Family Therapy Techniques and One-Year Clinical Outcomes among Adolescents in Usual Care for Behavior Problems

Abstract

Objective:

This study tested whether use of family therapy (FT) techniques predicted long-term clinical outcomes in usual care for adolescents enrolled in either family-based or non-family-based treatment for conduct and substance use problems.

Method:

Participants included 70 adolescents (53% female; mean age 15.4 years) from diverse backgrounds (64% Hispanic, 16% African American, 11% multiracial) assessed at baseline and 3-, 6-, and 12-month follow-up. Utilization of FT techniques was assessed for 802 therapy sessions via a therapist-report tool that collected post-session data on delivery of core treatment techniques of the FT approach.

Results:

Latent growth curve modeling was used to examine individual change over 12-month follow-up. FT technique scores, averaged across treatment sessions, were included as a predictor in latent growth models, controlling for adolescent age, sex, and ethnicity. More extensive use of core FT techniques predicted significant decreases in adolescent-reported delinquency and externalizing behavior and marginal decreases in substance use and parent-reported externalizing for cases in both family and non-family treatment.

Conclusions:

Greater use of FT techniques in routine care was associated with better long-term adolescent outcomes, regardless of whether clients attended services featuring family therapy or an alternative treatment approach.

Public Health Significance:

This study highlights the effectiveness of non-manualized family therapy interventions for adolescent behavior problems in usual care, even among providers who are not family therapists.

Ecological family therapy (FT) is an approach with strong empirical support for treating adolescent behavior problems that focuses on intervening directly with family members to repair intrafamilial relationships and addressing challenges encountered by adolescents and caregivers in key extrafamilial systems (Baldwin, Christian, Berkeljon, & Shadish, 2012). Despite their exceptional research portfolio, manualized FT models have not been widely adopted in frontline behavioral care. In response to implementation barriers commonly encountered with manualized treatments, experts in mental health, substance use, and behavioral health policy advocate an additional strategy to complement manual-driven dissemination: Focus on core elements of evidence-based interventions (EBIs) that represent a reduced set of intervention techniques common to multiple treatments for a given disorder (Chorpita, Daleiden, & Weisz, 2005; Chorpita & Daleiden, 2009).

The core elements strategy opens the door to investigating non-manualized EBIs that are practiced in everyday behavioral care. At this time, relatively little is known about how EBIs perform in clinically representative conditions. To advance the science of EBI dissemination and implementation, it is critical to learn more about whether and how non-manualized EBIs are delivered in usual care, and as importantly, whether EBI use is associated with meaningful improvement in the clinical disorders that prompt treatment referral (Garland, Bickman, & Chorpita, 2010; Weisz, Bearman, Santucci, & Jensen-Doss, 2017).

The current study advances EBI implementation science by testing whether delivery of core FT techniques implemented in usual care predicts long-term outcomes for adolescent behavior problems. To date, all studies of FT technique-outcome relations have focused on adherence to a manualized model, with “adherence” operationalized as utilization of manual-specific FT techniques. FT adherence-outcome studies have generally demonstrated significant results (Gillespie, Huey, & Cunningham, 2017; Hogue et al., 2008; Robbins et al., 2011), with each, in some form, demonstrating a positive correlation between adherence and outcome.

The current study is a secondary analysis of a randomized trial examining the effectiveness of non-manualized FT in usual care for adolescent behavior problems (Hogue et al., 2015). The current study extends these findings by examining direct associations between the use of specific FT techniques and clinical outcomes. Therapists provided post-session self-report data on use of core FT interventions; adolescent outcomes were assessed up to one-year post-baseline. We hypothesized that greater use of core FT techniques would predict greater decreases for the clinical outcomes assessed.

Method

Participants and Procedures

The study sample was drawn from clients enrolled in a parent randomized trial (Hogue et al., 2015). Clients from the parent trial were retained if (a) their therapist submitted at least one self-report checklist on FT technique use and (b) they completed at least one follow-up interview. The study sample was comprised of 70 adolescents averaging 15.4 years of age (SD = 1.5), 53% of whom were female. They self-reported Hispanic (64%), African American (16%), multiracial (11%), and other (9%) ethnicities. The most commonly diagnosed psychiatric disorders were: Oppositional Defiant Disorder (ODD) = 92%, Attention-Deficit/Hyperactivity Disorder = 77%, Conduct Disorder (CD) = 47%, Mood Disorder or Dysthymia = 44%, and Substance Use Disorder (SUD) = 20%. A total of 86% of the sample was diagnosed with more than one disorder.

The study was conducted under approval of the governing Institutional Review Board. Adolescents were referred from schools (80%), family service agencies (14%), juvenile justice or child welfare sources (3%), and other sources (3%). Participants were assessed at baseline and randomized to UC-FT versus UC-Other (described below) at the completion of baseline. Follow-up assessments were conducted at 3, 6, and 12 months following baseline with completion rates of 86%, 84%, and 87%, respectively.

The UC-FT condition consisted of a community mental health clinic that featured family therapy as the standard-of-care approach for behavioral interventions with youth. The site did not use a FT treatment manual or protocol of any kind. Fourteen therapists provided treatment for 39 participants. They ranged from 28 to 59 years; 7 were female, 7 were Hispanic American, and as a group they averaged 3.1 years (SD = 4.3) postgraduate therapy experience. Five did not provide demographic data.

UC-Other included a set of four clinics in order to sample outpatient treatment options widely available for teens. Two were community mental health clinics and two were outpatient clinics in child and adolescent psychiatry departments of teaching hospitals. Fifteen therapists provided treatment to 31 participants. These therapists ranged from 25 to 45 years of age; 11 were female, 9 were European American, and as a group they averaged 3.5 years (SD = 2.9) postgraduate therapy experience.

Measures

The FT technique measure is a therapist self-report scale that assesses delivery of discrete treatment techniques in routine care for adolescent conduct and substance use problems: Inventory of Therapy Techniques—Adolescent Behavior Problems (ITT-ABP; Hogue, Dauber, Henderson, & Liddle, 2014). The 8-item scale assesses the extensiveness (i.e., thoroughness and/or frequency) with which each technique was used in a just-completed session based on a 5-point Likert-type scale: 1 = Not at all, 2 = A little bit, 3 = Moderately, 4 = Considerably, 5 = Extensively. The seven FT techniques were: Targeted intervention efforts at a family member other than the adolescent in session; Discussed parental monitoring and family rules/caretaking with the adolescent and/or caregiver; Worked on enhancing communication and attachment between family members; Shared information about normative adolescent development; Discussed core relational themes that underlie everyday family events (e.g., love, trust, respect, independence); Worked individually with adolescent or caregiver to prepare for an in-session family interaction; Arranged, coached, and/or helped process interactions among family members in session.

Previous psychometric analyses provide support for its reliability and factor validity (Hogue et al., 2014), along with its criterion validity in comparison to observational ratings (ICC = .66) (Hogue, Dauber, Lichvar, Bobek, & Henderson, 2015). The FT scale score is computed by calculating the mean of all eight scale items; this constitutes the average extent of FT technique use reported for a given session.

Caregiver and adolescent reports of youth externalizing and internalizing symptoms were assessed via the Child Behavior Checklist (caregiver report) and Youth Self Report (Achenbach & Rescorla, 2001). For both the adolescent and caregiver versions, total scores on the externalizing and internalizing summary scales were analyzed. Adolescent delinquency was assessed with the Self-Report of Delinquency (Elliott, Huizinga, & Ageton, 1985); adolescents reported the number of times they engaged in various overt and covert delinquent acts since the previous assessment. Adolescent substance use was measured with the Timeline Follow Back Method (Sobell & Sobell, 1996), which assesses quantity and frequency of daily consumption of substances using a calendar and other memory aids.

Analytic Strategy

Latent growth curve (LGC) modeling (Duncan, Duncan, Strycker, Li, & Alpert, 1999) was used to examine FT technique-outcome relations. Missing data were handled with robust maximum likelihood estimation under the assumption that the data were missing at random (Little & Rubin, 2002). LGC was conducted using Mplus version 7.4 (Muthén & Muthén, 1998–2017). Due to the exploratory nature of the research, we report marginally significant effects (p < .10) along with those below the conventional p < .05 criterion. Treatment condition was included as a covariate, along with ethnicity, sex, and age, based on the results found in the parent trial, which indicated that UC-FT produced greater improvement than UC-Other on several outcomes, along with ethnicity, sex, and age. For all significant FT technique-outcome relations, we conducted an additional post hoc test of the condition by technique interaction. To control for therapist nesting effects we used the sandwich variance estimator (Diggle, Heagerty, Liang, & Zeger, 2002) available in Mplus, which produces corrected standard errors in the presence of nested data. Due to a “preponderance of zeroes” for delinquency and substance use, these outcomes were analyzed with two-part growth models (Brown, Catalano, Fleming, Haggerty, & Abbott, 2005). Such models are frequently used to analyze outcomes when a large proportion of the sample reports an absence of the given outcome variable, as they were in the parent trial (Hogue et al., 2015). For effect size estimates, we used β coefficients from LGC models estimating fully standardized effects.

Results

UC-FT therapists submitted 501 FT self-reports, and UC-Other therapists submitted 306 self-reports. Across the sample, internal consistency of the eight FT technique scale items was strong (Cronbach’s α = .80). There were 157 self-reports (104 for UC-FT, 53 for UC-Other) for which the given session was also rated by a nonparticipant coder using the same FT technique scale. The intraclass correlation coefficient representing the correspondence between therapist and observer report for these sessions was ICC = .66, which is considered good (Cicchetti, 1994). As expected, therapists in the UC-FT condition (M = 2.7, SD = 0.51) reported higher average technique scores than therapists in UC-Other (M = 1.9, SD = 0.57; t[68] = 5.98, p < .001). These scores indicate that FT technique scores in UC-FT fell between 2 (A little bit) and 3 (Moderately), whereas in UC-Other scores fell just below 2 (A little bit).

Results of LGC models in which FT technique use predicted client outcome across the 12-month follow-up are presented in Table 1; we report statistically significant results here. First, there was a significant relation between FT technique use and delinquency. Among those youth engaging in some delinquency, greater FT use was associated with greater decreases in delinquent acts (the continuous part of the two-part model; slope coefficient = −0.11, SE = 0.05, pseudo z = −2.21, p = .027, β = −0.62). Second, more extensive FT use was associated with larger decreases in adolescent-reported externalizing symptoms (slope coefficient = −1.40, SE = 0.53, pseudo z = −2.64, p = .01, β = −.68). Given the novelty of this research, we also report two marginal effects. Greater FT technique use was related to increasingly larger proportions of youth abstaining from substance use over follow up (the categorical part of the two-part model: slope coefficient = −.78, SE = 0.47, pseudo z = −1.65, p = .09, β = −.61). Greater FT use was also associated with larger decreases in caregiver-reported externalizing (slope coefficient = −1.66, SE = 0.92, pseudo z = −1.79, p = .07, β = −.89).

Table 1.

Coefficients and Standard Errors for Growth Parameters and Association with Family Therapy Technique Use by Treatment Outcome

Outcome	Growth Factor Mean		Growth Factor Variance		FT Technique Use
	Intercept	Slope	Intercept	Slope	Intercept	Slope
	Coeff (SE)	Coeff (SE)	Coeff (SE)	Coeff (SE)	Coeff (SE)	Coeff (SE)
Delinquency (Categorical)	N/Aa	2.64 (2.29)	2.92** (1.24)	0b	0.02 (0.60)	0.05 (0.39)
Delinquency (Continuous)	−0.76 (1.16)	0.41 (0.55)	0.17*** (0.05)	0b	0.31** (0.15)	−0.11** (0.05)
Substance Use (Categorical)	N/Aa	10.33***(2.61)	6.31* (3.79)	0b	2.50* (1.29)	−0.78* (0.47)
Substance Use (Continuous)	−1.29 (6.91)	−2.60* (1.35)	1.53***(0.40)	0b	0.87 (0.57)	−0.19 (0.16)
Adol. Report Internalizing	−13.92 (13.96)	5.34 (4.96)	54.60*** (10.15)	0b	−1.72 (2.07)	0.73 (0.60)
Adol. Report Externalizing	−16.10 (10.42)	14.95*** (3.36)	27.10*** (5.20)	0b	2.88 (1.97)	−1.40*** (0.53)
Parent Report Internalizing	−2.76 (8.36)	−4.78 (4.42)	29.75*** (10.39)	0b	2.50 (1.86)	0.24 (0.47)
Parent Report Externalizing	−8.51 (13.96)	−1.86 (4.74)	53.06*** (15.49)	0b	4.67* (2.61)	−1.66* (0.92)

Note. Growth Factor Mean indexes the mean growth trajectory collapsed across treatment conditions and adjusted for adherence-outcome relations (reported in column 6 and 7) and treatment condition (UC-FT vs. UC-Other). Growth Factor Variance indexes individual variation around the mean growth trajectory. Adol = Adolescent, Coeff = Regression coefficient, SE = Standard error

^aValue fixed to 0 for model identification, necessary for fitting latent growth curve models with categorical outcomes.

^bValue fixed to 0 to facilitate model convergence.

^*p < .10

^**p ≤ .05

^***p ≤ .01

There was one significant FT technique use by condition interaction. Results for the continuous part of the delinquency trajectories (slope coefficient = −0.21, SE = 0.09, pseudo z = −2.38, p = .02, β = −.61) suggested that greater FT use more strongly predicted reduced delinquent acts in the UC-FT condition than in UC-Other. However, when the two study conditions were analyzed separately, within-condition effects were not statistically significant for either, suggesting that the FT technique-outcome effects observed in analyses with the full sample were not primarily due to disproportionate influence from the UC-FT condition

Discussion

This study found that more extensive use of family therapy techniques by community therapists in routine behavioral treatment was associated with long-term gains in multiple clinical outcomes for adolescents. Specifically, greater use of core FT techniques during treatment predicted significant reductions in adolescent-reported delinquent acts and adolescent-reported externalizing symptoms at one-year follow-up, along with marginal associations with decreases in adolescent-reported substance use and caregiver-reported externalizing.

These results for non-manualized, core FT techniques in usual care are consistent with studies of manualized FT models in controlled settings (Hogue et al., 2008; Robbins et al., 2011), in which stronger adherence to model-specific techniques predict improved outcomes in conduct problems and substance misuse (Baldwin et al., 2012). Also, FT technique effects were found for espoused family therapists and non-family therapists alike, even though family therapists predictably delivered a stronger dose of core FT interventions by a diverse set of community therapists are encouraging for the youth behavioral care sector. That said, study results do not assert that non-manualized FT implemented in everyday practice is fundamentally equivalent to manualized FT implemented with extramural training and monitoring by model experts.

Strengths and Limitations

The sample featured community therapists operating in standard clinical settings, that is, without benefit of extramural resources of any kind. In addition, adolescent clients were diagnosed with an array of clinical disorders, and comorbidity was the norm. These are conditions of high ecological validity that affirm the generalizability of study findings to real-world practice. The therapist self-report measure used in the study was verified by comparison to the field’s gold standard, observational ratings (Hogue et al, 2015).

A primary study limitation is the small sample size, which prevented further investigation of intriguing research questions such as potential therapist differences in technique-outcome effects, differential effects for individual FT techniques or subsets of techniques, and more complex statistical procedures combining adolescent- and caregiver-reported data (e.g., dyadic latent growth curve modeling). The number of participating sites was too small to permit generalizable claims about the practice habits of providers, and only one clinic was staffed by credentialed family therapists. Indeed, the number of sites was too small to control or test for site differences, which are surely meaningful for understanding technique-outcome relations in routine settings. Study eligibility criteria required that caregivers agree at baseline to participate actively in treatment if requested, biasing the sample toward FT-amenable families. The strongest technique-outcome effects were registered for adolescent-report variables, perhaps reflecting common source effects; though notably there were also meaningful effects for caregiver report of externalizing symptoms. Therapists varied in the number of checklists submitted, which may inject some bias in the findings if therapists who used more FT techniques were more likely to submit checklists. Finally, because we did not experimentally manipulate FT delivery, we cannot state definitively that greater FT technique use was a cause of client improvements; it is possible, for instance, that more effective therapists in general were simply more likely to be generous in using FT techniques.

Study results illuminate findings of the parent trial (Hogue et al., 2015), in which family therapists outperformed non-family therapists for some adolescent behavior problems. The findings of this study suggest that the superior outcomes might be due to greater utilization of core FT techniques. It appears that non-manualized FT can work well by featuring many of the same techniques as “brand name” FT models for improving adolescent conduct and substance problems. Further, the findings support the predictive validity of the ITT-ABP, which is an important first step in establishing its clinical validity. Future research to expand its clinical utility should focus on the differential predictive validity of specific items and perhaps the sequencing of the techniques. While the overall study findings should be met with enthusiasm by therapists, supervisors, and administrators interested in incorporating FT into standard practice, there is still much work to be done to delineate the merits and limitations of fidelity to core FT interventions.

Appendix: Data Transparency Disclosure

Findings from the data collection have been reported in separate manuscripts. MS 1 (published) reports the primary outcome results of the RCT. MS’s 2–4 (published) report psychometric analyses of the fidelity instrument used as the measure of FT techniques in the current study. MS’s 5 and 6 (published) report secondary analyses of the RCT data, with MS 5 focused on predictors of treatment outcomes apart from treatment condition and MS 6 focusing specifically on youth reports of callous-unemotional traits and their relation to treatment outcome. MS 7 reports on whether one study condition, UC-FT, achieved a benchmark for FT adherence established by a manualized FT model, as well as benchmarks for global outcomes established in a meta-analysis of several FT models. In contrast, the current manuscript focuses on both study conditions (UC-FT and UC-Other) and is the first study from this data collection to examine whether FT technique use is related to treatment outcome. The current findings have not been previously published, and to our knowledge, no similar analyses of technique-outcome relations for non-manualized FT (or non-manualized EBIs of any kind) in usual care for adolescent behavior problems have been previously published.