Correlates of schizophrenia spectrum disorders in children and adolescents cared for in community settings

Authors


Abstract

Background:  This study examined the prevalence and correlates of schizophrenia spectrum disorders (SSD) among a national sample of 9006 children.

Methods:  Clinician-assigned diagnoses were used to divide the sample into two groups: children with SSD and children with other Axis I disorders.

Results:  Three percent of the sample had a SSD diagnosis. African American (OR=1.71, 95% CI: 1.11, 2.65) and Hispanic race/ethnicity (OR=1.96, 95% CI: 1.31, 2.94), a greater number of comorbid psychiatric diagnoses (three diagnoses, OR=2.22, 95% CI: 1.49, 3.31), a history of attempting suicide (OR=1.45; 95% CI: 1.05, 2.02), and past residential treatment (OR=1.59; 95% CI: 1.11, 2.28) were all associated with increased odds of SSD diagnosis.

Conclusions:  Although schizophrenia spectrum disorders in youth are rare, children with these disorders present with a distinct risk profile that may inform service planning and delivery and assist in identifying individuals early in the course of their illness.

Key Practitioner Message:

  •  Many children with SSDs treated in community settings have co-occurring disorders and enter systems-of-care with a history of restrictive service use and prior suicide attempts.
  •  These children may need comprehensive services and supports to meet their complex needs.

Schizophrenia spectrum disorders, such as schizophrenia and schizoaffective disorder, are costly conditions that have detrimental effects on quality of life (Schiffman & Daleiden, 2006; Wu et al., 2005). Onset of these disorders in childhood and adolescence is especially damaging; studies on the long-term course of individuals diagnosed with schizophrenia spectrum disorders (SSD) in childhood show that the majority of individuals have poor outcomes, including functional impairment and premature mortality (Lay, Blanz, Hartmann, & Schmidt, 2000; Remschmidt et al., 2007; Remschmidt, Schulz, Martin, Warnke, & Trott, 1994). While past research has examined characteristics of SSD in children and adolescents treated in specialized settings (Biederman et al., 2004; Caplan & Tanguay, 1996; Lay et al., 2000), less attention has focused on youth treated in community settings. An increased understanding of the characteristics of youth with SSD entering community based mental health services will be valuable in informing treatment planning and service delivery for these youth and their families.

Few studies have assessed the prevalence of schizophrenia spectrum disorders or associated symptoms in children and adolescents (Boeing et al., 2007; 1972; Remschmidt et al., 1994; Schiffman & Daleiden, 2006; Thomsen, 1996). One study of youth receiving services in a public mental health system found that 0.5% were diagnosed with a SSD (Schiffman & Daleiden, 2006). Other studies focused specifically on schizophrenia suggest that prevalence among children and adolescents in the community ranges from 1 to 10 per 100,000 (Remschmidt et al., 1994; Thomsen, 1996). Psychotic disorders are found in approximately 1% of youth in the community (Evans & Acton, 1972; Lohr & Birmaher, 1995), and psychotic symptoms in 4% to 8% of those referred to mental health services (Biederman et al., 2004; Caplan & Tanguay, 1996).

While correlates of SSD in children and adolescents have not been thoroughly investigated, the limited research suggests that children with SSD are typically older than those with other mental health conditions (Schiffman & Daleiden, 2006) with onset occurring earlier in males (Biederman et al., 2004; Remschmidt et al., 1994). This is consistent with the broader literature on schizophrenia that finds that onset typically occurs between age 15 and 45, with females experiencing onset 5-7 years later than males (Gross, 1997; Häfner, an der Heiden, Behrens, Gattaz, Hambrecht, Löffler, Maurer, Munk-Jørgensen, Nowotny, Riecher-Rössler, Stein, 1998). A study in Hawaii found that Asian children were more likely to receive an SSD diagnosis (Schiffman & Daleiden, 2006); however, little is known about ethnic differences in the prevalence of SSD among children in other parts of the United States. There is some evidence from the adult literature that suggests that African Americans are more likely to be diagnosed with schizophrenia compared to Caucasians (Minsky, Vega, Miskimen, Gara, & Escobar, 2003; Strakowski et al., 1996). This area needs further investigation.

A number of other characteristics have also been associated with schizophrenia spectrum disorders, including a family history of psychiatric hospitalization (Green, Padron-Gayol, Hardesty, & Bassiri, 1992), a family history of schizophrenia, and experiencing physical or psychological trauma (Read, Os, Morrison, & Ross, 2005). Early maternal stress and health behaviors have also been linked to offspring expression of psychosis (Spauwen, Krabbendam, Lieb, Wittchen, & Os, 2004). While not specific to youth, other established risk factors for SSD include experiencing social adversities in childhood, such as low socioeconomic status, living in single-parent households and receiving welfare (Wicks, Hjern, Gunnell, Lewis, & Dalman, 2005). Social adversity also has been associated with other mental disorders in children (Anda, Felitti, Bremner, Walker, Whitfield, Perry, Dube & Giles, 2006; Costello, Farmer, Angold, Burns, Erkanli, 1997). Youth with SSD or psychosis are also more likely to have comorbid mental disorders, greater functional impairment, and receive more restrictive services as compared to children with other mental health conditions (Biederman et al., 2004; Schiffman & Daleiden, 2006).

Although previous literature has identified characteristics of children with SSD, many studies have been limited by samples restricted to children in specialized clinic or hospital settings, a focus on narrow geographic areas, and an examination of a limited range of risk factors. Children that receive services in specialized clinics or hospitals may have more severe symptoms and may be less responsive to treatment than children treated in the community. Studies that focus on these populations may not generalize to youth served in the community. The present study extends previous research on child-onset schizophrenia spectrum disorders and examines the prevalence of SSD and the demographic, clinical, individual and family risk correlates of SSD among a national sample of children and adolescents entering community mental health systems-of-care from 1997 to 2005. Based upon associations established in the psychosis literature, we hypothesized that children with SSD would be older and more likely to have a family history of mental illness, be eligible for Medicaid and to have previously used intense psychiatric services than children without a SSD. Given the limited or conflicting existing research on race differences and family correlates (e.g., family functioning) of SSD in children, we did not have a priori hypotheses regarding these associations.

Methods

Data source

We conducted secondary data analyses using data from the national evaluation of the Comprehensive Community Mental Health Services for Children and Their Families Program (CMHI). Initiated in 1993, the CMHI funds communities to develop and enhance mental health systems-of-care for children and adolescents with serious emotional disturbances (Macro International Inc., 2007). The national evaluation of the CMHI examines youth and family outcomes, as well as the development and implementation of the system-of-care approach and service delivery practices. The evaluation collects descriptive information (e.g., demographic characteristics, clinical diagnosis, etc.) on all children referred into the program and more extensive information on a sub-sample of children and their families who agree to participate in a longitudinal study of the program. As part of the national evaluation protocol, funded communities obtained informed consent and assent from all participants in the longitudinal study. A comprehensive description of the national evaluation study design and procedures is found elsewhere (Macro International Inc., 2007).

Common to longitudinal studies, missing data was a concern and a multiply imputed dataset was created to allow multiple investigators to examine a broad range of research questions (Stuart, Azur, Frangakis, & Leaf, 2009). Missingness on variables included in this study ranged from 0% to 62%, with the majority of variables missing less than 10%. Using complete case analysis (i.e., ignoring the missing data) could lead to biased estimates and reduced power to detect differences. Multiple imputation is a principled method of addressing missing data whereby missing values are imputed by predicting the values based on the observed data (Rubin, 1987; Schafer & Graham, 2002). In this approach, a series of regression equations are estimated. A variable with missing data is regressed on other variables included in the dataset. That variable’s missing values are replaced with the predictions from that regression model. The regression model is run multiple times until the estimates become stable. The process is repeated for each variable with missing data and then it is repeated multiple times, generating multiple datasets. In the national evaluation data, ten imputed datasets were created. Analyses are simultaneously run across the individual imputed datasets and the results are combined accounting for variation within and across the imputed datasets (Azur, Stuart, Frangakis, & Leaf, 2011; Rubin, 1987).

Sample

Baseline data for children (n=9,006) and families participating in the longitudinal study were used. The sample includes children, ages 5-17 years, who entered services in one of 45 grantee sites from 1997-2005. The mean age was 12 years and the sample was predominately male (67%). Over half of the sample was Caucasian (52%), 23% African American, 12% Hispanic, and 6% American Indian/Alaskan Native.

The sample was divided into two groups: children with a SSD (defined as a DSM-IV diagnosis of schizophrenia, schizophreniform, schizoaffective disorder, delusional disorder, brief psychotic disorder, and psychotic disorder not otherwise specified) and children with a non-SSD (defined as any other Axis I diagnosis). Among the children in the SSD group, 16% had a schizophrenia diagnosis. Seventy-six percent of children with a SSD had a co-occurring disorder, including mood (28%), ADHD (21%), oppositional defiant (16%), conduct (5%), adjustment (4%) and anxiety disorders (4%). Among the children in the non-SSD group, diagnoses included ADHD (41%), mood (35%), oppositional defiant (29%), adjustment (11%), conduct (10%), substance abuse/dependence (7%), and anxiety disorders (5%). Comorbid Axis I diagnoses were present in 59% of children in the non-SSD group.

Measures

All measures were assessed at intake into the program.

Clinical characteristics.  Clinician assigned DSM-IV diagnostic codes were extracted from clinical case records and were used to classify children into SSD and non-SSD groups. This information was also used to create a variable that reflects number of diagnoses/diagnostic groups a child had.

The Child Behavior Checklist (CBCL), a valid and reliable instrument (Achenbach, 1991; Barkley, 1988), was used to assess caregiver perception of child internalizing and externalizing problems. We used the standardized T-scores for the internalizing and externalizing syndrome scales where higher scores indicate increased problems; a score of 64 or above suggests clinical levels of problems.

The Child and Adolescent Functional Assessment Scale (CAFAS) (Hodges, Lambert, & Summerfelt, 1994) was used to assess functional impairment across multiple domains. Scores 0-40 suggest minimal impairment, 50-90 moderate impairment, 100-130 marked impairment, and 140 or greater severe impairment (Hodges et al., 1994). The reliability and validity of the CAFAS has been widely established (Hodges et al., 1994; Hodges & Wong, 1996).

Behavioral and emotional strengths across multiple domains were assessed using caregiver report on the Behavioral and Emotional Rating Scale (BERS, α≥.83) (Epstein, 1999). Scores < 90 suggest below average strengths, 90-110 average strengths, and >110 above average strengths.

Demographic characteristics.  Caregivers reported on the child’s age, sex, race/ethnicity, eligibility for Medicaid, referral source, custody status, household income, and caregiver education level.

Psychosocial risk factors.  Caregivers reported on the child’s lifetime history of psychiatric hospitalization, suicide attempts, physical abuse, sexual abuse, running away, substance abuse, and being sexually abusive to others. Caregivers also reported on whether there was a lifetime history of domestic violence, mental illness in the child’s biological family, a biological parent who had experienced psychiatric hospitalization, a biological parent ever convicted of a crime, a family history of substance abuse and a biological parent ever receiving substance abuse treatment.

Family Characteristics.  The Family Resources Scale (α=.92-.95) assessed caregiver perception of the financial, physical and emotional resources available to the family in the last six months (Dunst & Leet, 1987; Heflinger, Northrup, Sonnichsen, & Brannan, 1998). Caregiver perception of current family functioning (e.g., communication, involvement) was assessed using the Family Assessment Device (α=.71-.92) (Byles, Byrne, Boyle, & Offord, 1988). Higher scores on both scales indicate greater perceived adequacy of family resources and family functioning, respectively. Caregiver strain (α=.73-.91) (Heflinger et al., 1998) was assessed using the total score of the Caregiver General Strain Scale, with higher scores indicating greater strain (Brannan, Heflinger, & Foster, 2003).

Past 12-month service use.  Caregivers were asked whether their child had received outpatient, school-based, inpatient/residential, day treatment, or substance use treatment services in the past year.

Statistical analyses

We tested bivariate differences in characteristics between children with and without a SSD. Then, we used random effects logistic regression to estimate the association between the independent variables and a SSD diagnosis, accounting for the clustering of children within program site. Variables were selected for inclusion in the multivariate logistic regression model by first identifying covariates with established associations in the literature (e.g. sex, age, Medicaid eligibility, family history of mental illness, psychiatric hospitalization). Forward and backward stepwise selection procedures were then used to identify additional variables. We used the variance inflation factor (VIF) to test for collinearity. The VIFs were less than 2.5, indicating collinearity was not a concern. The final regression model was estimated across the ten imputed datasets.

Given that differences may exist among children with SSD who are diagnosed at a young age and children who are diagnosed in adolescence, we also performed analyses stratified by age (5-12 years and 13-17 years). All significant associations were the same as for the full model, thus we only present findings from the model that includes both age groups. (The stratified analyses are available from the authors upon request). Analyses were performed using Stata 11.1 (StataCorp, 2009).

Results

Three percent of the sample received a SSD diagnosis at intake. Characteristics of children with and without a SSD are summarized in Table 1. Children with a SSD were significantly older than those without a SSD and were more likely to live in the Northeast or Midwest. Children with a SSD were also more likely to have three, or four or more comorbid Axis I disorders, more internalizing problems and greater functional impairment. A greater percentage of children with a SSD previously attempted suicide, had a history of psychiatric hospitalization, had a mental illness in their biological family and had received outpatient, school-based, day treatment and residential/inpatient services in the past year.

Table 1.  Characteristics of children with Schizophrenia Spectrum Disorders and non-Schizophrenia Spectrum Disorders
VariableSSD Diagnosis (N = 275)Non-SSD Diagnosis (N = 8731)p Value
n%n%
  1. aCaregiver report on CBCL bLifetime history cPast 12 months

  2. Percentages may not sum to 100% due to rounding. The p-values were generated from a logistic regression (for binary outcomes) or a linear regression (for continuous outcomes).

Demographic characteristics
 Age, mean (SD)12.94 (0.20)12.20 (0.03)<.001
 Males18065%587767%.559
Race/Ethnicity
 White11341%455752%.090
 African American8029%201323% 
 Hispanic5118%108512% 
 American Indian/Alaskan Native176%5556% 
 Other155%2206% 
Household income
 Less than $15,00012546%409947%.900
 $15,000-$24,9997025%200623% 
 $25,000-$49,9995219%197321% 
 $50,000 or greater2810%83310% 
Caregiver education level
 Up to high school graduate or GED11241%374843%.593
 Associates degree or some college10036%309835% 
 Bachelor’s degree or higher6323%188522% 
Eligible for Medicaid19069%614170%.691
Region of the country
 Northeast4817%125914%.040
 Midwest8531%221325% 
 South8129%304135% 
 West 6122%221825% 
Referral source
 Mental health agency8832%277032%.648
 School5420%166519% 
 Juvenile justice3613%121914% 
 Child welfare4416%116513% 
 Physical/substance use clinic52%1522% 
 Caregiver/youth186%86310% 
 Other3011%89710% 
Custody status
 Two parent family6625%219325%.757
 Mother12445%397045% 
 Father145%3864% 
 Adoptive/foster parents175%4615% 
 Grandparents146%5737% 
 Ward of state3011%7459% 
 Other104%4035% 
Clinical characteristics
Number of Axis I diagnoses
 One7427%355441%<.001
 Two9835%333738% 
 Three8330%164419% 
 Four or more207%1972% 
aInternalizing behavior problems, mean (SE)67.13 (0.74)64.91 (0.12).004
aExternalizing behavior problems, mean (SE)68.86 (0.68)69.41 (0.12).435
Child functional impairment
 Minimal155%7048%.004
 Moderate6423%251839% 
 Marked9233%297634% 
 Severe 10438%253229% 
Emotional/behavioral strengths
 Below average15255%489856%.827
 Average9434%294834% 
 Above average2910%88510% 
bPsychosocial factors
 Child risk factors
 Psychiatric hospitalization15255%269231%<.001
 Physical abuse7226%238227%.740
 Sexual abuse6223%205724%.732
 Previously run away9735%286933%.472
 Attempted suicide8531%144016%<.001
 Substance abuse5721%168419%.539
 Sexually abusive to others228%7268%.829
Family risk factors
 Domestic violence history12846%457352%.090
 Mental illness in biological family18166%510959%.039
 Parents ever in psychiatric hospital11542%365142%.985
 Parent convicted of a crime12947%436450%.342
 Substance abuse history in family17062%582067%.121
 Parental substance abuse treatment13951%455052%.755
cService use history
 Outpatient21478%630672%<.001
 School-based19270%538262%.025
 Day treatment6825%149317%.002
 Residential/inpatient14252%257529%<.001
 Substance use treatment249%89010%.493
Family characteristics
 Family resources, total score (SE)102.37 (1.39)103.35 (0.24).497
 Family functioning, mean (SE)14.07 (0.13)14.11 (0.03).723
 Caregiver strain, total score (SE)6.91 (1.28)6.40 (0.22).692

Results of the adjusted analyses are presented in Table 2. Increasing age (AOR=1.08, 95% CI: 1.01-1.14), African American race (AOR=1.71, 95% CI: 1.11-2.65) and Hispanic ethnicity (AOR=1.96, 95% CI: 1.31-2.94) were significantly associated with receiving a SSD diagnosis. Diagnosis of three (AOR=2.22, 95% CI: 1.49-3.31) and four or more (AOR=5.02, 95% CI: 2.77-9.09) comorbid diagnoses was also significantly associated with a SSD diagnosis. A higher internalizing score on the CBCL was associated with increased odds of SSD (AOR=1.02, 95% CI: 1.01-1.04), while a higher score on the externalizing subscale was associated with reduced odds of SSD (AOR=0.97, 95% CI: 0.96-0.99).

Table 2.  Multivariate logistic regression predicting schizophrenia spectrum disorders among children entering mental health services, n=9,006
CharacteristicAOR95% CI
  1. Abbreviations: AOR= adjusted odds ratio, 95% CI= 95% confidence interval. AOR in bold are significant at p<.05. 1Reference group is no.

Demographic
 Age1.081.01, 1.14
 Female (reference: male)1.120.81, 1.55
 Race/ethnicity (reference: Caucasian)
 African American1.711.11, 2.65
 Hispanic1.961.31, 2.94
 American Indian/Alaskan Native1.950.91, 4.17
 Other 1.180.57, 2.43
Income (reference: <$15
 $151.140.77, 1.68
 $250.870.57, 1.33
 $500.840.46, 1.55
Region (reference: Northeast)
 Midwest1.010.52, 1.95
 South1.000.53, 1.88
 West1.080.54, 2.18
Referral Source (reference: mental health agency)
 School1.370.91, 2.05
 Juvenile justice1.020.61, 1.71
 Child welfare1.300.79, 2.13
 Substance abuse clinic/provider1.090.34, 3.52
 Caregiver/self0.830.45, 1.55
 Other 1.360.82, 2.26
Custody status (reference: two parent family)
 Biological mother0.920.60, 1.42
 Biological father1.230.56, 2.72
 Adoptive parents1.300.70, 2.42
 Grandparents0.830.39, 1.77
 Ward of state1.100.55, 2.20
 Other0.840.36, 1.95
 Eligible for Medicaid10.920.64, 1.33
 Clinical characteristics
  Number of Diagnoses (reference: 1 diagnosis)
   2 Diagnoses1.340.95
   3 Diagnoses2.221.49
   4 or more diagnoses5.022.77
   CBCL Internalizing 1.021.01, 1.04
   CBCL Externalizing0.970.96, 0.99
Child functional impairment (reference: minimal)
 Moderate1.220.61, 2.45
 Marked1.420.69, 2.92
 Severe 1.760.82, 3.80
Psychosocial characteristics
 Previous psych hospitalization11.631.10, 2.40
 Past suicide attempt11.451.05, 2.02
 History of substance abuse11.050.69, 1.61
 Child sexually abused10.690.48, 0.99
 History of domestic violence10.780.57, 1.07
 Family history of mental illness11.410.99, 2.00
 Substance abuse in family10.800.59, 1.11
 Parent received substance abuse treatment10.970.62, 1.52
 Parents psych hospitalization11.040.76, 1.42
Service use history
 Day treatment past year10.990.71, 1.37
 Residential/inpatient treatment past year11.591.11, 2.28
 Substance use treatment past year10.580.32, 1.04
Family characteristic
 Caregiver strain1.000.99, 1.00

Children and adolescents with SSD had a 1.63 times increased odds (95% CI: 1.10-2.40) of previous psychiatric hospitalization, a 1.45 times increased odds (95% CI: 1.04-2.02) of a previous suicide attempt, a 1.59 times increased odds of receiving residential treatment in the past year (95% CI: 1.11-2.28), and a 0.69 times decreased odds of previous sexual abuse (95% CI: 0.48-0.99).

Discussion

This study assessed the prevalence and correlates of schizophrenia spectrum disorders in a large sample of children and adolescents referred to mental health systems-of-care programs. Three percent of youth in the program had a SSD. As expected, this is less than the 4-8% prevalence of psychotic symptoms in referred children (Biederman et al., 2004; Caplan & Tanguay, 1996); however, it is much greater than the 0.5% prevalence of SSD found among youth in Hawaii (Schiffman & Daleiden, 2006). The difference in SSD prevalence comparing our study with the Hawaii study may be due to a number of factors. Shiffman and Daleiden (2006) state that their approach to dealing with missing information and classifying cases into SSD and non-SSD groups may have led to a conservative prevalence estimate. Differences in demographic and geographic distributions of the samples likely also contributed to the difference in prevalence. Additional research is needed to assess the prevalence of SSD among children treated in U.S. community settings.

Thirty-one percent of children with a SSD had a history of attempting suicide. While this is consistent with research suggesting that adolescents with SSD are at risk for suicide (Falcone et al., 2010; Shoval, Zalsman, Apter, Diller, Sher & Weizman, 2007), the findings are concerning. Suicide is the third leading cause of death among youth 15 to 24 years old (Centers for Disease Control, 2007). The presence of a SSD and a co-occurring disorder, common in this study sample, also increases suicide risk (Falcone et al., 2010; Shoval et al., 2007). As part of routine intake and on-going assessment procedures in the systems-of-care program, clinicians may want to assess depression and suicidality, provide effective treatments for the underlying disorders, and have a process in place to support youth who are experiencing suicidal thoughts.

Co-occurring disorders were common in the sample. Over half of children with a non-SSD diagnosis and 73% of children with a SSD had at least one additional disorder. Other studies have found that 80-99% of children with SSD or psychotic symptoms also have co-occurring disorders, with ADHD, bipolar disorder and depression among the most common comorbidities (Biederman et al., 2004; Falcone et al., 2010). While the high comorbidity found among children with SSD in our sample may reflect difficulty in differential diagnosis, it also emphasizes the severity of mental illness experienced by these children and the complexity of care that is needed.

Higher scores on the internalizing subscale of the CBCL and lower scores on the externalizing subscale were associated with a SSD diagnosis. These results are consistent with the core features of psychotic disorders, which frequently involve anxiety, social withdrawal, and somatic complaints and less frequently include aggressive and delinquent behavior (McClellan & McCurry, 1999). However, prior researchers have found non-specific symptoms, including conduct disorders, to be present in youth with psychosis (Häfner & Nowotny, 1995).

Children with a history of sexual abuse had decreased odds of SSD. In other words, they were more likely to have a non-SSD psychiatric disorder. Perhaps this association reflects the type of disorders included in the comparison group. Child abuse is associated with mood and anxiety disorders (Heim, Shugart, Craighead, & Nemeroff, 2010), which represented 40% of the disorders among children in the non-SSD category.

In addition to the clinical and service use correlates of SSD, race and ethnicity were also associated with a SSD diagnosis. African American and Hispanic youth had increased odds of a SSD diagnosis. While race differences in SSD have been identified in adult populations (Minsky et al., 2003; Strakowski et al., 1996), few studies have investigated such differences among children. Our finding is consistent with elevated admissions rates for psychosis in Black adolescents in the United Kingdom (Tolmac & Hodes, 2004) and with the increased incidence of SSD found in African Americans adults in the U.S. (Bresnahan et al., 2007; Kendler, Gallagher, Abelson, & Kessler, 1996). These race and ethnicity differences should be interpreted carefully. African Americans and Hispanics are more likely to be misdiagnosed with schizophrenia (Jones & Gray, 1986; Mukherjee, Shukla, Woodle, Rosen, & Olarte, 1983) and it has been suggested that cultural biases of mental health practitioners, as well as greater presentation of psychotic symptoms among racial/ethnic minority groups with mood disorders (Adebimpe, Klein, & Fried, 1981), may contribute to differences in the prevalence and/or assignment of SSD diagnoses. Experiences associated with immigrating to a new country, such as stress, lack of social support and discrimination, are thought to contribute to the increased risk of schizophrenia found in first and second generation adult immigrants (Karlsen & Nazroo, 2002; Cantor-Graae & Selten, 2005; Weiser et al., 2008). While immigration status of the current study sample is unknown, it’s possible that some of the Hispanic youth are first or second generation immigrants and it is a contributing factor to the increased prevalence of SSD found in this group. Further investigation is needed to determine the causes of racial/ethnic differences in the prevalence of SSD diagnoses. Attention should be devoted to exploring the role of clinician cultural competence and other environmental and social factors associated with SSD diagnoses in minority youth.

Contrary to expectations, children with SSD and children with other psychiatric disorders had equivalent ratings of behavioral and emotional strengths, family resources, family functioning, and caregiver strain. Given the severity of schizophrenia spectrum disorders, we had thought that children with SSD and their families would experience greater strain than other children and families in the program. Both groups of children and their families entered into services with few strengths, severe impairment, and limited resources. This suggests that, irrespective of diagnosis, the systems-of-care programs are serving children and families with serious problems and these families may benefit not only from traditional mental health services, but also supportive services and resources that facilitate coping with a child who has a mental illness.

Limitations of the study should be noted. Misdiagnosis of schizophrenia in children is common (McKenna et al., 1994). Diagnostic misclassification in this sample is possible and information on the validity and reliability of the diagnoses was unavailable. The diagnoses should be viewed as reflecting common clinical practice in systems-of-care settings rather than research derived constructs. Also, we do not have information about the date of onset of mental illness in relation to enrollment into this study. The sample included youth referred to systems-of-care programs across the U.S. and does not reflect the general population of youth. While the results are not generalizable to youth in all community settings, the findings do provide an improved approximation to the general referred child population than studies drawn from specialized treatment settings.

Conclusion

Schizophrenia spectrum disorders are rare, but not uncommon in youth referred to mental health systems-of-care in the U.S. Enhanced awareness and early identification of SSD is important, as treatment during the initial phase of the illness may improve functional and clinical outcomes, including quality of life (McGlashan & Johannessen, 1996). Children with SSD may present with a distinct risk profile that could be used to inform procedures aimed at early identification of individuals, as well as to inform and target service planning and delivery. Further research examining these characteristics in a longitudinal sample is needed to elucidate these relationships.

Acknowledgement:

This work was supported by the following National Institute of Mental Health grants: 5R01MH075828, 5T32 MH014592, and 5T32MH019545. The authors would like to thank Dania Joseph for her assistance in reviewing the literature for a prior version of this manuscript. The authors have declared that they have no competing or potential conflicts of interest.

Ancillary