The effects of emotion-focused skills training on parental mental health, emotion regulation and self-efficacy: Mediating processes between parents and children

Abstract Objective: Emotion-Focused Skills Training (EFST) is a short-term parental intervention based on humanistic principles. While studies have demonstrated the efficacy of EFST in alleviating child mental health symptoms, the mechanisms by which this happens is less clear. The present study investigated whether program participation led to improvements in the parents’ own mental health, emotion regulation, and self-efficacy, and compared two versions of EFST: one experiential involving evocative techniques, and one psychoeducational involving didactic teaching of skills. Further, this study investigated whether improvements in parent outcomes mediated the effects on children’s mental health. All parents received 2-days group training and 6 h of individual supervision. Methods: 313 parents (Mage = 40.5, 75.1% mothers) of 236 children (ages 6–13, 60.6% boys) with mental health difficulties within the clinical range and their teachers (N = 113, 82% female) were included. Participants were assessed at baseline, post-intervention, and 4-, 8- and 12-months follow-up. Results: Multilevel analysis showed significant improvements over time on all parental outcomes with large effects (drange0.6–1.1, ps < .001), with fathers benefitting more in terms of emotion regulation and self-efficacy (ps < .05). Significant differences were found between conditions on parental mental health and self-efficacy (all p’s > .05). Cross-lagged panel models showed indirect effects of child symptoms at post-intervention on all parental outcomes at 12-months follow-up (βrange0.30–0.59, ps < .05). Bidirectional associations were observed between children’s mental health symptoms and parental self-efficacy (βrange0.13–0.30, ps < .05). Conclusion: This study provides support for the effect of EFST on parent outcomes and the reciprocal relationship between the mental health of children’s and their parents. Trial registration: ClinicalTrials.gov identifier: NCT03807336.


Data Transparency Statement
The data reported in this manuscript have not been previously published but form part of a larger data collection.Improvement in children's externalizing and internalizing symptoms has been the subject of a previous paper, describing a randomized clinical dismantling study (Ansar et al., 2022).Its results showed statistically and clinically significant reductions in children's internalizing and externalizing symptoms over time with strong effects on both outcomes (Ansar et al., 2022).In addition, a parallel qualitative study, employing a reflexive thematic analytic approach, has been conducted based on 15 parents who participated in the program.The paper (Ansar et al., 2021) focused on exploring parents' experiences on receiving the program, their perceived changes in everyday life, and their views on the effect of the program.Parents reported that they gained practical wisdom, experienced a new calmness, and felt less anxious in their role as parents.
Mental health problems in children and adolescents have been increasing over the recent decades with up to one in five children experiencing such difficulties, causing significant burden on individuals and their families (Erskine et al., 2015;Whitney & Peterson, 2019).Most mental health problems experienced in adulthood commence in childhood and adolescence (Beslky et al., 2020;Kessler et al., 2012;Yap et al., 2014), and externalizing and internalizing problems are amongst the most prevalent life-time mental disorders (Catalano & Kellogg, 2020).Externalizing problems tend to be more observable around 6-7 years, reaching a peak around 13 years (Bongers et al., 2004), while the age of onset of internalizing problems is around 8 years, and tend to increase as children move into adolescence (Wolff & Ollendick, 2006).The comorbidity rates during the primary school years exceed those expected by chance (Bitsko et al., 2013); Fanti & Henrich, 2010;Wang & Liu, 2021), making this period particularly suited for transdiagnostic treatments (Willner et al., 2016).
Several studies have shown that parents play a key role in their children's mental health and wellbeing (Belsky et al., 2020;Compas et al., 2017).Parental distress reduces parents' emotional responsiveness towards their child, and highly stressed parents tend to use ineffective and invalidating parenting strategies (O'Connor et al., 2020;Rayce et al., 2020).In addition, parents and their children's mental health are linked to each other, as children of depressed and anxious parents are more likely to develop internalizing and externalizing problems (Arikan & Kumru, 2021).Although there is solid empirical basis for claiming that parental factors affect children's mental health symptoms (i.e., parent-driven effects) (Eckshtain et al., 2019;Galbally & Lewis, 2017;Verhage et al., 2016), the reverse is also likely the case.Indeed, some studies suggest that children's mental health and symptomatology mediate parental mental health (i.e., child-driven effects) as much or even more than the reverse (Culp, 2010;Morris, Criss, et al., 2017;Serbin et al., 2015), suggesting a bidirectional relationship between the two.
Interventions that target parental skills are considered among the most effective strategies to prevent and treat mental health problems in children (Poole et al., 2018;Reedtz & Klest, 2016;Thomas et al., 2017).Notably, programs aimed at enhancing parental skills to alleviate children's symptoms, have demonstrated additional effects on the parents' own mental health, beyond what has been observed in studies of treatments focusing solely on either parent or child (Perrin et al., 2020).As such, parenting interventions can be seen as a form of mediational therapy, in which the child is targeted indirectly by assisting parents in developing skills to help their child and improving their own mental health as well (van Aar et al., 2017).The additional effect on the mental health of the parents (Galbally & Lewis, 2017), indicates that changes in the parent may go via two routes: (1) via the acquisition of parental skills, and (2) via changes in parental mental health.Given the research that suggests a reciprocal relationship between children's and parental mental health, it is useful to understand the mechanisms by which they influence each other (Sandler et al., 2011).
One hypothesized way by which parents influence their children's mental health is by helping them make sense of and regulate their emotions (Kehoe et al., 2020).Hence, one proposed mechanism is emotion regulation.Parents' sensitive attunement to the child's affects has been shown to help children regulate emotions (Crnic & Ross, 2017), and children's emotion regulation capacities have been found to mediate externalizing and internalizing problems (Morris, Houltberg, et al., 2017).Thus, emotion regulation processes might be understood as a leverage point for parental intervention programs.
Another potential route to change could be the acquisition of skills.Parental self-efficacy: the degree to which parents feel confident and competent in dealing with their child's emotions and behaviors (Bandura et al., 2003), might be one way to tap into skills acquisition.Several studies have shown that parents' perception of themselves in terms of their capacity to emotionally guide their children, mediates the development of mental health Psychotherapy Research 519 symptoms in their children (Lafrance et al., 2015;Strahan et al., 2017).However, how these mechanisms work to ameliorate symptoms, is less clear.Investigating potential mediators may provide a clearer picture of how the change in children's mental health symptoms takes place (Kazdin, 2007).
Although both maternal and paternal symptomatic distress has been found to predict children's mental health difficulties, parenting intervention research has largely excluded fathers (Panter-Brick et al., 2014).This gap in the research prevents any conclusions concerning the relative importance of fathers versus mothers.In addition to fathers tending to be underrepresented in parenting programs, research also finds that they are less likely to benefit from program attendance compared to mothers (Wells et al., 2016).In this context, examining the effects of both paternal and maternal involvement may give valuable information about how the program effects parents on both genders.
There is a growing interest in transdiagnostic emotion-focused parenting interventions for the treatment of children's mental health symptoms (Ehrenreich et al., 2018;England-Mason & Gonzalez, 2020).One such program is the recently developed intensive, short-term program, Emotion-Focused Skills Training (EFST) (Dolhanty et al., 2022;Lafrance et al., 2015).EFST is founded in emotion theory and research (Izard, 2002), Emotion-Focused Therapy (Greenberg, 2015), emotion coaching programs (Gottman et al., 1997), and humanistic psychotherapy (Elliott et al., 2004).EFST aims to increase parents' capacity to respond adaptively to their child's emotions as well as working with the parents' own emotional understanding and expression.A central assumption in EFST is that parents and children influence each other reciprocally through their relational bond (Greenberg, 2015).Emotion-focused programs have been found to improve children's mental health (Foroughe et al., 2019;Kehoe et al., 2020), reduce parental stress and burden (Lafrance et al., 2015), improve parents' ability to regulate their children's emotions (Yap et al., 2014) and enhance parental self-efficacy (Strahan et al., 2017).
Experiential techniques, such as two-chair dialogues, are considered essential in Emotion-Focused therapy, and research suggests that such techniques aid emotional processing and add to the effectiveness of psychotherapy (Elliott et al., 2013;Goldman et al., 2006).Although it is possible that this also applies to EFST, studies have not yet explored this assumption.The primary analysis connected to the current study, a randomized clinical dismantling study, tested the efficacy of EFST in alleviating symptomatic distress in children aged 6-13 (Ansar et al., 2022).Parents of 236 children between 6 and 13 years with externalizing (EXT) and/or internalizing (INT) problems within clinical range were randomly allocated to one of two treatment conditions: one experiential condition (n = 120) involving both didactic psychoeducation and explicit facilitation of emotion activation (using emotionally evocative techniques and two-chair interventions), and one psychoeducational only condition (n = 116) comprising solely of a didactic psychoeducational format without the explicit use of experiential techniques.In this study, only data from the one parent who initiated contact was included.Results showed that EFST significantly reduced parent-reported EXT and INT symptoms, as well as teacher-reported EXT but not INT symptoms.No significant differences between the conditions were found.The results are promising; however, we do not know how the program affects the mental health of the parents and how changes in the child occurs.Few existing studies have examined the reciprocity of perceived changes in children's and their parent's mental health by comparing two conditions and assessed the role of fathers as well as mothers.These questions will be addressed in the current study.

Aims and Research Hypotheses
The present study included data from parents and their children to investigate the effect of Emotion-Focused Skills Training (EFST) on the mental health (i.e., symptomatic distress), emotion regulation, and self-efficacy in parents of children aged 6-13 experiencing mental health difficulties within the clinical range, disaggregating effects by parental gender.Further, this study investigated whether evocative experiential techniques (EXP) enhanced the treatment effect compared to a psychoeducational format without two-chair interventions (PE), and whether there were indirect associations between mental health difficulties in children, and parental self-reports on mental health, emotion regulation, and self-efficacy, as reported by parents and the children's teachers.Based on prior research and theory, the following hypotheses were posed: 1 (a) EFST is effective in improving parental mental health, emotion regulation, and self-efficacy; (b) parental gender may impact the effectiveness of the intervention, with mothers gaining more than fathers; (c) the experiential condition will outperform the psychoeducational condition, and (2) The improvements in parent outcomes (their mental health, emotion regulation, and self-efficacy) and children's mental health will display a reciprocal pattern such that parental outcomes and children's mental health symptoms will mutually influence each other.

Study Design and Procedure
This is a secondary analysis of a previously reported RCT (ClinicalTrials.govregistration: NCT03807336), whose methodology has been reported in depth (Ansar et al., 2022).The study was approved by the regional committee for medical health and health research ethics (case 2018/754), and the treatment was administered at two different outpatient clinics at the Institute of Emotion-Focused Therapy in the two largest cities in Norway (Oslo and Bergen).Between August and December 2018, a total of 657 parents were recruited through newspaper features, advertisements, primary schools, public health services and clinical referrals.Caregivers who responded were pre-screened via telephone interviews conducted by the clinical experts in the project.Participants who were deemed eligible for treatment were invited to the clinic where they completed the initial assessment comprising a face-to-face interview and an online symptom questionnaire (BPM-P; Achenbach et al., 2011).Inclusion criteria were primary caregivers of children (6-13 years) who exhibited difficulties within the clinical range on at least one of the Internalizing or Externalizing scales of the Brief Problem Monitor for parents (BPM-P), with a cutoff score of 65, based on previous research and their T scores on BPM-P, adjusted to age, sex, and ethnicity (Achenbach et al., 2011).Of those who were included, both fathers and mothers were encouraged to participate in the program.Parents were excluded if the parent or the target child was already receiving any kind of psychotherapeutic treatment.Children with pervasive developmental disorders (such as severe autism or Asperger syndrome, information on which was obtained in the assessment interview) were also excluded, as these children require systematic treatment beyond a short-term parental program.After the intake interview, parents who agreed to participate were randomly allocated to one of the treatment conditions.To avoid experimenter biases interfering with the randomization, an electronic randomizer resource was used. 1 See Figure 1: CONSORT flowchart.All outcome measures were reported by parents and the child's main teacher at baseline, post-intervention, and at 4-, 8-, and 12months FU.The program was free of charge.

Participants
Parents and teachers.A total of 313 parents (M age = 43.2,SD = 5.6) were found eligible and included into the program.Although both parents were encouraged to participate, the sample consisted of 234 mothers (74.8%, [M age = 42.8,SD = 6.7]) and 79 fathers (25.2% [M age = 43.6,SD = 4.8]) seeking help for their children (N = 236, ages 6-13 (M age = 8.9, SD = 2.2), 143 boys [60.6%] and 93 girls [39.4%]) with EXT and/or INT problems within the clinical range.In 78 of the cases, (33.1%) both parents participated together in the program.Most of the parents were partners (74.8% married/cohabitants), 12.4% were separated/divorced, and 11.8% were single.The average educational level in the sample was higher than the population average 2 : 12.5% had completed upper secondary education (μ = 60.9%),5.3% had completed tertiary vocational education (μ = 3.1%), and 82.2% had completed higher education (μ = 36%).The average family income was less than $39' for 6.2%; $39-61 for 29.8%, and $83' or above for 64% of the sample, which is at the population average (μ = $98').Among one third of the caregivers (N = 118, 37.9%) had not received any earlier psychiatric treatment for their children (μ = 25%), while 163 (52.1%) had received earlier treatment one time or more due to children's mental health issues.Two thirds of the sample (N = 163, 69.1%) showed both INT and EXT difficulties within clinical range, with a moderate significant correlation (T ≥ 65, r = .453,p = .001),reflecting the comorbidity in the sample.No significant baseline differences were found between the EXP and PE samples in age, gender, social status, or mental health symptoms.Descriptives are presented in Table I.The primary school teachers of all children were invited to report on the children's mental health symptoms, however, only about half of the teachers chose to participate (N = 113, 82% female).No other teacher data was collected.
Therapists.Thirteen therapists (two men and eleven women) provided the treatment.Eleven were clinical psychologists, and two were family therapists with 5-15 years of clinical experience (M = 10.4,SD = 3.8).All therapists were trained in EFST, involving four-day group training, 15 hrs of supervision, and at least one year of experience conducting the program.All therapists received an additional day of training prior to the trial to promote treatment integrity.The group training was led by four of the therapists.The therapists received weekly supervision from two of the authors and at least 10 hrs of expert supervision during the study.After every session, the therapist filled in a session-form to check whether they had completed all components of the session according to the manual.

Treatment
The manualized program comprised a two-day group-training and six hours of supervision for Psychotherapy Research 521 either one or both parents of the included child.The group training was delivered in groups of 20-36 participants and individual supervision within a 3-4 weeks' time frame.The entire program was implemented within 12 weeks.The rationale for the program, group training, and supervision plan was described in a treatment manual for the therapists (Dolhanty et al., 2022), and involves training in four core parenting skills: (1) Validation of emotion.In the first phase, parents are taught to identify and strengthen their innate ability to attend to their child's emotions and to name, validate, and meet their child's needs.Additionally, parents are trained in empathy skills and how to differentiate between emotional states.(2) Enhancing motivation.Here, parents are aided in working through their own problematic emotions that can get in the way of helping their child (i.e., rejecting anger, fear, or self-blame) to enhance their emotion-processing capacity and increase their motivation to meet their child's needs.(3) Resolving interpersonal injuries.
Here, the goal is to help parents express responsibility in repairing past emotional injuries in relation to the child, thus attempting to alleviate feelings of rejection or self-blame in the child.(4) Boundaries.By strengthening new learning, parents are given training in how to set sound and flexible boundaries, enabling them to allow the child to rely on them.
Parents were randomly allocated to one of two EFST conditions.The experiential (EXP) condition included experiential tasks (evocative empathy and two-chair dialogues) to access the parent's problematic emotions, such as fear and self-blame.The goal was to facilitate the participants' emotional processing through awareness, acceptance, making sense of their own emotions, and to facilitate understanding the emotional reactions of the child.In the psychoeducational (PE) condition, parents received the same EFST skills program as in EXP, except for using experiential tasks and two-chair dialogues.In group-training and supervision, participants were informed of the treatment rationale didactically with active therapeutic engagement, guidance, and supervision to increase understanding of how emotions work, awareness, and acceptance of their Psychotherapy Research 523 own and their child's emotions, and motivation to take responsibility and engage in new ways of relating to their child.After group training, parents were assigned individual supervision where they could opt to continue the work on one of the four core skills that best suited their situation to help them adjust the skills learning to their child's individual needs.Therapists were instructed to include work with all four skills during the supervision period.

Measures
Parent outcomes.Outcome Questionnaire-45.The OQ-45 (Lambert et al., 2004) is a 45-item self-report measure that uses a 5-point Likert-type scale (ranging from "never" to "almost always"), designed to track changes over the course of psychotherapy.The instrument yields a total score, and three subscales: symptomatic distress, interpersonal relations, and social role.The OQ total score ranges from 0 to 180, with higher scores indicating greater distress.Scores at or above 63 indicate clinical distress.The test-retest reliability for non-treatment samples has been estimated at .84 and internal consistency reliability at .93 (Lambert et al., 2004).Internal consistency for the current sample was good (α = .93).
Difficulties in Emotion Regulation Scale (DERS) (Gratz & Roemer, 2004) is a 36-item scale that uses a 5-point Likert scale (1 = almost never, 5 = almost always) assessing emotion regulation difficulties across six domains: emotional non-acceptance, difficulties controlling impulsive behaviors and engaging in goal-directed behaviors, limited access to effective regulation strategies, and lack of emotional awareness and clarity.The DERS has been found to have good test-retest reliability and construct and predictive validity (Gratz & Roemer, 2004).Internal consistency in the current study was good (α = .88).
Parenting Sense of Competence (PSOC) (Johnston & Mash, 1989) is a 16-item scale measuring parents' satisfaction, interest, and efficacy in their parental role.Each statement is rated using a 6-point Likert scale (1 = strongly disagree, 6 = strongly agree).The total scores range from 16 to 96, with higher total score indicating a greater sense of competence in parenting.The PSOC has been validated in a normative sample comprising parents of school-age children and shows good psychometric properties in terms of internal consistency, test-retest reliability, and convergent validity (Johnston & Mash, 1989).Internal consistency within the current sample was good (α = .90).
Children's outcomes.Brief Problem Monitor for ages 6-18 (BPM) (Achenbach et al., 2011).BPM for parents (BPM-P) and their teachers (BPM-T) is a 19-item scale (0 = not true, 1 = somewhat or sometimes, 2 = very often true) to assess children's mental health symptomology.Both BPM-P and BPM-T are widely used instruments that evaluate a broad array of behavioral and emotional child difficulties and have a total score and three subscales: Internalizing problems (anxiety, depression, and withdrawal); Externalizing problems (aggression and rule-breaking behavior); and Attention/Hyperactivity problems (Chorpita et al., 2010).There is extensive evidence supporting the psychometric properties of both measures (Achenbach et al., 2011).The internal consistencies for the present sample were good (α = .82on BPM-P and α = 0.88 on BPM-T).The current study included only the total scores for each instrument.

Statistical Analyses
All data were analyzed in accordance with the intention-to-treat-principle, i.e., all cases included were analyzed.Data analyses were performed using SPSS version 28.0 (IBM Corp, 2021) for descriptive statistics and multilevel analyses (using the Linear Mixed Modeling option), and its add-on structural equation modeling module, Analysis of Moment Structures (AMOS, version 26) for the cross-lagged mediation analyses (Collier, 2020).Chi-square (χ 2 ) tests were applied for categorical data and ANOVA and t-tests were used for baseline differences between conditions.
To investigate research hypothesis 1; (a) EFST is effective in improving parental mental health (OQ-45), emotion regulation (DERS), and parental selfefficacy (PSOC), (b) the assumption that mothers will gain more than fathers, and (c) that the EXP condition will outperform the PE condition, multilevel growth curve modeling (MLM) was utilized.An important reason for using MLM is to account for non-independence in the data due to nesting.Failing to account for data dependency could result in an underestimation of the standard errors, which could lead to an inflated Type I error rate (Raudenbush & Bryk, 2002).MLM is also robust in allowing for missing observations, which is a typical problem in longitudinal clinical research (Hox, 2010).Following the procedures recommended by Singer and Willett (2003), we built models successively adding variables and interaction terms according to our hypotheses.These variables acted as covariates in the adjusted regression models.Because repeated measures (level 1) were nested within parents (level 2), we used a two-level hierarchically nested growth model to analyze the effect of EFST.We also tested for between-therapist effects using a three-level growth model with repeated measurement occasions (level 1) nested within parents (level 2) who were nested within therapists (level 3).However, the variability (ICC) due to being treated by one therapist rather than another was nonsignificant, and the three-level model resulted in poorer model fit.Hence, we dropped the third level and opted for a MLM growth curve model with two levels.
In our modeling procedure, Model 1 (the "null model") included only a fixed intercept and a fixed slope for time (representing all measurement points including follow-up data).In model 2, we added "parent gender" as a dummy-coded variable (with male = 1 and female = 0) to investigate the potential differential effects of outcomes between mothers and fathers (using an interaction term: parent gender *slope).In model 3, "treatment condition" was added (with EXP = 1, PE = 0) to investigate the potential differential effect of outcomes between conditions (treatment condition * slope).Akaike information criterion (AIC) was used to compare the overall fit of the models on a smaller-is-better basis (Akaike, 1974).An alpha level of 0.05 was chosen for all tests, and all tests were two-tailed.Log-linear and quadratic modeling of time were also tested; however, none of these improved the model fit (i.e., AIC was 8028 vs. 7609).Hence, we opted for linear modeling of time.We report the results of our most parsimonious and best fitting model to facilitate interpretation.The time intervals between the measurement waves were identical from post-treatment to one-year follow up (4 months).Random intercepts and slopes were included, but only random intercepts were significant; thus, we discarded random slopes from the models.
To investigate research hypothesis 2, whether parental mental health, emotion regulation and parental self-efficacy are putative mediators (i.e., possible mechanisms of change) when receiving EFST, a three-wave cross-lagged panel modeling (CLPM) for longitudinal mediation analyses was performed (Arbuckle, 2019;Little et al., 2007), following Kazdin's criteria's (Kazdin, 2007).CLPM are considered ideal for examining temporal associations between variables as they take into account various sources of error, such as the stability of the variables and prior associations (Lüdtke & Robitzsch, 2021).This also allows for a simultaneous estimation of the indirect effects of mediators as it accounts for other sources of variance (e.g., autocorrelations between variables across time).Establishing a timeline with three time points are considered optimal for testing the direction of mediation effects (Kazdin & Nock, 2003).Multiple mediators were included separately to investigate how one mediator account for the change, and to determine more precisely which variables influence each other over time (Cole & Maxwell, 2003).
For the purpose of the present study, two main CLPMs were tested: one based on teacher-reported child symptoms (BPM-T; Figure 5), one based on parent-reported child symptoms (BPM-P; Figure 6); and their relationship to the three mediators: (A) parental mental health (OQ-45); (B) difficulties in emotion regulation (DERS); and (C) self-efficacy (PSOC) across three waves at baseline (T1), postintervention (T2) and 1 year follow-up (T5).Full information maximum likelihood procedure was employed (Arbuckle, 2019).Skewness and kurtosis levels for all variables were within the levels recommended for CLPM's (Tabachnick & Fidell, 2014).Cutoff values suggested in the literature were considered to detect collinearity based on correlations (r ≥ .85 or r ≥ .90)and were all below the levels considered problematic for multicollinearity (Hair et al., 2010).Standard errors and 95% bias-corrected CI calculations (Lenhard & Lenhard, 2016) were estimated using the AMOS bootstrapping procedure (based on 1000 iterations), without accounting for missing data.An alpha level of .05 was chosen for all tests, and all were two-tailed.The fit of the CLPMs was assessed using multiple model fit indices; the chisquared (χ 2 ) statistics to detect the best-fitting model compared to the thresholds of CFI (≥ 0.95), RMSEA (≤ 0.05), TLI (≥ 0.95), and SRMR (≤.08) (Hu & Bentler, 1999).Regarding direct effects, we considered regression coefficients of .1 as small, .3 as medium, and .5 as large (Cohen, 1992).For the indirect mediation effects, we considered .01 as small, .09as medium, and .25 as large (Kenny, 2020 3 ).

Attrition and Missing Data
The number of respondents in all measurement occasions is presented in the CONSORT diagram (Figure 1).In the sample, 92.3% of the parents (mothers and fathers) completed the whole intervention (T1; group training and individual supervision).The overall missingness in the three follow-up occasions was larger than expected (29.7% at T2, 51.2% at T3, and 60.2% at T4). Missingness was analyzed following the procedure outlined by Buuren (2018) and assumed missing at random (MAR), with condition, time, and previous symptom severity predictive of missingness.However, we could not refute the assumption of missing not at random (MNAR) as the pattern of missingness.Based on parent reports, age, gender and condition were not significant predictors of outcome variable missingness (all ps > 0.5); however, time and children's pretreatment symptom severity were χ 2 [311] = 58.37 (p = .003).Based on teacher reports, the dropout rate was less than 20% (10.6% at T2, 12.3% at T4, and 17.7% at T5) and Psychotherapy Research 525 there were no significant predictors of outcome variable missingness, such as age, gender, condition and/ or symptom severity (all p's > 0.5).

Multilevel Growth Curve Modeling
Table III and Figures 2-4 depict all parental outcomes.With regard to modeling parental mental health (reported by OQ-45), the fixed intercept in the sample was 57.25 (SE = 1.29, 95% CI 54.66; 59.72), which suggests a non-clinical range for the sample as a whole (clinical cutoff ≥ 65).The coefficient was significant (p < .001),showing that the symptom scores varied significantly at baseline.In Model 1, we investigated linear effects of time (treatment period) in an "empty" model (with no covariates other than the time variable, which was coded 0, 1, 2, 3 and 4, with 4 months' intervals between each).This model indicated a significant reduction on the OQ-45 total score; the estimated fixed growth curve was b = -1.83,SE = .21,p < .001,indicating that, on average, parents reported statistically significant reductions in total mental health symptoms over time.Model 2 shows that the gender variation in intercepts was significant [est.= −15.35,SE = 3.52, p < .001],meaning the initial symptoms differed significantly between genders, with women showing more symptom severity than men; however, adding time as a covariate yielded no significant gender differences.Model 3 shows that the variation in intercepts between conditions was insignificant [est.= 3.22, SE = 3.1, p = .31],suggesting successful randomization.The interaction between slope and condition was significant, in favor of the EXP [b = -1.24,SE = .55,p = .02],meaning that parents in the EXP condition had greater outcome than in the PE.
Regarding parental emotion regulation difficulties (reported by DERS), the fixed intercept in the sample was 78.30 (SE = 1.44, 95% CI 76.05, 80.55), and they differed significantly at baseline (p < .001),suggesting an initial within-person variation in DERS scores.Model 1 indicated significant reduction in emotion dysregulation across time; b = -2.04,SE = .27,p < .001.Model 2 shows that the Psychotherapy Research 527 variation in intercepts between genders was significant [est.= -6.97,SE = 2.92, p = .017],again pointing to significant differences between the genders at baseline, with mothers showing higher scores than fathers.However, the interaction between slope and gender was significant in favor of fathers [b = -   III), indicating a better outcome for the EXP condition.For DERS, Model 2 had the best fit, for OQ45 and PSOC, Model 3 had the overall best fit (the deviance score was significantly reduced from Model 2 to 3).For details, see Table III and Figures 2-4.

Cross-Lagged Panel Models
In order to test hypothesis 2, whether parental change (measured by OQ-45, DERS, and PSOC) mediated changes in children's mental health symptoms (reported by BPM-P and BPM-T) and vice versa, cross-lagged panel models were conducted.Based on teacher-reported child symptoms (BPM-T), the fit indices suggested that all three models had good to excellent fit to the data, indicating that all parameter estimates can be interpreted.Figure 5 presents the fit indices, correlations, and the standardized regression coefficients of all study variables.Model A-parental mental health outcome (OQ-45) showed two significant indirect paths from OQ-45 T0 to BPM-T T1 (β = 0.18, b = 0.60, SE 0.03, p = 0.05), indicating that parental mental health at baseline significantly predicted teacher reported children's mental health symptoms at post-intervention, and from BPM-T T1 to OQ-45 T4 (β = 0.59, b = 1.20,SE 0.26, p < .001),suggesting that teacherreported child symptomatology at post-intervention significantly predicted parental mental health outcomes at one-year follow-up.In model B-parental difficulties in emotion regulation (DERS), two significant indirect child-driven paths were found (i.e., the extent to which changes in children's mental health symptoms predict changes in parental emotion regulation) from BPM-T T0 to DERS T1 Psychotherapy Research 529 (β = 0.32, b = 0.83, SE 0.35, p = .047),and from BPM-T T1 to DERS T4 (β = 0.48, b = 0.95, SE 0.38, p < .001).Also, in Model C-parents' sense of competence (PSOC), two significant negative unidirectional child-driven paths were found: from BPM-T T0 to PSOC T1 (β = -0.48,b = -0.72,SE 0.36, p = .049)and from BPM-T T1 to PSOC T4 (β = -0.51,b = -0.70,SE 0.24, p < .001),indicating that teacher-reported child symptoms at baseline and post-intervention mediated parental emotion regulation and self-efficacy at post-intervention and one year after the intervention (Figure 5).
Figure 6 shows the results of the CLPMs for parent-reported child symptomatology (BPM-P) and parental OQ-45 (Model A), DERS (Model B) and PSOC (Model C).Overall, all three models showed good model fit (Figure 6).Two indirect parent-driven associations were found: Model A from OQ-45 T0 to BPM-P T1 (β = 0.11, b = 0.44, SE 0.02, p = .047)and Model B from DERS T0 to BPM-P T1 (β = 0.11, b = 0.48, SE 0.20, p = .044),indicating that parental mental health and emotion regulation are positively related to child outcomes.From a child-driven perspective, two unidirectional In sum, the strongest associations between the three waves (pre, post, and follow-up) were found from teacher-reported child mental health symptoms at post-intervention to parental mental health symptoms, difficulties in emotion regulation, and parental self-efficacy at one years' follow up (β range = 0.48-0.59),indicating that teacher-reported child Psychotherapy Research 531 symptoms mediate parental mental health, emotion regulation and self-efficacy (Figure 5).Parentreported child mental health symptoms showed the same pattern, with the strongest associations from post intervention to one year's follow-up with (β range = 0.30-0.48),both depicting a clearer picture of child-driven effects, i.e., parental emotional change depended on the perceived improvements in children's symptoms (Figure 6).

Discussion
EFST has been found to be effective in reducing children's mental health symptoms (Ansar et al., 2022).However, little is known about how the program works.Since the research on EFST is at a preliminary stage, mediating processes that could account for program effects and alternate pathways that may influence the long-term effects of such parenting programs, are important in understanding possible mechanisms of change (Sandler et al., 2011).The current study examined the effect of EFST on parent outcomes, the differential effect of gender and treatment condition, and investigated whether the effects of EFST on child outcome were associated with parents' improving in their own mental health, emotion regulation strategies, and self-efficacy.Additionally, we also tested whether the changes in parents were associated with child symptom reduction and vice versa.The study included longitudinal data with multi-informant ratings (mothers, fathers, and teachers), enabling close tracking of various effects of the program.
In accordance with the first research hypothesis, multilevel analyses indicated that parental mental health, self-efficacy, and emotion regulation significantly improved over time with large effects on all outcome variables.Effects on parental mental health and emotion regulation were greatest immediately after the intervention and sustained at one year's follow-up.Concerning parental self-efficacy and parent-reported child symptoms, the effects were largest at one-year follow-up, pointing toward a potential sleeper effect (van Aar et al., 2017).Our findings correspond well with those obtained in other studies, suggesting that skills training programs enable parents to maintain the learned skills over time (i.e., Foroughe et al., 2019;Jones & Prinz, 2005;Strahan et al., 2017).The sustained and delayed effect on self-efficacy and child symptoms indicates that it may take time for the parents to become more secure in responding to their child's emotions in a healthier way, and similarly, the children may need time to get used to being at the receiving end of the newly acquired parenting skills.
Increased feelings of efficacy may help parents deal with difficult situations more effectively, and parental satisfaction with initial mastery might influence changes in children.This may also increase the child's feelings of self-efficacy with respect to the processing of painful emotions, which is in accordance with the model (Dolhanty & Lafrance, 2019).If child improvements meet parents' expectations, they might become more motivated to maintain their improved parenting skills, resulting in a spillover effect of gains or delayed improvements, rather than fade-out effects (van Aar et al., 2017).
The large effects on parental mental health and emotion regulation, which point towards an alternative route of change, that is via reduction in child symptoms, are surprising, considering the relatively short intervention.We suggest two potential explanations for this: First, the advertisement for the study emphasized the crucial role of parents in promoting psychological well-being in their children, and thus participants may have entered the trial with positive expectations, which may have bolstered the results.Second, due to the lack of a no-treatment condition (waitlist or TAU) or an alternative evidence-based treatment, there is a possibility that the effects could be due to regression toward the mean or the passage of time and resulting maturity, rather than being caused by the intervention.
Against our prediction, we found that fathers gained more from the program than mothers with regard to emotion regulation abilities and self-efficacy.Previous research has shown that mothers benefit more from such programs than fathers (Panter-Brick et al., 2014;Wells et al., 2016).Gender differences in emotional awareness and expression are well known, and women generally display more complexity and differentiation in their articulations of emotional experiences than men (Barrett et al., 2000).It has been proposed that men seem more reluctant to seek help on their own and often only go to therapy when they are prompted to do so by their partner (Greenberg & Goldman, 2008).Consequently, our findings may indicate that the EFST format is well-suited for fathers.One possible explanation could be that fathers have relatively more to gain from a program that focuses on understanding their own and others' emotions than mothers, but this is mere speculation.Additionally, the advertisements for the study emphasized the crucial role of both parent's in promoting wellbeing in children.Moreover, as the present gender sample predominantly consisted of mothers (74.8%), the participating fathers may have initially been more open to take in new knowledge and more willing to work on their parenting style and may not be representative of fathers in general.
Although the results show that parental outcomes improved in both conditions, the multilevel analysis supported the hypothesis that the EXP condition is superior to the PE condition with regards to fostering parental mental health and self-efficacy, but not emotion regulation.Other studies also suggest that the addition of EXP components to the relational conditions enhance treatment outcomes (Elliott et al., 2013;Goldman et al., 2006).Experiential interventions have also been found to be related to deeper emotional processing and stronger outcome (Pascual-Leone & Yeryomenko, 2017).One reason could be that the experiential interventions allow for a task-focused environment that provides a quicker way to identify core maladaptive emotions, which can be painful and unpleasant, but which seem to promote healing (Pascual-Leone & Greenberg, 2007).
The second research hypothesis, that a reciprocal relationship exists between parents' and children's' outcomes, was supported, as the cross-lagged panel models indicated both child-and parent-driven indirect effects.The findings are in line with previous studies suggesting bidirectional associations between parental mental health and children's psychological functioning (parent-driven effects) (Eckshtain et al., 2019).Some researchers have proposed that studies which show that a program leads to a change in parents, which in turn are associated with subsequent changes in child outcomes, provide particularly strong evidence for the causal effect of parenting (Sandler et al., 2011).Previous research has also identified associations between symptomatic distress in children and parents' mental health (Rayce et al., 2020), emotion regulation (Cloitre et al., 2019;Morris, Houltberg, et al., 2017) and self-efficacy (Bandura et al., 2003;Jones & Prinz, 2005), implying that such mediators may be possible mechanisms of change, and as such, a good target for future parental programs.Thus, our findings are in line with previous research, highlighting the potential benefits of parent-based interventions to shape the trajectories of children's mental health and improving parental mental health and wellbeing.
Interestingly, the strongest associations were found between teacher-reported child symptoms at post intervention and improvement in parental mental health, emotion regulation difficulties, and self-efficacy at one year's follow-up.We presently don't have knowledge about why different informants' ratings of childhood psychopathology often are discrepant from one another (De Los Reyes & Kazdin, 2005).Despite this, we interpret this finding to be related to the possibility that teachers see the children more as they are, while the perception of the child by the parents is biased or influenced by the status of their own mental health.Arguably, this finding of multi-informant associations (i.e., teacher and parents) provides additional support for the effectiveness of the EFST intervention and speaks to the validity of the results.While parents improved on the three dimensions assessed, the findings suggest that the strongest mediation effect goes in the opposite direction (child-driven effects), suggesting that when children change, parents improve.Most research have focused on how the parents impact the child; however, the opposite direction seems to be the strongest in our findings.This might indicate that parental distress and emotion regulation difficulties may be reduced due to perceived improvements in their child.The findings seem to be an extension of Beatrice Beebe's research, highlighting the impact of the child on the parent (Beebe et al., 2010).Some existing research suggests a similar pattern, implying that children's symptoms predict the parents' functioning (Culp, 2010;Morris, Criss, et al., 2017).
Another interesting finding was that only parental competence and child symptoms showed a bidirectional relationship across the three waves.In the treatment study design, the children were "treated" indirectly through the treatment of their parents, and because child outcomes were reported via parents and teachers who interacted with the children, the child symptoms are actually interactional outcomes.One possible explanation for the reciprocal relation is that parents who felt more competent in managing their child's mental health symptoms were more willing to put into action what they had learned to deal with the current situations rather than avoiding the problem.This is supported in a qualitative study, related to the current study.When receiving EFST, parents reported an increased willingness to use their innate ability, being less anxious about their role as parents, experiencing a new calmness inside them with renewed motivation to do what is needed (Ansar et al., 2021).

Strengths and Limitations
This study expands prior knowledge about possible links between symptomatic distress in children and parental emotional change.The present research has some important strengths, such as the use of a prospective longitudinal sample of children with multi-informant ratings, allowing for the control of measurement errors in the reports of mothers, fathers, and teachers, decreasing the likelihood that the observed effects are caused by shared method variance.Also, the use of multilevel analysis and Psychotherapy Research 533 structured equation modeling allow the disentangling of between-person processes in parent-child transactions by relying on standardized parent measures and adds to the parenting literature that is heavily focused on narrower aspects, i.e., parenting styles, parent/child characteristics or cross-sectional tests of mediation that cannot establish directionality (Cole & Maxwell, 2003).
Notwithstanding these strengths, the present research has some important limitations that should be noted, and the findings should be interpreted with caution.First, as this study relied on parental report, it may have been subject to recall bias and social desirability.Although independent observations from teachers were used to counterbalance this, directly observing parent-child interactions would most likely have captured nuances in a more fine-grained fashion.Also, as the included measures were all parent-rated questionnaires and did not include any child reports, we cannot determine how this would translate to more direct measures of the child's personal experience of their own mental health.Second, given that the parents were not recruited into the study based on their own mental health distress, but rather that of their children, this limits the possibility of examining the clinical significance of the findings on parental mental health.
Third, as the present population was predominantly monocultural (95.8% white), care should be taken when attempting to generalize these results.Additionally, as the present gender sample predominately consisted of mothers (74.8%), the results may not be generalizable to all fathers of children with mental health difficulties.Further, the missingness in follow-up data was not random but systematically confounded with children's pretreatment symptom severity and condition, as shown in the primary analysis (Ansar et al., 2022).Future research should focus on deciphering why differences in missingness occur and the implications for clinical practice.Finally, neither did the mediation models nor the linear models include child moderators (i.e., socio-economic background, child age and gender), which is a potential limitation.Thus, our study does not provide a full test of causality, because effects may be confounded by third variables that were not controlled for.The lack of measurement or observation of the parents' acquisition of skills or proficiency (i.e., adherence or competence measure) is another study limitation.Including such measures could help us further investigate whether the degree to which children change depend on the parents' acquisition of skills, or the degree of change in parents' psychological functioning (or both).

Clinical Implications
The results from this study provide initial evidence that EFST is beneficial for the parents and extend the current understanding by identifying an interplay between changes in children's mental health outcomes and the amelioration in parental mental health, emotion regulation and self-efficacy, also supported by teacher ratings.As the primary analysis showed statistical and clinical improvements in children's mental health symptoms (Ansar et al., 2022), this may suggest that the benefits are cumulative as parents apply the skills learned.Moreover, fathers' presence may also be an element that contributed to the effect, emphasizing why it is important to include fathers in preventive interventions to foster healthy child development.The study underlines the parallel interactional associations between children and their parents' mental health outcomes, highlighting that improving children's mental health may be powerful enough to, in turn, have downstream effects on parents' own mental health and emotional well-being.

Figure 1 .
Figure 1.CONSORT-participant flow from enrollment for original RCT study to the current study.

Figure 2 .
Figure 2. Mean of predicted values on Outcome Questionnaire across time by condition and gender.

Figure 3 .
Figure 3. Mean of predicted values on emotion dysregulation across time across time by condition and gender.

Figure 4 .
Figure 4. Mean of predicted values on parents sense of competence across time by condition and gender.

Table I .
Achenbach & Rescorla, 2001) parents and children in both conditions.Outcome questionnaire, DERS Difficulties in Emotion Regulation, PSOC Parents sense of competence, BPM-P Brief Problem Monitor for parents, EXP, experiential condition, PE, psychoeducational condition.INT internalizing symptoms and EXT externalizing symptoms, (T scores 65 or higher on Brief Problem Monitor for parents;Achenbach & Rescorla, 2001).
* Due to mental health issues.

Table II .
Mean scores and effect sizes (Cohen's d) of differences in means across assessment waves.
Notes.n sample size, SD standard deviation, d Cohen's d.Abbreviations: OQ45 Outcome Questionnaire-total symptoms, DERS Difficulties in Emotion Regulation-total score, PSOC Parents' Sense of Competence-total score, BPM-P Brief Problem Monitor for Parents (parental measures of their children's symptoms), BPM-T Brief Problem Monitor for Teachers (teacher measures of the children's symptoms), Internalizing symptoms in children, Externalizing symptoms in children, Total symptoms in children.

Table III .
Multilevel growth curve model for primary outcome variables across T0-T4.p < 0.05; * * p < 0.01; * * * p < 0.001.Estimates for parent outcome of mental health symptoms measured by OQ45-Outcome Questionnaire 45, DERS-difficulties in emotion regulation and PSOC-parents' sense of competence.Estimates (Est.)95% confidence intervals (CI) with lower and upper bound and standard error (S.E.).AIC Akaike information criterion.Model 1: Intercept and slope, no covariates.Variance components covariance matrices show significant differences in growth curves between primary outcome variables and slope on OQ45, DERS and PSOC.Model 2: adding gender improve model fit on DERS and PSOC, but not OQ45.Model 3: adding condition improve model fit on OQ45 and PSOC, but not DERS. *