Mental Health Service Use in a Community Head Start Population FREE

THE MENTAL HEALTH needs of young children are of growing concern in the field of primary care. In particular, children in the Head Start population are considered to be at high risk for psychosocial problems due to concomitant risk factors associated with poverty. The Head Start program was founded in 1965 to improve the social competence and school readiness of low-income children. Incorporated into its original mission statement, mental health was seen as the cornerstone of Head Start concepts such as prevention, outreach, early detection, and family involvement.^1- 2 Providing appropriate mental health services to young children and families in the Head Start program has remained a challenge because of increasing diversity among families, poorly integrated service systems, resource difficulties, and managed care restrictions.²

Service utilization research has begun to identify predictors of mental health service use among children and adolescents; however, reviews indicate conflicting findings about many of the variables associated with higher service use. For example, low-income families were found to be overrepresented in studies of referrals to mental health services in the Ontario Child Health Study.³ Other research suggests that low income is associated with low service use due to restrictions in access to mental health care.⁴ Some research indicates that middle-income families are also underserved because of ineligibility for subsidized service or the inability to afford services.⁵ The consensus appears to be that whereas lower-income families are at higher risk for mental health problems due to difficult social circumstances exacerbated by poverty, low–middle-income families may not be adequately covered by insurance. Other factors associated with lower child mental health service use include younger age,¹ minority status,⁴ and female sex.⁶ Little research has focused on early childhood (<5 years) in terms of child mental health service use. Studies that do include this age group indicate low rates of service utilization for young children despite a significant rate of behavioral and/or developmental difficulties, which places them at increased risk for later psychiatric diagnoses.^3,7 Across multiple epidemiological studies, it is estimated that 10% to 15% of preschool children have behavioral or emotional problems.^8- 9 Children from socioeconomically disadvantaged families are at higher risk for developing behavior problems.¹⁰

Another factor related to the need for and use of child mental health services is maternal mental health. Several studies have consistently found that the mental health of the child's primary caregiver (usually the mother) is a strong predictor of behavior problems in children.¹¹ Women, and mothers in particular, are at high risk for depression, with rates as high as 40% in mothers of preschool children.¹²

The purpose of this study was to examine both demographic and family factors associated with child-focused mental health service utilization in a group of preschool children. By understanding these variables, service needs can be identified and potential ways to improve access and utilization by low-income families can be devised.

This study investigates child-focused family mental health service utilization by Head Start children and families enrolled in the Starting Early Starting Smart (SESS) program. Based in Montgomery County, Maryland, this project was conducted at one of 12 sites taking part in a national research demonstration program funded by the Substance Abuse and Mental Health Services Administration and the Casey Family Program. A primary objective of the SESS program was to inform practice and policy to improve mental health and related services for families, particularly those with low incomes.

Enrollment was based on eligibility for Head Start services in Montgomery County, a large suburban county outside of Washington, DC. The study included 290 families. Of the enrolled Head Start children, 52% were boys and 48% were girls. Many of the families in this sample were immigrants: 43% were from Latin America, 22% were from non-Latino (African, Caribbean, or Asian) immigrant families, and 35% were from nonimmigrant families. At the time of data collection, the children ranged in age from 3.2 to 5 years with a mean ± SD age of 4.3 ± 0.3 years. Most respondents were biological mothers (91.3%) or biological fathers (6.25%). The predominant languages spoken in the home were English (45.1%) and Spanish (44.1%). About one quarter (27.5%) of the respondents had completed 12th grade (n = 80). The mean household income was $1491 per month, and 62% of the primary caregivers were employed. Although the SESS study is longitudinal in design, preliminary data presented in this article relate only to child and parent functioning and service use.

Intervention and comparison groups were randomized by school rather than by individual subjects. Prior to the enrollment of the first cohort of families, 4 schools in the county with comparable demographics were selected to participate in the study. Two were randomly designated as intervention schools, and the other 2 as comparison schools. All families with children enrolled in Head Start in those 4 schools during the 1998-1999 academic year were informed about the study and invited to participate. Caregivers who expressed interest in participating were visited in their homes by a research assistant, who explained the study in greater detail and obtained consent from the target child's primary caregiver. The process was repeated with 6 schools in the 1999-2000 academic year. The families enrolled from these schools constituted the study's second cohort. Across the 2 cohorts, 46 families were approached but refused to participate. Most (43%) of these families cited lack of time as their reason for not participating. Excluded from both cohorts were 17 families in which none of the adult members spoke English or Spanish well enough to give informed consent.

Data were collected by trained research assistants during home-based interviews. The interviews took place as close as possible to enrollment in the Head Start program and were conducted in either English or Spanish according to the preference of the respondent.

The Preschool and Kindergarten Behavior Scales (PKBS)¹³ are a rating instrument for use in evaluating problem behavior patterns in preschool-aged children. This norm-referenced, standardized instrument was developed for use in assessing young children in a variety of settings and by different informants. The PKBS is a broad-based scale that comprises a checklist of 42 problem behaviors commonly seen in the preschool-aged group. It is a written parent report measure designed to assess behavioral and emotional problems in young children. The respondent is asked to rate the frequency of each behavior on a 4-point scale ranging from "never" (0) to "often" (3). The PKBS yields raw scores that are combined into 3 global scores: an internalizing score for emotional problems, an externalizing score for behavior problems, and a total problems score. Higher scores indicate greater emotional and behavioral problems. Internalizing problems include symptoms of depression, social withdrawal, anxious and inhibited reactions, and the development of somatic problems. Externalizing problems include behaviors that are aggressive, defiant, disruptive, oppositional, and hyperactive. The PKBS was completed separately by parent and teacher. Children were considered to have clinically significant behavior problems if they scored within the moderate-deficit or severe-deficit range of functioning. Reliability coefficients for the PKBS Total Problem behavior scale for this sample were 0.91 for parent-completed scales and 0.96 for teacher-completed scales.

Parents completed the Brief Symptom Inventory (BSI).¹⁴ The BSI is a self-report measure that asks respondents to rate the degree of distress experienced in the past 7 days from various psychological symptoms on a 5-point scale (0, "not at all," to 4, "extremely"). The BSI consists of 53 items and takes about 15 minutes to complete. The primary symptom dimensions are somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. The BSI yields raw scores for individual scales and subscales that are then converted to T scores (mean ± SD standard score, 50 ± 10). T scores of 63 or higher on 2 of the 6 clinical subscales meet the criteria for a clinical case. Reliability coefficients for each subscale of the BSI for this sample ranged from 0.69 to 0.80. The Parenting Stress Index–Short Form (PSI-SF)¹⁵ is a 36-item self-report measure of the magnitude of stress in a parent-child system, as perceived by the parent or primary caregiver. Because of time constraints and similarity of items with the BSI and PKBS, only 1 subscale was analyzed (Parent-Child Dysfunctional Interaction [P-CDI]). The PSI-SF yields percentile scores for subscales, which are then converted into standard scores (mean ± SD, 100 ± 15) for data analysis purposes. Parents who obtained a subscale score at or higher than the 90th percentile were considered to be experiencing clinically significant levels of stress within the parent-child relationship. The P-CDI subscale focuses on the parent's perception that the child does not meet the parent's expectations and that parent-child interactions are not rewarding for the parent (eg, "My child seems to cry or fuss more than most children," "My child generally wakes up in a bad mood," or "My child does not like to be cuddled or touched"). Reliability for this subscale was 0.88 for our sample.

The Conflict Tactics Scale¹⁶ (CTS) is a 19-item self-report instrument that measures individual responses to conflict within the family. The scales are intended to measure physical aggression in the marital or adult relationship. A dichotomous variable (physical aggression) was created indicating the presence or absence of violence in the adult relationship. Violence was considered to be present if affirmative responses were given to 1 or both of the following criteria: (1) 3 or more incidents of any violent act, and/or (2) 1 incident of severe violence as defined in the CTS manual. The reliability coefficient for the CTS Violence scale was 0.79 for this sample. The Home Observation for Measurement of the Environment¹⁷ (HOME) is a 45-item interview and observation measure. Home visits were made to assess subjects according to the HOME Inventory, which includes the following domains: parental affection (responsivity), presence of materials to facilitate or stimulate learning, and emotional and verbal reciprocity in the child's home environment. Subscales of the HOME include learning material, language stimulation, physical environment, academic stimulation, modeling, variety, and acceptance. The reliability coefficient for the HOME scale was 0.82 for this sample. Scores on the HOME were dichotomized into the categories "at risk" and "other" for our study. Families with a score at or below the lowest quartile of the total SESS sample (n = 82) were considered to be within the clinically "at risk" range.

Instruments and their administration (eg, the use of short forms) were selected by consensus of the SESS cross-site steering committee to ensure national uniformity in measures across all subsites.

The Services Access and Utilization Scale (SAUS) was developed for the SESS cross-site study to evaluate social and behavioral health service use. The SAUS is a semistructured interview that includes questions about service need and use for the respondent and target child in the past 12 months.

Families were considered to have received child-focused services if a positive response to 1 or both of the following items was obtained: (1) use of parent education groups, and/or (2) consultation with a professional to receive help specifically on parenting or managing child behavior.

Parents were considered to have received services if a positive response to any 1 or more of the following items was obtained: (1) attended sessions with a professional for problems with emotions or stress; (2) participated in a support group to help with emotions or stress; (3) attended a substance abuse treatment program (such as Alcoholics Anonymous); and/or (4) was admitted to a psychiatric facility for mental health and/or substance use issues.

Risk factors for psychosocial adversity were identified using criteria from the Annie E. Casey Family Risk Index¹⁸ based on data collected during the SAUS interviews. The Family Risk Index comprises 6 demographic factors: (1) child is not living with 2 parents; (2) household head did not complete high school; (3) family income is below the federal poverty line; (4) child is living with parents who do not have steady, full-time employment; (5) family is receiving welfare benefits; and (6) child does not have health insurance. Whereas adjustment to 1 of these risk factors may be challenging, prior research indicates that children living in families with 1 risk factor are not likely to differ significantly on behavioral outcomes from those with none. Except for a maximum of 10%, all children in Head Start live in families whose income is below the state poverty line. Although likely, it was not certain that many of the children in this sample would meet the federal poverty line cutoff because of the more stringent (ie, lower income level) federal poverty criteria. The federal cutoff was therefore included as a potential risk factor. In combination, the presence of multiple risk factors can have an additive and damaging effect on child development. For this investigation, each family was given a cumulative risk factor score by summing each risk factor present (1-6) to form a composite index. Given the relatively small number of families (n = 21) with 4 or more risk factors, our study deviated slightly from the methods used by the Annie E. Casey Foundation by considering families with 3 or more risk factors to be at significant risk. This step was taken to include a larger sample of families while maintaining the sensitivity of the Annie E. Casey Family Risk Index.

Frequency distributions and the mean and SD of the measures were examined. An exploratory analysis was performed to identify potential variables that might act as mediators of service use. Users of services were compared with nonusers on categorical variables with the Fisher exact test.

Univariate odds ratios (ORs) between predictors (demographic risk, child characteristics [sex], and family measures [HOME, BSI, PSI-SF, and CTS]) and child- and parent-focused mental health utilization outcomes were computed using logistic regression analyses. Univariate ORs were computed to analyze the relationship between predictors for the following variables: (1) demographic predictors and PKBS outcome variables (parent and teacher ratings of internalizing and/or externalizing behavior problems); (2) demographic adult and child mental health predictors of child-focused family mental health service utilization; and (3) demographic adult and child mental health predictors of parent-focused mental health service utilization. Subsequent to univariate regression analyses, variables meeting the following criteria (P<.001) were included in multivariate logistic regression equations to determine which ones most effectively predicted scores on measures of internalizing and externalizing behavior problems. Data were analyzed using SAS version 8.1 (SAS Institute Inc, Cary, NC) and SPSS version 10.0 (SPSS Inc, Chicago, Ill) statistical software.

Data were collected on 290 children enrolled in the SESS study at school entry in 1998 and 1999 in 8 schools in Montgomery County. Of the total sample, 133 children (45.8%) were Latino, 73 (25.1%) were African American, and 15 (5.2%) were white. The children ranged in age from 3.2 to 5 years (mean ± SD age, 4.3 ± 0.3 years). Most respondents were biological mothers (91.3%) or biological fathers (6.25%). The predominant languages spoken in the home were English (45.1%) and Spanish (44.1%). With regard to parent educational level, 80 caregivers (27.5%) had completed 12th grade. Risk factors for psychosocial adversity were identified using criteria from the Annie E. Casey Family Risk Index based on data collected during the SAUS interviews. Frequency counts for the 290 children on whom these data were collected were as follows: (1) child is not living with 2 parents (n = 166; 57%); (2) household head did not complete high school (n = 7; 2%); (3) family income is below the federal poverty line (n = 140; 48%); (4) child is living with parents who do not have steady, full-time employment (n = 83; 29%); (5) family is receiving welfare benefits (n = 87; 30%); and (6) child does not have health insurance (n = 49; 17%).

As indicated in Table 1, 29% of this sample of children entering Head Start met the criteria for behavioral problems, as rated by their parents (n = 84). For teachers, the rate was much lower (12%; n = 32). Prevalence rates for internalizing and externalizing diagnoses were similar (22% and 17%, respectively, for parent report, and 9% and 5%, respectively, for teacher report).

There was low concordance between parents' and teachers' ratings of child behavior problems (κ = 0.02; P = .75) (Table 2 and Table 3). Differences between parent and teacher externalizing and internalizing behavior rating scores were not significant, indicating that differences in behavioral ratings were not attributed to the severity of behavior problems observed (ie, teachers rating only very severe problems as clinically elevated). Low parent-teacher concordance on child behavior ratings appears to be consistent with previous research.^19- 20

As presented in Table 4, 43% of the sample met the criteria for caseness on the BSI. Caseness is defined in the BSI manual as 2 or more elevated scales and refers to the value or score on the screening measure that defines a positive case (ie, the presence of parental mental health problems). Prevalence rates for individual scales were as follows: paranoid ideation (46%), interpersonal sensitivity (38%), obsessive-compulsive (28%), depression (27%), anxiety (21%), and hostility (21%). As indicated in Table 5, 16% of the sample were in the clinically elevated range on the P-CDI subscale of the PSI-SF.

Using univariate logistic regression to predict the risk of behavior problems reported by parents, the significant variables were parent mental health (caseness on the BSI), physical aggression between parents (CTS), and elevated P-CDI scores on the PSI-SF. Elevated HOME scores and demographic risk were significant at P = .10. Using multivariate logistic regression, the only significant predictors of parent ratings were BSI caseness (OR = 3.7; 95% confidence interval [CI], 2.1-6.5; P<.001) and P-CDI score (OR = 4.3; 95% CI, 2.1-8.5; P<.001) (Table 6). Significant variables predicting teacher behavior problem ratings were child's sex (male) (OR = 2.7; 95% CI, 1.2-6.0; P = .02) and elevated HOME scores (OR = 2.8; 95% CI, 1.3-5.8; P = .01) in a multivariate model.

Using multiple logistic regression to predict child-focused family mental health utilization, the significant variables were parent-focused mental health use and clinically elevated HOME scores (Table 7). Child behavior problems (parent- or teacher-rated) were not associated with child-focused family mental health utilization.

The design of this study was limited in several ways. Like most studies of this type, much of the data collected were from a single source: the primary caregiver (>90% were mothers). The use of teacher ratings lessens this problem; however, further supportive data would be helpful, especially considering the low rate of agreement between parents and teachers regarding what constitutes problem behavior. Whether this is due to real-life situational differences or differences in perception is a question that has long been asked in developmental research. Although the cross-sectional design of this study is appropriate for generating and testing preliminary hypotheses, a prospective longitudinal design is required to fully explore how the parent-child relationship, child behavior problems, and child-focused mental health service use are causally related. Given the high rate of parent-reported behavioral problems in this age group, it is possible that these difficulties are transient and/or within normal limits for this population.

Findings from this study underscore the importance of evaluating parent mental health because of its central role in both the development and identification of behavioral problems in young children. Screening tools that can be easily administered during an office visit are most likely to be useful to community providers.^21- 22

Although not new to this field, these findings emphasize the fact that many children have unmet mental health needs before they enter elementary school. In this Head Start population, nearly 1 in 3 children were identified as having clinically significant behavioral problems by their parents. However, almost 80% of identified children did not receive child-focused mental health services. A similar picture emerged when examining teacher ratings. Whereas nearly 1 in 8 children were rated as having clinically significant behavioral problems, more than 90% of these children did not receive services. In part, these unmet needs may be due to low concordance between a child's key caregivers regarding behavior problems. For example, in this Head Start population, parents and teachers frequently disagreed on whether a child truly had behavioral problems. This may be partially explained by situational differences in the behavior of young children. Moreover, different factors may influence parents' opinions about their child's behavior. Whereas teacher ratings appear to be related to sociodemographic factors (eg, sex or home environment), those of parents were linked to their own emotional distress and parent-child relationship factors. Thus, some of the difficulty concerning adequate access to and delivery of child-focused mental health services may be due to different perceptions of a child's behavior by critical gatekeepers.

The literature suggests that one of the predictors of adult mental health service use is internal distress,²³ which acts as a powerful motivator to access and utilize services. Similarly, parental motivation to make and keep appointments for mental health services would seem to be motivated by the level of distress or burden they feel as a result of their child's behavior. Future research is needed to operationalize the concept of parental stress and its relationship with access to and utilization of services.

Consistent with previous research, parent mental health problems were strongly related to parent perceptions of child behavioral problems. In this Head Start population, a relatively large proportion (43%) of caregivers surveyed met the criteria for clinically elevated levels of distress as measured by the BSI. In part, this seemed to be due to a significant number (46%) of caregivers with elevations on the paranoid ideation and interpersonal sensitivity subscales of the BSI. The paranoid ideation scale represents paranoid behavior including suspiciousness, hostility, grandiosity, and fear of loss of autonomy. The interpersonal sensitivity dimension focuses on feelings of personal inadequacy and inferiority. Items on this scale are based on concepts such as self-deprecation, self-doubt, and marked discomfort during interpersonal interactions.

These subjective feelings, in particular, may add to caregivers' perceptions of behavioral and emotional problems in their children and may be related to a greater sense of inadequacy and alienation. In turn, this may influence parents to seek out assistance (parent- or child-focused) from mental health professionals.

Another interesting finding in this preliminary study was the role of the child's home environment as measured by the HOME scale. Both teacher ratings of child problem behavior and utilization of child-focused mental health services were predicted by the quality of the home environment. Children living in a home with parental affection (responsivity), materials to facilitate or stimulate learning, emotional and verbal reciprocity, and developmentally appropriate stimulation were nearly twice as likely not to be rated by teachers as having clinically significant behavior problems. Moreover, children living in these homes were twice as likely to receive child-focused mental health services. These findings suggest that increased parental sensitivity to the developmental needs of their children at home is associated with utilization of child-focused mental health services and reduced behavioral problems in preschool.

In conclusion, even though parents identified mental health problems in themselves and their children in this study, the gap between need and use remains. Pediatricians in primary care are in a unique position in terms of being able to address education regarding early identification of behavior problems, appropriate referrals for treatment, and issues of restrictions in access to behavioral health, mental health, and substance use services. This leads to a challenging corollary for many community physicians. As research highlights the need for community level screening of mental health problems in children and their parents, the health service community is lacking in qualified professionals to meet the identified need.

Accepted for publication March 15, 2002.

This study was supported by grant 5 U1H SPO7983 (Jill Joseph, MD, PhD) from the US Department of Health and Human Services (DHHS), Washington, DC; the Substance Abuse and Mental Health Services Administration (SAMHSA) and its 3 centers—the Center for Mental Health Services, the Center for Substance Abuse Prevention, and the Center for Substance Abuse Treatment, Rockville, Md; and Casey Family Programs, Seattle, Wash.

This study would not have been possible without the contributions of staff from DHHS, the Office on Early Childhood at SAMHSA, and Casey Family Programs; the SESS principal investigators, project directors, and researchers; and the parent representatives who helped design and supervise the data collection. We thank Children's National Medical Center for institutional support and the following individuals for help with the preparation of this manuscript: Cheng Shao, MPH, Carolin Frey, PhD, Juanita Wiley, BA, Melissa Rocklen, BA, Wendy Albright, BA, Erin Anne Smith, BA, Mara Kailin, BA, Maia Coleman, BA, Katherine Marshall, BA, and Carla Jenkins, BA. Thanks are also extended to 2 anonymous reviewers for their helpful comments.

The content of this publication does not necessarily reflect the views or policies of DHHS or Casey Family Programs, nor does mention of trade names, commercial products, or organizations imply endorsement by the US government. Responsibility for the content of this report rests solely with the named authors.

Pediatricians are becoming aware of the disparity between the mental health needs and mental health service use of young children. This study focuses on the unique needs of younger children (<5 years old) and the importance of their home environment, relationships with their primary caregivers, and parents' mental health. Understanding the behavioral health needs of children as they are commonly seen in general pediatric practice is essential to primary care physicians. The context of Head Start and other community preschool programs represents a potential venue for collaboration between primary care physicians and school-based programs. Suggestions from this study include adding screening instruments for use by community physicians as a tool in the identification of behavioral problems in young children. Developing appropriate services to meet the referral needs of these children and families is an urgent policy matter.

Corresponding author and reprints: Michelle New, PhD, Department of Psychology, Children's National Medical Center, 111 Michigan Ave NW, Washington, DC 20010-2970 (e-mail: mnew@cnmc.org).