Youth Depression in the Family Context: Familial Risk Factors and Models of Treatment

Abstract

Research on parent risk factors, family environment, and familial involvement in the treatment of depression in children and adolescents is integrated, providing an update to prior reviews on the topic. First, the psychosocial parent and family factors associated with youth depression are examined. The literature indicates that a broad array of parent and family factors is associated with youth risk for depression, ranging from parental pathology to parental cognitive style to family emotional climate. Next, treatment approaches for youth depression that have been empirically tested are described and then summarized in terms of their level of parent inclusion, including cognitive–behavioral therapy, interpersonal therapy, and family systems approaches. Families have mostly not been incorporated into clinical treatment research with depressed adolescents, with only 32% of treatments including parents in treatment in any capacity. Nonetheless, the overall effectiveness of treatments that involve children and adolescents exclusively is very similar to that of treatments that include parents as agents or facilitators of change. The article concludes with a discussion of the implications of these findings and directions for further research.

Developmentally, youth are known to have close ties to their families of origin (Diamond, Siqueland, & Diamond, 2003; Steinberg, 2005). As a result, there is a general belief in the field of psychology that it is important to include parents and families in the treatment of depressed youth (Clark, Beck, & Alford, 1999; Curry, 2001) however, empirical guidelines for how to do so are lacking. Consensus about the need for, and how to measure the effects of, parental involvement in treatment is elusive (Goodman & Gotlib, 1999). Still, achieving a better understanding of the multidimensional nature of risks for depression and how parent involvement in treatment relates to those risks are the priorities for the area of youth depression.

Recently, a national group of experts was called to summarize and prioritize the state of research on risk and treatment for youth depression. Specifically, the National Institutes of Mental Health (NIMH) invited a workgroup of leaders in the field of depression to clarify the state of information about the risks for and treatment of depression, and to make recommendations on where the field needs to focus research endeavors (Costello et al., 2002). Regarding risks for depression, the workgroup recommended further research into both family environment and genetics. As summarized by Costello and the NIMH workgroup, multiple pathways to depression are evident. The workgroup emphasized the multidimensional nature of depression, along with the paucity of research into the early pathways and interactions that cause depression. Although research has not identified all of the parent or family environment factors that contribute to youth risk for depression and the mechanisms through which they operate, it is becoming clear that “there is a considerable familial component to unipolar depression” (Costello et al., 2002; p. 533). The workgroup focus on both environment and biology is a clear indication that research in psychosocial parent and family factors as well as biological pathways of risk are important.

The current state of the literature reflects a need to review and synthesize literature on parent and family risk factors, along with how they have been addressed in treatment research related to youth depression. This review will address that need by offering a synthesis of the psychosocial family risk factors associated with youth depression, followed by an integration of available treatment outcome studies and their inclusion or exclusion of parent components, concluding with points of consensus for clinicians and specific suggestions for areas for further research to advance our knowledge of parent and family factors in youth depression.

PARENT AND FAMILY RISK FACTORS FOR YOUTH DEPRESSION

Parent Factors

Family psychiatric history and parental depression, primarily maternal depression (Birmaher et al., 1996; Burge & Hammen, 1991; Downey & Coyne, 1990), have been associated with a child’s risk for developing depression (Beardslee et al., 1996; Weissman & Jensen, 2002). Effects can span three generations, from grandparent to parent to child (Warner, Weissman, Mufson, & Wickramaratne, 1999). Daughters are at greater risk for depression than sons, but even sons appear to have lower educational achievement if their mothers were depressed during their childhood (Ensminger, Hanson, Riley, & Juon, 2003). The understanding of parent-to-child risk has evolved over time, moving from a focus on maternal depression and linear links to more complex interactive models including maternal, paternal, and other family factors (reviewed in Downey & Coyne, 1990; Goodman & Gotlib, 1999). Measuring traits of multiple family members in an interactive risk equation reflects current research design in the field. Consideration of characteristics of youth as they interact with parent depression is also important (Jaser et al., 2005).

Parental Depression

In one of the early studies of parental and youth depression, Fendrich, Warner, and Weissman (1990) studied children and adolescents with and without a depressed parent. They examined several psychosocial factors, and concluded that parent depression was the most important risk factor for several types of youth psychopathology, including depression. Parent depression was more important than family discord, low cohesion, and “affectionless control” (high overprotection combined with low warmth). This kind of a multidimensional approach to studying parent–child factors in depression—including measures of parental psychopathology and family climate simultaneously—has now become standard practice, although other authors have revised the emphasis on a single factor (parent pathology) in favor of approaches that incorporate combinations of parental or family risk factors (e.g., Miller, Warner, Wickwamaratne, & Weissman, 1999).

In a comprehensive and widely cited review by Goodman and Gotlib (1999), four mechanisms that may explain the complex transmission between mother and youth depression were presented: (1) genetics, (2) neuroregulation difficulties that impact affect regulation, (3) exposure to negative maternal affect and behaviors, and (4) stress and the environmental context within which the youth lives. Since the publication of this seminal article, several other studies that fit within Goodman and Gotlib’s framework within the psychosocial mechanisms of this transmission (mechanisms 3 and 4, above) have been published and are included in this review.

Maternal Factors in Youth Depression

Although a growing number of studies exist that include mothers and fathers in the research design, the literature is skewed toward an emphasis on mothers. Historically, negative interactions between mothers and children and the presence of depression in mothers have been viewed as primary risk factors in youth depression (Kane & Garber, 2004). The impact of these factors is now being researched within a larger context. For example, in a 10-year follow-up investigation of depressed and non-depressed mothers and their children, Miller et al. (1999) reported that maternal depression alone did not place children at risk for depression. Instead, risk for childhood depression was related to a combination of factors, including maternal depression, low maternal emotional availability, and high maternal control, along with low self-esteem in the children. This study supported the value of adopting a multidimensional approach in understanding parent depression as a risk factor in child depression.

A similar multidimensional approach guided a longitudinal study of families recruited near the time of their child’s birth, who were followed through the child’s adolescence (Hammen, Shih, Altmann, & Brennan, 2003). Consistent with Goodman and Gotlib’s fourth proposed mechanism of parent child transmission (stress and the environmental context), Hammen and her colleagues included measures of environmental stress and interpersonal support. Maternal history of depression appeared to relate to greater symptoms of depression and poorer adaptive interpersonal functioning in adolescence (Hammen et al., 2003). In contrast, adolescents with depressive symptoms and non-depressed mothers were more often experiencing stress related to episodic environmental events, and did not display the same chronic interpersonal deficits that the adolescents with depressed mothers displayed. Youth with depressive symptoms and non-depressed mothers also had more social and interpersonal strengths. Hammen and her colleagues’ investigation highlighted two important factors; the first is that different forms of adolescent depression may have more or less interpersonal impairment associated with them, and the second is that awareness of the family history or parent (maternal in this case) depression can possibly inform treatment emphasis (Hammen et al., 2003).

In one of the few studies that examined maternal depression as a moderator of youth risk, Brennan, LeBrocque, and Hammen (2003) assessed 816 adolescents and found that maternal depression interacted with parental psychological control, maternal warmth, and involvement to buffer youth from depression. Specifically, youth with depressed mothers were more likely to be resilient to depression if mothers exhibited low psychological control, high warmth, and low over-involvement. In most cases, the parenting qualities had the same direction of effect but less strength for children of non-depressed mothers, compared to children of depressed mothers. An important implication of the study is that incorporating parenting behaviors into case conceptualization may be as important as including parental psychopathology history (Brennan et al., 2003). Depressed mothers experience parenting difficulties, but those difficulties can arise from many sources, depression being but one source (Downey & Coyne, 1990).

Paternal Factors in Youth Depression

In the most recent decade of research, more attention has been focused on fathers, shifting from the earlier focus on mothers alone. In a study conducted by Sanford et al. (1995), parent variables were studied in relation to the course of adolescent depression over a 1-year period, with family relationships measured from a semi-structured interview about social and academic functioning. Results suggested that paternal and maternal variables contributed in separate ways to the adolescent’s depression status at 1-year follow-up. According to Sanford and his colleagues, less positive involvement of fathers and higher conflict with fathers distinguished adolescents who had persistent depressive symptoms from those with remitted depressive symptoms at follow-up. Youth who had persistent depression at follow-up were also less likely to be responsive to both paternal and maternal discipline (Sanford et al., 1995).

Kane and Garber (2004) recently conducted an important meta-analysis of paternal factors in youth psychopathology. Their conclusions highlighted the role of fathers’ psychopathology and father–child conflict in relation to youth depression, moving beyond the traditional focus on the role of mothers. Fathers’ depression was related to both youth depression and greater family conflict levels. Kane and Garber (2004) noted that it is important to consider the unique impact of paternal depression on youth risk for depression, which may consist of biological and psychosocial components. Mezulis, Hyde, and Clark (2004) demonstrated similar interest in examining father factors in the development of emotional and behavior problems in early childhood, in joint consideration with maternal depression. Mezulis et al. (2004) reported that fathers become particularly important in child outcomes when mothers are depressed. Namely, fathers’ depression exacerbated risk for child internalizing problems in the families with a depressed mother (Mezulis et al., 2004), even when time spent with the child was controlled. Clearly, there could be a greater biological risk for those youth.

Paternal warmth has also been investigated as a risk for youth depression at different developmental stages. In a study conducted with college students, Alloy et al. (2001) examined depressive cognitive vulnerability and risk for depression over a 2.5-year period. Mothers and fathers of participants were included in the study. The emotional warmth and acceptance from fathers predicted which participants eventually developed depression (Alloy et al., 2001). However, results may not apply to very young children, when depression is relatively uncommon. For example, in one study, the relationship between paternal warmth and internalizing symptoms in kindergarten-age children did not support the protective effect of paternal warmth (Mezulis et al., 2004). In a retrospective study with adults, Jorm, Dear, Rodgers, and Christensen (2003) reported results that may shed light on the inconsistency of how paternal warmth relates to youth depression across different studies. Adults provided retrospective accounts of their family experiences, as well as measures of emotional functioning. Participants who reported higher levels of warmth from fathers compared to mothers indicated that they experienced high levels of emotional distress and family conflict. The participants who reported high levels of maternal warmth or equivalent mother and father warmth reported low levels of emotional distress and family conflict. Paternal warmth may not be a buffer for youth depression when maternal warmth is low, particularly when paternal warmth is greater than maternal warmth (Jorm et al., 2003). Clearly, fathers are important, even if there is not a clear consensus on how fathers contribute within an interactive set of risks for the development of youth depression.

Parental Cognitive Style

In addition to parent pathology, parent involvement, and parent warmth, the risk of depression increases with maladaptive cognitive styles, such as negative attributions (bad things happen due to internal, stable, and global reasons; e.g., it’s my fault, it’s always my fault, and things will not change) (Alloy et al., 2001) or negative views of the self, world, and future (Stark, Schmidt, & Joiner, 1996). Parents’ influence on their child’s cognitive style is another area of potential transmission of risk. Stark et al. (1996) assessed youth depression, parents’ cognitive style, and youth cognitive style in 133 youth. Results of that study indicated that parents’ view of self, world, and future was related to their child’s view of self, world, and future, and that children’s negative views of self, world, and future were related to the severity of the youth’s depressive symptoms. The findings were more powerful in same-gender parent–child pairs, particularly the relationship of mothers’ cognitive style to daughters’ cognitive style and risk for depression. More recently, in a prospective investigation, researchers investigated cognitive style related to older adolescents (college students) and their families of origin. Alloy et al. (2001) examined parental cognitive style, perceived parenting practices, college-age adolescents’ cognitive style, and the adolescents’ development of depressive symptoms over a 2.5-year period. Youth depression vulnerability was related to the cognitive style and parenting dimensions of both their mothers and fathers. A stronger association was found between the cognitive styles of youth and their mothers (compared to their fathers), but emotional warmth from fathers was also significant. However, the specific mechanisms or reasons why this might be the case, particularly with the parent gender effect, remain unclear. In general, these results call for a greater understanding of the role of parents’ inferential messages and attributions of causality that they convey to their children, including the impact of mothers and fathers separately, in research and treatment studies.

Family Environment and Parenting Practices

Family Climate

Consistent with a multidimensional approach to parental depression in child risk studies, researchers have studied family environment and additional parent traits related to youth depression. Family emotional climate is another contributing factor in child depression. In a study by Nomura, Wickramaratne, Warner, and Weissman (2002), family relationship variables of high discord, low cohesion, and high affectionless control were all important predictors of general child pathology, including depression. Not surprisingly, parental depression was a strong predictor of child affective disorders of depression and anxiety. However, affectionless control was highly predictive of depression in youth of non-depressed parents. Family discord and low cohesion were also important predictors of child pathology, but not of child depression; those factors instead predicted substance use in youth. Overall, the results highlighted that parent pathology is important, but so is the broader family environment in predicting youth depression (Normura et al., 2002).

There is evidence in the literature suggesting that negative family interactions contribute to youth depression (Goodman & Gotlib, 1999; Sheeber & Sorenson, 1998), with family support and conflict being two particularly important factors. Stice, Ragan, and Randall (2004), for example, reported that youth perception of low parental support was related to adolescent depression. Several studies have included measures of family conflict, and found it to predict risk for depression among youth (Kane & Garber, 2004; Marmorstein & Iacono, 2004). Family conflict has been found to be a risk factor for youth depression in both middle-class Anglo-American (white) and low-income urban African-American samples (Sagrestano, Paikoff, Holmbeck, & Fendrich, 2003).

The relationship between family environment and youth depression is not necessarily direct, and may instead be mediated by other factors. Some researchers have been examining youth coping to negative family environment or parent depression as important factors in predicting youth outcome (Compas, Conner-Smith, & Jaser, 2004; Langrock, Compas, Keller, Merchant, & Copeland, 2002). Using a coping style approach, Jaser et al. (2005) examined youth risk and parent depression, finding that youth’s style of coping with parental intrusiveness and negativity mediated the relationship between a negative family environment and youth depression among children with depressed parents. Youth who employed an active coping style designed to halt negative thinking, such as using positive thinking, distraction, and restructuring, appeared to adapt better to living with depressed parents. Similarly, youth who responded to the negative environment with less active means appeared less well-adjusted, and they experienced more ruminative thinking, undesired emotional or physiological arousal, and intrusive thoughts (Jaser et al., 2005).

Specific Environmental or Situational Factors

In addition to parental depression and family climate, several additional factors have been studied as risks for depression. First, certain parenting practices, such as harsh parental discipline, served to increase risk for depression, whereas use of parental monitoring and inductive reasoning practices reduced child risk of depression (Kim & Ge, 2000).

Applying a methodology that included genetic risks and parent traits, Eley et al. (2004) examined characteristics of parents and family factors, such as body mass index, family stress, and education level of parents. Each of these variables—more overweight parents, higher levels of family stress, and lower educational achievement by parents—predicted depressive symptoms in adolescents 1 year later. The authors noted that the trait of high body mass index (overweight) parents could be related to a common third factor predicting depressive symptoms in adolescents (lack of exercise or poor diet, for example). If such risks are modifiable by changes in lifestyle, parent coping, and parent cognitions, then including treatment strategies to address some of these variables could reduce risk for depression in the adolescents (Eley et al., 2004). However, another study that included a large sample of predominantly African-American participants, followed over 25 years time, produced a different set of findings. Ensminger and her colleagues (2003) studied educational achievement, maternal depression, and depression in youth, following children from first grade until adulthood. They noted a clear relationship between maternal education, poverty, and daughters’ depression, but maternal depression remained the strongest predictor (Ensminger et al., 2003). Another study examining income and depression, based on a naturalistic experiment resulting from the opening of a tribal casino, found that moving out of poverty did not change rates of depressive symptoms among American Indian children, although other psychiatric symptoms, most notably conduct and oppositional defiant disorder symptoms were decreased (Costello, Compton, Keeler, & Angold, 2003). Quite obviously, the role of factors such as parental education level, poverty, parental obesity, and stress in creating risk for depression, warrants further study.

Attachment

As Bowlby (1980) originally described attachment, comfort in times of stress is important; the function of attachment behaviors is to seek safety from caregivers in times of stress. A lack of felt security would lead to increased distress (Bowlby, 1980). Research in the area of youth depression supports Bowlby’s original tenets. Shirk, Gudmundsen, and Burwell (2005) assessed 168 young adolescents, and their perceptions of maternal availability and youth support-seeking. According to Shirk et al. (2005), in times of higher stress, mothers’ availability was important as a buffer for depression in youth. In low-stress times, maternal availability was not as clearly related to risk for depression. Shirk et al.’s (2005) findings appear consistent with the idea that if adolescents feel increased stress and perceive their caregivers to be unavailable, they may be more vulnerable to emotional distress and depression.

One classic framework for explaining risk for depression, incorporating attachment and relational experiences with cognitive interpretations for the individual, is the interactional view of depression put forth by Coyne (1976). In this interactional approach, excessive reassurance-seeking is coupled with rejection by others, resulting in an increasingly negative interpersonal experience, increased reassurance-seeking, and an unpleasant ongoing cycle and greater depression. A clear model with data to support an interactional model of depression remains unavailable (Coyne, 1999), but several studies have noted links between perceptions of parental perceptions and emotional availability, theoretically related to attachment, and depression (Armsden, McCauley, Greenberg, Burke, & Mitchell, 1990; Burge et al., 1997; Cole-Detke & Kobak, 1996; Rudolph, Hammen, & Burge, 1997). In a recently published article, Abela et al. (2005) reported that severity of depressive symptoms was related to perceptions of parenting that are consistent with insecure attachment, but the relationship was moderated by the youth’s excessive reassurance seeking. Attachment did not predict depression alone, but the combination of insecure attachment and excessive reassurance seeking predicted risk of more severe depression and prior history of depression in youth (Abela et al., 2005).

Interactions of Parent Psychopathology with Child Temperament

Temperament has also been studied in relation to child depression and parent factors (Mufson, Nomura, & Warner, 2002). Parent psychopathology, together with child temperament variables of low adaptability, high irritability (negative emotionality), and increased activity, influenced eventual child diagnosis of depression (Mufson et al., 2002). There was not a single parent-pathology or child-temperament predictor for youth depression and the authors highlighted the fit between parent psychopathology and child temperament to understand youth risk for depression. Although not specifically related to depression, in a study of divorce effects on children, Lengua, Wolchik, Sandler, and West (2000) also addressed the importance of considering temperament in youth, together with parenting behaviors, when predicting symptoms of emotional distress. Based on a sample of 231 families, children with temperament characterized by low positive emotionality coupled with parental rejection predicted risk for depression. Also, youth with greater impulsivity experienced greater symptoms of emotional distress when their parents exhibited inconsistent discipline. These findings highlight the interaction and fit between parenting and youth temperament to explain risk for youth distress, particularly in times of family stress.

Summary of Parent and Family Risk Factors

To summarize, the literature indicates that there are several parent and family risk factors associated with youth depression. They are broad in scope, including parental cognitions, parental pathology, parenting behaviors of warmth and emotional availability, individual coping with the family environment, and family conflict. Results clearly indicate that mothers and fathers, for different reasons, are important to consider when conceptualizing youth risk, and possibly treatment, for depression. In addition, outcomes beyond depression in youth, such as educational achievement, are important for future research (Ensminger et al., 2003).

TREATMENT APPROACHES

Available treatments used in intervention studies for youth depression generally fit into one of the following categories: cognitive–behavioral therapy, interpersonal psychotherapy, and family systems approaches. Each treatment category represents diverse guiding assumptions, and there is considerable variation with regard to intervention techniques used within some of these categories; however, the general treatment approaches described here cover the current empirical base in the literature with depressed youth. This review does not provide an in-depth discussion of all treatment models, but instead gives an overview of the basic goals and tenets of these therapy approaches which form the foundation of our understanding of treatment efficacy. It is noteworthy that each general approach addresses one or more of the risk areas mentioned in the previous section.

Cognitive–Behavioral Therapy

Within the cognitive–behavioral framework, there are a few different approaches. For example, “cognitive therapy” (Beck, 1995; Beck, Rush, Shaw, and Emery, 1979; Clarke et al., 1999) places secondary emphasis on emotions, behaviors, and the reality of the environment, and primarily hinges on the individual’s perceptions and resultant cognitions (Coyne & Gotlib, 1983). Essentially, situations are presumed to activate underlying thoughts, assumptions, and information processing. The thinking patterns directly influence the emotional reaction and behaviors in a chain of connected experiences. According to Beck (1995), depressed persons maintain a negative schema of themselves, and seek information from that distorted point of view within their surroundings and interactions. The bias in information processing, in turn, serves to filter out schema-inconsistent information. Clarke and his colleagues (1999) theorized that early childhood experiences contributed to the development of dysfunctional self–other schemas, causing an interpersonally based vulnerability to depression, which included cognitive and relational components. Stark, Humphrey, Laurent, Livingston, and Christopher (1993) reported that parents communicate negative messages about the child’s self, world, and future to the child in subtle ways, through interactions and comments, which predisposes a child to adopt those views.

There is also a more behavioral approach within the cognitive–behavioral framework. Lewinsohn’s classic behavioral model (Lewinsohn, Youngren, & Grosscup, 1979) emphasized the role of “response contingent positive reinforcement” (p. 297) in depression. Namely, depressed persons do not engage in pleasant activities, and do not receive reinforcement for behaviors, even if a reinforcer is available. In an important study by Cole and Rehm (1986), a behavioral and self-control theory-driven model illuminated the role of parental response to child behaviors and depression. Cole and Rehm (1986) reported that depressed children received less frequent rewards and their mothers held higher standards for praise than non-depressed children. Lewinsohn’s CBT program for depressed youth, “Coping with Depression,” incorporates the behavioral aspects of pleasant events and reinforcement, while also addressing negative cognitions and interpersonal skills deficits (Lewinsohn, Clarke, Hops, & Andrews, 1990).

Concepts from traditional cognitive–behavioral approaches have been incorporated into cognitive–behavioral family therapy (CBFT; Dattilio, 2001). In CBFT, family schemas would contain information about how parents and children interact, relating back to the parents’ family of origin and experiences. Family schemas would influence how parents perceive and interact with their children and families, shaping their children’s family schemas in turn (Dattilio, 2001). CBFT would address maladaptive family schemas, as opposed to the focus on individual schemas in traditional cognitive–behavioral therapy.

Interpersonal Psychotherapy for Adolescents

Interpersonal psychotherapy for adolescents (IPT-A; Mufson, Moreau, Weissman, & Klerman, 1993) is a well-defined, manualized treatment. The guiding theory and goals of this approach are more uniform than cognitive–behavioral and family systems approaches. Mufson and her colleagues developed IPT-A based on assumptions that depression is maintained within an environment characterized by interpersonal and social functioning difficulties. IPT-A does not assume that relationships or any specific risk causes youth depression, but that recovery can be facilitated once the interpersonal relationships are reconstructed to provide better social support. Treatment emphasis is on rebuilding relationships and resolving interpersonal conflicts. Reminiscent of Bowlby’s work (1980) on attachment, depression is seen as a natural consequence of the loss of social connectedness, attachment disruptions, or difficulties establishing emotional autonomy in relation to primary caregivers. Therefore, the unique emphasis of IPT-A is to develop skills that will help the depressed adolescent regain the sense of attachment and improve the quality of social relationships (Mufson et al., 1993).

Family Systems

Gotlib (1990) summarized a systemic interpersonal approach to understanding depression. From this point of view, the family is a system that seeks to maintain its current functioning, or homeostasis. The depressive symptoms would either be related to that family member’s role in maintaining the homeostasis of the family system, or a sign that the current family system is not adequately meeting the needs of the family (Leslie, 1988). Classic family systems authors include Haley (1990), Minuchin (Minuchin & Fishman, 1981), and Bowen (see Becvar & Becvar, 1996). The approaches used by each school of family systems therapy vary somewhat (Cotrell & Boston, 2002). However, the general goal across systemic therapy approaches is to alter patterns of communication and behavior, to assist all family members in meeting their needs and the family’s goals without psychologically distressing symptoms. A common assumption underlying systemic therapy is that depression develops to serve some function within that system.

There are two specific types of family systems approaches that have specifically been evaluated with depressed youth. One is the systemic-behavioral family therapy (SBFT; Brent et al., 1997), and the other is the attachment-based family therapy (ABFT; Diamond, Reis, Diamond, Siqueland, & Isaacs, 2002; Diamond et al., 2003). The components of SBFT include principles from systemic family therapy, functional family therapy, and behavioral therapy. As described by Cotrell and Boston (2002), SBFT includes skill-building in communication, incorporating reinforcement, and distinct aspects of cognitive restructuring.

Attachment-based family therapy incorporates family systems and attachment theory in a manualized treatment for depression in youth, based on the premise that a “negative family environment inhibits children from developing the internal and interpersonal coping skills needed to buffer against the family, social, and community stressors that can cause or exacerbate depression (Diamond et al., 2003; p. 1191).” Diamond et al. (2003) designed ABFT to assist the adolescent in the task of developing autonomy from the family, while simultaneously maintaining healthy communication and positive relationships in the family as a whole. Treatment progresses through two stages. First, the therapist assists the youth in identifying and confronting troubled family relationships. Next, the goal is to encourage developmentally appropriate and parent-supported autonomy for the youth, working with the family as a unit (Diamond et al., 2003).

TREATMENT STUDIES AND PARENT INVOLVEMENT

In general, the treatment outcome literature for child depression lags significantly behind that of other child disorders (Compton et al., 2004) and treatments may be somewhat less effective (Weisz, McCarty, & Valeri, in press). For example, although family-based therapies have been shown to be effective for treating other child and adolescent disorders, including conduct disorder, substance abuse, and schizophrenia, there is a paucity of family-focused treatments for depression (Diamond, Serrano, Dickey, & Sonis, 1996). This discrepancy is surprising in light of “increasing evidence that interpersonal factors, particularly family relations, play a critical role in the development and maintenance of child and adolescent depression” (Diamond et al., 1996). Despite an increased focus in recent years, family interventions are still relatively sparse in the literature to date, and the level of involvement of families is very diverse within the research base. At the time of a review by Birmaher et al. (1996), only one study, conducted by a team led by Lewinsohn (Lewinsohn et al., 1990), included families in the intervention for depressed youth. Since that time, the literature appears to continue to cite family difficulties as part of the risk, with eight additional treatment studies including parents in treatment in some capacity.

Levels of Parental Involvement

Treatments vary dramatically in the ways they incorporate parent involvement. An operational definition of parental involvement provides a clear way to communicate about research in the field (see following for definitions of various levels of parental involvement within treatment). To further quantify the extent to which parent involvement has been incorporated into clinical trials targeting depressive symptoms among youth, data from a recent meta-analysis of outcome studies (Weisz et al., in press) were regrouped and analyzed.

Weisz and his colleagues applied rigorous analytic methods to the largest set of studies on youth depression to date, including dissertations. To be included in the meta-analysis, a study had to meet the following criteria: (a) participants selected because of elevated levels of depressive symptoms, formal diagnosis of major depressive disorder, dysthymic disorder, or research diagnostic criteria diagnoses of minor or intermittent depression; (b) random assignment of participants to at least one active treatment group and at least one untreated, waitlist, minimally treated, or active placebo control group; (c) participants of mean age younger than 19 and (d) intervention intended by the investigators to target depressive symptoms or disorder (for a full description of the methodology, see Weisz et al., in press). Although case studies and within group pre-to post-comparisons have been useful in developing treatments, such studies do not demonstrate efficacy and were not included in the meta-analysis. The final sample included 35 separate studies comprising 44 treatments.

To facilitate a synthesis of effectiveness for each level of family involvement, effect sizes were calculated for each treatment following Smith and colleagues (1980) and averaged within each category. The mean effect size and the number of treatments with an effect size greater than .50 for each level of family involvement are reported in the next section. Benchmarks for effect size values have been estimated with .20 indicating a small effect, .50 indicating a medium effect, and .80 indicating a large effect (Cohen, 1988).

Exclusive Adolescent Treatment

The vast majority (68%) of treatments for adolescent depression have focused exclusively on adolescents without parental involvement. These treatments include: group based therapies (CBT, cognitive restructuring, relaxation, group support), and individual cognitive behavior therapy. These adolescent-only treatments have been tested in a range of different settings including schools and school-based clinics, outpatient clinics, day treatment and juvenile detention settings, and they have included both clinically depressed and subclinical populations. The mean weighted effect size across treatments that were delivered exclusively to the adolescents was .45, and 14 of the 30 treatment studies generated at least “medium” effect sizes.

Exclusive Parent Treatment

There are many studies of adult treatment of depression, but few of them have directly assessed the children as parents received treatment. Strategies using parent-only treatment were not included in the Weisz et al. (in press) meta-analysis. However, the research question deserves attention, so the few available studies are included in this review. Sanford et al. (2003) reported the design of a parent-education intervention for families with a depressed parent. The children were assessed for depression in that study, but data about effects were inconclusive due to the small sample size. The design itself is promising for future research, however. An ongoing study by McCauley, Garber, Diamond, and Schloredt (2005) is finding that positive changes in the level of parental depression as a result of treatment delivered to one parent are associated with positive changes in family environment and child psychopathology, at least by parent report. Child-reported changes in psychopathology appear to depend upon the degree of change in parental depression over course of treatment. In cases where the parent experiences large decreases in depressive symptoms, children also report lower levels of externalizing and depressive symptoms (McCauley et al., 2005).

Parents as Partner

A handful of the treatment outcome studies (18%) available from the meta-analysis conducted by Weisz et al. (in press) have educated parents about depression and the goals of treatment (parents as partners), with frequency of parental inclusion ranging from one session to “as needed” to monthly family educational sessions. These eight treatments were cognitive, cognitive–behavioral, or interpersonal in orientation, and half utilized symptomatic as opposed to diagnosed samples of adolescents. On average, treatments with parent partners yielded a small positive benefit (weighted mean ES = .25), with two of these eight treatments generating an effect size greater than .50, including one interpersonal and one CBT trial (Mufson, Weissman, Moreau, & Garfinkel, 1999; Stark, 1990).

Parents as Agents of Change

The smallest number of treatment outcome studies included in the meta-analysis (n = 5, 11%) examined parents as agents of change, defined as explicitly including parents in the treatment process. In each of these cases, the parent received a similar or equal dose of treatment as did the youth themselves, ranging from 8 to 14 hr of intervention. Of the five treatments, four of which were tested with diagnosed youth, two were a parent addition to a standard CBT curriculum, one was parental inclusion in all CBT sessions, and two were family therapies. These treatments generated a weighted mean ES of .40, which is strikingly similar to the estimate of ES found for adolescent-only treatment. Two of the treatments generated effect sizes larger than .50, including one CBT package and one of the family therapies.

Descriptions of Treatments that Included Family Components

An example of the parents as therapy partners concept was described in a pilot study involving youth with elevated depression by Asarnow, Scott, and Mintz (2002). In addition to a nine session CBT group (“Stress Busters”) that the youth participated in without their parents, there was one multi-family meeting in which the participants showed their families a videotape demonstrating their use of CBT skills. The purpose of this session was to educate parents and to encourage them to support the learning that occurred in group sessions. The overall treatment package yielded a small, positive effect (ES = 0.25), and 94% of the parents rated the intervention as helpful. Sixty percent of the parents felt that the brief family session was adequate to their needs, whereas the other 40% felt more extended family sessions would have been helpful. This intervention presents a creative way to involve adolescents in educating their families about treatment. It also provides useful data on the extent to which families wish to be involved in treatment, although the specific benefits of the family education component above and beyond the nine CBT sessions remain unassessed at this time. Another model of parent involvement in CBT entails monthly family meetings lasting between 1 and 1.5 hr during which parents are taught how to encourage their child to use his or her new CBT skills and to engage in more pleasant family activities (Stark, 1990).

IPT includes parents-as-partners in therapy to facilitate achievement of the treatment goals (Mufson, Gallagher, Dorta, & Young, 2004). Parents are included in an initial visit that provides psychoeducation about depression and explains the IPT treatment model. They are asked to give input on the history of the problem, to review the treatment parameters, and to commit to support the adolescent’s participation in the intervention. Parents are also brought in as needed in the middle phase of treatment to support the adolescent’s attempts at new ways to communicate and problem-solving interpersonal situations. The goal is to help facilitate a change in the quality and style of their interactions at home if that is the identified problem area, or to support the adolescent’s work in another identified area. Finally, parents are included in a session at the end of treatment to discuss current status of symptoms, the experience of treatment itself, and to evaluate if more treatment is needed. The number of sessions including parents is purposefully limited in the IPT model “because clinical experience suggests that adolescents frequently prefer to work individually. They are more resistant to significant parental involvement, such as in family therapy, as often are parents” (Mufson et al., 2004).

Examples of using parents as agents of change include two studies that have evaluated a cognitive–behavioral intervention, the Coping with Depression course, in comparison to the same course with the addition of a parent group (Clarke, Rohde, Lewinsohn, Hops, & Seeley, 1999; Lewinsohn et al., 1990). Parent sessions focused on learning the same skills and techniques taught in the adolescent group, including cognitive restructuring, communication, problem-solving skills, and conflict resolution in an effort to “promote parental acceptance and reinforcement of the expected positive changes in their teenagers” (Lewinsohn et al., 1990, pp. 388–389). In addition, coping skills to address family problems were presented, with joint parent–adolescent meetings incorporated into the parent curriculum for the second study only (Clarke et al., 1999).

In the first study, the adolescent and parent group tended to show more improvement than the adolescent-only group, but the results did not attain statistical significance. However, calculations of effect size suggested larger benefit for the combined approach (ES for adolescent only = 0.68; ES for adolescent + Parent = 1.31; Weisz et al., in press). Although the second study also found no significant differences between groups for remission rates of adolescent depression (Clarke et al., 1999), the authors caution that their findings are not definitive. Clarke et al. (1999) considered several reasons for the lack of group differences, including less emphasis on developing the parent component (compared to the adolescent component) and inconsistent attendance of parents (fathers in particular). As is noted in the review of risk factors, fathers and their parenting may exert a particularly important influence on adolescent mental health, particularly with girls, who are at greater risk for developing depression (Alloy et al., 2001). Importantly, parent and family conflict, which appears consistently in the literature as a risk factor for adolescents’ depression, was addressed in the parent treatment arm by Clarke et al. (1999), but it is unclear whether or not family conflict was actually changed as a result of treatment.

Parents have also been incorporated into treatment as agents of change through the family therapy modality. Brent et al. (1997) conducted a study comparing CBT to systemic behavior family therapy (SBFT). Both treatment components included family psychoeducation about depression and treatment, but only SBFT included full parental involvement as agents of change, as described earlier. The trial of 78 adolescents yielded small effect sizes for both CBT and SBFT (.36 and .18, respectively). The CBT group was found to fare better at post-test than the SBFT group in terms of depressive symptoms, rate of MDD, and parents’ perception of treatment credibility, but the two treatment groups did not differ on measures of suicidality or functional impairment (Stein et al., 2001), and were not reliably different from each other at 2-year follow-up (Birmaher et al., 2000).

A second study of family therapy for adolescent depression utilized an attachment approach (attachment based family therapy). Adolescents aged 13–17 years with major depressive disorder were randomly assigned to ABFT, in which they and their parents attended 12 weekly therapy sessions, or to a minimal contact control (Diamond et al., 2002). ABFT was found to be successful in decreasing the rate of depression, depression severity, anxiety symptoms, hopelessness and suicidal ideation and also in improving attachment to mothers. The overall effect size for this treatment was .72, which is in the medium to large range.

One example of a more flexible approach to parent involvement was utilized in the Treatment for Adolescents with Depression Study (TADS), a large-scale investigation comparing CBT, SSRI medication, and combined medication/therapy for treating depressed adolescents (Treatment of Adolescents with Depression Study Team, 2003; Treatment of Adolescents with Depression Study Team, 2004). In the therapy protocol, there is flexibility in having individual or family CBT sessions for the first 12 weeks (up to 14 sessions). Minimally, parents were initially engaged in psychoeducation about depression and the TADS rationale, as well as joint goal-setting. There were additional family sessions that were focused on addressing identified parent–adolescent concerns. Although flexibility of this manualized approach is a strength, information on the degree of family participation, the specific family members participating, and number of family sessions provided is needed before conclusions can be drawn about the true impact of parental involvement.

Ongoing Studies

In a 5-year study initiated in 2002, Stark and his colleagues began investigating the effectiveness of a CBT-group only compared to a CBT-group plus parent training condition for depressed young adolescent girls (Stark & Sander, 2004; Stark et al., in press). In this model of parent involvement, parents attend parent group meetings, work individually with the therapist, and receive their own workbook. Individual meetings between the parent and therapist allows the therapist to address the specific parent behaviors with their child who is depressed, and the therapist can directly model and encourage positive parent–child interactions (Stark et al., in press). Data are not yet available with sufficient power to compare the group differences in treatment response, but the combined rates of active-treatment groups’ improvement suggest, as found in other literature, that the CBT and CBT plus parent training are highly effective in reducing depressive symptoms in young adolescents. The forthcoming dissemination of information from that study may provide additional information to the literature, following the earlier work of a similar design by Clarke et al. (1999).

Finally, in the case of a depressed parent and youth, the idea of parents in parallel treatment (separate therapists and treatment goals for parent and child), deserves consideration. Studies using such an approach have not been identified, although Sanders and McFarland (2000) reported that a cognitive–behavioral intervention that integrated the treatment of parental depression with teaching parenting skills was helpful in reducing both maternal depression and child disruptive behavior. This twofold approach addressing both parental psychopathology and the parent–child subsystem could hold promise toward intervening with child depression. As an example, in an ongoing study targeting parenting practices in families with a history of maternal depression, there are plans to examine the addition of parenting components to the traditional treatment offered to mothers who are depressed (Riley, 2005).

Moderators and Mediators of Treatment Outcome

Understanding why and for whom treatment works can serve as a basis for maximizing treatment effects and ensuring that critical features are generalized to clinical practice (Kazdin & Nock, 2003). In psychotherapy research, this is often accomplished through formal testing of mediation and moderation (Baron & Kenny, 1986). Identifying moderators and the mechanisms through which they may operate can improve treatment outcomes by providing better triage of patients to treatments from which they are likely to benefit (Kazdin & Nock, 2003).

Research examining moderators of treatment efficacy—factors that interact with treatment in predicting outcome—is far from developed, with Weisz et al.’s recent meta-analysis (in press) identifying only three depression treatment studies that have included any test of potential moderators. These three studies suggest that certain family factors, such as family conflict and parent involvement in treatment, are associated with response to treatment, whereas others, such as parental/familial depression, are not. The first study to conduct moderator analyses used data from Lewinsohn and his colleagues’ treatment outcome study, and found that better outcome on self-reported depression was predicted by parent involvement in CBT treatment (Clarke et al., 1992). A second study examined recurrence among participants in Brent et al.’s (1997) treatment outcome study comparing cognitive behavioral and systemic behavioral family therapy to non-directive supportive treatment. Elevated parent–child conflict predicted lack of recovery, chronicity, and recurrence of depression even after controlling for depression, suggesting that conflict with parents is “not merely an epiphenomenon of depression” but is a true predictor of the course of depression following treatment (Birmaher et al., 2000). The third study to assess moderation examined family history of depression as a moderator of treatment outcome among adolescents who received cognitive–behavioral therapy; however, family history of depression among first-degree relatives was unrelated to remission of depression (Jayson, Wood, Kroll, Fraser, & Harrington, 1998). These moderator analyses suggest that parent involvement and behavior are more important than parental history in predicting youth response to treatment.

Though the moderator literature is scant, even more striking is the relative in attention to the question of what change processes underlie improvement (Kazdin & Nock, 2003). It is ironic that we have managed to develop efficacious treatments without being able to say with certainty what produces change (Shirk, 2005). The literature on treatment mediators for depression—processes that are influenced by treatment and that in turn influence outcomes—is even smaller and therefore the role of family factors is as of yet unresolved. In their recent review of psychosocial interventions for youth, Weersing and Weisz (2002) concluded that although 42 of the 67 studies they reviewed included measures of potential mediating variables, only six formally tested for mediation, and only one of these focused on adolescent depression. That study (Kolko, Brent, Baugher, Bridge, & Birmaher, 2000) found some evidence for treatment specificity but not of mediation. Specifically, family therapy was found to be superior to CBT in improving parent-rated family conflict and, to some extent, parent–child relationship problems at the 2-year follow-up. However, such changes were not found to mediate treatment effects, nor were changes in cognition. One other study has emerged testing mediators of treatment (Kaufman, Rohde, Seeley, Clarke, & Stice, 2005) since that time, but this study exclusively assessed cognitive and behavioral mediators, rather than family factors.

SUMMARY AND IMPLICATIONS FOR TREATMENT AND RESEARCH

Evidence is accumulating for a few points of consensus on parent and family risk factors for youth depression. First, parental depression is clearly linked to youth depression. The relationship is not causal or linear, but reflects an interactive, multidimensional set of factors. Second, parent–child relationships, the interactions of child temperament and child coping with family environment, parent pathology, and the impact of situational and environmental stress on the family system are all contributing factors to youth depression. Third, lack of parental warmth and availability is a consistent risk for youth depression. Stress, depression, marital conflict, and social support are just a few of the important factors that may influence parental warmth and availability. The field is addressing many of these environmental, relational, and ecological-systemic factors in current research.

In general, the evidence base for the impact of including parents in treatment is largely lacking, as families have by and large not been incorporated into clinical research with depressed adolescents. Only 32% of the depression treatments tested in clinical trials in a recent meta-analysis (Weisz et al., in press) included parents in any capacity. The overall effectiveness of treatments that work solely with children and adolescents is very similar to that of treatments that include parents as agents of change. Yet, this finding likely does not tell the full story as there are treatments with different degrees of family involvement ranging from none at all to full inclusion of the family in therapy, such as in attachment based family therapy, that have demonstrated preliminary efficacy. Overall, the popular notion that it is important to include parents in adolescent treatment has not yet been supported empirically in the treatment of depression. However, ruling out parent involvement as unnecessary is premature. Future work focusing on characteristics of parents and youth that are particularly well-suited to parental inclusion, including mediator and moderator designs, would be important in guiding empirically-supported clinical delivery of treatment where a “one size fits all” approach is neither appropriate nor possible.

Clinical Implications

There are three primary implications for clinicians, based on this synthesis of literature. First, families should be included in the assessment process, regardless of whether they are included in the treatment process. Valuable information can be gleaned about the general family atmosphere, specific relationships, availability, and affect within the family that can impact the child (Costello et al., 2002; Fendrich et al., 1990; Hammen et al., 2003). Next, fathers and mothers are both important. As supported by Alloy et al.’s research (2001), and by Sanford et al.’s study (1995), relationships and experiences with fathers are important for adolescent risk. The field would benefit from increased knowledge if researchers and clinicians alike incorporate working with fathers more regularly and documentation of the fathers’ roles continues. Third, consideration of a child’s ecological setting, including support, relationships with peers and parents, and overall social networks is necessary. Based on the demonstrated effectiveness of treatments targeting social connections, such as IPT and ABFT, attention to the environment within which the youth functions appears to be important to integrate in treatment planning.

Research Directions

Based on this review of research, we suggest five priority areas for future investigations. First, continuing the standard practice of multidimensional design but adding a new focus on interactions of risks and outcomes related to parent involvement in treatment will be useful. Research that employs methods to measure interactions of parent and youth factors would inform literature on parent participation and treatment outcomes at a more sophisticated level of understanding. The current consensus in the literature about the moderate impact of parent factors on youth risk for depression in tandem with the findings that there may be effective treatments at all levels of parent participation may be surprising, but is not necessarily inconsistent or unexpected. Having descriptions of parent involvement levels is helpful, but is not sufficient in understanding outcomes. Documentation of parent level of participation oversimplifies the potential for interacting variables of parent risks, which could explain the seemingly contradictory results in the current literature base. Level of parent participation, parental warmth, child temperament, and interpersonal functioning of parents and children are just a few of the variables to include in these designs. Understanding interactions of parent and youth factors would further allow for testing combinations of risk factors and parent involvement in order to produce the greatest benefit for the youth. This research could even occur with existing data, using methods such as meta-analysis. With cooperation of investigators, research teams could consolidate data that may have been collected but were not published in traditional outcome studies, allowing the creation of a large shared database of a broad range of risk factors to facilitate analysis of multiple risk factors in relation to outcomes.

Next, ongoing research about the specific components of treatment that produce successful results will inform future treatment design. Certain components could be the essential elements of symptom reduction or relapse prevention, and diverse approaches may in fact contain common elements. Many treatments include attention to relationships, perceptions, skill-building, and coping strategies. A comprehensive examination of the components of empirically supported treatments would allow for more understanding of what constitutes effective treatment. Access to specific treatment manuals, in order to quantify and operationally define treatment components, as well as data about therapist skill, adherence and treatment integrity, would be critical parts of the design. These data are likely available, and could be gathered in a central location with cooperation from original investigators.

Third, long-term outcome studies are necessary. Measuring short-term symptom reduction in a pre-post-treatment design is important but does not provide a complete understanding of treatment response, both because different treatments may show equivalent results when examined longer term (e.g., Birmaher et al., 2000) and because the episodic nature of depression may argue for a longer-term follow up (Weisz et al., in press). Lasting changes, relapse rates over time, and the rewards of learning new relationship skills, for example, may not be evident in short-term outcome designs. Excluding some treatments based solely on poor short-term outcome results could also misrepresent the number of effective treatments.

Fourth, more studies of mediators and moderators of treatment outcome are critical in order to facilitate a better understanding of how therapy works and what predicts effectiveness. Kazdin and Nock (2003) provide several recommendations for studying the mechanisms of change in psychotherapy studies, relevant to the study of child depression. Among their recommendations are: (1) the need to use theory as a guide to select potential mechanisms for focus of study, (2) the importance of assessing more than one potential mechanism of change, as well as possible confounding variables, and (3) replication of observed effects in different studies, samples, and conditions (Kazdin & Nock, 2003).

Fifth, specific racially and culturally sensitive information is needed. Only 13 of the 35 outcome studies reviewed by Weisz et al. (in press) reported detailed information on the race and ethnicity of their samples. Only a few research groups are focusing on culture-specific groups, such as Latino populations (Rossello & Bernal, 1999) and African American groups (see Miranda et al., 2003). The outcomes and risks appear generally consistent with the literature regarding cultural-majority populations, but the understanding is incomplete with so few studies representing the area. In order to recommend treatment practices with confidence, both risk and outcome studies need to include and report findings as they relate to specific minority and cultural groups.

In general, the practices in recent and current research are addressing the need to understand multidimensional, interactional, and complex relationships between youth depression and parent and family factors. The development of efficacious treatments for depression has lagged behind that of other child and adolescent disorders, with the need for more research on parental inclusion in treatment in particular. To date, evidence has not yet indicated that parental involvement is necessary for successful treatment of youth, but more studies addressing the inclusion of parents in treatment and specific predictors of success will allow for a better understanding of how it is best to incorporate the family in treating depressed youth.