Video Abstract

Video Abstract

Close modal
BACKGROUND AND OBJECTIVES

Housing insecurity is associated with adverse effects on child growth and development cross-sectionally; less is known about its cumulative, long-term effects. This study describes longitudinal experiences of housing insecurity during childhood from infancy (age 1 year) to adolescence (age 15 years) and examines their associations with adolescent health outcomes.

METHODS

Using data from the Future of Families and Child Wellbeing Study, we created a composite measure of housing insecurity using 5 indicators (eg, skipping a rent or mortgage payment, eviction) for participants at ages 1, 3, 5, 9, and 15 years. We used group-based trajectory modeling to identify distinct patterns of housing insecurity, sociodemographic predictors of these patterns, and how these patterns relate to adolescent health outcomes.

RESULTS

We identified 3 trajectories of housing insecurity from infancy to adolescence: secure, moderately insecure, and highly insecure. Adolescents who experienced moderately and highly insecure housing had decreased odds of excellent health (adjusted odds ratio, 0.81; 95% confidence interval [CI], 0.69–0.95; adjusted odds ratio, 0.67; 95% CI, 0.50–0.92, respectively) and more depressive symptoms (adjusted incidence rate ratio, 1.05; 95% CI, 1.02–1.08; 1.13; 95% CI, 1.08–1.19, respectively) than adolescents with secure housing. Adolescents who experienced highly insecure housing reported significantly higher anxiety symptoms (adjusted incidence rate ratio, 1.05; 95% CI, 1.003–1.113).

CONCLUSIONS

Housing insecurity starting in infancy was associated with poorer adolescent health outcomes. These longitudinal patterns emphasize the need for novel screening mechanisms to identify housing insecurity when it emerges, as well as policies to prevent housing insecurity and its associated health outcomes.

What’s Known on This Subject:

Housing is widely recognized as a social determinant of health. In children, housing insecurity has been associated with a variety of negative physical and mental health outcomes. Existing studies typically examine housing insecurity at a single point in time.

What This Study Adds:

We contextualize the cumulativeness and long-term impact of housing insecurity in childhood by describing longitudinal patterns of housing insecurity experienced from infancy to adolescence and how these patterns prospectively relate to adolescent health outcomes.

Housing insecurity is the experience of not having a stable, affordable, and safe home.1  Distinct from housing quality, housing insecurity extends beyond the physical home to the intangible “feeling of home”: one’s lived experience of stability where they reside.2–4  No universal criteria defines housing insecurity yet, but researchers have used 1 or more of the following hardships to indicate its presence: overcrowding, difficulty paying rent or mortgage, eviction, homelessness, or frequent moves.1,2,5,6 

Housing is increasingly recognized as an adverse social determinant of health given its associations with physical and mental health.1,3,7  It may influence child health directly or through disruptions to caregiving or other hardships (eg, inconsistent health insurance coverage, scrutiny from welfare systems).8–10  In children, it is associated with attention problems, internalizing and externalizing behaviors, and decreased wellbeing.2,11,12  In parents, housing insecurity is associated with poorer health, increased depressive symptoms, and parenting stress.6,13 

Housing insecurity can occur concurrently with, and adversely alter, trajectories of healthy child growth and development.5,6,14  Life course researchers have shown that prolonged, cumulative experiences of hardship and instability (eg, stress from unpredictability) during key stages in childhood can lead to long-term health effects.5,15,16  Examining housing insecurity as a longitudinal pattern with child development may provide unique insight into how they intersect and influence adolescent outcomes, highlighting optimal periods for intervention.13,17–20 

Studies identifying longitudinal patterns of housing insecurity have focused mainly on adults (eg, urban fathers, vulnerably housed individuals).21–23  Kang identified 4 distinct trajectories of housing stability in low-income households with and without children: chronically unstable, mostly unstable, mostly stable, and consistently stable.21  Aubry also identified 4 longitudinal trajectories of housing stability in vulnerably housed adults: moderate to low, moderate to high, high to moderate, and high sustained housing stability.22  Less is known about longitudinal experiences of housing insecurity for children and their relationships with adolescent health outcomes.21 

To address this gap, we use longitudinal data from the Future of Families and Child Wellbeing Study (FFCWB), an ongoing national birth cohort study.24  We assessed indicators of housing insecurity (eg, skipping rent, eviction) and applied group-based trajectory modeling (GBTM) to identify patterns of housing insecurity across childhood from infancy (age 1 year) to adolescence (age 15 years). We hypothesized that: (1) we would identify distinct, varying patterns of housing insecurity with 1 mostly secure group; (2) certain sociodemographic characteristics would be associated with patterns of insecurity; and (3) patterns reflecting increased housing insecurity would have increased odds of poor health outcomes in adolescence.

We performed a secondary analysis of FFCWB data. As previously described, 4898 participants were enrolled at birth across 20 large, nationally representative US cities with varying local welfare support systems.24  Nonmarital births were oversampled (3:1) to represent births in each city, leading to an oversample of individuals from minoritized racial and ethnic groups with financial disadvantage, to whom nonmarital births disproportionally occur.24,25  Birthing parents (individuals who gave birth) were interviewed after delivery (baseline) and when their children were 1, 3, 5, 9, and 15 years old. Our institution’s institutional review board deemed this study exempt from human subjects review because these were publicly available, deidentified data.

Housing Insecurity

We created an index measure of housing insecurity using the following indicators based on previous studies: skipping a rent or mortgage payment, doubling up (sharing housing with other families), eviction, homelessness (spending at least 1 night in the past year sleeping in a shelter, car, abandoned building, or another place not meant for residence), or moving more than once during any year since the last timepoint.1,2,10,21–23,26  We created a count of these indicators, from 0 (secure) to 5 (highly insecure), to measure the severity of housing insecurity experienced at each timepoint (ages 1, 3, 5, 9, 15 years).

Adolescent Health Outcomes

The primary outcomes at age 15 years were self-reported overall health (SRH) and depressive and anxiety symptoms. SRH was measured with 1 Likert-scaled question: “In general, how is your health?” with 5 answer choices from “poor” (1) to “excellent” (5).27  This question has demonstrated validity across several contexts as a predictor of mortality and overall health.28–30  We dichotomized SRH by coding “very good” (4) and “excellent” (5) as 1. Dichotomized versions of this measure yield results comparable to its original ordinal form.31 

Depressive symptoms were measured with an adolescent-tailored adaptation of the 5-item Center for Epidemiologic Studies Depression Scale.32,33  Anxiety symptoms were measured with 6 modified questions from the Brief Symptom Inventory 18, Anxiety Subscale.34  Responses were scored on a 4-point scale and summed into overall scores for each outcome. Depression and anxiety scores ranged from 5 to 20 and 6 to 24, respectively, with higher scores indicating more symptoms.

Sociodemographic Characteristics

At baseline, birthing parents reported their age (years), racial and ethnic identity (white, non-Hispanic; Black, non-Hispanic; Hispanic; other), level of education (eg, less than high school, high school/equivalent), household income (Federal Poverty Level [FPL] proximity), nativity (United States born or not), relationship status with the child’s nonbirthing parent (eg, married, cohabitating), household size (number of people in household collected at age 3 years), the child’s biological sex, and first-born status (yes/no).24 

Analytic Sample

We included participants with at least 1 housing measure (N = 4714, 96.2% of the main sample) in our analytic sample. Nearly 75% (3476/4898) of participants had housing measurements at 4 or more timepoints. Incomplete case analysis was preferred over a complete case approach to reduce bias from loss to follow-up, which would likely be higher in families experiencing housing insecurity.35  Sensitivity analyses were conducted in a subsample of children with complete data across all 5 timepoints.

Group-Based Trajectory Modeling

GBTM is a statistical methodology that uses longitudinal data (eg, clinical or behavioral measures) to identify clusters of individuals who follow similar developmental, behavioral, or experiential trajectories over time.36–38  Trajectory models estimate group patterns and may not represent each individual experience exactly. GBTM is useful for modeling incomplete data because it accommodates all available information and provides maximum-likelihood estimations of model parameters.39,40 

Following previously outlined model-selection procedures, we used GBTM to identify distinct trajectories of housing insecurity from ages 1 to 15 years using our housing insecurity index measure.36,41  A zero-inflated Poisson model was chosen because the housing insecurity index measure had excess zeroes (70% or more) at every timepoint.41  The number of trajectory groups was determined using Bayesian information criterion. Several polynomial variations and orders were tested, with an a priori zero-term to represent a “consistently secure” trajectory of housing.36  The final model maximized Bayesian information criterion and entropy, contained group posterior probabilities that exceeded the recommended 0.70 threshold, and demonstrated consistency between observed and model-estimated group membership probabilities.39,41  Participants were assigned to the trajectory group in which they had the highest probability of membership.

Hypothesis Testing

Groupwise comparisons were made using analysis of variance or χ2 testing, depending on variable type. Logistic regression was used to model SRH. As demonstrated previously, the distributional nuances (eg, overdispersion, variance patterns) of depressive and anxiety symptom scores were not suitable for linear regression.42  Poisson regression was used to model anxiety symptoms. Negative binomial regression was used to model depression symptom scores because its variance exceeded the mean.43  Baseline sociodemographics significantly associated with the housing groups were considered as covariates.44  The child’s biological sex was included a priori because of its relationship with adolescent health.45,46  Final models were selected in a backward stepwise manner and model assumptions (eg, distribution of outcome, independence of observations) were adequately met.

Tests were 2-tailed and α was set to .05. Statistical analyses were completed using Stata (version 17).47  GBTM procedures were completed using the “traj” package.48 

The final group-based trajectory model contained 3 distinct trajectories of housing insecurity experienced from ages 1 to 15 years: no insecurity (“secure”); intermittent-moderate insecurity (“moderately insecure”); and high insecurity in early childhood, trending toward security in adolescence (abbreviated to “highly insecure”) (Fig 1). Most children were classified into the secure (2230/4714, 47.3%) and moderately insecure (2188/4714, 46.4%) groups, whereas 296 (6.3%) children were classified into the highly insecure group. Model summary statistics are in Supplemental Table 4.

FIGURE 1

Trajectories of housing insecurity experienced by focal child from ages 1 to 15 (n = 4714).

FIGURE 1

Trajectories of housing insecurity experienced by focal child from ages 1 to 15 (n = 4714).

Close modal

Table 1 describes the sociodemographics for the FFCWB sample overall and by housing group. Educational attainment and household income were higher in the secure group and more parents in the secure group were married (34.6%). The highly insecure group had the largest proportions of parents who had not completed high school (51.5%) and who were living below the FPL (46.3%).

TABLE 1

Baseline Sociodemographic Characteristics Overall and by Housing Group (n = 4714)

VariableSampleSecureModerately InsecureHighly InsecureP
n = 4714n = 2230n = 2188n = 296
Parent characteristicsn (%)n (%)n (%)n (%)
Birthing parent’s age, y; mean (SD) 25.3 (6) 26.4 (6.3) 24.3 (5.6) 23.6 (5.46) <.001a 
Race/ethnicity     <.001a 
 White, non-Hispanic 993 (21.1) 558 (25.1) 364 (16.7) 71 (24)  
 Black, non-Hispanic 2242 (47.7) 936 (42.1) 1161 (53.2) 145 (49)  
 Hispanic 1284 (27.3) 624 (28) 591 (27.1) 69 (23.3)  
 Other 184 (3.9) 107 (4.8) 66 (3) 11 (3.7)  
Education     <.001a 
 Less than high school 1615 (34.3) 635 (28.5) 828 (37.9) 152 (51.5)  
 High school or equivalent 1433 (30.4) 655 (29.4) 699 (32) 79 (26.8)  
 Some college or technical education 1153 (24.5) 549 (24.6) 547 (25) 57 (19.3)  
 College degree or higher 508 (10.8) 390 (17.5) 111 (5.1) 7 (2.4)  
Household income     <.001a 
 0%–49% FPL 883 (18.7) 336 (15.1) 474 (21.7) 73 (24.7)  
 50%–99% FPL 810 (17.2) 339 (15.2) 407 (18.6) 64 (21.6)  
 100%–199% FPL 1220 (25.9) 520 (23.3) 609 (27.8) 91 (30.7)  
 200%–299% FPL 731 (15.5) 324 (14.5) 363 (16.6) 44 (14.9)  
 300%+ FPL 1070 (22.7) 711 (31.9) 335 (15.3) 24 (8.1)  
Parental relationship     <.001a 
 Married 1149 (24.4) 772 (34.6) 350 (16) 27 (9.1)  
 Cohabitating 1717 (36.4) 704 (31.6) 885 (40.5) 128 (43.2)  
 Visiting 1229 (26.1 507 (22.8) 639 (29.2) 83 (28)  
 Friends 286 (6.1%) 120 (5.4) 142 (6.5) 24 (8.1)  
 No/limited contact 332 (7) 126 (5.6) 172 (7.9) 34 (11.5)  
People living in household     <.001a 
 <3 1355 (32.2) 589 (30) 674 (34) 92 (34.3)  
 4–5 2033 (48.3) 1029 (52.5) 894 (45.1) 110 (41)  
 >6 822 (19.5) 342 (17.5) 414 (20.9) 66 (24.6)  
US born 3928 (83.5) 1772 (79.6) 1884 (86.4) 272 (91.9) <.001a 
Focal child characteristics      
Biological sex     .94 
 Male 2459 (52.2) 1159 (52) 1147 (52.4) 153 (51.7)  
 Female 2255 (47.8) 1071 (48) 1041 (47.6) 143 (48.3)  
First-born 1802 (38.4) 865 (38.9) 834 (38.2) 103 (34.8) .39 
VariableSampleSecureModerately InsecureHighly InsecureP
n = 4714n = 2230n = 2188n = 296
Parent characteristicsn (%)n (%)n (%)n (%)
Birthing parent’s age, y; mean (SD) 25.3 (6) 26.4 (6.3) 24.3 (5.6) 23.6 (5.46) <.001a 
Race/ethnicity     <.001a 
 White, non-Hispanic 993 (21.1) 558 (25.1) 364 (16.7) 71 (24)  
 Black, non-Hispanic 2242 (47.7) 936 (42.1) 1161 (53.2) 145 (49)  
 Hispanic 1284 (27.3) 624 (28) 591 (27.1) 69 (23.3)  
 Other 184 (3.9) 107 (4.8) 66 (3) 11 (3.7)  
Education     <.001a 
 Less than high school 1615 (34.3) 635 (28.5) 828 (37.9) 152 (51.5)  
 High school or equivalent 1433 (30.4) 655 (29.4) 699 (32) 79 (26.8)  
 Some college or technical education 1153 (24.5) 549 (24.6) 547 (25) 57 (19.3)  
 College degree or higher 508 (10.8) 390 (17.5) 111 (5.1) 7 (2.4)  
Household income     <.001a 
 0%–49% FPL 883 (18.7) 336 (15.1) 474 (21.7) 73 (24.7)  
 50%–99% FPL 810 (17.2) 339 (15.2) 407 (18.6) 64 (21.6)  
 100%–199% FPL 1220 (25.9) 520 (23.3) 609 (27.8) 91 (30.7)  
 200%–299% FPL 731 (15.5) 324 (14.5) 363 (16.6) 44 (14.9)  
 300%+ FPL 1070 (22.7) 711 (31.9) 335 (15.3) 24 (8.1)  
Parental relationship     <.001a 
 Married 1149 (24.4) 772 (34.6) 350 (16) 27 (9.1)  
 Cohabitating 1717 (36.4) 704 (31.6) 885 (40.5) 128 (43.2)  
 Visiting 1229 (26.1 507 (22.8) 639 (29.2) 83 (28)  
 Friends 286 (6.1%) 120 (5.4) 142 (6.5) 24 (8.1)  
 No/limited contact 332 (7) 126 (5.6) 172 (7.9) 34 (11.5)  
People living in household     <.001a 
 <3 1355 (32.2) 589 (30) 674 (34) 92 (34.3)  
 4–5 2033 (48.3) 1029 (52.5) 894 (45.1) 110 (41)  
 >6 822 (19.5) 342 (17.5) 414 (20.9) 66 (24.6)  
US born 3928 (83.5) 1772 (79.6) 1884 (86.4) 272 (91.9) <.001a 
Focal child characteristics      
Biological sex     .94 
 Male 2459 (52.2) 1159 (52) 1147 (52.4) 153 (51.7)  
 Female 2255 (47.8) 1071 (48) 1041 (47.6) 143 (48.3)  
First-born 1802 (38.4) 865 (38.9) 834 (38.2) 103 (34.8) .39 

FPL, federal poverty level.

a

P < .05.

None of the children categorized into the secure group experienced any indicators of housing insecurity during the study. Table 2 and Fig 2 show frequency of housing insecurity indicators and prevalence of each individual indicator over time, respectively.

TABLE 2

Frequency of Housing Insecurity Indicators Experienced at Each Timepoint for the Moderately and Highly Insecure Groups

Indicators reported in the moderately insecure group
Timepoint, yAge 1Age 3Age 5Age 9Age 15
n = 2039n = 1986n = 1961n = 1744n = 1421
 None 1057 (51.8%) 1313 (66.1%) 1267 (64.6%) 980 (56.2%) 1108 (78%) 
 1 indicator 758 (37.2%) 592 (29.8%) 592 (30.2%) 622 (35.7%) 280 (19.7%) 
 2 indicators 177 (8.7%) 78 (3.9%) 85 (4.3%) 100 (5.7%) 27 (1.9%) 
 3 or more indicators 47 (2.3%) 3 (0.2%) 17 (0.9%) 42 (2.4%) 6 (0.4%) 
Indicators reported in the highly insecure group 
Timepoint, y Age 1 Age 3 Age 5 Age 9 Age 15 
 n = 267 n = 271 n = 258 n = 226 n = 154 
 None 37 (13.9%) 22 (8.1%) 42 (16.3%) 78 (34.5%) 105 (68.2%) 
 1 indicator 69 (25.8%) 67 (24.7%) 81 (31.4%) 88 (38.9%) 36 (23.4%) 
 2 indicators 84 (31.5%) 104 (38.4%) 63 (42.4%) 35 (15.5%) 8 (5.2%) 
 3 or more indicators 77 (28.8%) 78 (28.8%) 72 (27.9%) 25 (11.1%) 5 (3.2%) 
Indicators reported in the moderately insecure group
Timepoint, yAge 1Age 3Age 5Age 9Age 15
n = 2039n = 1986n = 1961n = 1744n = 1421
 None 1057 (51.8%) 1313 (66.1%) 1267 (64.6%) 980 (56.2%) 1108 (78%) 
 1 indicator 758 (37.2%) 592 (29.8%) 592 (30.2%) 622 (35.7%) 280 (19.7%) 
 2 indicators 177 (8.7%) 78 (3.9%) 85 (4.3%) 100 (5.7%) 27 (1.9%) 
 3 or more indicators 47 (2.3%) 3 (0.2%) 17 (0.9%) 42 (2.4%) 6 (0.4%) 
Indicators reported in the highly insecure group 
Timepoint, y Age 1 Age 3 Age 5 Age 9 Age 15 
 n = 267 n = 271 n = 258 n = 226 n = 154 
 None 37 (13.9%) 22 (8.1%) 42 (16.3%) 78 (34.5%) 105 (68.2%) 
 1 indicator 69 (25.8%) 67 (24.7%) 81 (31.4%) 88 (38.9%) 36 (23.4%) 
 2 indicators 84 (31.5%) 104 (38.4%) 63 (42.4%) 35 (15.5%) 8 (5.2%) 
 3 or more indicators 77 (28.8%) 78 (28.8%) 72 (27.9%) 25 (11.1%) 5 (3.2%) 
FIGURE 2

Prevalence of individual housing insecurity indicators experienced from ages 1 to 15 years for the (A) moderately insecure and (B) highly insecure groups.

FIGURE 2

Prevalence of individual housing insecurity indicators experienced from ages 1 to 15 years for the (A) moderately insecure and (B) highly insecure groups.

Close modal

Most children in the moderately insecure group experienced 1 or 0 housing insecurity indicators at each timepoint. More than half in the highly insecure group experienced 2 or more housing insecurity indicators until age 5 years. Housing insecurity was most prevalent at ages 1 and 3 years, respectively, for the moderately and highly insecure groups (Table 2). The most common indicators for both insecure groups were skipping rent and doubling up (Fig 2). These indicators peaked at age 9 years for the moderately insecure group. In the highly insecure group, most indicators began to decrease in prevalence at ages 3 or 5 years. At age 15 years, more than two-thirds (68.2%) of the highly insecure group did not experience any indicators (Table 2).

Approximately 73% (3437/4714) of adolescents had at least 1 outcome measure available. Unadjusted and adjusted comparisons of outcomes at age 15 years between the housing groups can be found in Table 3. The housing groups had significantly different SRH, and depressive and anxiety symptom scores at age 15 years.

TABLE 3

Unadjusted and Adjusted Analyses of Outcomes at Age 15 y Based on Housing Group

OutcomenSecureModerately InsecureHighly InsecureP (Unadjusted)
Excellent or very good self-reported health 
 Unadjusted, n (%) 3434 1120 (74.4%) 1195 (70%) 146 (66.1%) .003a 
 Adjusted OR [95% CI] 3427 Ref. 0.81 [0.69–0.95]a 0.67 [0.50–0.92]a  
Depression score 
 Unadjusted mean (SD) 3383 7.7 (2.9) 8.1 (3) 8.8 (3.3) <.001a 
 Adjusted IRR [95% CI] 3376 Ref. 1.05 [1.03–1.08]a 1.14 [1.08–1.20]a  
Anxiety score 
 Unadjusted, mean (SD) 2250 10.5 (3.3) 10.8 (3.2) 11.2 (3.2) .03a 
 Adjusted IRR [95% CI] 2248 Ref. 1.02 [0.99–1.05] 1.06 [1.01–1.11]a  
OutcomenSecureModerately InsecureHighly InsecureP (Unadjusted)
Excellent or very good self-reported health 
 Unadjusted, n (%) 3434 1120 (74.4%) 1195 (70%) 146 (66.1%) .003a 
 Adjusted OR [95% CI] 3427 Ref. 0.81 [0.69–0.95]a 0.67 [0.50–0.92]a  
Depression score 
 Unadjusted mean (SD) 3383 7.7 (2.9) 8.1 (3) 8.8 (3.3) <.001a 
 Adjusted IRR [95% CI] 3376 Ref. 1.05 [1.03–1.08]a 1.14 [1.08–1.20]a  
Anxiety score 
 Unadjusted, mean (SD) 2250 10.5 (3.3) 10.8 (3.2) 11.2 (3.2) .03a 
 Adjusted IRR [95% CI] 2248 Ref. 1.02 [0.99–1.05] 1.06 [1.01–1.11]a  

Adjusted analyses controlled for baseline household income, parent racial/ethnic identity, and child's biological sex. CI, confidence interval; IRR, incidence rate ratio; OR: odds ratio; Ref., reference group.

a

P < .05.

More adolescents in the secure group reported “very good” or “excellent” SRH than those in the insecure groups. Adolescents in the moderately and highly insecure groups had a 19% (95% confidence interval [CI], 0.69–0.95) and 33% (95% CI, 0.5–0.91) lower odds, respectively, of reporting “very good” or “excellent” SRH than those in the secure housing group (Table 3).

Adolescents in the highly insecure housing group reported more depressive symptoms on average than their secure and moderately insecure counterparts. Adolescents in the moderately and highly insecure groups reported depressive symptoms at rates of 1.05 (95% CI, 1.03–1.08) and 1.1 (95% CI, 1.08–1.2) times higher, respectively, than adolescents in the secure group (Table 3).

Of the 3 housing groups, adolescents in the secure housing group reported the lowest anxiety symptoms. Individuals in the highly insecure group reported anxiety symptoms at a rate 1.06 times higher than the secure group (95% CI, 1–1.1) (Table 3). Anxiety symptoms reported in the moderately insecure group did not significantly differ from the secure group.

In a sample of US children with socioeconomic disadvantage followed from infancy to adolescence, we identified 3 distinct longitudinal experiences of housing insecurity: secure, moderately insecure, and highly insecure. The rate of housing insecurity (52.7%) in the FFCWB sample, which we selected to examine long-term health effects of specific material hardships, reflects a similar rate reported in another study of low-income households with children.49  Baseline educational attainment, marital status, and household income were associated with more secure trajectories. At age 15 years, adolescents in the secure group had better SRH, and lower anxiety and depressive symptoms compared with those who experienced insecure housing. These findings contribute knowledge about patterns of housing insecurity throughout childhood and their associations with adolescent health.

Our study is among the first to identify trajectories of housing insecurity in a sample of children. Although similar to the 4 trajectory models of housing insecurity identified by Kang and Aubry, our model did not contain trajectories of sustained or increasing insecurity, reinforcing evidence that households with children were less likely to experience chronically unstable housing.21,22  The absence of these “mostly unstable” experiences for FFCWB participants could be due to increased access to safety net programs for households with children. Parents may also describe housing insecurity differently than individuals without children. Future research may employ qualitative methods to understand the lived experiences of families with children who have fluctuating housing insecurity and identify the mechanisms that protect them from more severe experiences of insecurity.

Although our trajectories are not an exact representation of each individual child’s experience, estimating them generated novel information about childhood housing insecurity and allowed us to identify priorities for intervention and future research. In line with life course research, even though children in both insecure groups experienced periods of security, those periods did not diminish the cumulative effects of insecure housing on adolescent health. Because the highest proportions of children in the insecure groups experienced housing insecurity in the first 3 years of life, early childhood may represent a vulnerable period to screen and help families experiencing housing hardships. The reduction of insecurity indicators for the highly insecure group over time, though encouraging, may be partially attributed to loss of follow-up in the FFCWB study, particularly from those experiencing housing insecurity. Children with the highest degree of insecurity may fall through the cracks because of their ever-changing housing situations. Children experiencing fluctuating housing insecurity may also be harder to identify, especially because there are not yet validated screening tools for clinicians to use in practice.50  Our results highlight the potential long-term consequences if current public health measures do not identify children experiencing housing insecurity and underscore the need for innovative screening efforts beyond just health systems.

Our work extends current literature showing that deleterious effects of housing insecurity may begin as early as the first year of life. Studies connecting experiences of housing insecurity during various periods of child development to health outcomes later in life have had inconsistent results, which may be due to the child’s age at the time of the analysis and who is reporting the information (eg, child versus parent).9,13,17–20  We demonstrated that prolonged experiences of housing insecurity spanning the first 15 years of life, regardless of severity, were predictive of poorer SRH and more symptoms of depression in adolescence. Adolescents in the highly insecure group had higher anxiety symptoms, despite the gradual improvement of this trajectory after age 3 years. Therefore, highly insecure housing early in life may have lasting effects regardless of experiences later in childhood, further reinforcing our recommendation for interventions as early as possible. Future studies should examine child psychiatric symptoms longitudinally with housing experiences to clarify the role of housing in mental health across developmental stages.

Though socioeconomic characteristics were significantly different among the housing groups, the differences were not distinct enough to provide any “contextually significant” information about risk factors for housing insecurity. For example, many households in both insecure housing groups had incomes that were well above the FPL, strengthening previous evidence that financial hardship alone is not a reliable indicator of material hardship.51  Controlling for racial and ethnic identity in our adjusted models showed that housing disparities likely had additional health effects exacerbating the well-documented health disparities experienced by racial and ethnic minority groups. Consequently, we contextualized housing insecurity as a social determinant of health existing independently and comorbidly with other drivers of health disparities. Risk factors for housing insecurity are likely rooted in larger systemic issues (eg, systemic racism, discrimination of female-headed households), and future studies should consider this broader context.45,52,53 

Though the children in FFCWB were born >20 years ago, the context is the same or worse for families with children today; the national rate of 1-parent households has increased, the cost of living is higher, and individuals’ buying power has not caught up.54,55  The expiration of pandemic-era federal assistance programs that may have lessened families’ growing financial burden, and the twofold increase in the supplemental child poverty measure from 2021 to 2022 leaves more families in need, fewer resources for providers to offer, and increased urgency for solutions.56,57 

Our study has several limitations to consider when interpreting the results. Our composite measure of housing insecurity is imperfect because of the dichotomization of each individual indicator; a child who spent 1 night or several months in a place not meant for residence were counted equally. Housing hardships are likely underrepresented because housing insecurity may limit participant ability to participate in long-term research, biasing the FFWCB sample toward participants with secure housing. Moreover, indicators and outcomes could have been underreported because of social desirability bias or mistrust of research entities from historical mistreatment of minoritized racial and ethnic groups. Alternatively, because these data come from individuals from marginalized racial and ethnic groups and with lower incomes who are already disadvantaged, they may overestimate housing insecurity rates as they exist in the general population.

Housing insecurity is a complex, systemic issue with unique implications for children. Examining housing insecurity over years helped detect relationships with adolescent health outcomes not observed in previous cross-sectional analyses. Our findings extend evidence of the cumulative nature of housing insecurity, therefore emphasizing the importance of early intervention for families with children experiencing housing insecurity.

Housing insecurity is preventable and addressable through policy and public health intervention. Future work is needed to validate a universal measure for housing insecurity and implement screening and referral procedures for families with young children to appropriate services. In addition to traditional settings for screening, health systems would benefit from partnering with public health professionals to screen for housing insecurity in community spaces (eg, churches, recreational centers), where they may be able to engage parents of children living in overcrowded or precarious housing. Screening is only half of the battle, and providers need more resources to offer once families experiencing housing insecurity are identified. Policy efforts should prioritize affordable housing accessibility and financial support programs for families with children since skipping rent and doubling up were the most common indicators for the insecure groups. Stakeholders should seek to understand how pandemic-era financial assistance programs succeeded in diminishing child poverty and how they can be adapted into permanent protections for families with young children.

Ms Pierce and Dr Duh-Leong conceptualized and designed the study, carried out the analyses, reviewed the analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Mendelsohn substantially contributed to the conception of the study and analyses, reviewed the analyses, and reviewed and revised the manuscript; Drs Smith and Johnson substantially contributed to the conception of the study and analyses and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Dr Duh-Leong acknowledges support from National Institutes of Health/NIEHS (K23ES035461) and the Sala Elbaum Pediatric Research Scholars Program.

CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no potential conflicts of interest to disclose.

FFCWB

Future of Families and Child Wellbeing Study

FPL

federal poverty level

GBTM

group-based trajectory modeling

SRH

self-reported health

1
Swope
CB
,
Hernández
D
.
Housing as a determinant of health equity: a conceptual model
.
Soc Sci Med
.
2019
;
243
:
112571
2
Bess
KD
,
Miller
AL
,
Mehdipanah
R
.
The effects of housing insecurity on children’s health: a scoping review
.
Health Promot Int
.
2023
;
38
(
3
):
daac006
3
Rolfe
S
,
Garnham
L
,
Godwin
J
,
Anderson
I
,
Seaman
P
,
Donaldson
C
.
Housing as a social determinant of health and wellbeing: developing an empirically-informed realist theoretical framework
.
BMC Public Health
.
2020
;
20
(
1
):
1138
4
Karjalainen
PT
.
House, home and the place of dwelling
.
Scand Housing and Planning Res
.
2007
;
10
(
2
):
65
74
5
Green
KA
,
Bovell-Ammon
A
,
Sandel
M
.
Housing and neighborhoods as root causes of child poverty
.
Acad Pediatr
.
2021
;
21
(
8S
):
S194
S199
6
Sandel
M
,
Sheward
R
,
Ettinger de Cuba
S
, et al
.
Unstable housing and caregiver and child health in renter families
.
Pediatrics
.
2018
;
141
(
2
):
e20172199
7
Martin
P
,
Liaw
W
,
Bazemore
A
,
Jetty
A
,
Petterson
S
,
Kushel
M
.
Adults with housing insecurity have worse access to primary and preventive care
.
J Am Board Fam Med
.
2019
;
32
(
4
):
521
530
8
Carroll
A
,
Corman
H
,
Curtis
MA
,
Noonan
K
,
Reichman
NE
.
Housing instability and children’s health insurance gaps
.
Acad Pediatr
.
2017
;
17
(
7
):
732
738
9
Marçal
KE
,
Maguire-Jack
K
.
Housing insecurity and adolescent well-being: relationships with child welfare and criminal justice involvement
.
Child Abuse Negl
.
2021
;
115
:
105009
10
Covington
LB
,
Rogers
VE
,
Armstrong
B
,
Storr
CL
,
Black
MM
.
Toddler bedtime routines and associations with nighttime sleep duration and maternal and household factors
.
J Clin Sleep Med
.
2019
;
15
(
6
):
865
871
11
Ziol-Guest
KM
,
McKenna
CC
.
Early childhood housing instability and school readiness
.
Child Dev
.
2014
;
85
(
1
):
103
113
12
Coley
RL
,
Leventhal
T
,
Lynch
AD
,
Kull
M
.
Relations between housing characteristics and the well-being of low-income children and adolescents
.
Dev Psychol
.
2013
;
49
(
9
):
1775
1789
13
Hatem
C
,
Lee
CY
,
Zhao
X
,
Reesor-Oyer
L
,
Lopez
T
,
Hernandez
DC
.
Food insecurity and housing instability during early childhood as predictors of adolescent mental health
.
J Fam Psychol
.
2020
;
34
(
6
):
721
730
14
Sandel
M
,
Sheward
R
,
Ettinger de Cuba
S
, et al
.
Timing and duration of pre- and postnatal homelessness and the health of young children
.
Pediatrics
.
2018
;
142
(
4
):
e20174254
15
Russ
SA
,
Hotez
E
,
Berghaus
M
, et al
.
Building a life course intervention research framework
.
Pediatrics
.
2022
;
149
(
suppl 5
):
e2021053509E
16
Frederick
TJ
,
Chwalek
M
,
Hughes
J
,
Karabanow
J
,
Kidd
S
.
How stable is stable? Defining and measuring housing stability
.
J Community Psychol
.
2014
;
42
(
8
):
964
979
17
Keen
R
,
Chen
JT
,
Slopen
N
,
Sandel
M
,
Copeland
WE
,
Tiemeier
H
.
Prospective associations of childhood housing insecurity with anxiety and depression symptoms during childhood and adulthood
.
JAMA Pediatr
.
2023
;
177
(
8
):
818
826
18
Fowler
PJ
,
Henry
DB
,
Marcal
KE
.
Family and housing instability: longitudinal impact on adolescent emotional and behavioral well-being
.
Soc Sci Res
.
2015
;
53
:
364
374
19
Crumé
HJ
,
Nurius
PS
,
Fleming
CM
.
Cumulative adversity profiles among youth experiencing housing and parental care instability
.
Child Youth Serv Rev
.
2019
;
100
:
129
135
20
Gilman
SE
,
Kawachi
I
,
Fitzmaurice
GM
,
Buka
L
.
Socio-economic status, family disruption and residential stability in childhood: relation to onset, recurrence and remission of major depression
.
Psychol Med
.
2003
;
33
(
8
):
1341
1355
21
Kang
S
.
Severe and persistent housing instability: examining low-income households’ residential mobility trajectories in the United States
. Available at: https://www.tandfonline.com/doi/full/10.1080/02673037.2021.1982871. Accessed May 24, 2024
22
Aubry
T
,
Agha
A
,
Mejia-Lancheros
C
, et al
.
Housing trajectories, risk factors, and resources among individuals who are homeless or precariously housed
.
Ann Am Acad Pol Soc Sci
.
2021
;
693
(
1
):
102
122
23
Geller
A
,
Curtis
MA
.
A longitudinal examination of housing hardships among urban fathers
.
J Marriage Fam
.
2018
;
80
(
5
):
1176
1186
24
Reichman
NE
,
Teitler
JO
,
Garfinkel
I
,
McLanahan
SS
.
Fragile families: sample and design
.
Child Youth Serv Rev
.
2001
;
23
(
4-5
):
303
326
25
James
S
,
McLanahan
S
,
Brooks-Gunn
J
.
Contributions of the Fragile Families and Child Wellbeing Study to child development
.
Annu Rev Dev Psychol
.
2021
;
3
(
1
):
187
206
26
Geller
A
,
Franklin
AW
.
Paternal incarceration and the housing security of urban mothers
.
J Marriage Fam
.
2014
;
76
(
2
):
411
427
27
Ware
JE
 Jr
,
Sherbourne
CD
.
The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection
.
Med Care
.
1992
;
30
(
6
):
473
483
28
Schnittker
J
,
Bacak
V
.
The increasing predictive validity of self-rated health
.
PLoS One
.
2014
;
9
(
1
):
e84933
29
Idler
EL
,
Benyamini
Y
.
Self-rated health and mortality: a review of twenty-seven community studies
.
J Health Soc Behav
.
1997
;
38
(
1
):
21
37
30
Johnson
SB
,
Wang
C
.
Why do adolescents say they are less healthy than their parents think they are? The importance of mental health varies by social class in a nationally representative sample
.
Pediatrics
.
2008
;
121
(
2
):
e307
e313
31
Manor
O
,
Matthews
S
,
Power
C
.
Dichotomous or categorical response? Analysing self-rated health and lifetime social class
.
Int J Epidemiol
.
2000
;
29
(
1
):
149
157
32
Radloff
LS
.
The CES-D Scale: a self-report depression scale for research in the general population
.
Appl Psychol Meas
.
1977
;
1
(
3
):
385
401
33
Perreira
KM
,
Deeb-Sossa
N
,
Harris
KM
,
Bollen
K
.
What are we measuring? An evaluation of the CES-D across race/ethnicity and immigrant generation
.
Soc Forces
.
2005
;
83
(
4
):
1567
1601
34
Derogatis
LR
,
Savitz
KL
. The SCL–90–R and Brief Symptom Inventory (BSI) in primary care. In:
Maruish
ME
, editor.
Handbook of Psychological Assessment in Primary Care Settings
.
Lawrence Erlbaum Associates Publishers
;
2000
:
297
334
35
Little
TD
,
Jorgensen
TD
,
Lang
KM
,
Moore
EWG
.
On the joys of missing data
.
J Pediatr Psychol
.
2014
;
39
(
2
):
151
162
36
Nagin
DS
.
Group-Based Modeling of Development
.
Harvard University Press
;
2005
37
Nagin
DS
.
Analyzing developmental trajectories: a semiparametric, group-based approach
.
Psychol Methods
.
1999
;
4
(
2
):
139
157
38
Nagin
DS
.
Group-based trajectory modeling: an overview
.
Ann Nutr Metab
.
2014
;
65
(
2-3
):
205
210
39
Nagin
DS
,
Odgers
CL
.
Group-based trajectory modeling in clinical research
.
Annu Rev Clin Psychol
.
2010
;
6
:
109
138
40
Ray
JV
,
Sullivan
CJ
,
Loughran
TA
,
Jones
SE
.
The impact of missing risk factor data on semiparametric group-based trajectory models
.
J Dev Life Course Criminol
.
2018
;
4
(
3
):
276
296
41
Cook
SH
,
Wood
EP
,
Scott
M
,
Pierce
KA
,
Kapadia
F
,
Halkitis
PN
.
Predicting trajectories of substance use during emerging adulthood: exploring the benefits of group-based trajectory modeling for zero-inflated outcomes
.
TPM Test Psychom Methodol Appl Psychol
.
2020
;
27
(
3
):
291
311
42
Gajos
JM
,
Miller
CR
,
Leban
L
,
Cropsey
KL
.
Adverse childhood experiences and adolescent mental health: understanding the roles of gender and teenage risk and protective factors
.
J Affect Disord
.
2022
;
314
:
303
308
43
Cameron
AC
,
Trivedi
PK
.
Regression Analysis of Count Data
.
Cambridge University Press
;
1998
44
Phelan
JC
,
Link
BG
,
Tehranifar
P
.
Social conditions as fundamental causes of health inequalities: theory, evidence, and policy implications
.
J Health Soc Behav
.
2010
;
51
(
1
suppl
):
S28
S40
45
Campbell
OLK
,
Bann
D
,
Patalay
P
.
The gender gap in adolescent mental health: a cross-national investigation of 566,829 adolescents across 73 countries
.
SSM Popul Health
.
2021
;
13
:
100742
46
Mauvais-Jarvis
F
,
Bairey Merz
N
,
Barnes
PJ
, et al
.
Sex and gender: modifiers of health, disease, and medicine
.
Lancet
.
2020
;
396
(
10250
):
565
582
47
StataCorp
.
Stata Statistical Software: Release 17
48
Jones
BL
,
Nagin
DS
.
A Stata plugin for estimating group-based trajectory models
. Available at: https://www.andrew.cmu.edu/user/bjones/refpdf/ref3.pdf. Accessed May 24, 2024
49
Cutts
DB
,
Meyers
AF
,
Black
MM
, et al
.
US housing insecurity and the health of very young children
.
Am J Public Health
.
2011
;
101
(
8
):
1508
1514
50
Chung
EK
,
Siegel
BS
,
Garg
A
, et al
.
Screening for social determinants of health among children and families living in poverty: a guide for clinicians
.
Curr Probl Pediatr Adolesc Health Care
.
2016
;
46
(
5
):
135
153
51
Schenck-Fontaine
A
,
Ryan
RM
.
Poverty, material hardship, and children’s outcomes: a nuanced understanding of material hardship in childhood
.
Children (Basel)
.
2022
;
9
(
7
):
981
52
Desmond
M
.
Eviction and the reproduction of urban poverty
.
Am J Sociol
.
2012
;
118
:
88
133
53
Cornelissen
S
,
Hermann
A
.
COVID-19 and racial inequalities in housing: pre-pandemic and pandemic pathways to housing insecurity
.
J Urban Aff
.
2023
;
1
23
54
United States Census Bureau
.
National single parent day: March 21, 2023
. Available at: https://www.census.gov/newsroom/stories/single-parent-day.html. Accessed January 21, 2024
55
Bailey
P
.
Addressing the affordable housing crisis requires expanding rental assistance and adding housing units
. Available at: https://www.cbpp.org/research/housing/addressing-the-affordable-housing-crisis-requires-expanding-rental-assistance-and#_ftn1. Accessed January 21, 2024
56
Shrider
EA
,
Creamer
J
.
Poverty in the United States: 2022, current population reports
. Available at: https://www.census.gov/content/dam/Census/library/publications/2023/demo/p60-280.pdf. Accessed September 11, 2023
57
U.S. Department of the Treasury
.
Assistance for American families and workers
. Available at: https://home.treasury.gov/policy-issues/coronavirus/assistance-for-American-families-and-workers. Accessed September 11, 2023

Supplementary data