Abstract
Importance Eviction is a pressing public health concern that disproportionately impacts disadvantaged families. Understanding how caregivers’ stress about eviction relates to specific childhood psychiatric issues across development is important.
Objective To examine associations between eviction stress and caregiver-reported child depression, anxiety, attention-deficit/hyperactivity disorder (ADHD), and behavioral problems, adjusting for sociodemographic factors.
Design Cross-sectional analysis of a nationally representative survey.
Setting United States.
Participants Over 36,000 caregivers and children from a national survey.
Main Outcomes and Measures Caregiver-reported child depression, anxiety, ADHD, and behavioral problems. Generalized linear mixed models were used to test associations with eviction stress.
Results Eviction stress was associated with 12-35% increased odds of depression and anxiety. Relationships were moderated by age, with stronger associations for younger children. No significant relationships emerged between eviction stress and ADHD or behavioral problems after adjustment.
Conclusions and Relevance This study provides new insights into how eviction-related stress differentially impacts internalizing versus externalizing psychopathology across child development. Prospective longitudinal research is still needed to fully understand these complex relationships over time. Findings underscore the importance of policies and interventions to address housing instability and its mental health consequences for children.
Eviction is a major public health concern that disproportionately impacts disadvantaged families.
This nationally representative study examined associations between caregiver eviction stress and specific childhood mental health outcomes.
Eviction stress was associated with 12-35% increased odds of child depression and anxiety, with stronger relationships for younger children.
No significant relationships emerged between eviction stress and child attention-deficit/hyperactivity disorder or behavioral problems after adjusting for sociodemographic factors.
These findings provide new insights into how eviction-related stress differentially impacts internalizing versus externalizing psychopathology across child development.
Prospective longitudinal research is needed to further understand these complex relationships over time.
The results underscore the importance of policies and interventions to address housing instability and its mental health consequences for children.
Eviction and housing instability presents significant threats to the well-being of millions of Americans each year1,2. It is estimated that more than 3.6 million evictions are filed annually, with millions more people threatened with removal from their rental homes3. This is of incredible import for public health and public policy, as adults under threat of eviction report and evince multiple negative physical and mental health outcomes2,4,5. Notably, eviction and the threat of eviction are not equally distributed across demographics– it disproportionately affects communities of color, the economically marginalized, and families with children in their home6. Despite these alarming facts, there are many open questions related to the impacts of eviction on child mental health where deeper investigation is needed, especially in regards to the etiology and occurrence of different psychiatric issues.
Research finds that eviction, threat of eviction, and housing instability have significant health impacts on families and children, including premature birth, low birthweight, maternal depression, and more parenting stress7–10. Housing instability and eviction have been linked to decreased social support11, food insecurity12,13, increased conflict in the home14, and harsh parenting15–19. As such the multiple challenges associated with eviction may cascade to significantly impact child development and mental health20–23. For example, housing instability and experiencing multiple moves in childhood is associated with more externalizing problems24 – or aggression, rule-breaking, and other disruptive behaviors that are expressed outwardly through actions25. Similarly, household and residential changes in adolescence were related to depression and internalizing problems, or issues that are inwardly directed, with distress in internal thoughts and feelings26. Examined collectively, research generally suggests eviction impacts child mental health, but further study is needed to understand the consequences of these experiences on child and adolescent functioning.
While existing research has provided insights into links between housing instability and child mental health, important gaps remain in fully understanding these relations. First and foremost, while individuals and families can be formally evicted via court order, informal evictions are actually much more common. Informal eviction is the most common reason (∼72%) for a forced move27, as landlords may pressure, intimidate, or deceive a renter to leave without an official court proceeding28. It is important to think about psychological pressures and concerns related to eviction to more fully operationalize this issue. Second, additional work is needed to examine eviction’s associations with mental health in relation to children’s age. Eviction at specific points in development could have larger impacts on health; for example, younger children may be more impacted by stress and experiences as: they are more dependent on caregivers and substantial brain development is occurring earlier in time29,30. Age may moderate the impact of eviction given that different forms of psychopathology show differential onsets in childhood versus adolescence. Finally, there is significant variability in how the impact of eviction is isolated, and what confounding variables are used in adjusting statistical models. Nearly all projects include child demographics, while only some projects adjust for other developmental challenges common to families who face eviction (i.e., food insecurity; low birth weight). Moving forward, it will be critical to think about psychological pressures and anxiety related to eviction, while comprehensively adjusting statistical models and considering potential interactions with age.
To overcome these limitations, this study will probe relations between caregivers’ stress about being evicted and indicators of four forms of psychopathology– depression, anxiety, attention-deficit/hyperactivity disorder (ADHD), behavioral/conduct problems. Specifically, we will examine: the main effect of eviction stress; and the interactions between eviction stress and child age to potentially understand differential developmental impacts of mental health. We will probe these associations using a base set of demographic covariates, and also a wider set of potential confounding variables. Given the wide-reaching impacts of eviction, we expect to find: 1) Caregiver eviction stress will be positively associated with risk for these four types of mental health problems; and 2) the association between caregiver eviction stress and child mental health problems will be moderated by age.
Method
Participants
We leveraged the 2022 National Survey of Children’s Health (NSCH), an ongoing population-based survey that collected responses from over 50,000 caregivers in the United State31. NSCH is a cross-sectional, high-quality survey, administered by the U.S. Census Bureau. Parents or other knowledgeable adult caregivers answered questions about one specific child (ages 0–17 years old) under their care. These caregivers completed a large battery of questions about their child, their demographics, their family circumstances, and multiple other domains.
Measures
To measure eviction stress, caregivers were asked one question about how often they were worried or stressed about being evicted, foreclosed on, or having their house condemned during the past year. Responses included “Always, Usually, Sometimes, Rarely, Never”. We reverse the coding of this item, so that higher scores indicated greater concern and stress about these issues. Related to mental health, we focused on caregiver reports of whether their children currently had, or they were previously told that their children had the following conditions: depression, anxiety, ADHD, or behavioral/conduct problems. We recoded these questions to be a binary indicator, with 1=“Currently has condition” and 0=“Does not have condition” or “Ever told, but does not currently have condition”.
Regarding statistical covariates, we adjusted models for a basic set of demographic factors, including child sex, child age, child race/ethnicity, family structure, caregiver’s highest education level, and household poverty level based on federal guidelines and caregiver-reported family income (as noted in Table 1). We also more stringently adjusted our statistical models for factors correlated with, but independent of, eviction stress. These included: the number of places the child lived in the past year, if the child ever experienced homelessness, premature birth weight, low birth weight, poor maternal physical health, food insecurity over the past year, and exposure to adverse childhood experiences. We employed these two sets of covariates motivated to balance under– and over-adjustment biases. While it is critical to rule out potential confounding variables, “overadjustment” in social science can skew estimates away from the true total causal effect, prevent consistent estimation of associations, and introduce additional errors (e.g., collider-stratification)32,33.
Statistical Analyses
We ran separate generalized linear mixed models with a logistic link function, where each form of psychopathology (binary indicator) was entered as the dependent variable; in base or stringently adjusted models, eviction stress and different covariates (noted above) were entered as independent variables, and we included a random factor for geographic location (50 states in the US, plus the District of Columbia). These multiple statistical adjustments were a way to address potential sources of bias. For parsimony, we included the interaction between eviction stress and age in all models. We calculated adjusted odds ratios and 95% confidence intervals (CI) for our independent variables of interest (eviction stress; eviction stress X age). For significant interactions, we tested differences for the simple slopes of the association between eviction stress at different values of child age in relation to each form of psychopathology. Analyses only included participants with complete data for all covariates across the two sets of covariates (Analytic N=36710). All statistical analyses were conducted using R version 4.4.034,35. Additional information about the statistical packages used are noted in our supplement. These supplemental materials also included statistical models: probing eviction X race interactions and re-running models after utilizing data imputation techniques for missing independent variables.
Results
Models Examining The Presence Of Depression or Anxiety
Our statistical models with a base set of covariates found several factors significantly associated with the presence of depression, including family structure, race/ethnicity, caregiver education, poverty status, and sex assigned at birth (as noted on the left side of Table 2). Eviction stress was significantly associated with higher odds of current depression, with an adjusted OR of 1.35 [CI=1.27-1.44], z=9.329, p<.001. The interaction of eviction stress X age was significant (z=– 2.398, p=0.016, as shown in Figure 1, Panels A and C). The slope for eviction stress was continually significant, but varied, at different levels of age (at lower levels of age –1 SD,CI=0.26-0.48, z=6.43, p<.001; at higher levels of age +1 SD CI=0.19-0.27, z=11.38, p<.001). More stringent model adjustment attenuated some of these connections, as shown on the right side of Table 2. Eviction stress was again significantly associated with higher odds of depression, with an adjusted OR of 1.12 [CI=1.04-1.19], z=3.122, p=0.02. The interaction of eviction stress X age was significant (z=-2.239, p=0.025, as shown in Figure 1, Panels B and D). For these models, the slope for eviction stress was significant in younger participants, but not for older participants (–1SD age CI=0.06-0.30, z=2.9, p=0.004; at higher levels of age +1 SD CI=0.00-0.09 z=1.83, p=0.068).
The top of the figure shows moderation graphs of eviction (horizontal axis) by age (green, blue, and red lines). The bottom panel depicts the slope of the relations between eviction stress and depression at varying levels of age. The left side is base adjusted models, while the right side shows more stringently adjusted models.
The left side of the table shows adjustment using a base set of covariates, while the right side has a more stringently set of model adjustments.
Similar to depression, the base adjusted model found anxiety was related to many sociodemographic factors (as noted on the left side of Table 3). Connected to our primary hypotheses, greater reported eviction stress was significantly associated with higher odds of reported anxiety, with an adjusted OR of 1.26 [CI=1.22-1.31], z=12.792, p<.001. The interaction of eviction stress X age was significant (z=-2.049, p=0.04, as shown in Figure 2, Panels A and C). Again, the slope for eviction stress was continually significant, but varied, at different levels of age (at lower levels of age –1 SD,CI=0.21-0.33, z=8.47, p<.001; at higher levels of age +1 SD CI=0.16-0.23, z=11.69 p<.001). More stringent model adjustment attenuated relations between anxiety and sociodemographic factors (as noted on the right side of Table 3). Paralleling results from depression, even in stringently adjusted models, eviction stress was again significantly associated with higher odds of anxiety, with an adjusted OR of 1.12 [CI=1.01-1.08], z=2.734, p=0.006. The interaction of eviction stress X age was not significant (z=-1.757,p=0.079, as shown in Figure 2, Panels B and D).
The top of the figure shows moderation graphs of eviction (horizontal axis) by age (green, blue, and red lines). The bottom panel depicts the slope of the relations between eviction stress and anxiety at varying levels of age. The left side is base adjusted models, while the right side shows more stringently adjusted models.
The left side of the table shows adjustment using a base set of covariates, while the right side has a more stringently set of model adjustments.
Focusing on ADHD, we see increased incidence of this disorder was related to family structure, level of caregiver education, poverty status, and sex assigned at birth (see Table 4, left side). Eviction stress was related to greater rates of ADHD, OR=1.19 [CI=1.15-1.23], z=9.503, p < 0.001. The interaction of eviction stress X age was not significant (z=-1.69, p=0.09, as shown in Figure 3, Panels A and C). When adjusting for a more stringent set of covariates, there were connections between ADHD and different covariates (e.g., having low birthweight, childhood adversity, see Table 4, right side). Of important note, in these models, ADHD was not related to eviction stress (OR=1.04, CI=1.00-1.08, z=1.777, p=0.07) or the interaction of eviction stress X age (z=-1.45, p=0.14, as shown in Figure 3, Panels B and D).
The top of the figure shows moderation graphs of eviction (horizontal axis) by age (green, blue, and red lines). The bottom panel depicts the slope of the relations between eviction stress and ADHD at varying levels of age. The left side is base adjusted models, while the right side shows more stringently adjusted models.
The left side of the table shows adjustment using a base set of covariates, while the right side has a more stringently set of model adjustments.
Finally, regarding behavioral/conduct problems, in our base adjusted models, we see increased incidence of this disorder was related to race/ethnicity, family structure, and some indices of poverty status (as noted on the right side of Table 5). Eviction stress was related to greater rates of behavioral/conduct problems, OR=1.22 [CI=1.18-1.26], z=11.142, p < 0.001. The interaction of eviction stress X age was not significant (z=-0.188, p=0.85, as shown in Figure 4, Panels A and C). Stringent adjustment did not change the majority of these associations, but there were significant connections between behavioral/conduct problems and different covariates (e.g., low birth weight, premature birth, food insecurity, childhood adversity, as shown on the right side of Table 5). Again, of important note and in contrast to results for internalizing issues, behavioral/conduct problems were not related to eviction (OR=1.02, CI: 0.98,1.06, z=0.894, p=0.372) or the interaction of eviction stress X age (z=0.217, p=0.828, as shown in Figure 4, Panels B and D).
The top of the figure shows moderation graphs of eviction (horizontal axis) by age (green, blue, and red lines). The bottom panel depicts the slope of the relations between eviction stress and behavioral/conduct problems at varying levels of age. The left side is base adjusted models, while the right side shows more stringently adjusted models.
The left side of the table shows adjustment using a base set of covariates, while the right side has a more stringently set of model adjustments.
Figure 5 shows associations between eviction stress and different forms of psychopathology for stringently adjusted statistical models (in Figure 5). Our supplemental materials also probe eviction X race interactions and re-runs models after utilizing data imputation techniques for missing independent variables.
DISCUSSION
This study explored the association between caregivers’ eviction stress and children’s mental health, namely depression, anxiety, ADHD, and behavioral/conduct problems. We probed whether these mental health outcomes were related to the main effect of eviction stress, as well as the interaction of eviction stress and age. We saw the strongest relations between eviction-related stress and depression and anxiety, two forms of internalizing psychopathology. Eviction-related stress was related to a 6-35% increase in incidence of these types of disorders. Relations between eviction, ADHD, and behavioral/conduct problems were less robust, with models adjusting for a base set of covariates finding relations; however, these associations were not significant when adjusting estimates using a more stringent set of potential confounders. We saw some notable relations between mental health and the interaction of eviction stress and age; specifically, relations between eviction stress and depression and anxiety were strongest in younger samples, often becoming non-significant in older participants. Examined collectively, these findings provide valuable information to researchers interested in the role of eviction in child development, and in particular how this may influence mental health.
These results fit with past findings that eviction and eviction-related stress may significantly impact development. Our findings connect to work noting increased depression in adolescents experiencing household and residential changes26. We, however, did not see strong relations between eviction stress and externalizing problems. This diverges from past work noting that housing instability and experiencing more than three moves in childhood was associated with aggression and other disruptive behaviors24. This may be in part driven by our statistical modeling choices and the use of a stringent set of covariates. In both sets of models, we see that eviction stress is related to internalizing issues; however, associations between eviction stress and externalizing were only present in models adjusted for a base set of covariates and not in more stringently adjusted models. Externalizing problems were more associated with a child’s perinatal risk (i.e., low birth weight; premature birth), but not all projects have controlled for these variables. Eviction’s impact may be through more indirect developmental pathways, as it may contribute to perinatal risk, and this is then associated with externalizing challenges. As such, future research is needed to comprehensively understand relations between eviction and externalizing problems.
Reflecting on our work, this cohort was nationally representative and had a good deal of diversity in age, family structure, and socioeconomic status. Related to age, many past studies have only examined a very confined age range (e.g., middle childhood), but our work had participants from young infants to 17 years of age. With family structure, some past publications have used cohorts oversampling unmarried parents in large U.S. cities36. While appropriate, one has to use statistical weights to have a truly nationally representative sample and make estimates generalizable to all families. Finally, connected to socioeconomic status, it is important to note that while eviction is concentrated in families at or below the federal poverty line6, eviction and eviction-related stress still occurs for those above the federal poverty line (as detailed in our Supplement). Examined collectively, this dataset allows for a more inclusive assessment of how the stresses associated with eviction are linked to child well-being.
Our project is, however, not without limitations. First, the work is cross-sectional in nature. Moving forward, longitudinal studies are needed to richly isolate the developmental impacts of eviction. Such designs could be particularly informative related to the age X eviction interactions that we described here. We find that at the highest levels of age in the cohort, eviction stress was often no longer significantly related to the presence of different disorders. This result, though very interesting, could be indexing conflicting, developmental phenomena. One possibility is that eviction and eviction-related stress occurring early in life have outsized influences on development. Gaps in critical developmental skills may emerge early in childhood due to experiences before 5 years of age37. Alternatively, adolescents spend increasingly more time outside the home38,39 and the eviction-related stress of caregivers may have less “opportunity” to affect youth’s mental health. Prospectively following participants and seeing how families deal with eviction and eviction-related stress will be important to fully understand potential interactions between eviction and development. In terms of other limitations, we used caregiver self-reports of mental health challenges; our work would have been strengthened by getting independent measures or clinical interviews of potential psychopathology. Use of multiple informants (e.g. teachers, clinicians) would have provided a more comprehensive picture of the child’s behaviors and functioning across different contexts (i.e., home; school40. This type of information could be advantageous to help guide treatment and intervention strategies.
Limitations notwithstanding, here we find that eviction-related stress is associated with an increased incidence of poor mental health in childhood. Given the profound long-term economic and social impacts of these problems41, it will be critical for communities to think about lessening housing precarity and better supporting families facing eviction. This could take many forms including rental assistance, legal aid for tenants, eviction diversion programs, and expanded social safety nets42,43. With millions facing evictions annually, those in public health and public policy must continue to push for expansion of these programs to reduce inequities and this key social determinant of child health.
Conflicts of Interest and Financial Disclosures
None to report
Funding/Support and Role of Funder/Sponsor
The work was supported by internal funds provided to Dr. Hanson by the University of Pittsburgh; that university played no role in any element of the work
Author Contributions
Dr Hanson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Data Sharing Statement
All data is publicly available from the Census website: https://www.census.gov/programs-surveys/nsch/data/datasets.html
Disclaimer
The views expressed in this article are those of the authors and do not necessarily reflect the official policies of the US Department of Health and Human Services (HHS) or the Health Resources and Services Administration (HRSA), nor does mention of HHS or HRSA imply endorsement by the US government.
Data Availability
Data is available from the Health Resources and Services Administration (HRSA) of the U.S. Department of Health and Human Services (HHS).
