, discrimination by landlords, and accessibility obstacles for large or racialized families [8, 52]. Another plausible explanation might be the emergence of COVID-19 after 2019, which created higher unemployment rates, exacerbated by the ongoing rise in rental costs [54, 55]. Our findings highlight that while humanitarian migrants experience greater housing security, a significant proportion remain vulnerable to housing insecurity long after initial settlement. These findings underscore the need for sustained support beyond the initial resettlement phase to address the housing insecurity experienced in the long term by a significant proportion of humanitarian migrants. Policy and programmatic responses need to focus on building pathways to stable employment, enhancing the availability of affordable housing options, and strengthening social support networks, particularly in the context of economic downturns such as pandemics and prevailing cost-of-living pressures [56–59]. Furthermore, the potential compounding effects of exogenous factors such as economic downturns and increases in rental costs on housing security for these populations must be considered in long-term settlement planning and policy responses.

The simultaneous trajectories of housing insecurity, psychological distress, and PTSD peaking early after arrival suggest a close interrelationship between housing stability and mental health outcomes [8, 31, 60]. Research consistently finds that unstable, poor-quality, or unaffordable housing exacerbates refugees' and migrants' stress, anxiety, and trauma [8, 31, 60]. Improvements in housing conditions are linked to better mental health [61–63], while chronic housing insecurity is a strong risk factor for psychological distress and PTSD [64]. This evidence supports the view that secure, affordable, and adequate housing is a key social determinant of psychological well-being for humanitarian migrants [8]. In contrast to the trends demonstrated for housing insecurity and mental health, self-rated general health changed minimally throughout all waves. This aligns with evidence to suggest that poor general health remains at high levels among refugees in the early years of their resettlement [65] and might be caused by unattended health states [66] and the cumulative effect of pre- and post-migration stressors [65, 67]. This implies that early access to stable and affordable housing and simultaneous mental health care is critical to relieving psychological trauma and PTSD in humanitarian migrants. Culturally competent healthcare, building on this, is critical to meeting ongoing unmet health needs during post-resettlement.

In this study, humanitarian migrants experiencing housing insecurity had higher odds of reporting psychological distress compared to those in stable housing. Previous research has consistently demonstrated that housing insecurity is one of the primary causes of chronic stress, which in itself is an established risk factor for a range of mental disorders, such as depression and anxiety [48, 68–72]. National data from the Specialist Homelessness Services (SHS) in Australia reported that about 39% of people aged 16–85 with a history of homelessness had a mental health issue in the last year, nearly double the rate of the general population [73]. Another Australian study also suggests that housing insecurity leads to a deterioration in mental health [74]. In addition, a nationally representative survey in the United States during the COVID-19 pandemic reported that those who were housing insecure reported significantly higher rates of moderate to severe psychological distress than those with stable homes [48]. One possible explanation is the stress of potential displacement, financial hardship, and disruption of social networks, all of which can enhance feelings of insecurity, loneliness, and powerlessness [68, 69]. These stressors not only increase the risk of psychological distress but can also hinder recovery from mental ill-health, particularly where there is poor social support and access to health services [70, 75, 76].

Migrants experiencing housing insecurity in this study had higher odds of reporting PTSD compared to those who were not. The association of housing insecurity with PTSD supports previous findings that unstable housing status increases the frequency of trauma-related mental illness [77]. In addition, housing insecurity may lead to displacement, exposure to violence, or homelessness, all of which are significant risk indicators for the development of PTSD [77, 78]. Chronic stress related to ongoing housing instability can act as a chronic trauma, heightening both risk and severity of PTSD symptoms, especially in already vulnerable groups [69]. Therefore, there is a strong need for comprehensive support systems that address housing needs as well as mental health care among migrants, including proactive screening, culturally appropriate care, and multidisciplinary programs that support stable housing and mental health care. While humanitarian migrants demonstrate improvement in housing and health outcomes over time, there are persisting vulnerabilities, particularly in housing insecurity and its mental health consequences. These call for specific policy responses that prioritise access to affordable and secure housing, availability of culturally and linguistically appropriate tenancy support, and inclusion of trauma-informed mental healthcare as part of housing interventions. Structural barriers, such as discrimination in the rental market and restrictive eligibility for housing assistance, also need to be addressed systematically [79].

Finally, poor self-rated health was significantly higher among individuals experiencing housing insecurity compared to those who were not. There is broader evidence that people who experience housing insecurity are more likely to be in poorer health due to such factors as disrupted healthcare access, poor housing conditions, and the stress of dealing with housing uncertainty [8]. A study analyzing Washington State Behavioral Risk Factor Surveillance System data found that housing-insecure survey respondents were twice as likely to report bad or fair health status and avoiding going to the doctor due to cost [80]. This may be because housing insecurity may undermine physical health by reducing access to health-promoting resources such as stable housing, social support, and health care services [81]. Furthermore, housing insecurity can limit an individual’s ability to access regular health check-ups, exacerbating chronic disease, which leads to poorer self-reported health [82]. These findings emphasize the value of structural interventions that expand access to safe and affordable housing, including rent subsidies and supportive housing programs, as part of public health strategies to promote long-term health outcomes among humanitarian migrants.

Strengths and limitations of the study

This is the first study to investigate the relationship between housing insecurity and health outcomes using 10-year longitudinal data among humanitarian migrants. A key strength of this research is the longer data-collection period, which enables a full picture of long-term trends and associations. The use of Generalised Estimating Equations (GEE) is also a specific strength, as this robust modelling method is well-suited to handle the correlated nature of repeated measures within individuals, thereby providing more precise and generalizable estimates. Sometime-varying covariates, such as settlement experiences, may be endogenous, potentially influencing estimates. To assess robustness, we conducted sensitivity analyses excluding this variable, and the main associations remained unchanged. Loss to follow-up may introduce selection bias if dropouts systematically differ from those who remained in the study; therefore, we conducted attrition analysis. These findings suggest that individuals with housing insecurity or health status at baseline were not systematically more likely to have greater attrition over time. Another strength is the nationally representative humanitarian migrant sample, making the study relevant to guiding policy and practice. However, the study has several limitations. Notably, although the BNLA is a longitudinal cohort survey, the observational design does not allow definitive causal inference. The study's operationalisation of housing insecurity presents a methodological limitation. For instance, by relying on limited housing security variables available in the BNLA datasets, our approach may not capture the multidimensional nature of housing insecurity, potentially overlooking important dimensions including housing quality, overcrowding, and neighbourhood safety. Furthermore, although BNLA includes derived measures of time since arrival, these were coded differently across study groups and could not be harmonised into a single comparable covariate for final estimates. Thus, the subsequent waves of the BNLA survey should cover broader dimensions of housing insecurity and a similar categorisation of variables. Some potential confounders, including employment status, pre-migration trauma, and social support, were excluded due to sample size limitations, though sensitivity analyses were done for the rest of the confounders with a small sample size. Furthermore, some time-varying covariates may be endogenous, potentially influencing estimates. The reliance on self-reported data may introduce recall or social desirability bias. Future datasets will be benefitted by incorporating more objective data including linked administrative data or clinician-administered assessments to compliment self-report data. Although the longitudinal design facilitated examination of changes over time, the study remains observational. Similarly, the estimated associations should not be interpreted as causal effects, and residual confounding cannot be ruled out.

Conclusions

Housing insecurity, psychological distress, and PTSD declined over time but showed a slight increase in later waves, highlighting persistent vulnerabilities among humanitarian migrants. However, self-rated health remained stable throughout the ten years. Our study presents strong evidence of the longitudinal associations between housing insecurity and (psychological distress, PTSD, and self-rated health). Addressing housing insecurity through targeted public health and housing policies may improve health outcomes for humanitarian migrants. These findings offer essential information for policymakers and practitioners by demonstrating that housing insecurity is significantly associated with poorer mental and physical health outcomes among humanitarian migrants.

Abbreviations

AOR

Adjusted Odds Ratio

BNLA

Building a New Life in Australia

PTSD

Post-traumatic stress disorder

Authors’ contributions

HBE, KB, RL and GC conceived the idea. HBE extracted the data, conducted the analysis, and prepared the original draft of the manuscript. GAT, KB, RL, BD and GC assisted in the data analysis and interpretation. KB, RL and GC GAT critically reviewed the manuscript.

Funding

The authors did not receive any funding for this study.

Data availability

The availability of the data set for this study was obtained from the Australian Institute of Family Studies data sets using the website [https://dataverse.ada.edu.au/dataverse.xhtml?alias=bnla]. https://dataverse.ada.edu.au/dataverse.xhtml?alias=bnla.

Declarations

Ethics approval and consent to participate

The BNLA dataset is made available to approved researchers (our approved application number is #294715). Ethical approval for the original BNLA study was granted by the Australian Institute of Family Studies Human Research Ethics Committee. All information is found here: https://aifs.gov.au/building-new-life-australia/ethics-approval. All Participants in the study provided written informed consent. All methods were carried out in accordance with the Helsinki Declaration.

Consent for publication

Not applicable since it is secondary data.

Competing interests

The authors declare no competing interests.