Better together: A review and recommendations to optimize research on family involvement in CBT for anxiety and related disorders

Abstract

Anxiety and related disorders (ARDs) occur in an interpersonal context. Individuals with ARDs respond well to individual cognitive behavioral therapy (CBT); however, there is room for improvement. As such, family members may be included to “enhance” treatment outcomes, yet findings from studies examining family involvement in CBT for ARDs are equivocal. The present paper (a) identifies methodological considerations for explaining inconsistent outcomes among CBT for ARDs with family involvement, and (b) reviews factors that affect outcomes of CBT for ARDs with family involvement including levels of involvement in treatment (e.g., number, duration, and spacing of sessions) and characteristics of who is involved in treatment (e.g., family member cognitions and cultural factors). Limitations of the literature and recommendations for future research are discussed. Researchers should focus on conducting studies that can test not whether but for whom and how family involvement can contribute to improved outcomes above and beyond individual CBT for ARDs.

Anxiety and related disorders [ARDs; i.e., DSM-5 anxiety disorders, obsessive-compulsive disorder (OCD), and posttraumatic stress disorder (PTSD)] are traditionally conceptualized and, at least among adult patients, treated individually (e.g., Clark, 1999). Yet ARDs often occur within an interpersonal context, and treatment researchers have turned to investigating whether involving a family member (FM; typically a parent or partner/spouse) in treatment offers advantages over individual approaches. The majority of this research has focused on cognitive-behavioral therapy (CBT) for anxious youth, with parents as the identified FM; yet, a growing number of studies examine CBT for adults with ARDs who involve their partner/spouse or another close FM in treatment. The existing body of literature on family involvement in CBT for ARDs, however, does not allow for firm conclusions. The evidence for whether family involvement in treatment boosts outcomes varies widely across studies, likely owing to differences in study methodology (e.g., types of outcome measures), treatment format (e.g., group or individual), focus of the intervention (e.g., accommodation), and extent of family involvement in treatment (e.g., number of sessions attended). Accordingly, the focus of this conceptual overview is threefold: first, to survey the varying methods used in this diverse body of literature, and second, to identify moderators of treatment outcome where such information can be derived. Third, we review some limitations of existing studies and offer recommendations for future research, with a lens towards clarifying the very research questions that are asked in order to maximize the ability to draw conclusions.

CBT for ARDs

CBT, which has the strongest empirical basis in the treatment of ARDs (e.g., Carpenter et al., 2018; Lee et al., 2019), derives from the conceptual model of the individual’s experience of anxiety and its maintenance (e.g., Clark, 1999). The primary techniques used in CBT for ARDs are cognitive restructuring, exposure, and exposure with response prevention (ERP; e.g., Clark, 1999). Randomized controlled studies and meta-analyses indicate the efficacy and effectiveness of individual CBT for ARDs (e.g., Mayo-Wilson et al., 2014). Although many adults and youth benefit from CBT, a substantial proportion (9–50%; Taylor et al. 2012) either refuse, drop out prematurely from, do not respond to, or fail to maintain long-term gains following CBT. Examining ways to improve CBT for ARDs is an important research objective.

Given that ARDs often occur within an interpersonal context, there are a number of reasons that implementing CBT from an interpersonal approach (i.e., including a FM in treatment) is a promising avenue for addressing limitations of CBT and interpersonal maintenance factors of ARDs. First, skills (e.g., response prevention) may be taught to aid the FM in minimizing behavior that maintains anxiety (e.g., symptom accommodation; Calvocoressi et al., 1995; Lebowitz et al., 2016). Second, CBT with family involvement can address family-level challenges such as poor communication. Addressing these factors could reduce ARD symptomatology and improve quality of life. Third, a well-informed FM can model healthy coping skills (e.g., how to manage high levels of anxiety) based on empirically-supported CBT techniques. Fourth, CBT with FM involvement may enhance treatment gains or promote long-term maintenance of benefits; for example, the FM could reinforce homework completion. Fifth, family members can have a common and shared knowledge to support the treatment tools learned in therapy. Specifically, it is helpful for families to share an understanding of psychoeducation about anxiety (so as to not further support avoidance), share knowledge about how to conduct exposure, and gain specific instruction regarding how to help.

Though unlikely, involving a FM in CBT for ARDs could attenuate outcome. In some instances, feelings of shame and guilt accompany the disclosure of ARD symptoms, such as taboo-related obsessional thoughts in OCD or self-blame in PTSD. As such, the presence of a FM may initially impede disclosure. In other cases, FMs may hinder progress via criticism or refusal to support their loved one (e.g., Emmelkamp et al., 1980). Some argue that including a FM inadvertently reinforces codependency among partners or hinders a child’s autonomy (or self-efficacy) if the patient attributes their success in therapy to their FM’s participation (e.g., Depestele et al., 2015). Conversely, FMs may become upset when treatment disrupts said codependency (e.g., Barlow et al., 1981) and therefore may be motivated to undermine the treatment. In these cases, a clinician might reasonably exclude a family member. Logistically, including a FM may be complicated due to scheduling challenges and the time commitment (if sessions are longer). Childcare responsibilities, work obligations, or personal health issues may preclude FMs from participating in CBT with their relatives (Reynolds et al, 2013). Additionally, clinicians may not have been taught/trained to include family members and may feel uncomfortable doing so. With these potentials and perils in mind, we next identify and discuss variations in key methodological parameters across this literature that contribute to inconsistent findings regarding FM involvement in CBT for ARDs.

A Note Regarding Developmental Considerations

Although there are common elements of interpersonal involvement across relationships, developmental considerations must be made with regard to treatment delivery. We recognize that a parent-child relationship is qualitatively different from a marital (or other adult-based) relationship, and this, too, must be acknowledged in treatment development and rationale. Parents have a unique (and legal) obligation to keep their child safe and inevitably play a role in logistical aspects of their child’s treatment (e.g., driving children under age 16 to sessions). Clinical observations suggest that interventions with FM involvement are more common among children (most treatments have at least some/minimal attention paid to parental involvement). Comer and colleagues (2019) concluded in a recent Evidence Base Update that, in the treatment of early childhood disorders, parental/family involvement must be included for successful treatment and that there is no evidence in the treatment of early childhood anxiety that individual child-focused CBT is effective. Despite these developmental considerations, the points discussed in this paper apply to empirical studies of FM involvement in CBT for ARDs across age groups.

Procedural Variability in Studies of CBT with FM Involvement

FM’s Role in Treatment

FMs’ varied roles in CBT for ARDs may contribute to inconsistent findings. Baucom and colleagues (1998) defined three conceptually-based approaches to FM involvement: (a) family-assisted interventions, (b) disorder-specific family interventions, and (c) general couple or family therapy focused on issues aside from the ARD. Family-assisted interventions serve the purpose of aiding the patient’s improvement by using the FM as a “coach” or surrogate therapist without directly targeting the relationship. In a disorder-specific intervention, the therapist works with the patient and FM to target aspects of the relationship that are affected by (or that affect) the patient’s ARD with the aim of altering roles and responsibilities regarding how to handle anxiety symptoms. Lastly, general family or couple therapy treats relationship, which can exacerbate ARD symptoms. In this format, the FM is a “co-client” in therapy.

Kendall (2006) identified three roles for parental involvement in CBT for pediatric ARDs. He suggested that parents could be involved as (a) consultants that provide information to the therapist, (b) collaborators that provide information to the therapist, aid the child with learning new skills as a “coach” throughout treatment, and assist with exposures, or (c) co-clients who jointly work on skills to manage their own anxiety. As outlined in pediatric CBT manuals (e.g., Coping Cat; Kendall & Hedtke, 2006), parents most often serve in the collaborator role (Pereira et al., 2016). Yet, including a parent as a collaborator (e.g., helping with homework) does not consistently lead to superior results compared to individual CBT (Kendall et al., 2008).

Piacentini and colleagues (2011) defined family treatments as those that contain structured intervention sessions focused on changing family dynamics (e.g., Siqueland et al., 2005), rather than treatments that include FMs in a less structured formats or less frequent manner (e.g., as a consultant who checks in after a session). Some fit this definition: in disorder-specific couple therapies (e.g., Abramowitz et al., 2013) and “parent as co-client” models, FMs are involved as co-clients who receive therapy designed to target their specific concerns (e.g., improving communication) in addition to the patient’s ARD symptoms. As co-clients, FMs have an opportunity to address their feelings about the patient with an ARD (e.g., Nauta et al., 2003), their own difficulties with anxiety, and the relationship. By Piacentini and colleagues’ logic, however, an intervention in which the FM attends sessions to observe the therapist or serves as a surrogate coach without addressing the relationship (e.g., Whiteside & Jacobsen, 2010) would not be considered a family treatment. Given that a FM’s anxiety may hamper the implementation of CBT (e.g., exposure practice; Creswell et al., 2008), interventions that directly address FMs as co-clients and require a more active role may lead to more durable changes than passive FM interventions (perhaps an “insufficient dose”). This distinction may help explain why some treatments do not find effects.

Single Family versus Multi-Family (Group) Format

CBT for ARDs with FM involvement is often delivered in an individual format (e.g., one couple at a time), yet some researchers have evaluated treatment in a group format. Although individual formats may provide a more personalized intervention, the social aspect of group treatment may help to normalize experiences with anxiety, increase peer support, and encourage peer learning (e.g., Fischer et al., 2013). For example, in a study by Barrett and colleagues (2004), children and adolescents were randomized to individual family-based CBT, group family-based CBT, or a waitlist control. Although both active family conditions yielded significant reductions in OCD symptoms (with no significant between-group differences), patients in the group condition evidenced larger reductions in depression and anxiety symptoms. Parents and children in the group condition also reported that the social support and mutual understanding that they received was helpful.

Van Noppen and colleagues (1997) tested multifamily (group) CBT for OCD in comparison to a group format without FMs. The multifamily program included discussions about family reactions to OCD symptoms and behavioral contracts (i.e., positive reinforcement schedules) for exposure tasks. Group members were encouraged to share contact information and support each other throughout the therapy process. Results revealed that the multifamily group behavioral therapy showed slightly larger effect sizes than the group therapy that did not include FMs. Further, more patients in the multifamily condition reported healthier role functioning at posttest and follow-up. The researchers suggested that multifamily group cohesion affected motivation to complete treatment goals; however, motivation was not explicitly assessed. There appears to be some indication that group interventions targeting FM involvement may hold some added benefit beyond individual family approaches.

Separate, Joint, and Hybrid Session Formats

Family involvement in CBT for ARDs differs according to whether the FM participates in treatment sessions with the patient or is seen separately (without the patient present) by the therapist. Potential advantages to jointly attending all therapy sessions include emphasizing open communication and establishing a team approach to brainstorming and problem solving (e.g., planning exposure practices). Indeed, Barrett and colleagues (1996) found added benefits from a family-based CBT anxiety management treatment (in which parents and their children attended all sessions together) relative to individual CBT. In treatment protocols where couple-focused skills are a prominent feature (e.g., couple-based CBT for OCD; Abramowitz et al., 2013), each partner’s presence may be critical to learn and practice such techniques.

Alternatively, if the content intended for the patient and the FM are relatively distinct, then joint sessions may be less efficient than separate FM- and patient-specific sessions. Findings regarding the effectiveness of separate sessions are mixed. In Nauta and colleagues’ (2003) protocol, which included a 12-week CBT intervention for children and a separate seven-session cognition training program for their parents, parents and children met with separate therapists at separate times. There were no significant differences in children’s improvement in anxiety symptoms between children whose parents did or did not receive the parental component. In Barrett and colleagues’ (2004) trial comparing individual CBT to group family-based CBT for pediatric OCD, children in the family-based condition received the majority (13 of 16 sessions) of treatment separately from their FM (a parent or sibling). This study failed to find a significant difference in pediatric OCD symptoms between treatment conditions. As a counter example, a significant effect was observed among adult FMs seen separately in Thompson-Hollands and colleagues’ (2015) brief family intervention (BFI) for OCD such that patients whose FMs received the intervention improved more quickly than those whose FMs did not receive the BFI.

Some family involvement in treatment entails separate sessions with little active involvement. Rather, the family members receive psychoeducation sessions or reading materials (i.e., bibliotherapy; Rapee et al., 2006) as the main FM component. For example, Ost and colleagues (2015) tested whether an educational course to parents of children with social anxiety would lead to improved outcomes in comparison to no participation. The authors concluded, however, that the parent involvement did not contribute to improve outcomes for children. Although further testing is required to understand this finding, the separated format, in combination with the relatively inactive session content (i.e., simply attending a course), may have played a role in the lack of effect. Similarly, written materials and bibliotherapy for FMs may confer some benefit but remains inferior to standard group treatment (Rapee et al., 2006).

Still, many treatments—including some treatments labeled as “individual” and some labeled as “joint”—include a hybrid of time spent together and apart. In some “family” treatments, parents were only included for a portion (i.e., the final 25%) of each session (e.g., Silverman et al., 1999), or family treatment involved combinations of sessions (e.g., adolescent alone or parent alone) on a case by case basis (Siqueland et al., 2005). Conversely, the individual CBT conditions in Kendall and colleagues’ (2008) study, which outperformed an explicitly family-based condition, included parent sessions. Despite the conceptualization (or label) as an “individual treatment” and relatively few parent sessions, parent behavior could be meaningfully affected by participating in just two sessions, in addition to other potentially unaccounted for involvement (e.g., reviewing the child’s therapy materials), which is difficult to measure.

Empirical questions remain. Do separate therapy sessions fail to integrate active involvement or foster opportunities for teamwork, which may account for some added benefit of FM involvement? Are joint sessions particularly important for therapy involving children or addressing certain skills? It likely depends. Joint family sessions might not be appropriate if the family exhibits high levels of hostility or if the therapist lacks family- or couple-based training. Further, high numbers of joint sessions incur increased costs (e.g., time) to families, and could therefore contribute to higher treatment dropout. To date we do not know the incremental value of each FM-involved session, though presumably there is a point at which FM involvement is so limited as to offer no benefit and a point at which further FM sessions produce no greater effect.

Treatment Foci and Targets

Family involvement in CBT for ARDs can have different foci, such as psychoeducation, reducing accommodation, cognitive restructuring, and contingency management. Study findings highlight the benefit of identifying specific, active treatment targets (e.g., family communication; Shortt et al., 2001) and actively addressing said targets. Psychoeducation, which entails offering disorder-specific information about the conceptualization and treatment of an ARD, is a powerful tool that could help to normalize the experience of anxiety, provide accurate knowledge about the disorder, increase support for treatment, and reduce stigma. Öst and colleagues (2015) tested whether including parents as participants in a group educational course (8 90-minute sessions) would improve outcomes for children with social anxiety over and above no parental participation but found little evidence of such an effect. Perhaps FMs need to be actively involved in treatment rather than simply attending a psychoeducational course alongside their relative.

Other studies of CBT with FM involvement have addressed FMs’ behaviors (e.g., accommodation) and thoughts (e.g., maladaptive beliefs about anxiety). For example, a number of studies have identified accommodation reduction as a central focus (e.g., Lebowitz et al., 2014). Results from studies with children (Lebowitz et al., 2014) and adults (Thompson-Hollands et al., 2015) revealed statistically significant reductions in accommodation following a FM-involved treatment; the study by Thompson-Hollands and colleagues also demonstrated that changes in accommodation temporally preceded changes in patients’ disorder-specific symptoms, and that the reverse was not true, providing strong evidence for changes in accommodation as a mechanism. In another trial focused on using cognitive restructuring to challenge maladaptive FM beliefs, Nauta and colleagues (2003) implemented a seven-session cognitive parent training intervention in addition to individual CBT for the anxious child. During later stages of treatment, parents were taught to identify and restructure automatic thoughts related to negative feelings toward their anxious child. Yet, this intervention did not emerge as statistically superior to individual CBT (and the proposed mechanism of cognitive change was not measured). Collectively, these findings suggest that addressing specific FM responses or reactions may offer promising outcomes for family involvement, but that not all behavioral targets are equally impactful.

Contingency management involves applying the principles of operant conditioning to increase or decrease a behavior (e.g., a parent rewarding an anxious child for practicing an exposure). A gradual transfer of control (TOC; transferring knowledge and skills) from the therapist to the FM and then to the patient occurs (Silverman & Kurtines, 1996). These techniques have been examined in parent-child dyads with results indicating that contingency management and TOC contribute to improvements in ARD symptoms (e.g., Silverman et al, 1999). A meta-analysis (Manassis et al., 2014) found that the rate of ARD remission continued to improve over time among treatments focused on contingency management and TOC strategies.

The findings suggest an emerging picture regarding the potential importance of getting FMs actively involved in working toward select foci in CBT, as opposed to merely observing sessions or receiving psychoeducation. It is not surprising that generic or passive FM involvement does not seem to enhance individual treatment outcomes, as active involvement in CBT (e.g., engagement in ERP, homework completion) on behalf of the patient predicts outcome for ARDs in general (Glenn et al., 2013). Despite plausible rationales for many approaches, it remains difficult to predict what changes in the FM will precipitate changes in the patient, relationship, and experience of treatment. It would be useful to operationalize and assess how active FM involvement must be to meaningfully affect outcome. Further, many CBT treatments (e.g., Thompson-Hollands et al., 2015) contain components that address a combination of targets. Findings regarding how FMs should be involved and the behaviors to target remain inconclusive, and researchers haven’t yet compared approaches (e.g., targeting accommodation vs. communication) via dismantling studies.

Assessment Strategies

Variability with regard to how outcome is assessed across studies of family involvement in CBT for ARDs precludes overarching conclusions about “advantages” or “disadvantages” of said involvement. Commonly used primary outcome measures include the presence or absence of categorical DSM diagnoses and disorder-specific symptom severity for the identified patient, levels of family distress and functioning, and levels of symptom accommodation. Secondary outcome measures often assess (for the FM, identified patient, or both) coping, relationship satisfaction, quality of life, and comorbid symptoms (e.g., depression). Some researchers have also included measures of attendance, attrition, treatment feasibility, treatment satisfaction, interpersonal bonding, FM self-efficacy, and FM anxiety. The patient, their FMs, or the clinician (or combinations thereof) may serve as informants or provide self-report data. Given the heterogeneity of assessment measures, their foci, and how the data are collected (e.g., multiple informants, self-report), it is not surprising that results vary.

Who completes measures.

Best research practices entail the use of multiple informants (Hunsley & Mash, 2007), yet many studies do not include multiple reporters at each assessment point. A review of 12 couple-based interventions for panic disorder found that only three-quarters collected self-report data from the partner (Byrne et al., 2004). A meta-analysis examining parental involvement in CBT for pediatric anxiety disorders found that only 9 of 16 studies included an assessment of parental functioning at post-treatment (Thulin et al., 2014). With no opportunity to measure whether FM behavior and attitudes have shifted, researchers cannot evaluate whether their interventions that address FM behavior and attitudes have achieved meaningful change in the targeted mechanisms.

Given their differing perspectives, informants’ reports often diverge (de los Reyes et al., 2015). Such disagreement complicates the interpretation of study findings and renders results from just one informant potentially unreliable. Barlow and colleagues (1981), for example, detailed the relationship satisfaction trends of a couple wherein the wife’s marital satisfaction rose considerably following CBT for agoraphobia, while the husband’s marital satisfaction gradually decreased as his wife’s phobia remitted (the study authors attributed this decrease to the husband’s’ discomfort with his wife’s increasing independence). In another study comparing individual CBT for ARDs to CBT with family involvement (e.g., sessions dedicated to family alliance building), adolescents reported increases in perceived parental acceptance and warmth at post treatment despite no significant changes in parent-reported acceptance and warmth (Siqueland et al., 2005). It is important to consider that indeterminate findings with regard to FM involvement may be due to differing perceptions of treatment outcome and differing perspectives according to who is assessed. Taken together, multiple informants are key to truly understanding the full impact and mechanisms of action for an intervention with FM involvement.

Matching treatment targets with assessment measures.

Outcome measures must align with hypothesized outcomes and mechanisms. A mismatch between treatment targets, intervention strategies, and assessment measures reduces the interpretability of data. If the goal is to decrease ARD symptoms by targeting a FM’s accommodation, both ARD symptoms and accommodation should be measured. Reynolds and colleagues (2013), for example, compared parent-enhanced CBT and individual CBT for pediatric OCD. Although the parent-enhanced condition explicitly addressed accommodation, no measures of parent accommodation were included. Similarly, Nauta and colleagues (2003) concluded that there were no significant differences in outcome between an individual CBT intervention for children with anxiety disorders and an intervention that included cognitive parent training. The researchers, however, did not assess changes in parental cognitions and therefore could not discern whether the parent training program successfully altered parental cognitions. Conclusions such as these hamper the ability to make specific (and generic) conclusions about FM involvement.

Longer term follow-up assessment.

Studies suggest that greater family dysfunction predicts attenuated long-term outcome following CBT for ARDs (Barrett et al., 2005). To test this, however, long-term assessment (i.e., one year or greater) is necessary. Unfortunately, many studies of CBT for ARDs with family involvement do not include long-term (i.e., greater than one year) follow up data (e.g., Öst et al., 2015); this precludes overarching conclusions about attenuated outcome or potential benefits. Studies that conclude equivalency (or inferiority) of FM involvement at post-treatment without consideration of long-term outcomes may overlook a potentially important benefit of family-based CBT. In a review, Gibby and colleagues (2017) found that youth receiving CBT without family components had poorer long-term (i.e., two years or longer) outcomes in two of four studies in which this question was tested.

Potential Moderators of Treatment Outcome

Although many studies suggest the efficacy of FM involvement in CBT for ARDs, there is variability in outcomes. Accordingly, we next consider factors (methodological and otherwise) that may moderate treatment outcome. Where possible, we rely on RCTs and meta analyses.

Level of FM Involvement

Studies and meta-analyses have examined how the degree of FM involvement in treatment (defined in various ways) relates to outcome for various ARDs. Thompson-Hollands and colleagues (2014) conducted a meta-analysis of treatment studies for adults and youth with OCD in which a FM was involved. They concluded that family-inclusive treatments had a large effect on OCD symptoms, despite significant variability in the degree to which a FM was involved. Moreover, across studies, the number of sessions of family inclusive treatment was a significant predictor of outcome: patients did better the more sessions in which a FM was present. However, the meta-analysis included only studies in which there was some level of family involvement, precluding a direct comparison against individual treatments.

In a meta-analysis of treatment studies for mixed childhood anxiety disorders, In-Albon and Schneider (2007) did not find significantly different outcomes between child-focused therapy and family-focused treatment (defined as active involvement of a parent in at least four treatment sessions). Of note, In-Albon and Schneider excluded studies of youth with OCD, and the categorical definition of FM involvement was broad. Similarly, Manassis and colleagues (2014) conducted a meta-analysis of RCTs of CBT for anxious children that classified parental involvement in treatment as “limited” (parental involvement in less than 50% of sessions or in only a short portion of each session) or “active” (parental involvement in greater than 50% of sessions). Level of parental involvement was unrelated to the presence of an ARD diagnosis following treatment. Again, this lack of significant difference may be due to the authors’ broad definition of “limited” parental involvement.

Finally, Reynolds and colleagues (2013) compared a manualized CBT program for youth with OCD in which one or both parents attended all 14 weekly treatment sessions to an identical program in which parents attended only three sessions. At post-treatment, both formats produced large effect sizes, and there was no significant difference in OCD symptom severity. Yet, patients in the high parental involvement condition scored significantly lower on measures of general anxiety and depression at post-treatment than did those in the low parental involvement condition. This suggests that greater family involvement was associated with additional opportunities for generalizing learning and applying treatment principles to related concerns.

Spacing and Duration of Sessions

The frequency and length of treatment sessions in CBT with FM involvement varies widely across the literature, ranging from weekly 50-minute sessions to daily multi-hour sessions; such variables may influence outcome. (A related question concerns the feasibility of FMs regularly attending treatment sessions, especially those that occur with greater frequency than once per week.) Some authors have suggested that closely spaced (i.e., more frequent) sessions translate to a greater family focus on the individual’s ARD, which might enhance outcome (Storch et al., 2007). To this end, several open trials of intensive CBT for OCD with family involvement have been conducted (e.g., Whiteside & Jacobsen, 2010); however, the lack of a comparison condition precludes conclusions about the effect of treatment intensity on outcomes.

With regard to other ARDs, Hardway and colleagues (2015) tested the efficacy of a 20 hour, 6-day intensive CBT treatment for children and adolescents with PDA and comorbid depression with and without parent involvement. Youth in both conditions demonstrated reduced symptoms of PDA and depression at post-treatment. (Although parental involvement did not moderate outcome of intensive CBT for PDA, interestingly younger adolescents benefitted more than older adolescents when their parents were not involved directly in therapy.) Ollendick and colleagues (2015) compared an intensive one-day treatment for youth with specific phobias with and without parent involvement and also found no differences at post-treatment. In this case, the brevity of treatment may not have provided sufficient opportunity for a FM’s attendance to make a difference. Although inconclusive due to the small number and variation in focus and format, the findings suggest it is possible to include a FM in an intensive treatment format.

Characteristics of the FM and Family Environment

Given the complex interpersonal processes that maintain ARDs, it is important to consider how interpersonal processes may mediate CBT and how characteristics of participating FMs (e.g., demographic characteristics and FM beliefs, traits, psychopathology) might influence outcome. It is important to consider family dynamics and culture.

Demographics.

With regard to FM gender, research suggests mothers are more commonly involved and show higher levels of attendance than do fathers in CBT for ARDs (e.g., Barrett et al., 2004; Pereira et al., 2016) with FM involvement. These findings align with a broader literature suggesting that mothers are more engaged with their child’s environment (e.g., spend more time with their child) than are fathers. In further examining the independent contributions of mothers and fathers to outcomes of CBT for pediatric anxiety, Podell and Kendall (2011) found that mothers better mastered session content (per therapists’ ratings) than did fathers, but individual parental attendance and engagement did not significantly predict child anxiety outcomes at post-treatment. Interestingly, father attendance and engagement in session were significantly associated with mother-reported outcomes related to the child’s internalizing and externalizing symptoms. Greater combined parental engagement, however, was a significant predictor of less child-reported anxiety at post-treatment. This finding supports the notion that parental inclusion and engagement can facilitate improved treatment gains for children. Given these gendered findings among parents, it is important to consider similar implications in adults (i.e., CBT involvement among male versus female partners).

Additional research is necessary to extend these findings across diagnoses and with larger samples. Further, there are no data with respect to whether socioeconomic status (SES) predicts outcome of family involvement in CBT for ARDs, although factors such living in a single parent household can predict attrition in CBT for pediatric ARDs (Kendall & Sugarman, 1997). Given that demographic factors such as gender and SES may impact FM involvement in CBT for ARDs and that involvement, in turn, may impact treatment outcome, it is crucial to not only identify which factors affect involvement, but also design treatments that actively address potential disparities (e.g., promote involvement among fathers or parents of lower SES).

FM psychopathology and family distress.

Lower levels of FM anxiety facilitate exposure, and individuals whose FMs encourage them to face anxiety provoking stressors are twice as likely to complete treatment (Pereira et al., 2016; Meis et al., 2019). Conversely, anxious FMs might impede CBT by promoting avoidance behaviors, accommodating anxiety symptoms, and withdrawing from difficult exposure tasks. Mehta (1990) observed that a partner’s consistent low anxiety and high frustration tolerance had a positive effect on treatment outcome for adults with OCD. Specifically, “firm” partners were more successful in family-based behavioral treatment for OCD than anxious and inconsistent partners. In concert with this finding, Kendall and colleagues (2008) reported that children with an anxious mother showed less improvement on a measure of internalizing symptoms (e.g., anxiety, depression) from pre-treatment to follow-up as compared to children with a nonanxious mother.

Relationship conflict not only serves as a chronic, diffuse stressor, but also disrupts provision of a consistent, supportive message about treatment to one another or to a child in treatment (e.g., feeling “pushed too hard” by a partner in treatment; Barlow et al., 1981). Greater family conflict may limit the effectiveness of certain treatment methods or goals. The degree of hostility and family conflict also impacts FM involvement in CBT for ARDs; for example, it may hamper the implementation of therapy techniques at home. In fact, adult patients’ perception of their partner as critical and angry was a predictor of poorer CBT outcome (e.g., Chambless & Steketee, 1999). In a study examining high expressed emotion (EE) among parents as a predictor of treatment outcome among children with OCD, Przeworski and colleagues (2012) found that greater pre-treatment EE was associated with greater pre-treatment OCD symptom severity, but not symptom reduction at post-treatment. Greater EE was, however, associated with impaired functioning in related domains (e.g., school functioning) as reported by the parent. In line with these observations, Öst and colleagues (2015) advocate for future research that examines the benefits or risks of including a FM in treatment among dyads that have a “disabling” relationship. On one hand, including highly critical FMs in treatment may sabotage efforts; on the other hand, it can be crucial to include hostile relatives to develop effective communication skills. Fortunately, it appears that some FM characteristics are malleable. For example, Gar and Hudson (2009) found that maternal EE decreased over group CBT for youth with ARDs and FM involvement.

Family member cognitions.

On a related note, FMs who believe that “anxiety is dangerous and to be avoided at all costs” (i.e., anxiety sensitivity; Reiss & McNally, 1985) tend to make efforts to protect themselves and their loved ones from experiencing anxiety (e.g., by engaging in symptom accommodation; Wu, McGuire, & Storch, 2016), which can interfere with treatment. Results also reveal that parents with more negative beliefs about their child’s experience of anxiety (e.g., ‘‘If my child gets too nervous, it could be harmful’’) were less likely to participate in a group CBT intervention for ARDs (Pereira et al., 2016). Research suggests that FMs who hold positive expectancies about treatment (e.g., “therapy will result in positive outcomes”), as opposed to negative expectancies, are more likely to engage in treatment, and their loved ones are more likely to experience greater improvement (Nock & Kazdin, 2001). Lewin and colleagues (2011) found that patient gender (female) and increased parental anxiety symptoms were associated with pessimism about treatment outcome. Given the potential impact on treatment outcome, expectancies represent an important target for assessment and treatment.

Limitations of the Literature and Recommendations

Aggregating across the literature on CBT with family involvement for ARDs, a number of methodological limitations are worthy of mentioning. In general, sampling issues, study design issues, and treatment implementation issues impair researchers’ ability to identify benefits of family involvement and should be foci for future research.

Sampling Issues

Overall, there is a dearth of research regarding interventions with ethnic minorities or same gender romantic relationships. Many studies cite, as a limitation, sample homogeneity with regard to race, ethnicity, and socioeconomic status (i.e., predominately White middle-upper class samples; e.g., Hardway et al., 2015). Additionally, the vast majority of research regarding couple-based CBT for ARDs includes heterosexual couples. It is important to consider diffuse stressors (e.g., racism, homophobia) that may increase family distress and how these factors may impact couple- and family-based treatment (i.e., whether treatments may be differentially effective for various diverse groups). More broadly, cultural factors influence parenting methods, relationship expectations, perceived importance of healthcare, and therapy seeking behaviors (e.g., Bornstein, 2012; Bui & Takeuchi, 1992). For example, individuals from some ethnic minority groups perceive more barriers to treatment and are less likely to complete treatment. A family-oriented treatment may be beneficial in collectivist cultures where family relationships are of primary importance and interrelatedness is heavily emphasized (Mehta, 1990). Researchers should prioritize the recruitment of non-White families and same-gender couples to test whether existing family-based interventions are efficacious.

Further, some studies (e.g., Abramowitz et al., 2013) have included participants with a single diagnosis (e.g., OCD) while others (e.g., Kendall et al., 2008) have included participants with a variety of ARDs. Although ARDs are conceptualized and maintained in similar ways (Clark, 1999), certain diagnoses (or associated behaviors) may be more or less amenable to family involvement. Researchers should carefully consider whether they expect a particular FM-involved treatment to target disorder-specific processes/features and should explicitly describe whether they would expect the intervention to have similar effects in other populations (if they present outcomes from a diagnostically homogeneous sample).

Study Design Issues

A number of previous studies contain small samples, which might contribute to the inconclusive findings presented in this review (e.g., six couples in Barlow et al., 1981; ten children in Lebowitz et al., 2014). Such studies are underpowered to detect between group differences. A small sample size precludes analyses of moderators and mechanisms, which for the reasons previously outlined are the most likely to yield useful findings. Additionally, within small datasets outliers may be difficult to detect and can exert undue influence on analyses.

Although numerous studies address treatment outcomes in patients with ARDs and family members, few studies explicitly query the ways in which family members want to be involved or assess the acceptability of involvement in treatment. Given that many family members may want to be involved, it is important to find out how to do so in a way that is acceptable to the client, clinician, and FM. As an example, Chavira and colleagues (2017) conducted qualitative interviews to better understand parental attitudes toward CBT and modes of service delivery. Understanding stakeholder preferences and measuring treatment acceptability is vital for improving family involvement and its impact on treatment.

Regarding measurement issues, we recommend the formal assessment of baseline family characteristics (e.g., psychopathology, beliefs, support, accommodation). By collecting data on these variables of interest, researchers may be able to study predictors of outcome and eventually determine who would benefit from an individual or family-based treatment approach. A thorough assessment approach is the only way to conclusively determine if an FM-involved treatment is having an effect on the proposed mechanisms of interest, and also whether there are subgroups of participants/families for whom the intervention is more or less effective.

A lack of longer-term follow-up assessments in many studies may contribute to equivocal findings, as differences in treatment with or without FM involvement may not emerge immediately following treatment. CBT is robust and effective in evidencing symptom reduction at post-treatment, yet FMs may play a key role in maintaining gains months later (see e.g., Cobham et al., 2010). We recommend that researchers incorporate multiple repeated assessments to identify potential patterns of change and temporal relationships between variables. Certain ingrained relationship patterns (e.g., communication) may not be realized immediately or may require extended time to shift. This approach of repeated, longitudinal assessment will mutually support nuanced analyses and the ability to determine when changes occur.

Lastly, many studies lack a control condition or comparison to previous studies (e.g. benchmarking). This weakens arguments regarding the potential benefit of FM involvement over and above the effect of standard CBT. CBT is highly effective for the treatment of ARDs (e.g., Carpenter et al., 2018), and it is implausible to assume that family involvement would so weaken the impact of CBT that patients would not demonstrate at least moderate improvement. Nowadays, an open trial in which a FM-inclusive CBT condition merely shows overall improvement in patients’ disorder-specific symptoms is hardly noteworthy. To advance the quality of our studies and begin answering the important question of whether or not the addition of FM involvement meaningfully enhances CBT outcomes beyond what is achieved in individual treatment, researchers must include either an individually-delivered control condition or an alternative FM-inclusive condition that differs on some important variable (e.g., number of FM sessions, focus of FM behavior change, group versus individual FM component, etc.). Alternatively (or additionally), researchers could opt to measure a hypothesized mediating variable of the FM component and demonstrate that the treatment results in changes in this mediator and that these are associated with subsequent changes in disorder-specific symptoms.

Treatment Implementation Issues

By design, cognitive-behavioral treatments are typically composed of multiple modules (e.g., psychoeducation, exposure therapy); treatments are often delivered as a “package,” and this applies to the FM components of many studies as well. Given the varied components within treatment packages, it is important to consider whether researchers are truly comparing two different treatments when contrasting a treatment with FM involvement to an individual protocol. Kendall and colleagues (2008) acknowledged overlap between treatment conditions (family CBT and a family-based education/support active control) in which the same therapists treated families in all conditions. Specifically, two thirds of the active control manual contained CBT content, and tapes from the active control highlighted the unintentional inclusion of CBT techniques (i.e., “bleeding”). (The researchers concluded that CBT “bleeding” into the active control condition might have been associated with the lack of difference between conditions.) We recommend using a dismantling approach whereby active treatment components are isolated to identify which ingredients (e.g., psychoeducation, adjusting relative’s cognitions) are essential. It may be that individual and family-based treatments are more alike than they are different/unique, so a dismantling study can help to elucidate crucial components. A periodic assessment approach (Barrett et al., 2004) suggests that CBT components may be individually successful (i.e., contribute to clinically significant change), and it is crucial to test these components separately or in a counterbalanced order. Further, a formal assessment conducted at the beginning of treatment may aid clinicians in developing a personalized family treatment (e.g., determining what to emphasize in a modular CBT intervention).

Limitations of the Current Paper

In addition to the individual studies’ limitations, the current paper is not without limitations. Although the paper intentionally combined studies related to varying ARDs across ages and relationships, this method introduces extra variability into an already complex matter. Further, the current paper did not submit the included studies to a systematic quality assessment (e.g., Delphi consensus method). Thus, the paper may appear to assign a perfect quality-score to each trial by seemingly ‘weighting’ study outcomes equally. Additionally, as noted in the Limitations of the Literature section, many of the studies are open trials that could not undergo systematic evaluation that is typically used for RCTs. Despite these limitations, this paper overviews existing research to date among CBT trials for ARDs with FM involvement to help understand why the literature is seemingly inconclusive and outlines future directions.

Conclusion

Although a great number of interventions have been piloted, the literature regarding FM involvement in CBT for ARDs is inconsistent regarding the potential added benefit of components that incorporate or target FMs. Methodological variations, combined with study limitations outlined above, support the three-fold intention of this paper to guide researchers in understanding why these inconsistencies exist and planning research to study neglected areas and clarify findings. Heterogeneous methodology and myriad potential (but typically unexplored) moderating factors preclude overarching conclusions regarding “enhanced” outcomes when including FMs in CBT for ARDs. In fact, the very pursuit of such broad conclusions may stymie forward progress in the field. Given the potential futility of focusing on categorical conclusions (e.g., “family involvement enhances CBT outcomes”), future research should focus on developing and testing studies that can examine for whom and how FM involvement can contribute to improved outcomes above and beyond the contributions of individual CBT for ARDs. In order to rally families to come together and fight mental illnesses jointly, we must be able to offer tailored empirical guidance with a strong methodological foundation. Family involvement in CBT for ARDs represents a promising area for future research and clinical care; however, thoughtful study design is essential for avoiding equivocal conclusions.