To examine the health status, risk-taking behaviors, and access-to-care issues of shelter-based homeless adolescents compared with a domiciled adolescent population from the same large US city.
The samples consisted of 109 youth (aged 12-17 years) in emergency shelters and 1010 youth using school-based inner-city clinics.
Adolescents completed a health history, which was followed by a physical examination.
Homeless youth began sexual activity at an earlier age (median, 12 vs 13 years for homeless vs school-based youth), were less likely to have used birth control at their first sexual experience, and were twice as likely to have ever been pregnant. Oral and anal sex, same-sex activities (boys only), multiple sex partners in the past 30 days, depression, and substance use behaviors were reported more often by the homeless sample. The homeless youth were twice as likely to have visited an emergency department in the past 12 months. After adjustment for other risk factors, homelessness was an independent predictor of depression (adjusted odds ratio [OR], 7.0; 95% confidence interval [CI], 3.9-12.6), emergency department use in the past 12 months (adjusted OR, 1.8; 95% CI, 1.2-2.7), and history of pregnancy (adjusted OR, 2.6; 95% CI, 1.3-5.2) in the final logistic regression models.
This study confirms and extends past research indicating that homeless youth exhibit more risk-taking behaviors and suffer from poorer overall health than do nonhomeless youth.
ESTIMATES OF the number of homeless adolescents in the United States range from 500000 to 2 million.1 Estimates of the number and characteristics of homeless youth vary between localities, in part a reflection of true differences and in part due to differences in definitions and study methods. Recent studies reveal that homeless youth suffer from poor health status compared with the general adolescent population2,3; homeless youth have high rates of sexually transmitted diseases, human immunodeficiency virus infection, pregnancy, depression, and injuries.4- 11 The often overlooked subset of homeless youth that includes gay, lesbian, and bisexual identified individuals seems at particularly high risk for depression and suicide.12
Adolescents in general have unique barriers to access of appropriate health care and exhibit poor patterns of preventive primary health care.13 Health services for adolescents have largely been planned by adults, and many fail to recognize and accommodate the age-appropriate behaviors of adolescents.14 Although not well documented, it seems that homeless adolescents have even greater problems with access to health care, including a more profound lack of insurance or payment source, anxiety about issues of confidentiality, and confusion about the ability to give legal consent for care.2,15,16
However, the health status data for homeless youth are limited, with most being either anecdotal or clinic based. Clinic-based data collection represents a sampling bias because it does not include the nonseekers of health care. Medical records usually do not systematically collect the same data nor do they collect data in the same way, thus introducing instrumentation bias.17 Previous studies of homeless adolescents also suffer from a lack of an appropriate nonhomeless comparison group, making interpretation of findings difficult.
This study examines the health status and health care needs of shelter-based homeless adolescents and compares those adolescents with a domiciled adolescent population from the same city using a standard health history instrument. Health history questionnaires and physical examinations were conducted with both groups. Differences in risk-taking behaviors, access issues, and documented health problems of the 2 groups were analyzed.
The samples consisted of 109 homeless adolescents living in emergency shelters and 1010 domiciled adolescents using school-based health clinics (SBHCs) in Baltimore, Md. The homeless adolescents were accessed at the only 2 youth shelters in Baltimore. The shelters serve approximately 140 youths aged 12 to 17 years per year, with 80% being girls and 98% being from the poor inner-city areas of the city where the shelters are located. The SBHCs included in the study are in 6 different junior and senior high schools in Baltimore and are operated by the Baltimore City Department of Public Health.
Between January 1, 1994, and August 31, 1994, shelter-based youth were offered the health assessment. Only one youth refused to participate in the study, and more than 80% of all eligible shelter residents during this period were captured by the final sample. The remaining 20% were "missed" because they left the shelter before they could be approached for participation in the study. An analysis of the demographic profiles of missed youth during the first 3 months of the study showed no notable differences from those who did participate. Data from the health assessments of the SBHCs were obtained for the period from September 1990 to June 1992. School-based health clinic data from more recent adolescent health assessments were not available. Consent for inclusion in the study was received from all shelter-based youth. Approval for the study was obtained from the Institutional Review Board of The Johns Hopkins University School of Hygiene and Public Health, Baltimore, and the Baltimore City Department of Public Health.
A standardized, self-administered health history using closed-choice questions was used with both samples. The health history form has a separate male and female version, its readibility tested at the sixth grade level, and it has been tested and refined with more than 4000 Baltimore youth. The survey was adapted for use in the shelter-based youth by eliminating questions irrelevant to their living situation (ie, current family configuration) and by adding questions pertaining to homelessness (ie, places lived during the past 6 months). A midlevel provider nurse practitioner or physician assistant (J.E. for the shelter-based youth) reviewed each health history with the youth and then performed a physical examination on the youth at the SBHCs (for adolescents using SBHCs) or at the shelter or a community-based clinic (for homeless adolescents).
A standardized abstracting form was used by the providers for abstracting medical record data. The completed health history and physical examination forms were stripped of personal identifiers (ie, names were replaced with study numbers), and data were entered into a computer program (Epi Info Version 5, USD Inc, Stone Mountain, Ga).18 The data files were transferred to the university-based mainframe computer, checked for inconsistencies and incomplete data, cleaned, and analyzed using a software system (SAS, SAS Institute Inc, Cary, NC).19
Self-reported variables included demographics (age, race, sex, and living situation), risk behaviors (age at their first sexual experience, number of partners in the past month, condom use, and drug use), markers for problem behaviors or outcomes (repeating a grade, prior arrest, or prior pregnancy), markers for access to care (emergency department [ED] use in the past 12 months and source of last care), and positive behaviors (exercise, church attendance, and being employed). Depression status was obtained from provider diagnosis. Pregnancy incidence was too low overall to allow for further analysis, so history of previous pregnancy was used instead.
Differences between the 2 groups for behaviors and International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10)20 clusters were examined using χ2 and 2-tailed t tests. The Breslow-Day statistic was used to assess whether stratification provided significantly (P≤.05) different results for the stratifying variable21; if confounding was present, bivariate statistics were stratified. After examining differences within the 2 groups for outcomes of interest (ever pregnant, current diagnosis of depression, and ED use in the past 12 months) by bivariate statistics and determining that the patterns were similar, the combined data set (homeless and school-based youth) was also examined using χ2 and 2-tailed t tests.
Risk factors found to be significant to at least P=.10 were entered into a logistic regression model for predicting the outcomes of interest. These variables included risk and positive behaviors, as well as markers for problem behaviors. Demographic variables and several other variables not found to be significant, but which were deemed important explanatory or control variables (ie, age of menarche for ever pregnant), were entered into the models. The best subsets selection, along with purposeful selection of variables in the software system (SAS), was used to build the models, using the change in χ2 as the criterion. The models were first constructed with all variables except homelessness and then homelessness was added to the final model.
The 2 samples had the same sex distribution (69% girls and 31% boys), and the mean age of subjects in both samples was 14.4 years. However, the samples did differ by ethnic or racial composition, with the homeless sample having a higher proportion of white youth (21% for the homeless adolescents vs 13% for the school-based adolescents). Thus, most youth in both samples were black (79% for the homeless adolescents and 87% for the school-based adolescents). For current living situation, the survey questions necessarily differed for the 2 populations to more accurately capture this variable. All the shelter-based youth were living away from home, while only 9% of the SBHC group was not living with a mother or father (or both). It is probable that most of this 9% were living with extended family and were not homeless at the time of the survey. Of the homeless youth, 62% reported only foster care and shelter living within the past 6 months. The remaining 38% reported various living arrangements within the past 6 months, including doubling up with friends or lovers and living in cars or parks.
Markers for risk-taking behaviors were much higher in the homeless youth. These markers included having been arrested and having repeated a grade (Table 1). The homeless youth also exhibited higher levels of risk behaviors, such as sexual initiation and substance use. Protective and social support factors of engaging in regular exercise and attending church were significantly lower in the homeless sample.
The girls in both groups did not differ by age of onset of menses, with the average age being 11.4 years for both. The 2 groups did not differ by average Tanner stage for either sex. The homeless youth started sexual activity on average a year earlier (median, 12 vs 13 years for homeless vs school-based youth) and were more likely ever to have been pregnant (Table 1).
Figure 1 shows the difference between the 2 samples for age at their first sexual experience. The largest difference between the 2 samples was for sexual initiation in the 0- to 9-year-old age category (odds ratio [OR], 4.0; P<.001). In addition, oral and anal sex and same-sex activities (same-sex activities for boys only) are reported more often by the homeless youth (Table 1). Multiple sex partners (≥2 partners) within the past 30 days is reported by a higher percentage of homeless youth (19% vs 4% for homeless vs school-based youth; OR, 6.0; P<.001).
Cumulative percentage of adolescents according to age at their first sexual experience for homeless and school-based clinic youth, Baltimore, Md.
Experience with drug use is uniformly higher in the homeless youth (Table 1). The most commonly used substances for both groups of youth (listed in order of descending frequency) are cigarettes, alcohol, and marijuana. Other drugs, including crack, heroin, glue, and speed, are used by considerably fewer youth overall, but use of these substances is higher in the homeless. Not only do proportionately more of the homeless youth smoke, but they also smoke more heavily, averaging more than 3 cigarettes per day compared with 0.5 per day for the smokers in the SBHC sample. A higher percentage of girls than boys report cigarette use in the SBHC sample; the sex difference for smoking is reversed in the homeless sample. The homeless youth report much higher rates of previous physical abuse, ever having been forced to have sex, and homicidal and suicidal ideation (Table 2). In addition, more than 75% of the homeless youth report that they have witnessed a shooting or a stabbing incident, and 27% say they have hurt themselves on purpose sometime in the past. These questions were not asked of the SBHC youth.
The homeless youth report much higher rates of currently having asthma and seizure disorders (Table 2). The homeless youth were more likely to report being depressed in the past 6 months and ever to have had suicidal ideation. The homeless youth were asked to rate their overall health status on the health survey form; the self-rating of health status question was not included in the SBHC survey. Almost 30% of the homeless youth rated their health as fair or poor, and only 34% rated their health as excellent.
The homeless youth were twice as likely to report having had an ED visit within the past 12 months (44% vs 28% for the homeless vs the school-based youth; OR, 2.1; P< .001). Youth in the SBHC group were more likely to report having used school clinics or a physician's office as the source of last health care. In contrast, homeless youth most often reported having used hospital EDs or clinics as the source of last health care.
The main predictors of ever having been pregnant included older age, ever having repeated a grade, and ever having been forced to have sex (Table 3). After controlling for these plus age at the first sexual experience, age of menarche, and race, homeless girls were significantly more likely to have ever been pregnant.
Predictors for current diagnosis of depression included an interaction term for ever having wanted to kill oneself and having been depressed in the past 6 months (Table 3). This interaction term was used after examining the multicollinearity for these 2 variables. After controlling for these variables plus age, race, and sex, homelessness remained a significant predictor of depression.
The final model for ED use in the past year included having been forced to have sex, ever having had a sexually transmitted disease, current report of asthma, and black race (Table 3). After controlling for these variables plus age and sex, homelessness again remained a predictor of ED use, although not as strongly predictive as for the other outcomes of interest.
The study results indicate that the shelter-based homeless youth engage in more risk-taking behaviors and less positive health behaviors than do SBHC youth from the same city. The combination of starting sexual activity earlier, forced sex experiences, having higher rates of multiple sex partners, and increased drug use behaviors places the shelter-based youth at increased risk for unintended pregnancy, sexually transmitted diseases, and human immunodeficiency virus infection or acquired immunodeficiency syndrome. The results have important implications for health services for homeless youth and highlight areas for future research.
Most of the difference between the samples in age at the first sexual experience occurred in the 0- to 9-year-old age stratum, and it is likely that forced sex occurred in this age group. Many in the homeless sample (boys and girls) asked one of us (J.E.) for clarification of the question of "age you first had sex," whether it was the first time they were forced to have sex or the first time they chose to have sex. It is important for primary care providers to be aware of the high rates of previous forced sex experiences among female and male homeless youth, as sexual abuse of boys is often overlooked. Because the average age of menarche in the homeless girls was 11.5 years and a high percentage reported no birth control use at the time of their first sexual experience, it is apparent that the homeless girls are at high risk for unplanned and early pregnancy. Based on multivariate analyses, girls who are homeless, who have been sexually abused, and who have failed a grade are at highest risk for pregnancy and should be targeted for early intervention strategies.
The much higher rates of ED and hospital clinic source of last health care by the homeless supports the hypothesized increased barriers to access to appropriate primary health care for homeless youth. Emergency department use is a marker for poor access to primary health care, and use of hospital clinics probably indicates a higher reliance on medical assistance health insurance among homeless youth. Given the fact that homeless youth seem to have much higher rates of injuries and physical or sexual abuse or both, higher use rates for EDs may be appropriate. However, homelessness remained a predictor of ED use after controlling for other factors such as abuse. Thus, it is possible that homeless youth also rely on EDs for routine health care. School-based health clinics in Baltimore have been found to reduce self-report ED use by approximately 20%22; this may have affected somewhat the higher report of ED use among homeless youth. Nevertheless, access to comprehensive and appropriate primary health care for homeless adolescents could prevent crisis-oriented and costly ED care, as well as provide other stabilizing health and social services.
There are several limitations of the study that are important to keep in mind. The SBHC sample may not be representative of school youth and probably is not representative of nonhomeless youth. Likewise, the homeless sample may not be representative of homeless youth in Baltimore. Previous studies suggest that youth using SBHCs report somewhat more health problems than nonusers, at least for selected health problems.23 Other studies have shown that street-based homeless youth may differ from shelter-based homeless youth.24 Further attempts to identify and include out-of-shelter homeless youth in the homeless sample would strengthen future research efforts. Additionally, research about the health needs of different subgroups of homeless youth is recommended.
Despite these limitations, the study allows for further insight into the health needs of homeless youth by the use of a more complete health assessment as well as a more appropriate comparison group. The results support the conclusions of previous research that shelter-based homeless youth suffer from poorer overall health, have more profound problems with access to primary health care, and exhibit more risk-taking behaviors than do in-school youth.
Providers in any setting serving adolescents should be alert for the possible signs of risk for homelessness among youth they serve. Providers can ask appropriate questions pertaining to the living situations of youth, identify homeless or at-risk youth, and refer them to social service personnel and agencies. In addition, service providers for homeless youth should know the major mental and physical health issues and environmental threats for these youth. Supplemental health education and primary health care services should be offered through shelters and in locations accessible to homeless youth. These services should be tailored to the lifestyles, needs, and priorities of the youth.
Accepted for publication August 14, 1997.
Editor's Note: If home is where the heart is, and you have no home...—Catherine D. DeAngelis, MD
Reprints: Jo Ensign, FNP, DrPH, Department of Psychosocial and Community Health, University of Washington, Box 357263, Seattle, WA 98195-7263 (e-mail: firstname.lastname@example.org).
Ensign J, Santelli J. Health Status and Service UseComparison of Adolescents at a School-Based Health Clinic With Homeless Adolescents. Arch Pediatr Adolesc Med. 1998;152(1):20-24. doi:10.1001/archpedi.152.1.20