Homelessness and Health Care Access After Emancipation: Title and subTitle BreakResults From the Midwest Evaluation of Adult Functioning of Former Foster Youth FREE

Approximately 18% of the 550 000 children in the foster care system are between the ages of 16 and 18 years¹ and thus approaching the age of emancipation from the foster care system, which occurs between ages 18 and 21 years, depending on state policy.² Roughly 20 000 youth “age out” or are “emancipated” from the foster care system annually without being reunited withfamilies.³ These youth are at high risk of homelessness^{4 - 6} and have poor health outcomes,^{7 - 11} including high rates of drug and alcohol use,⁹ unplanned pregnancies,^{11 - 12} and poor mental health outcomes.^{9 - 10}

Foster care is a risk factor for adult homelessness^{4 ,13 - 18} ; an estimated 12% of all homeless adults were in the foster care system as youth.¹⁹ The association between the foster care system and adult homelessness is particularly stark among young adults. The last national survey of homeless adults found that 2.3 to 3.5 million Americans have an episode of homelessness annually, 5% of whom were 18 to 20 years old. Approximately 40% of homeless adults aged 18 to 20 years were in the foster care system as youth.¹⁹ Among homeless persons, those with a history of foster care are more likely to have longer spells of homelessness, mental illness and substance abuse, and children in foster care than other homeless adults.¹³

Homelessness, both among adolescents and adults, is associated with poor health outcomes. Homeless adolescents have higher rates of substance abuse,^{20 - 22} high-risk sexual behavior,²³ victimization,²⁴ unplanned pregnancy,²⁵ and mental health^{22 ,26} and health problems than housed adolescents.²⁷ They have poor access to ambulatory care and higher rates of emergency department use.²⁸ Homeless adults have high rates of morbidity^{29 - 31} and mortality,^{32 - 35} high rates of acute health care service use,^{36 - 39} and low rates of preventive care.^{36 ,40 - 42} Housing instability, including paying more than 50% of income on housing, frequent moves, or difficulty paying rent or mortgage, is also associated with poor access to ambulatory care and high rates of acute care use.⁴³

We examined a prospective cohort of older adolescents who were aging out of the foster care system to examine both the incidence of homelessness in the period immediately following emancipation and whether homelessness is associated with poor health care access and outcomes. We hypothesized that young adults who were recently emancipated from the foster care system are at high risk of homelessness and housing instability and that among emancipated young adults, those who experienced either homelessness or housing instability would have worse access to health care and health outcomes than those with stable housing.

We conducted an analysis of factors associated with access to health care, use of ambulatory and inpatient hospital services, and health outcomes among participants in the Midwest Evaluation of the Adult Functioning of Former Foster Youth data set (MEAF).⁴⁴ The MEAF study is a longitudinal study of foster youth as they age out of the child welfare system and transition to adulthood in 3 Midwestern states (Illinois, Iowa, and Wisconsin). The MEAF study is conducted by the Chapin Hall Center for Children at the University of Chicago, in collaboration with the University of Wisconsin Survey Center and the public child welfare agencies of Illinois, Iowa, and Wisconsin.⁴⁴

The child welfare agencies supplied researchers with the names of eligible foster youth. All participants were foster youth who were 17 or 18 years of age, had been in foster care for at least 1 year prior to turning 17 years of age, and were placed in foster care because of abuse or neglect or because they were adjudicated delinquents. Youth were excluded if they had developmental delay or severe mental illness, were incarcerated or in a psychiatric hospital, had an out-of-state placement, or had run away or were missing during the data follow-up period. In Iowa and Wisconsin, the survey included all eligible youth; in Illinois, the survey included a random sample of 67% of the youth.

Researchers at the University of Wisconsin Survey Center administered the questionnaires in person. Researchers conducted baseline interviews with 749 youth (response rate, 94.7%) between May 2002 and March 2003. Follow-up interviews took place between March and December 2004; 643 youth (85.8% follow-up rate) participated, of whom 290 (45.1%) were still in the foster care system and 345 (53.7%) had been emancipated. Eight subjects who were missing housing status at follow-up were excluded from analysis, leaving 635 subjects. For analyses of health care access, we only included the 345 subjects who had been emancipated from the foster care system at the time of follow-up. We used deidentified data to conduct this study. All participants provided signed informed consent prior to each interview. Participants were given a cash payment of $25 for completing the baseline interview and $30 for completing the follow-up interview.

For emancipated participants, we categorized housing in 3 categories: homeless, unstably housed, and stably housed. We categorized participants who had been emancipated from the child welfare system as having had an episode of homelessness if they stated that they had been homeless (defined as, for 1 night or longer, they slept in a place where “people weren't meant to sleep or slept in a homeless shelter or didn't have a regular residence in which to sleep”) at any point since being emancipated; as unstably housed if they reported moving 3 or more times or they reported spending more than 50% of their income on rent since emancipation; and stably housed if they reported moving fewer than 3 times and spent less than 50% of their income on rent since emancipation. All still-in-system participants were assumed to be stably housed.

Demographic variables included sex, race, state, emancipation status, and education level. Based on participant self-report, we classified participants as white, black, or other. We categorized state as Illinois, Iowa, or Wisconsin. For the analyses of health outcomes, because the emancipated participants almost exclusively came from 1 state (Illinois), we categorized participants from Illinois as still in system or emancipated. We categorized educational attainment as high school graduate (high school diploma or GED [General Education Development]) vs not a high school graduate.

Participants rated their health at baseline as excellent, very good, good, fair, or poor. We dichotomized the report of health status to fair/poor vs good/very good/excellent.

We assessed the mental health of participants using the 1998 World Mental Health Composite International Diagnostic Interview⁴⁵ (CIDI). The CIDI is a highly structured interview that generates psychiatric diagnoses according to criteria listed in the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition)⁴⁶ (DSM-IV). We classified participants who met criteria for lifetime alcohol abuse or dependence on the CIDI scale at follow-up as having a drinking problem. We classified participants who met criteria for lifetime substance abuse or dependence at follow-up as having a drug problem. We classified participants as having (lifetime) depression or posttraumatic stress disorder (PTSD) if they met criteria for these diagnoses.

Our dependent variables included 3 measures of access to care: (1) being uninsured, (2) unmet need for medical care, and (3) no ambulatory care visits in the prior year; and 2 health outcomes at follow-up: (1) fair or poor health status and (2) pregnancy in the follow-up period.

We classified participants as being uninsured if they responded no when asked if they had health insurance at the time of the follow-up interview. We classified participants as having unmet need for medical care if they reported an inability to get needed medical care between the baseline and follow-up interviews. We dichotomized ambulatory care use (last physician or nurse visit > 1 year ago or ≤ 1 year ago) based on self-reported length of time since last ambulatory health care visit during the follow-up interview.

Participants rated their health as excellent, very good, good, fair, or poor at the time of follow-up. We classified those participants who reported fair or poor health status at follow-up as having fair or poor health status and used this as a dependent variable. We classified female participants as having a pregnancy if they reported having been pregnant at any time since the baseline interview.

We used Pearson χ² to test for differences in proportions between groups in Table 1. We used binary logistic regression to test for bivariate associations with the dependent variables. For the analyses of health care access, we excluded the still-in-system participants from the analysis. We constructed stepwise multivariable models using multivariable logistic regression. We used binary logistic regression for all multivariate analyses. We considered all variables that were associated with the outcome at α < .15 in the bivariate models. We began constructing each model with the 3-level housing status variable and then added, singly and in order, the covariates health status, demographic variables (sex, race, state [including emancipation status for health care outcome analyses], and education), mental health status (drinking problem, drug problem, and depression and PTSD), and insurance status. Covariates remained in the models if they were significantly associated with 1 or more outcomes after adjusting for all the other covariates in the model at that step. Goodness of fit was evaluated using the Hosmer-Lemeshow test. We used SAS version 9.1 (SAS Institute Inc, Cary, North Carolina) for all analyses.

The institutional review boards at the University of California, San Francisco; University of Chicago; and the University of Wisconsin, Madison, approved the study.

Of the 635 study subjects, 47.7% were male, 51.7% were black, and 32.3% were white. More than one-fifth (21.8%) had not graduated from high school. More than half (60.5%) of the participants were from Illinois, 28.8% from Wisconsin, and 10.7% from Iowa.

About half (45.7%) of the participants were still in the foster care system and most (96.6%) of those still in the system were from Illinois. Overall, 49 participants (7.7%) reported an episode of homelessness, 136 (21.4%) were unstably housed, and 160 (25.2%) were stably housed. Forty (6.3%) were adjudicated delinquents. Among those who had been emancipated, 14.2% had experienced homelessness and 39.4% were unstably housed. The mean length of homeless episodes was 27.6 days and the median was 7 days.

At follow-up, 14.8% of participants met criteria for a drinking problem; 13.6%, for a drug abuse problem; 8.5%, for depression; and 13.7%, for PTSD. In bivariate models, problems with both alcohol and drugs were highly associated with housing status, with a decreasing rate of problems with better housing status. Whereas nearly half (46.3%) of homeless participants met criteria for drinking problems, only 23.5% of stably housed, 12.8% of unstably housed, and 7.9% of those still in the system had drinking problems. Similarly, 46.3% of homeless participants met criteria for having drug problems compared with 20.0%, 10.7%, and 7.9% of stably housed, unstably housed, and still-in-system participants, respectively.

Among emancipated participants, there was a progressive worsening of insurance coverage and unmet need for care with worsening housing status, with stable housing associated with the best access (Table 2).

The majority (76.6%) of participants with an episode of homelessness were uninsured, as compared with 53.0% of unstably housed and 46.5% of stably housed emancipated participants. Only 2.5% of still-in-system participants were uninsured. In multivariable analysis of emancipated participants, participants with episodes of homelessness (adjusted odds ratio [AOR], 3.41; 95% confidence interval [CI], 1.52-7.63) and those who were unstably housed (AOR, 1.47; 95% CI, 0.89-2.44) had higher odds of being uninsured than stably housed participants. Being male, being in good health, and having a drinking problem were also associated with being uninsured.

The inability to obtain needed medical care was reported by 40.8% of participants with an episode of homelessness and 24.3% of unstably housed participants, compared with 14.5% of the stably housed participants and 4.8% of those still in care (P < .001). In a multivariable model of the emancipated participants, participants with an episode of homelessness had elevated odds (AOR, 3.26; 95% CI, 1.40-7.56) and unstably housed participants had marginally elevated odds (AOR, 1.74; 95% CI, 0.93-3.26) of reporting unmet need for medical care compared with the stably housed participants. Meeting criteria for PTSD and being uninsured were significantly associated with unmet need for medical care. Living in Illinois was associated with decreased odds of unmet need.

The last reported ambulatory care visit was more than a year ago for 45.8% of participants with an episode of homelessness, 37.5% of unstably housed, and 31.3% of stably housed emancipated participants, compared with only 9.0% of youth still in the foster care system. In multivariable analysis of emancipated participants, housing status was not associated with having had an ambulatory care visit in the prior year. Being uninsured, male, white, in good health at baseline, and from Wisconsin were all associated with increased odds of not having had a visit in the prior year (Table 2).

Overall, 12.1% of participants reported having fair or poor health status and there were no statistically significant differences in health status based on housing status. Meeting criteria for depression, having fair or poor health status at baseline, and not having a high school degree were associated with fair or poor health at follow-up (Table 3).

Of the 289 women in the study, 111 (38.4%) became pregnant between the baseline and follow-up interviews. Half (53.3%) of the homeless participants became pregnant, compared with 43.9% of unstably housed, 45.2% of stably housed, and 31.5% of the still-in-system participants. In multivariable analysis, there was no association between housing status and pregnancy (Table 3). Illinois participants who were emancipated had elevated odds of pregnancy compared with Illinois still-in-system participants. Having insurance at follow-up was associated with lower odds of pregnancy as was meeting criteria for depression.

Young adults emancipated from the foster care system had high rates of homelessness within the 18 months following emancipation. More than half (53.6%) of emancipated youth experienced either homelessness or unstable housing in the follow-up period and 14.2% experienced an episode of homelessness. While it is difficult to estimate what proportion of young adults experience homelessness, it is estimated that only 1% of American adults have an episode of homelessness annually¹⁹ and the lifetime prevalence of homelessness is approximately 7%.⁴⁷

Recently emancipated youth face particular challenges establishing stable housing because most exit the child welfare system without financial resources or family ties to provide assistance. Recognizing these and other challenges that emancipated foster youth face on exiting the child welfare system, Congress enacted the Foster Care Independence Act (FCIA) in 1999, which allocates $140 million annually nationwide for housing stipends and social services (mental health services, life skills, employment preparation) to ease the transition from foster care to adulthood.⁴⁸ However, only an estimated 40% of eligible youth receive FCIA services,⁴⁹ and states can only use up to 30% of FCIA funds to provide housing, leaving many emancipated former foster youth without resources to cope with homelessness and poor health outcomes soon after emancipation.

We found high rates of poor access to health care among all emancipated youth, with an experience of homelessness being associated with significantly higher odds of being uninsured and of having unmet need for health care. While approximately 30% of young adults in the general population report an episode of being uninsured over the course of the past year,⁵⁰ we found that more than half of emancipated participants were uninsured, with rates ranging from 46.5% of the stably housed to 76.6% of the participants with an experience of homelessness. More than one-fifth (22.0%) of emancipated participants reported unmet need for medical care (14.5% of stably housed youth, up to 40.8% of those with homelessness), compared with approximately 12% of young adults in the general population.⁵⁰

In the general population, almost half of young adults receive health care coverage under their parents' employer-sponsored insurance but emancipated youth often do not have that option.⁵¹ While the majority of children in the child welfare system are eligible for Medicaid,⁵² states are not required to provide Medicaid to emancipated youth. The FCIA gives states the option of automatically extending Medicaid to emancipated youth until they are 21 years of age, but only a few states have done so⁴⁹ and none of the states in the study had done so.

While this could explain the poor access to care observed in all the emancipated participants, the particularly poor access to health care associated with homelessness and, to a lesser extent, unstable housing, could be explained by the principle of competing priorities, shared risk factors, or effect-cause. Competing priorities on limited economic and social resources complicate the receipt of health care in homeless and unstably housed populations by forcing people who cannot meet their basic needs to prioritize these needs (food, housing) above health care.^{43 ,53} Participants with housing problems may not have had discretionary income to spend on payments for medications or clinic visits, may not be able to attend to health care needs because their time or energy may be going toward securing income or housing, or may not have been able to relinquish earned income by missing work for attending to their medical needs.

The relationship between poor access to health care and housing problems may be related to shared risk factors. For example, substance abuse, mental health problems, and poverty may be related to both housing and access problems. However, we found that the association between housing status and poor access was independent of substance abuse and mental health problems. Both income and employment status were not found to be related to the outcomes of interest (data not shown). Finally, the relationship between housing problems and poor health outcomes may be bidirectional, with health problems and medical expenses complicating the ability to remain in housing and housing problems complicating the receipt of appropriate health care. Out-of-pocket costs for health care and health problems that interfere with work may cause emancipated youth to lose their housing.

We found high rates of pregnancy among all participants, with increased rates for emancipated participants. Young adults with poor access to care are at higher risk of unintended pregnancy. While insurance was associated with pregnancy in these women, this was almost certainly because pregnant women with incomes less than 133% of the federal poverty level are eligible for Medicaid throughout their pregnancy and for at least 60 days afterward.⁵⁴ While there was not a statistically significant association between homelessness and pregnancy, there was a trend in that direction; because of small sample size, we may not have had the power to detect a true difference.

Pregnancy among emancipated youth has far-reaching implications. Former foster youth who become parents are more likely to have children in the foster care system than other young adults, and homeless adults with children are likely to have their children enter foster care.^{55 - 56} Young children of homeless adults placed in foster care are more likely than those without homeless parents to have siblings in foster care⁵⁶ and less likely to be placed with family members.⁵⁶

While there was not an association between housing problems and fair or poor health at the follow-up period, we did find high rates of self-reported fair or poor health among all participants. Whereas less than 4% of similarly aged young adults in a national study reported fair or poor health, between 10.3% (still in system) to 22.4% (homeless) of participants reported to be in fair or poor health.⁵⁰

Our study has several limitations. Few of the respondents were actually homeless at the follow-up interview; we defined those as homeless who had experienced any homelessness after emancipation. This is in keeping with the experience of young adult homelessness, which tends to be marked by short, episodic experiences of homelessness, rather than by chronic homelessness.⁵⁷ This would likely bias our results toward the null, in that participants were not continuously homeless. It is possible that some of the still-in-system participants experienced housing instability or episodes of homelessness during their stay in foster care; we did not have information on this and classified all still-in-system participants as stably housed. Our limited sample size may have impeded our ability to detect true differences. We do not know if the effects of emancipation are age dependent, that is, if the still-in-system youth will experience similarly poor outcomes on emancipation or whether delaying emancipation is associated with improved outcomes postemancipation.

Recently emancipated foster youth are at high risk for poor housing and poor health care access, which are associated with one another. Strategies to improve health outcomes among emancipated youth should address both their lack of health insurance and their risk of housing instability and homelessness.

Correspondence: Margot B. Kushel, MD, UCSF, Box 1364, San Francisco, CA 94143 (mkushel@medsfgh.ucsf.edu).

Accepted for Publication: April 25, 2007.

Author Contributions:Study concept and design: Kushel, Yen, and Courtney. Acquisition of data: Yen and Courtney. Analysis and interpretation of data: Kushel, Yen, and Gee. Drafting of the manuscript: Kushel and Yen. Critical revision of the manuscript for important intellectual content: Kushel, Yen, Gee, and Courtney. Statistical analysis: Yen and Gee. Obtained funding: Kushel, Yen, and Courtney. Administrative, technical, and material support: Courtney. Study supervision: Courtney.

Financial Disclosure: None reported.

Funding/Support: Dr Kushel was supported by the Stuart Foundation, the Agency for Healthcare Research and Quality grant K08 HS011415-03, and by the Hellman Family Award for early career development. Dr Yen was supported by the Stuart Foundation.

Additional Contributions: Amy Dworsky, PhD, assisted with the research project and Katherine Au, BA, assisted with the preparation of the manuscript.