Introduction

The results of randomized clinical trials (RCTs) have demonstrated that child-focused cognitive-behavioral therapy (CBT) significantly reduces anxiety in 50–80% of youth (e.g., Barrett et al. 1996a; Kendall et al. 1997; Walkup et al. 2008). However, some children respond less favorably to treatment than others (for review, see Silverman et al. 2008). Attention has turned to ways to improve response rates, and studies have examined the effects of various treatment components and modalities including group format, modularized treatment, and the addition of parent/family involvement.

A growing body of literature suggests that parents play an important role in the development and/or maintenance of child anxiety and may also have an impact on treatment outcome (Ginsburg et al. 2004; Wood et al. 2003). Some studies have demonstrated that parents engage in certain anxiety-enhancing behaviors, either due to their own anxiety or in response to child factors, and this makes them less likely to assist in their children’s adaptive cognitive, social, and emotional development (Ginsburg and Schlossberg 2002). Studies that have examined levels of parental anxiety (e.g., Ginsburg and Schlossberg 2002) found that among parents diagnosed with an anxiety disorder, 60% of their children met diagnostic criteria for an anxiety disorder. In addition, 80% of anxious youth have parents with an anxiety disorder (e.g., Silverman et al. 1988). The specificity of relationships between child and parent anxiety disorders (Hughes et al. 2009) identified, for example, a significant relationship between prevalence rates of maternal panic disorder and child separation anxiety disorder. Mothers of anxious children were also three times more likely to meet criteria for any anxiety disorder and three times more likely to meet criteria for a specific phobia compared to mothers of non-disordered youth (Hughes et al. 2009).

Research on social learning (e.g., modeling) and operant conditioning (e.g., reinforcement) has identified parenting factors associated with the maintenance of anxiety in youth (e.g., Muris et al. 1996). As Rapee (1997) postulated, the parents’ excessive protection and over-control conveys to the child that the world is a dangerous place and fosters the child’s anxiety. Parents also play a role in a child’s avoidance. In a study examining children’s responses to ambiguous situations, Barrett et al. (1996) found that anxious parents increased avoidant responses in their children, a finding that is consistent with an anxious cognitive style being reinforced within the family context. Parental behavior has also been found to be related to children’s emotional experiences. Hudson et al. (2008) found that mothers of anxious youth displayed greater intrusive involvement than mothers of non-anxious youth in situations in which the child was discussing negative affect. Parents of anxious youth also reported significantly worse family functioning than parents of non-anxious youth. Fathers of anxious youth reported worse problem solving and affective involvement compared with fathers of children with no psychological disorders (Hughes et al. 2008), and mothers of anxious youth reported significant problems with family functioning and behavior control.

Research suggests that parental and familial factors affect the probability of successful treatment outcome. Southam-Gerow et al. (2001) reported that higher levels of maternal self-reported depressive symptoms were associated with a less favorable treatment response. Likewise, Berman et al. (2000) found that the following parental symptoms successfully differentiated a treatment failure group from positive responders: higher global severity ratings on the Symptom Check List; higher levels of self-reported depression (using the Beck Depression Inventory); and higher levels of fear (using the Fear Questionnaire). Child reports of family dysfunction and parental frustration at pretreatment were also significant predictors of limited outcomes. Children who perceived more problems in their families were less likely to improve with treatment (Crawford and Manassis 2001). It seems reasonable to hypothesize that addressing parental psychopathology and family functioning in treatment could lead to greater child treatment response rates. Theorists have argued that because children are incredibly reliant on their family environment, an improved model of treatment would be one that is premised on an interpersonal conceptualization of youth anxiety and that aims to employ interventions at the social (e.g., familial) rather than individual levels (Barrett 2000). Consistent with this notion, including parents in the treatment of child externalizing disorders (e.g., oppositional, aggressive, antisocial behavior) has been documented as beneficial to treatment success (Kazdin 1997; Patterson et al. 1982).

Several investigators have added parental/family involvement to CBT for youth with anxiety disorders and compared its efficacy to individual child CBT and/or waitlist control (e.g., Barrett et al. 1996a, b; Berman et al. 2000; Bogels and Siqueland 2006; Cobham et al. 1998; Crawford and Manassis 2001; Kendall et al. 2008). These trials vary in the extent and manner in which parents are involved (Barmish and Kendall 2005). In the treatment of anxious youth, as described by Kendall (2006), parents can be consultants (e.g., parents provide information to make a diagnosis for the child), collaborators (e.g., parents arrange the treatment schedule, assist with the child’s acquisition of coping skills and support the child during in vivo exposures), and/or co-clients (e.g., parents learn skills to manage their own anxiety and become appropriate models for their children, parents are taught ways to reduce family conflict, and parents learn to reduce negative reinforcement of their child’s anxious behavior). Unfortunately the way that parents have been involved across trials is not uniform making it difficult to draw conclusions about the utility of parental involvement in treatment.

Results of evaluations of parental inclusion have been mixed. In some, the inclusion of parents in treatment did not produce superior effects on child anxiety outcomes but did show added gains on some symptom questionnaires, global ratings of functioning, and diagnostician’s severity scores (Barrett et al. 1996a, b; Cobham et al. 1998, Mendlowitz et al. 1999; Wood et al. 2006). Other trials found no significant differences between child-focused CBT and Family CBT (e.g., Nauta et al. 2003). Kendall et al. (2008) compared individual CBT, Family CBT and a family education/support/attention (FESA) control group. Individual CBT outperformed Family CBT and the FESA on teacher reports of child anxiety whereas Family CBT outperformed individual CBT on child reports of anxiety when both parents had an anxiety disorder. Conclusions from these studies differ depending upon the outcome measure (i.e., child self-report, parent report, teacher ratings, clinician ratings, diagnostic interview) and whether or not parental anxiety was present. In a study by Bodden et al. (2010), both individual CBT and Family CBT were less effective if a parent had an anxiety disorder. Overall, the mixed findings may be attributed to factors including the type of outcome measure, the format of therapy (e.g., group, individual, or parent only sessions), the content of the Family CBT (e.g., directly targeting parental anxiety, increasing parent–child communication), and the diagnostic status of the parent. Given the mixed findings, the role of parental involvement in CBT for anxious youth merits further study.

Potential limitations in this area of research include (a) the lack of uniformity in the extent and manner of parental involvement, (b) parental involvement being misused as synonymous with maternal involvement, and/or (c) paternal ratings being collapsed with maternal ratings under the label of parental involvement. The last of these limitations has prevented the examination of the exclusive impact of father involvement. Several researchers (e.g., Duhig et al. 2000) have discussed how the scarcity of father involvement in psychological research studies is problematic given that fathers, like mothers, play a critical role in their child’s development, and in the development and maintenance of child internalizing and externalizing behavior problems (e.g., Connell and Goodman 2002). The scarcity of father involvement in psychological research has likely led to an incomplete picture of the familial context involved in child psychopathology.

The role of fathers has been neglected in investigations of internalizing disorders, particularly anxiety disorders. As highlighted by Bögels and Phares (2007), there is an abundance of literature providing evidence for the negative influence of paternal absence on children’s school achievement, aggression in boys, and heightened risk taking behaviors, but the role of paternal presence or absence in child anxiety has not received much attention. Numerous parental factors have been identified that may have an impact on the development and maintenance of childhood anxiety including modeling, overprotection, avoidance, and control, as well as parental psychopathology, expressed emotion, and marital conflict. Models of parental factors and child anxiety fail to acknowledge that mothers and fathers might influence children’s anxiety differently. Mothers and fathers may have different behaviors or attitudes that could promote anxiety or protect a child from anxiety.

Phares et al. (2005) posit a number of reasons why fathers are often not included in research. Many studies that involve parents almost exclusively involve mothers and therefore lead people to believe that mothers have a greater influence than fathers. There is also the assumption that because mothers may spend more time with their children they have more of an impact on them (Lamb 2004). Yet another faulty assumption is that it is easier to include mothers in research than fathers. Due to the lack of inclusion of fathers in research many parental models of child anxiety have been built predominately on mothers’ roles and do not take into account the different roles that mothers and fathers play for their children both in and out of treatment.

Some studies of youth with externalizing disorders report that higher paternal involvement in the child’s life is associated with less behavior problems (Amato and Rivera 1999) however there is little to no evidence for this relationship with fathers of youth with internalizing disorders. There have been studies examining the role of maternal and paternal depression and findings suggest that children of depressed fathers are at risk for internalizing problems and psychological maladjustment (Kane and Garber 2004; Phares and Compas 1992). Other studies have examined the relationship between fathers with anxiety disorders and rates of anxiety disorders in youth and found an increased prevalence of anxiety disorders in youth whose mother or father has an anxiety disorder. Observational studies have documented specific anxiety-enhancing behaviors utilized by fathers of anxious children including physical control, negative verbal feedback, and less explanation (Hudson and Rapee 2004).

Overall, mothers are included in psychological treatment for youth much more than fathers. A survey of clinicians found that fathers were included in 30% of therapy sessions whereas mothers were included in 59% (Duhig et al. 2002). When fathers are included in the treatment of youth with externalizing disorders there is evidence for better long-term effects (Coplin and Houts 1991; Webster-Stratton 1985). In a study of behavioral parent training that included both mothers and fathers, children who had both parents present in treatment showed significant maintenance of treatment gains compared to youth who only had mothers participate (Bagner and Eyeberg 2003).

Far less is known about the inclusion of fathers in the treatment of childhood anxiety disorders. Some studies suggest a relationship between maternal and paternal functioning and treatment outcome for anxious youth. For example, father’s self-reported somatization negatively predicted child outcome after treatment (Crawford and Manassis 2001). It is possible that fathers who experience somatic distress in response to psychosocial stress, model and encourage avoidant behavior for youth. Rapee (2000) found that higher levels of paternal anxiety, but not maternal anxiety, predicted worse outcome for youth who participated in a CBT program at post-treatment and one-year follow-up. Given this research the inclusion of fathers in treatment appears crucial. Research is needed to explore the role of both mothers and fathers in treatment outcome for anxious youth who receive CBT.

To this end, we examined the relationship between mother and father attendance and engagement in session, as well as parental psychopathology, and child treatment outcome when parents were included as co-clients in Family CBT for anxious youth. We hypothesized that (1) high ratings of mother and/or father attendance and engagement in session would be associated with better child outcome, (2) parental psychopathology would be associated with lesser outcomes, and (3) father engagement would have an association with child outcome. Research indicates that fathers have a unique influence on child functioning in both normative (Lamb 2004; Videon 2005) and abnormal development (Connell and Goodman 2002) and that this influence is distinct from that of mothers. Because mothers are traditionally the parent involved in the child’s daily activities, we hypothesized that having the father involved in weekly therapy would have a significant impact on the child’s treatment outcome.

Method

Participants

Participants were 45 anxiety disordered youth and their parents (45 mothers; 45 fathers) who had participated in an RCT (Kendall et al. 2008). In the RCT, 161 youth were randomly assigned to individual CBT, Family CBT, or the family education/support/attention control group. Participants were recruited through professional, school, and community sources, as well as through prints notices. Of the 161 randomized participants (from 231 screened), 53 received Family CBT. Of these participants, 8 were not included in the present study because both the mother and father were “not available to participate” in treatment (e.g., deceased, not involved in their child’s life). Child participants met criteria for a principal DSM-IV childhood anxiety disorder (Generalized Anxiety Disorder- GAD, Social Phobia- SP, or Separation Anxiety Disorder- SAD) based on the Anxiety Disorders Interview Schedule-Parent and Child Versions for DSM-IV (ADIS-C/P; Silverman and Albano 1997). Participants were excluded if they did not meet a minimum IQ requirement (IQ ≥ 80), were currently taking anti-anxiety or anti-depressant medications, or did not have at least one English-speaking parent.

Child participants (N = 45) were between the ages of 9 and 13 (M = 10.42; SD = 1.79), with a principal diagnosis of SAD (n = 9), Social Phobia (n = 12), or GAD (n = 24) based on the composite ADIS-C/P. In this sample, 55.6% of the participants were male; 80% were Caucasian, and 20% identified themselves as ethnic minorities (African American or Hispanic). Parents consisted of 45 mothers (ages 29–67; M = 40.70; SD = 6.83), and 45 fathers (ages of 32–63; (M = 42.09; SD = 6.18). Of these mothers and fathers, 44 were biological mothers, 1 was a foster mother, 42 were biological fathers, 2 were step-fathers, and 1 was a foster father. In this sample, 36 of the children resided with both parents, 8 lived with only the mother, and 1 lived with only the father. Approximately 79% of mothers were currently married, 7% were divorced, and 13.9% identified their marital status as other (separated, widowed, single); 92.3% of the fathers were married, 2.6% divorced, and 5.1% other (separated, widowed, single). Overall, 35.5% of the mothers and 18.2% of the fathers in this sample met criteria for a clinical anxiety or depression diagnosis (according to the ADIS-IV-L; DiNardo et al. 1994). Participants came from a variety of socioeconomic classes. Please see Table 1 for child participant information as well as a comparison of this subsample compared to the larger sample. There were no significant differences on age, diagnosis, socioeconomic status, or parent diagnosis between our sample and the larger study sample.

Table 1 Child and parent demographic and diagnostic information
Full size table

Measures

Children’s Diagnostic Status

Anxiety Disorders Interview Schedule for Children, DSM-IV edition, Child and Parent versions (ADIS-C/P). The ADIS-C/P (Silverman and Albano 1997) is a semi-structured interview for diagnosing anxiety disorders in youth (ages 6–18) and for screening other disorders. The ADIC-IV-C/P has demonstrated excellent reliability in symptom scale scores for separation anxiety disorder, social phobia, and generalized anxiety disorder and good–excellent reliability for deriving composite diagnoses (Silverman et al. 2001). Interviews were administered separately to parents and children and the diagnostician assigned a composite clinician severity rating (CSR) ranging from 0–8 (0 = no symptoms; 1–3 = sub-clinical levels; 4–8 = significant distress/impairment) based on the symptoms, distress, and interference reported.

Parent Report of Child Behavior and Symptoms

Child Behavior Checklist (CBCL; Achenbach 1991; Achenbach and Edelbrock 1991). The 118-item CBCL is a parent report of their youth’s behavioral problems and social competencies. The CBCL is scored 0–2 depending on the extent to which a particular statement describes their youth (i.e., 0 = not true, 1 = somewhat or sometimes true, and 2 = very true or often true). The CBCL has been standardized to obtain normative scores as such, standard t scores represent a youth’s standing in relation to other youth and determine whether elevated scores on a particular scale falls in a clinical range. The CBCL is one of the most extensively tested rating scales available and possesses excellent psychometrics. Internal consistency on the subscales range from .54 to .96. Test–retest reliability on the subscales range from .86 to .89. The CBCL also includes items that can form a separate internalizing score. The Internalizing factor includes the three syndromes of Withdrawn, Somatic Complaints, and Anxious/Depressed. Mothers and fathers independently completed the CBCL at intake, post-treatment, and follow-up about their child’s behavior over the past 6 months. The CBCL Total score as well as the Internalizing scale was used as a measure of parent-reported child anxiety symptoms.

Child Report of Anxious Symptomatology

Multidimensional Anxiety Scale for Children (MASC). The MASC (March et al. 1997) covers the major domains of self-reported anxiety in youth. The 39-item MASC has four factors: physical symptoms (e.g., tension), social anxiety (e.g., rejection), harm avoidance (e.g., perfectionism) and separation anxiety. The MASC factor structure has been cross-validated in clinical and population samples and holds across gender and age (March et al. 1997). The Total MASC score was used as a measure of child-reported anxiety symptoms.

Parents’ Diagnostic Status

Anxiety Disorders Interview Schedule for DSM-IV, Lifetime Version (ADIS-IV-L). The ADIS-IV-L (DiNardo et al. 1994) is a semi-structured interview to diagnose current and past episodes of adult anxiety and mood disorders. The instrument (a) includes assessments of current and past mood, somatoform, and substance abuse disorders and (b) provides a timeline to assess the duration and onset of current and past episodes. Reliability for the adult ADIS-IV-L diagnoses has been found to be good to excellent (DiNardo et al. 1993). Parental psychopathology (defined in this study as the presence of a current or past anxiety or depressive disorder) was determined based on the ADIS-IV-L.

Measures of Parental Engagement in Session

Session Summary Sheet (SSS). The Session Summary Sheet (SSS) is a therapist rated form developed for, and used in, several earlier studies (e.g., Kendall et al. 1997, 2008). The SSS gathers data on participant attendance and therapist ratings of processes that took place in each session. Therapists completed the SSS immediately after each session. The therapist used the SSS to rate the participants’ (i.e., child, mother, father) (a) level of involvement, (b) mastery of skill, and (c) therapeutic relationship. These items were rated on a Likert scale from 1 to 7 (with 1 = very poor and 7 = very good) similar to those of the Clinical Global Impression-Severity (CGI-S) and Improvement (CGI-I) Scales (Guy 1976). There were 16 session summary sheets completed for each parent in this study. The construct of mother and father engagement in session was based on a composite of the therapists’ SSS ratings (the mothers and fathers involvement in session, mastery of skill, and therapeutic relationship).

Therapist ratings of involvement were based on the parents’ observable interest in session (e.g., amount of smiling, eye contact, talking, making suggestions) as well as the parents’ willingness to participate in session activities. High levels of involvement were indicated by the parents initiating discussions, demonstrating enthusiasm in therapy-related tasks, and elaboration on the points made by the therapist. Therapist ratings of mastery of skill were based on how well the parent understood and applied the skill being addressed. High levels of mastery were indicated by whether the parents gave appropriate examples, if they were able to reiterate the skill in their own words, and if they completed the workbook in a way that revealed that they understood the session content. Therapist ratings of the therapeutic relationship were based on how well the therapists felt they connected to the client (e.g., mother, father, child). The therapist’s SSS ratings were informed by (a) how well the client liked coming to session, (b) the amount of information the client shared with the therapist, and (c) whether the client sought advice from the therapist.

To examine the reliability of the SSS of (a) parent involvement in session, (b) parent mastery of skill, and (c) parent therapeutic relationship, a separate and independent observer randomly selected and rated 30 videotapes (15 tapes from sessions 1–9 and 15 from sessions 10–16) and completed an SSS for each tape. Random selection was done by picking numbers from 3 groups: subject number, psychoeducational sessions (1–8), and exposure-task sessions (9–16). The combination of numbers selected was then used to select the tape to be rated (i.e., if the observer picked the numbers 1,012 and 6 then the tape from subject 1,012 session 6 was coded). Numbers were selected until there were a total of 30 ratable tapes (15 pre and 15 post exposure). The therapists’ and independent observer’s ratings on the SSS were compared and yielded an intra-class correlation coefficient (ICC) of .91. Consensus between the observer and the therapists’ ratings are supported by the ICC. An ICC of .91 indicated that the total variation in measurement on the SSS is due mainly to the target variables and not differences in raters.

Procedure

Following referral, a brief telephone-screening interview was conducted with each child’s parent. Parents and children signed informed consent/assent. Independent evaluators administered the ADIS C/P. Each child’s diagnostic status was determined from the composite of the parent and child interviews as recommended in the ADIS C/P. Participants were included if the interview yielded a child principal diagnosis of SAD, SP, or GAD with a composite CSR ≥ 4. If the child met inclusion criteria, parents were administered the ADIS-IV-L to diagnose current and past adult anxiety and depressive disorders. Parents and children completed self-report questionnaires. Following the assessment, the study coordinator used a predetermined schedule to randomly assign eligible participants to the Individual CBT, Family CBT, or Family education/support/attention condition. Participants were administered the same measures posttreatment and at one-year follow-up. Independent Evaluators were blind to treatment condition.

To examine the relationship between mother and father engagement in session and child treatment outcome, only participants randomized to the Family CBT group were included. Family CBT participants received 16, 60-75 min sessions of CBT, once per week except for illness or vacation. Session measures were completed weekly. Participants were informed that therapists would be completing rating forms but participants were not aware of the content these forms. The therapist completed the Session Summary Sheet separately for the mother and father after each session.

Treatment Program

CBT for anxious youth utilizes helps the child (a) recognize anxious feelings and somatic reactions to anxiety, (b) clarify cognition in anxiety provoking situations, (c) develop a plan to cope with the situation by modifying self talk as well as determining effective coping actions, and (d) evaluate performance and administer self-reinforcement. The manual-based Family CBT (i.e., Howard et al. 2000) adds a family perspective to the typical individual Coping Cat program (Kendall and Hedtke 2006). Treatment is divided into two 8-session parts. The first 8 focus on psychoeducation whereas the second 8 provide the child/family opportunities to practice newly learned skills in exposure tasks. Family CBT included affective education, familiarity of somatic responses to anxiety, coping modeling, and role-plays, and assignments.

Participants received 16, 60–75 min sessions over the course of a 16-week period using a family version of the Coping Cat. The therapist, child, and parent(s) met together for all sessions with the exception of sessions 4 and 9 where the therapist met with the parents and the child separately to provide them with a private opportunity to discuss issues with the therapist. Both parents were expected to participate in all sessions. Although there were no protocol requirements for what to do when a parent missed a session, in general the therapist reviewed the missed material with the parent in the next session and reminded them to attend future sessions. The Family CBT program allowed parents to learn the skills and strategies along with the child and highlights the importance of the transfer of control to parents and from the parent to the child, enlisting the parents as team members (see Silverman et al. 2005). Although the program uses the child’s Coping Cat workbook (Kendall and Hedtke 2006), the program includes more attention to contextual family factors. The program aimed to modify maladaptive parental beliefs and expectations, teach parents constructive responses to their child’s anxious distress, encourage parents to support the child’s mastery, and teach parents and children effective communication skills. Family CBT therapists served as models for the parents and the child, demonstrating adaptive responses to anxiety-related cues. When parents themselves were anxious, they were encouraged to apply the skills taught in therapy to cope with their own distress.

Treatment Training and Integrity

Therapists had a minimum of 2 years experience at the clinic working with anxious youth and their families prior to being trained in the Family CBT treatment protocol. A total of 22 therapists (3 males, 19 females) provided the Family CBT. Twenty therapists identified themselves as Caucasian, one as Indian, and one as Hispanic. Eighteen therapists were advanced doctoral students in clinical psychology with 2–3 years experience in the clinic and four therapists were Ph.D.’s with 4 years clinic experience and 4–5 years post-clinic experience. Family CBT training included studying written materials, workshops led by CBT supervisors, a CBT competency test, and supervised pilot cases.

All therapy sessions were recorded and a subsample (15%) was randomly selected for integrity ratings (as in Kendall et al. 1997). Treatment integrity was assessed by comparing session content to a list of session activities and goals specific to the administered Family CBT session. Results of the unannounced treatment integrity checks indicated actual session content matched proposed session content 92% of the time.

Results

Child Outcome

Child outcome was assessed using a child self-report measure (MASC), a parent report measure (CBCL- Internalizing scale), and the diagnostician’s ratings from the ADIS C/P, which were used to classify the child as a treatment responder or non-responder (based on whether or not the child met criteria for their primary anxiety disorder at post treatment and follow-up). We were interested in predictors of positive treatment response in relation to normal levels of functioning. To that end we used clinical severity ratings, standardized parent rating scales, and child self report to assess child functioning. Results indicate that participant’s anxiety symptomatology improved over the course of treatment according to parent, child, and diagnostician report (see Table 2).

Table 2 Means and standard deviations of diagnostician, child, and parent report at pretreatment, posttreatment, and follow-up
Full size table

Mother and Father Attendance and Engagement

Paired sample t-tests revealed significant differences between father and mother attendance [t(44) = −5.69, P < .01]. Mothers had a mean attendance of 15.63 (SD = .77) while fathers had a mean attendance of 9.43 (SD = 6.0). Correlations between engagement variables (involvement, mastery of skill, and therapeutic alliance) were all significant at P < .01 (see Table 3). Father involvement was highly correlated with father therapeutic relationship (r = .98) and father mastery of skill (r = .98). Given these high correlations involvement, mastery of skill, and therapeutic relationship were combined into one predictor variable labeled Engagement. There were significant differences between father and mother engagement [t(44) = −5.09, P < .01]. Fathers had a mean engagement rating of 11.14 (SD = 5.7) across all attended sessions while mothers had a mean engagement rating of 15.82 (SD = 2.5) across all attended sessions. Fathers had much more variability in engagement and attendance and it appears that even when fathers were present in session they did not have as high engagement ratings as mothers. Mother and father engagement, as well as mother and father attendance, were also combined into a combined parent engagement and a combined parent attendance variable to examine the relationship between combined parental engagement, attendance, and outcome. Combined parent engagement had a mean rating of 13.44 (SD = 3.63) and combined parent attendance had a mean rating of 12.66 (SD = 2.98).

Table 3 Correlations among mother and father engagement and attendance variables as obtained with the SSS
Full size table

Relation of Mother and Father Variables to Treatment Outcome

Bivariate Pearson’s and Kendall’s Tau B correlations were conducted to examine the relationship between individual mother and father attendance and engagement, combined mother and father attendance and engagement, and child outcome (Table 4). Results suggest that father attendance was significantly correlated with CBCL Total Post Score Mother Report (r = −.39, P < .05) and CBCL Internalizing Post Score Mother Report (r = −.46, P < .05). Higher ratings of father attendance were correlated with lower CBCL scores (less impairment). Mother Attendance was significantly correlated with treatment responder status at follow-up (r = .37, P < .05) suggesting that mother attendance is related to positive treatment response. Mother attendance was also significantly correlated with lower CBCL Total Scores at follow-up per father report (r = −.58, P < .01) with results suggesting that higher ratings of mother attendance are related to less behavioral symptoms according to father report. Father Engagement was significantly correlated with mother engagement (r = .34, P < .05) and mother-reported CBCL Internalizing Post Score (r = −.43, P < .05). The higher the ratings of father engagement the lower the CBCL Internalizing score according to mother report, suggesting that father engagement is associated with fewer internalizing symptoms. Combined parent attendance was significantly correlated with parent engagement (r = .85, P < .01), CBCL Total Post Score Mother Report (r = −.50, P < .01), and CBCL Internalizing Post Score Mother Report (r = −.47, P < .05) also suggesting that higher ratings of combined parent attendance are associated with lower scores on the CBCL.

Table 4 Correlations among child outcome, parent engagement, and parent attendance
Full size table

Correlations between mother and father psychopathology (defined as the presence of a DSM-IV anxiety and/or depression diagnosis), child outcome, and mother and father attendance and engagement were also explored. Results suggested no significant correlations. Surprisingly the presence of a maternal or paternal anxiety and/or depression diagnosis were not associated with child treatment outcome or mother and father engagement or attendance.

A series of multiple linear and logistic regression analyses examined the prediction of child outcome from mother and father attendance and engagement (see Tables 5, 6). Logistic regressions were conducted for categorical outcome (responder status) and linear regressions were conducted for continuous outcomes (CBCL, MASC). Pretreatment scores were entered into the first block of the regression and mother and father engagement and attendance was entered into the second block. Results indicate that mother and father engagement and attendance in session were not significant predictors of responder status at post-treatment or follow-up on the child-reported MASC, on the mother-rated CBCL Internalizing scale, or on the mother-rated CBCL Total score. Interestingly mother attendance was a significant predictor of the father-reported CBCL Internalizing score (ß = −.95, P < .05) at follow-up with results suggesting that the higher the rating of mother attendance the lower the father-reported CBCL symptoms on the internalizing scale. Examination of combined mother and father attendance revealed that combined engagement was a significant predictor of MASC scores (ß = −.63, P < .05) at post-treatment with higher ratings of engagement predicting lower MASC scores (less child reported anxiety symptoms). Combined attendance was a significant predictor of father-reported CBCL Internalizing scale scores (ß = −1.10, P < .05) at follow-up with higher ratings of combined attendance predicting lower father-reported CBCL Internalizing scores.

Table 5 Prediction of treatment outcome from mother and father engagement and attendance
Full size table
Table 6 Prediction of treatment outcome from combined mother and father engagement and attendance
Full size table

Discussion

We examined the relationship between mother and father engagement and attendance in session, mother and father psychopathology, and child treatment outcome in a sample of anxious youth who participated in a 16-session Family CBT program. Mother and father engagement in session was conceptualized as a combination of parental involvement, mastery of the session content, and a good therapeutic relationship, all rated by the therapist. Mothers had higher rates of attendance and higher ratings of engagement compared to fathers. Father attendance and engagement in session were significantly correlated with mother reported outcome on the CBCL. Mother attendance was significantly correlated with father reported outcome on the CBCL as well as responder status based on the independent evaluator’s diagnostic report. Consistent with discussions in Ginsburg et al. (2004) about parental impact on child outcomes, higher ratings of attendance and engagement were associated with better treatment outcome. Results of regression analyses indicated that mother attendance, as well as combined mother and father attendance, were significant predictors of father rated outcome on the CBCL. Combined mother and father engagement was also a significant predictor of child reported outcome on the MASC with higher ratings of engagement predictive of fewer anxiety symptoms.

Although some studies (Hudson et al. 2008) have found that mothers of anxious youth have more intrusive involvement, in this study, both mother and father attendance and engagement were associated with child gains. Higher ratings of mother and father attendance and engagement in session predicted improved child outcome. Interestingly results of correlational analyses suggest that the presence of a maternal and/or paternal anxiety or depression diagnosis was not associated with child outcome, parental engagement, or parental attendance. Parental psychopathology did not significantly impact the parents’ ability to attend session or engage in session, nor was parental psychopathology related to child outcome. In the larger RCT from which this sample was taken (Kendall et al. 2008) the Family CBT condition outperformed the Individual CBT condition when both parents had an anxiety disorder. It may be that Family CBT is particularly helpful for those parents with an anxiety disorder. Since the treatment includes parents as co-clients the therapists are able to help the parents cope with their own anxiety as well as model appropriate behavior for their children. Research suggests that parents with anxiety disorders may have a particular parenting style (e.g., excessive protection and over-control) which fosters the child’s anxiety and conveys to the child that the world is a dangerous place (e.g., Barrett et al. 1996a, b; Muris et al. 1996; Rapee 1997). Other studies have also shown that parental psychopathology is associated with less favorable child treatment outcome (e.g., Southam-Gerow et al. 2001). In our study parental psychopathology was not associated with poorer child treatment outcomes. It may be that including parents as co-clients in treatment enables them to learn better strategies for managing their own and their child’s anxiety.

Although literature on the inclusion of parents in treatment of anxiety in youth is mixed (Barmish and Kendall 2005), results of this study suggest that when parents are included and engaged in session children experience improved treatment gains. Consistent with this finding, we suggest that parents be given a companion document (Kendall et al. 2010) that describes the treatment that their child will be undertaking.

Given the emphasis on empirically-supported treatments, there is growing interest in, and need for, examinations of the processes (or mechanisms) of change within these treatments (Kendall 2009; Kendall and Ollendick 2004; Shirk et al. 2010). The present study highlights the merits of studying parent process variables. To date, most process research on child treatment has focused on child process variables with little emphasis on parent processes. One exception (Khanna and Kendall 2009) examined associations between in-session therapeutic components related to parent training and treatment outcomes and found “transfer of control” (Ginsburg et al. 1995) and contingency management to be particularly valuable for anxious youth who received Family CBT. As Shirk et al. (in press) noted, process research has not progressed as rapidly as would be preferred and much more work in this area is needed. There is empirical support for CBT for anxious youth (for a review see Edmunds et al. 2010; Silverman et al. 2008; Ollendick and King 2010) however we do not yet know about the inner mechanisms of change. Early process research with youth suggested that the therapeutic relationship is a modest predictor of child outcomes (Shirk and Karver 2003), and reductions in negative self-talk have been identified as a mediator of outcome in two studies (Kendall and Treadwell 2007; Treadwell and Kendall 1996). Given the wide range of potentially important process variables more research is needed. In the present study, results suggest that mother and father process variables (e.g., parent engagement in session) are significant predictors of child treatment outcome.

An area of particular interest is the parent–therapist relationship. In child-focused treatments, particularly when parents are included, an alliance is formed between the therapist and each parent as well as with the parental dyad (Friedlander et al. 1994). The relationship between the parent and therapist can experience obstacles. Parent(s) may be engaging in behavior that maintains the child’s distress and there may be disagreement between parents (consistent with reduced family functioning among families of anxious youth, Hughes et al. 2008). Research has shown that when mothers and fathers are included in sessions, mothers tend to speak more and when fathers are addressed more outcomes vary (Postner et al. 1971). These findings highlight the need to examine not only child process variables but also parent and family process variables.

In our study both mother and father attendance and engagement were associated with child gains—a finding of particular interest given that fathers are rarely included child treatment research (Phares et al. 2005a, b). The type of referral problem may dictate whether the father is included in treatment. Evidence-based behavioral parent training is more likely to include fathers (e.g., Bagner and Eyeberg 2003). Most outcome studies involving parents focus disproportionately on the mother’s role in treatment (McBride and Lutz 2004; Tiano and McNeil 2005). When fathers are included in therapy, studies do not suggest overwhelming improvement for the efficacy of treatment at termination, but there does appear to be some improvements at follow-up (e.g., Bagner and Eyeberg 2003). Father involvement in parent training was predictive of the maintenance of decreased child disruptiveness and parent stress (e.g., Bagner and Eyeberg 2003). In our study, high ratings of father attendance were predictive of outcomes and of maintenance of gains. Also important to note is the pattern of attendance for mothers and fathers in the present study. Overall, mothers had a higher average attendance and less variability across sessions compared to fathers. Father presence in session, although not as consistent as mother presence, accounted for variability in child treatment outcome.

The present study examined mother and father variables, and focused on parent processes that are rarely examined in child treatment outcome studies. An independent observer validated the therapists’ ratings of these variables. Nevertheless, potential limitations included a lack of measurement of family functioning, parental distress, and the quality of the inter–parental relationship. Accordingly, it is difficult to isolate whether, for example, father engagement predicts child outcome separate from an index of overall family functioning or marital discord. The generalizaibilty of the present findings are restricted by the relatively homogenous sample (i.e., mostly middle SES Caucasians). Future research could explore possible ethnic or SES differences in the role of mothers and fathers in child treatment.