INTRODUCTION

Couple therapy is an effective treatment for depression and relationship distress (Barbato et al., 2018), the two most common presenting problems couple and family therapists treat (Northey, 2002). Emotionally focused therapy (EFT) is one of the only evidence-based couple therapy models, and studies have shown that EFT can improve depressive symptoms and relationship satisfaction (Denton et al., 2012; Dessaulles et al., 2003; Wittenborn et al., 2019). Most EFT research has been conducted in North America, with few researchers examining EFT in other geographic regions, such as Asia (Ahmadi et al., 2014; Soleimani et al., 2015; Soltani et al., 2013, 2014). While therapists in Asia, including Taiwan, are increasingly using EFT in their practices (Asia EFT, n.d.), there have been no known studies of outcomes associated with EFT among couples in Taiwan. Therefore, the aim of this study is to empirically evaluate the effectiveness of EFT in Taiwan.

EMOTIONALLY FOCUSED THERAPY

A meta-analysis of 20 studies of EFT with 332 couples found medium to large treatment effects for relationship distress from pretest to posttest and pretest to follow-up (Spengler et al., 2022). It also found medium to large treatment effects for EFT versus alternative couple interventions. In addition, a previous meta-analysis of EFT studies from 1999 to 2017 found that improvements in relationship satisfaction during treatment were sustained following termination of treatment (Beasley & Ager, 2019). Wiebe et al. (2017) also found that couples who receive EFT reported significant growth in relationship satisfaction across a 2-year follow-up.

EFT consists of three stages: cycle de-escalation, restructuring of interactional positions, and consolidation and integration of therapeutic gains (Johnson, 2019). In the first stage, therapists track and identify negative interactional cycles and help couples access unacknowledged emotions underlying their negative interactions. In the second stage, therapists guide couples to respond to and support each other in an emotionally attuned way, which leads to changes in interactional patterns and new emotional engagement. In the third stage, therapists help couples explore and identify new solutions to their problems and integrate the changes they have made. Change in EFT occurs due to new corrective emotional experiences in therapy. When therapists can access each partner's emotions and identify the attachment needs underlying their negative interactional cycles, couples can create positive cycles of bonding and connection (Johnson, 2019).

The application of EFT in Taiwan

Wiebe and Johnson's (2016) review of EFT studies found a lack of research on EFT for couples outside of North America. In a small body of outcome studies among couples in Iran, EFT was shown to reduce depression and anxiety (Soltani et al., 2014), increase marital satisfaction (Soleimani et al., 2015), increase intimacy (Soltani et al., 2013), and decrease marital conflict (Ahmadi et al., 2014). In Taiwan, EFT research has been limited to a case study (Liu & Hung, 2019), a qualitative study of couples' experiences of receiving EFT (Sun, 2019), a clinical application (Wang & Wang, 2015), and the efficacy of EFT in the form of the psychoeducational “Hold me Tight” group (Sun & Liu, 2015). Even though there is limited research on the efficacy of EFT in Taiwan, therapists are increasingly seeking training in and delivering EFT (Asia EFT, n.d.).

Current study

While scholars and clinicians have argued that EFT is applicable in other cultural contexts because attachment needs are universally experienced (Ekman, 2007; Johnson, 2019), there have been no prior studies of the clinical outcomes associated with receiving EFT among couples in Taiwan. This single-arm pragmatic trial aims to explore the effectiveness of EFT for depressive symptoms and relationship distress among couples in Taiwan. Pragmatic trials are ideal for studying the effects of an intervention in a “real-world” environment to assess whether the intervention works under usual conditions (Loudon et al., 2015). A pragmatic trial was selected over an explanatory approach to assess whether EFT is effective when delivered by EFT therapists in Taiwan. Specifically, we aimed to test the following research questions:

RQ1.To what extent do depressive symptoms change across treatment?

RQ2.To what extent does relationship distress change across treatment?

METHOD

This study was approved by the Michigan State University institutional review board (study # 469).

Study design

This single-arm pragmatic trial (Thorpe et al., 2009) explored the effectiveness of EFT in a typical setting in Taiwan. A pragmatic clinical trial aims to examine whether an intervention works under usual conditions, and the recruitment procedures aim to include a heterogeneous sample of participants who reflect the real-world diversity of couples who seek therapy (Sox & Lewis, 2016; Thorpe et al., 2009). Since participants are typically assessed during routine clinical practice, clinicians are given more flexibility in how they apply the intervention (e.g., therapists and clients are given the choice of when to terminate couple therapy instead of being required to complete a specific number of sessions). The results of pragmatic trials are more relevant to usual practice; however, the emphasis on external validity instead of internal validity makes it more challenging to find treatment effects even when they exist (Godwin et al., 2003).

In this trial, we collaborated with therapists trained in EFT who were implementing the intervention with couples in their practices in Taiwan. Our data were collected between April 2018 and May 2020. WHO declared the COVID-19 outbreak a global pandemic in March 2020, and Taiwan recorded its first case of COVID-19 in April 2020; however, Taiwan maintained zero cases until December 2020, and all participants received in-person EFT throughout the study period (Thornton & Griffiths, 2020). Of the 17 couples in the study, all data for 16 couples were collected before the COVID-19 outbreak. Thus, we assume the impact of COVID-19 on the data collection to be minimal.

Procedures

Therapists, who were certified in EFT and were receiving supervision from EFT-certified supervisors, were identified through the Asia EFT website (www.asia-eft.com) in collaboration with the coordinator and director of the Asia EFT Center. Therapists were contacted to collaborate in this study if they (1) completed the EFT therapist certification requirements established by the International Centre for Excellence in Emotionally Focused Therapy (ICEEFT), (2) used EFT in their practice in Taiwan, (3) received supervision from EFT-certified supervisors, and (4) worked with Taiwanese couples. Each EFT couple therapist was contacted by email, and the study information was shared. Therapists interested in collaborating in the study were invited to an online meeting to explain the study in detail.

Therapists served as recruitment liaisons by inviting their couple clients to participate if they met minimal inclusion criteria, as is common in pragmatic trials. Participants were included in this study if they (1) sought couple therapy from a participating EFT-certified therapist in Taiwan, (2) were aged 18 or older, and (3) agreed to participate in the study. Couples were excluded from participating in this study if they (1) reported ongoing or past intimate partner violence, (2) reported suicidal thoughts or ideation in the past 3 months, or (3) reported ongoing infidelity. These exclusion criteria align with the criteria for who is best served by EFT (Johnson, 2019).

Once participants reviewed the study information and signed the consent forms, they were officially enrolled into the study. All participants received EFT in their therapist's practice setting and paid their therapists' existing fees, which cost $40 to $160 per 1-h session. Participants completed surveys at intake and after each therapy session. They were encouraged to complete surveys after each therapy session using their smartphones; however, two couples did not have smartphones, so their therapists provided paper-and-pencil surveys, which were then sealed in envelopes to protect the participants' confidentiality.

Each participant received a gift voucher for NT$100 (equal to approximately USD$3) after every three sessions for which they completed surveys until the termination of EFT. Therapists received a gift voucher for NT$700 (equal to approximately USD$23) to compensate them for their time as recruitment liaisons. All gift vouchers were sent to participants and therapists through email or text messages.

Participants

All participants in this study identified as different-sex couples. The mean age of participants was 33 years for male partners (SD = 7.90, median = 35, range: 19–47) and 32 years for female partners (SD = 7.47, median = 31, range: 20–45). Relationship duration ranged from less than 1 to 21 years, with an average of 7 years (SD = 6.70). Exactly 65% had no children, 24% had one child, and 12% had two children. Regarding educational attainment, 18% of male participants and 18% of female participants completed high school, 65% of males and 59% of females had bachelor's degrees, and 18% of males and 24% of females had graduate degrees. Participants' monthly household income was NT 75,000 (equal to approximately USD$2500). In Taiwan, the median monthly household income in 2021 was NT 77,416, which equals USD$2580 (Directorate-General of Budget, Accounting and Statistics, Executive Yuan, 2022).

The majority of couples in the study (65%) were moderately distressed at intake, with an average DAS score of 85.13 (range: 47–140, SD = 21.54). Most participants scored moderately depressed on the PHQ-9 at intake, with a mean score of 10.55 (range: 0–24, SD = 6.68). More specifically, 12% of both men and women were moderately depressed, 24% of both men and women were moderately to severely depressed, and 6% of men and 12% of women were severely depressed. Descriptive statistics are presented in Table 1.

EFT intervention

EFT was provided by 10 therapists who were certified in EFT by ICEEFT, including nine female therapists and one male therapist. All therapists held master's degrees and professional mental health licenses in Taiwan. Five had counseling psychologist licenses, four had clinical psychologist licenses, and one was a licensed social worker. Therapists had an average of 9 years (SD = 5.34; range: 4–22 years) of clinical experience. Therapists provided EFT in their practices and were given autonomy to determine the number of sessions needed to achieve positive outcomes based on the needs of each couple. All therapists were instructed to provide EFT according to the EFT manual (Johnson, 2019) and received supervision from an EFT-certified supervisor.

Participants (n = 17) completed 1–17 sessions (M = 7, SD = 4.16), and only 41% of couples (n = 7) completed more than eight sessions, which is the minimum recommended treatment length for EFT (Alan & Zeidlhack, 2021; Johnson, 2019). Among 17 couples, 29% of couples (n = 5) dropped out of the study; they were unable to continue their involvement in therapy due to various reasons. Three couples stopped attending therapy after two sessions without providing a reason; one couple stopped attending therapy after the first session because they ended their relationship, and one couple stopped attending therapy after the sixth session because the therapist observed an increase in violent behaviors and depressive symptoms and recommended the partners discontinue couple therapy and initiate individual therapy.

Measures

Dyadic Adjustment Scale (DAS)

The DAS (Spanier, 1976) is a measure of relationship distress that includes four subscales: consensus (13 items), satisfaction (10 items), cohesion (5 items), and affectional expression (4 items). The total score (range: 0–151) is obtained by summing all 32 scale items. Higher scores indicate lower relationship distress (Spanier, 1976). The DAS has demonstrated high reliability with a Cronbach's α coefficient of 0.96 (Spanier, 1976). The DAS has also been translated into Chinese and validated in various studies (Shek, 1995; Shek & Cheung, 2008; Shek et al., 1993). The Chinese-language version of the DAS was used for the current study (C-DAS; Shek et al., 1993), demonstrating high internal reliability (α = 0.91; Shek et al., 1993). For this study, the internal reliability of DAS was high across treatment and ranged from 0.92 to 0.98 for men and 0.87 to 0.98 for women.

Patient Health Questionnaire-9 (PHQ-9)

The PHQ-9 (Kroenke et al., 2001) assesses depressive symptoms on a 4-point Likert-type scale (0 = not at all to 3 = nearly every day). Participants rated the extent to which they had experienced each symptom in the prior 2 weeks. Examples of symptoms include “feeling down, depressed, or hopeless” and “poor appetite or overeating.” The total score (range: 0–27) is obtained by summing all nine scale items. Higher sum scores indicated higher depressive symptoms. The cutoffs for symptom severity (Kroenke et al., 2001) are 0–4 for minimal, 5–9 for mild, 10–14 for moderate, 15–19 for moderately severe, and 20–27 for severe. The PHQ-9 has been translated into Chinese and tested among Chinese populations with a high reliability of 0.96 and a high validity of 0.92 (Wang et al., 2014). The Chinese-language version of the PHQ-9 was used for this study. The internal reliability of PHQ-9 was high across treatment and ranged from 0.88 to 0.98 for men. For women, the reliability was acceptable, ranging from 0.56 to 0.96 for women (see Supplemental Table 1).

Material translation

This study was conducted in Taiwan and all participants spoke Mandarin. Therefore, all study materials, including the research information flyer and informed consent materials, were written in Mandarin by the first author, who is fluent in both Mandarin and English. All therapy sessions were conducted in Mandarin.

Data analysis

We conducted the following analyses in SPSS version 28 (IBM Corp, 2021). First, we conducted preliminary analyses to ensure that the data met the assumptions of the planned analyses. Second, we used dyadic multilevel growth modeling to examine the rates of change across treatment for depressive symptoms (RQ1) and relationship distress (RQ2). We used dyadic multilevel growth modeling because it addressed the range of completed sessions and small sample sizes (Kashy & Donnellan, 2008; Ledermann & Kenny, 2017; Planalp et al., 2017). Specifically, all couples (n = 17) in the study were included in the analysis. Couples completed 1–17 sessions (varying time points) and attended therapy with varying frequency (varying time intervals). Therefore, a multilevel modeling (MLM) approach was used because it can account for varying time points and intervals across and within dyads without increasing the number of missing values (Atkins et al., 2005; Ledermann & Kenny, 2017). Also, restricted maximum likelihood (REML; Ledermann & Kenny, 2017) was used to address the small sample size. In a clinical trial, clients may not complete the assessment for one session, but they might complete one at the next session, or clients may drop out of therapy. MLM uses all the data from each participant and can provide unbiased estimates if the data are missing at random (Atkins et al., 2005; Ledermann & Kenny, 2017; Schafer & Graham, 2002).

For these dyadic multilevel growth models, the partner's scores at each session of treatment (level 1) were nested within couples (level 2; Kashy & Donnellan, 2008; Kenny et al., 2006). This type of growth model is created by an intercept (initial score) and slope (rates of change), where the slope is grand mean centered. Due to the small sample size, we followed an iterative approach to build a parsimonious model (Hox & McNeish, 2020). We developed these models by starting with an undistinguished dyadic multilevel growth model to determine the number of random effects (i.e., intercept and/or random slope) in the model. Next, we evaluated whether the rates of change were linear or nonlinear (e.g., quadratic). Then, we used omnibus tests of distinguishability to determine if the model was distinguishable by sex. Chi-square difference (or deviance) tests guided our analysis in each step to determine better-fitting models where lower log-likelihood values indicated a more parsimonious model. These steps were followed for both depressive symptoms and relationship distress models.

Missing data

Participants (n = 17) completed a range of 1–17 sessions, and five couples (29%) dropped out of treatment. REML methods and Kenward-Roger standard error adjustments were recommended to handle missingness with small sample sizes (McNeish & Stapleton, 2016). Therefore, we used REML and Kenward-Roger standard error corrections to handle missing data in our models.

Sensitivity analysis

As this study is a pragmatic clinical trial, the completion of treatment was determined collaboratively by the participating couples and their therapists. Given the importance of assessing the potential impact of missing data on our final results, we conducted a sensitivity analysis (Thabane et al., 2013) that evaluated the results of couples who completed treatment or dropped out. First, we used pattern-mixture models, which are a common and widely utilized approach for analyzing longitudinal data when missingness is caused by participant dropouts (Daniels & Hogan, 2000). Specifically, we created a binary variable of those that dropped out of treatment (completed = −1, dropped out = 1). Particularly, those that dropped out of treatment (n = 5) did so after the first session (n = 1), second session (n = 3), and fifth session (n = 1). We then added dropout as a predictor to the final models of relationship distress and depressive symptoms to evaluate differences between couples who completed the treatment and those who dropped out.

RESULTS

Table 1 provides means and standard deviations of the variables of interest, including baseline scores of depressive symptoms and relationship distress.

Multilevel modeling

Depressive symptoms

We began with an undistinguishable dyadic model that included the initial levels (i.e., intercept) and linear rates of change (i.e., slope) of depressive symptoms as fixed effects. We used an autoregressive (1) matrix structure for the variance, covariance, and residuals because depressive symptoms for each week were highly correlated (ρ = 0.75, p <0.001). We started with the random effects for the initial levels of depressive symptoms, and the model fit improved by adding the random effects of linear rates of change of depressive symptoms (χ² = 12.90, p < 0.05). However, the random effects of the linear rates of change of depressive symptoms had a p = 0.21, which was above p = 0.05. For parsimony, we only included the random effect of the initial levels of depressive symptoms. Next, the model fit improved when adding quadratic rates of change for depressive symptoms (χ²[1] = 4.78, p < 0.05). Specifically, there was a small but notable quadratic slope (b = 0.03, p < 0.05), while the linear slope was not significant (b = −0.15, p = 0.09). Hence, we included both linear and quadratic rates of change for depressive symptoms. An omnibus test of distinguishability revealed that the model was not distinguishable by sex (χ²[5] = 8.64, p = 0.12). Particularly, the initial (p = 0.85), linear rates of change (p = 0.05), and quadratic rates of change (p = 0.66) for depressive symptoms did not differ by sex. Therefore, we continued with an undistinguishable dyadic multilevel growth model. Results from the final model revealed mild average initial levels of depressive symptoms (b = 8.57, p < 0.001) that varied across couples (σ² = 31.21, SE = 10.14, p < 0.05). There was no linear change in depressive symptoms (b = −0.15, p = 0.09), but there was a small quadratic rate of change in depressive symptoms across treatment (b = 0.03, p < 0.05) (Figure 1). Specifically, the quadratic slope showed an initial decline from session 1 to session 13, followed by a small incline toward the end of treatment (refer to Table 2 for full details).

Relationship distress

We began with an undistinguishable dyadic model that included the initial levels (i.e., intercept) and linear rates of change (i.e., slope) of relationship distress as fixed effects. Because relationship distress scores for each week were highly correlated (ρ = 0.75, p < 0.001), we used an autoregressive (1) matrix structure for the variance, covariance, and residuals since it assumes covariances are correlated and it factors in declining correlations over time (Leyland, 2004). We added in the random effects for the initial levels of relationship distress, and the model fit improved with the addition of random effects of linear rates of change for relationship distress (χ[2] = −19.02, p < 001). However, the random effects of the linear rates of change for relationship distress had a p = 0.87, which was above p = 0.05. For parsimony, we only included the random effect of the initial levels of relationship distress. Next, there was no difference in the model fit when the quadratic rates of change were included in the model (χ²[1] = 0.32, p = 0.57); therefore, we only included linear rates of change for relationship distress. Additionally, an omnibus test of distinguishability revealed that the model was not distinguishable by sex (χ²[4] = −3.53, p = 0.47). Specifically, the initial levels (p = 0.65) and linear rates of change (p = 0.77) did not differ by sex (women = 1, men = −1). As a result, we continued with an undistinguishable dyadic multilevel growth model. Results from the final model revealed participants had moderate initial levels of relationship distress (b = 88.26, p < 0.001) that varied across couples (σ² = 430.16, SE = 138.61, p < 0.05). However, the linear rate of change in relationship distress remained stable and did not change across treatment (b = −0.09, p = 0.76).

Sensitivity analysis

The sensitivity analysis examined differences in relationship distress and depressive symptoms between couples who completed or dropped out of treatment. Adding dropout to the relationship distress model did not improve model fit (χ²[2] = −0.27, p = 0.87), and there was no difference in the initial (p = 0.90) and linear rates of changes in relationship distress (p = 0.69) by couples who completed or dropped out of treatment. However, adding dropout improved the model fit for the depressive symptom model (χ²[2] = −11.48, p = 0.003) where the initial (b = 8.31, p = 0.002) and quadratic rates of changes in depressive symptoms (b = 0.69, p = 0.03) differed by couples who completed or dropped out of treatment (see Supplemental Table 2). Specifically, this revealed that couples who dropped out of treatment started with higher initial depressive symptoms and had a steeper quadratic change in depressive symptoms across treatment compared to couples who completed treatment. This suggests that couples who dropped out of treatment began with higher depressive symptoms that declined steeply and then began to rise because no dropped-out couples remained in treatment past session 5. The higher initial depressive symptoms could be explained by a couple ending their relationship and another couple being referred to individual treatment due to violent behaviors and increased depressive symptoms.

DISCUSSION

Several studies have demonstrated the efficacy and effectiveness of EFT for improving depressive symptoms and relationship distress. The primary purpose of this study was to explore the effectiveness of EFT for depressive symptoms and relationship distress among couples in Taiwan. For depressive symptoms, our findings provide preliminary information regarding the use of EFT in reducing depressive symptoms. Over the course of EFT, we observed a quadratic change in the rate of change of depressive symptoms. Specifically, we noted an initial decrease in depressive symptoms from session 1 to session 13, followed by a small incline toward the end of treatment. It is important to highlight that more than half of the couples in our study completed EFT within 8 sessions, indicating that the majority of them experienced a decrease in depressive symptoms despite the slight increase observed from session 13 to session 17. Furthermore, only two couples remained in treatment after session 14, which may indicate that the most symptomatic couples continued treatment until the end. This finding is partially consistent with prior findings indicating that EFT improves depressive symptoms (Denton et al., 2012; Dessaulles et al., 2003; Wittenborn et al., 2019).

For relationship distress, the multilevel models revealed no change over time. This finding was unexpected because prior studies have found that EFT improves relationship distress (e.g., Denton et al., 2000; Johnson & Greenberg, 1985; Wittenborn et al., 2019), including when compared to a range of control groups (e.g., behavioral approaches, couple therapy as usual, wait-list). One prior study found that couples had an overall 0.39-point increase in relationship satisfaction each week during EFT (Dalgleish et al., 2015).

There are several possible reasons for the limited improvement in relationship distress over the course of treatment in this study. First, a highly plausible explanation is that couples did not receive enough therapy to allow EFT to facilitate change in couples' relationship distress. It is recommended that couples receive 8–20 sessions, though the appropriate “dose” often depends on the severity of the symptoms (Alan & Zeidlhack, 2021; Johnson, 2019). In our study, couples completed 1–17 sessions, with only 41% (n = 7) of couples completing more than eight sessions. Research has shown that couples made more intimate sharing in EFT after seven sessions (Greenberg et al., 1993), and completing stage two of EFT, including an important component of the blamer-softening event, often leads to increased relationship satisfaction (Burgess Moser et al., 2018). It is likely difficult for EFT to facilitate change if these couples do not receive enough therapy to enable changes to occur. Other possible explanations for this finding are that there was insufficient power to detect treatment effects due to the small sample, and treatment effects tend to be smaller in pragmatic versus explanatory trials, given the focus on external validity (McNeish & Stapleton, 2016).

The number of sessions couples attended may have been affected by the public health system in Taiwan. Taiwan's national health insurance provides universal coverage; however, counseling and therapy services in private practice are not covered by national health insurance (Guo et al., 2013; Ministry of Health and Welfare, 2021). In Taiwan, therapy costs are high, ranging from $40 to $160 per hour, depending on the geographic location of the private practice and the therapist's years of experience (Taiwan Counseling Psychologist Union, 2011). The high cost of couple therapy could have contributed to the low number of completed sessions. It may be the case that couples discontinued therapy when they started to feel better. The first major change process in EFT is de-escalating the negative cycle, which occurs in stage one of EFT (Johnson, 2019). In this change process, partners often experience new ways of interacting and are thought to feel newly hopeful about their relationship (Johnson, 2019). It is possible that couples began to feel some hope or relief and, given the high cost, decided to terminate therapy. Doing so would prevent couples from receiving the full “dose” of treatment and completing all three stages of EFT, which could alter the immediate and sustained impact of EFT (Burgess Moser et al., 2016).

Another potential explanation for the limited improvement in relationship distress involves cultural variation. EFT aims to facilitate emotional expression and communication of attachment needs in interpersonal relationships (Tsai & Levenson, 1997). Li (2006) found that tolerance (i.e., suppressing emotions) is considered an appropriate emotion regulation strategy in Chinese relationships. Moreover, Chinese men tend to use disengagement strategies, such as creating psychological distance from stressful situations or individuals, more often than Chinese women (Davis et al., 2012). Our study findings highlight the potential conflict between EFT's emphasis on exploring and expressing emotions and cultural expectations around emotion regulation, which may have contributed to the lack of change observed for relationship distress.

It is also possible that EFT may need to be culturally adapted to fit the needs of couples in Taiwan to facilitate change. Bernal et al. (2009) have recommended considering language, cultural patterns, meaning, and values when systematically modifying an evidence-based treatment. Chinese cultural patterns, such as tolerance in relationships, can affect how couples prefer to interact with one another. For example, tolerance is often used by Taiwanese couples to repress negative emotions and maintain harmony (Chang et al., 2020), and it predicts an individual's marital satisfaction and their partner's supportive behaviors (Chen & Li, 2007). However, EFT's main mechanism of change is the expression of emotions, which at times may conflict with Taiwanese values. Therapists may overcome this conflict by creating a safe space for couples to express, instead of tolerate, their inner emotions. Emotional vulnerability is essential for success in EFT, and therapists should guide clients through their fears with reassurance and confidence (Johnson, 2019). Future research investigating potential adaptations of EFT for couples in Taiwan could be an important next step. For example, therapists may need to use different strategies and spend more time accessing and expanding emotional experiences, as well as facilitating emotional expression, especially for men.

Limitations

There were several limitations in this study. First, only 41% of the couples completed more than eight EFT sessions. Couples are recommended to receive at least eight sessions or up to 20 or more based on the needs of the couple (Alan & Zeidlhack, 2021; Johnson, 2019). The pattern mixture model indicated that missing data at session 17 did not affect the model; however, the fact that most couples in the study did not have the recommended treatment “dose” could explain the lack of clinical improvement. Second, as a single-arm design, it was not possible to capitalize on the benefits of randomization or to make comparisons across groups. Third, the study was limited by the small sample size, which reduced the statistical power, restricting the likelihood of finding significant effects.

CONCLUSIONS

This study is the first to explore the effectiveness of EFT among couples in Taiwan. Our findings showed a significant quadratic change in depressive symptoms and no significant change in relationship distress. Limitations such as the small sample size and small treatment “dose” may have accounted for the limited change observed. Future research should continue to explore the effectiveness of EFT for couples in Taiwan with a larger sample size and longer treatment duration. Future research on potential cultural adaptations of EFT for couples in Taiwan is also needed.