Diversity of outcomes among adolescent children of mothers with mental illness.

Children of parents with mental illness are an at-risk population according to research on psychiatric outcomes using White, middle-class samples of depressed parents and infants and preschool children. The current study expands this evidence by exploring within-group heterogeneity across psychosocial outcomes, in a racially diverse, low-income sample of adolescent children of mothers with mental illness (N = 166). Using measures of mental health, academics, behavior problems, and social relationships--and employing cluster analysis methodology--we identified five meaningful subgroups of these youth. Two of five identified clusters evidenced mental health symptoms (15%) or possible behavioral problems (27%). The largest cluster (30%) appeared quite socially and academically competent; another cluster (22%) presented as average in their functioning but adult-oriented. A final small cluster (4.8%) was distinguished by members' social isolation. Cluster membership related to maternal substance abuse history, father's relationship to youth, and social support available to mothers. Implications for planning preventive interventions are discussed.

**********

When parents have a serious mental illness, it significantly increases the risk of a psychiatric disorder in their children (Erlenmeyer-Kimling et al., 1997; Weissman, Leaf, Bruce, 1987). In this post-deinstitutionalization era, a sizeable number of children are likely to grow up with a mother who has a serious mental illness; many of these women are single-handedly carrying out parenting responsibilities (Nicholson, Nason, Calabresi, & Yando, 1999). Further, as chronic mental illnesses are episodic, with episodes lasting up to 2 years, children of these parents are likely to experience more than one episode of parental mental illness influencing more than one developmental phase (Hammen, 1997).

At the same time, developmental research has indicated that not all children who grow up in high-risk situations show adverse outcomes. The term resilience has been used to encompass the "class of phenomena involving successful adaptation in the context of significant threats to development" (Masten et al., 1999, p. 143). Parental mental illness, in fact, does not always produce psychological or other problems for offspring (Arbelle et al., 1997; Wals et al., 2001). Some researchers have identified subgroups of resilient children--those whose parents have mental illness but who are competent with no mental health problems (Garber & Little, 1999; Scherer, Melloh, Buyck, Anderson, & Foster, 1996). Garmezy's (1987, 1993) original theory and research on resilience hypothesized that children's competency relates to psychosocial resources available; whereas children growing up in adversity often have fewer "good" resources available to them. If reasonably adequate resources are present, competency outcomes are generally satisfactory, even in the context of chronic, severe stressors. In research on parents with mental illness and their children, most studies have involved infants or toddlers. Unfortunately, the least amount of research has focused on adolescent offspring of parents with mental illness (Oyserman, Mowbray, Allen-Meares, & Firminger, 2000), who experience more risk factors and also are more likely to have access to psychosocial resources.

MATERNAL PSYCHIATRIC VARIABLES AS PREDICTORS OF ADOLESCENT OUTCOMES

Research has focused on the link between parental depression and children's mental illness (Boyle & Pickles, 1997). Adolescent children of parents with affective disorders have higher rates of diagnosed mental illnesses than do children of nonmentally ill parents (Beardslee, Keller, Lavori, Staley, & Sacks, 1993; Grigoroiu-Servanescu, Christodorescu, Totoescu, & Jipescu, 1991). In fact, by early adulthood, children with a depressed parent have nearly a 50% chance of experiencing an episode of major depression themselves (Beardslee, Versage, & Gladstone, 1998; Warner, Weissman, Mufson, & Wickramaratne, 1999), compared to the rate in the general population of 5% to 7% (Jellinek & Snyder, 1998). Parental depression has also been linked to adolescent diagnoses of anxiety disorders and conduct disorder (Boyle & Pickles, 1997; Wickramaratne & Weissman, 1998).

Although research has focused on parents diagnosed with depression, other diagnoses are often considered more severe in terms of their impact on functioning (Oyserman et al., 2000). Thus, children of parents diagnosed with bipolar disorder or schizophrenia may be at even greater risk than are children of parents diagnosed with depression (Goodman & Brumley, 1990). In fact, children whose parents have a bipolar disorder are 2.7 times more likely than other children to develop a mental disorder (Lapalme, Hodgins, & LaRoche, 1997), are at risk for a diagnosis of attention-deficit/hyperactivity disorder (ADHD; Chang, Steiner, & Ketter, 2000), and are also more likely to display a personality disorder (Grigoroiu-Servanescu et al., 1991). Chronicity of the mother's mental illness has also been found to significantly affect parenting, across diagnoses (Oyserman el al., 2000).

Although much of the research on adolescent children of mentally ill parents has examined risks for mental health problems or psychiatric diagnosis, less attention has focused on the functioning of these adolescents in academic, social, or behavioral domains. The available research does indicate that parental depression is significantly related to adolescent problems in school (Billings & Moos, 1983; Hammen et al., 1987), lower grade-point averages (Tannenbaum & Forehand, 1994), problems with peer interactions (Billings & Moos, 1983), reduced social competence (Thomas, Forehand, & Neighbors, 1995), teen substance use (Su, Hoffmann, Gerstein, & Johnson, 1997), and higher levels of adolescent internalizing and externalizing behavior problems (Thomas et al., 1995). Academic or cognitive difficulties, as well as emotional--behavioral difficulties, have been reported for adolescent children of mothers with a bipolar disorder (Hammen et al., 1987), schizophrenia (Arbell et al., 1997), and other diagnoses (Rutter & Quinton, 1987).

CONTEXTUAL VARIABLES AS RISK FACTORS FOR ADOLESCENT OUTCOMES

Individuals with serious mental illnesses usually experience concurrent difficulties with health, income levels, relationships, and other important life domains. Thus, mental illness is likely to be only a small part of the total risks that mothers and their children experience. Risks often include family disruptions and conflicts, single-parent status, social isolation, and financial and other stresses from living in impoverished conditions. These difficult life circumstances are often concomitant with chronic long-term mental illness and thus potentiate and exacerbate risk. However, to date, empirical evidence is insufficient to parse out and assess differential contributions of maternal psychiatric variables versus context (Oyserman et al., 2000). Further, many potentially significant contextual variables that could improve parenting and child outcomes have not been fully explored (e.g., children's separations from mother due to her mental illness; the satisfaction women gain from parenting).

LIMITATIONS OF CURRENT RESEARCH KNOWLEDGE

Few studies have examined psychosocial outcomes for teenage children of parents with bipolar disorders or schizophrenia. Investigators have not systematically compared adolescent outcomes across major categories of parental diagnosis. Evidence of the effects of parents' mental illness on their adolescent children has concentrated primarily on parental depression. Even in that research, there are significant limitations--involving mainly White, middle-class parents. Few investigators have included proportional representation of minorities in their samples, which raises questions about the applicability of findings to more diverse groups. Furthermore, analyses have not simultaneously taken into account SES or race. In fact, many investigators do not specify their participations' race, ethnicity, or economic levels. These omissions are important because the literature on parenting has suggested that poverty can increase stress on parents, decreasing the quality of their parenting (Samaan, 1998), and that the relationship between minority status and parenting is complex (Oyserman, 2003). Studies are also limited in that most do not analyze for differences in the influence of maternal mental illness on girls versus boys, an important omission, given the established gender differences in trajectories for achievement and autonomy and the consequences of parental separation in adolescence. Finally, most researchers have concentrated on the effects of parental mental illness on children's mental health; much less research has examined its effects in academic, social, and behavioral domains. Risk of mental health problems may or may not relate to other important outcomes or to successful engagement with school, social skills, peer relations, or behavior problems in the community (Luthar, Cicchetti, & Becker, 2000). Some children who are at-risk may be functioning well in some domains but not in others, indicating the need to examine a broad constellation of positive and negative outcomes, strengths, and problems. Further, Radke-Yarrow and Klimes-Dougan (2002) concluded that research on offspring of depressed parents has inappropriately examined problem outcomes as either present or absent and that assessments should take into account the form of the problem and any associated co-occurring disorders.

Our aim in the current research was to study diverse outcomes for teenage children of mothers with a serious mental illness and to identify predictors of outcomes from mothers' clinical history, as well as contextual features reflecting stress and resources. Previous gaps in the research were addressed by including a substantial number of minority youth in our sample, using measures that assessed outcomes of adolescence in all major life domains (academic, social, and behavioral, as well as mental health), and controlling for race/ethnicity, age, and gender as predictors of outcomes. To understand these diverse outcomes across multiple domains, we used a cluster analysis to identify similar groups.

METHOD

Sample

Participants were 166 teenage children and their mothers, participating in National Institutes of Mental Health-funded studies of mothers with serious mental illness and their adolescent children (see Mowbray, Oyserman, & Bybee, 2000). In terms of maternal demographics, 60% were African American, 32% non-Hispanic White, 6% Latina, and 2% other racial/ethnic groups. At the time of the youth interview, mothers were 40.5 years of age on average (SD = 5.8, range = 26-56 years); the median family income was $1,200 per month, with 53% of the participants living below the poverty line: 23.6% of mothers were married; 44.2% were separated, divorced, or widowed; and 32.1% were single (never married); 34.5% had less than a high school diploma, 21.8% had a high school diploma or GED, and 43.7% had some college education or more. Mothers' diagnoses were 23.3% schizophrenia/schizoaffective disorder, 52.3% depression, and 25.4% bipolar disorder. Youth were, on average, 15.0 years old (SD = 2.04); 87 were boys, 79 girls. At time of study recruitment, mothers had care responsibilities for an adolescent youth included in this study.

Procedures

Mothers were recruited from the public mental health system in two counties in southeast Michigan (including Detroit) and asked to participate in a longitudinal study about parenting and mental illness. Eligible mothers had received public mental health services for a mental illness that lasted at least 1 year and interfered with one or more major areas of functioning. Originally, 485 women were identified as meeting study criteria: 46 could not be contacted or scheduled and 59 refused to be involved, producing a 78.4% participation rate. Mothers were interviewed three times over a 5 year period (approximately every 20 months), using a structured questionnaire. The retention rate, overall for the study, was 87.5%. At the third interview for mothers with adolescent children, we requested permission to contact their children about participation in a study of adolescent transitions. The youth selected were mothers' youngest child in the age range of 11 to 17 years. The youngest child was chosen to increase the likelihood that children had not left school: in which case, interpreting the outcome measures across participants would have proved more problematic. Mothers were told that participation was voluntary and confidential and assured that we would not share any information about her situation with the youth, including her identification as a person with mental illness or recipient of mental health services. Written informed consent (guardian) and assent (youth) were obtained before the questionnaire administration. Youth were reimbursed for participating ($20 at T1 and $25 at T2). Interviewers were women ages 20 to 45; all had prior experience with children, were university students, or had undergraduate degrees. They received extensive 3-day interviewing training.

In total, 166 guardians (159 mothers and 7 other relatives) and youth consented and 39 refused (81%). The structured interviews with the youth were approximately 2 hours in length and took place in the privacy of participants' homes. Interviewers provided a snack and a break midway through the questionnaire. Participating …