Effectiveness of online family connections for family members of people with BPD: a naturalistic comparison with the face-to-face format

Abstract

Introduction

Borderline Personality Disorder (BPD) is a common condition with significant negative impacts on relatives due to its interpersonal instability and suicidal behaviors. Family Connections (FC) is currently the most empirically supported intervention for family members of individuals with BPD. However, access to FC remains limited. Online delivery could overcome geographic barriers, as digital formats have shown efficacy in psychotherapies for patients, though little is known about their effectiveness for family-oriented interventions. During the COVID-19 crisis, rapid adaptations of psychotherapeutic programs such as FC to online formats created an opportunity to collect data. This study aimed to compare the effectiveness of the online FC with face-to-face format.

Method

This multicenter naturalistic study extended a previous project involving French-speaking FC participants. The program was first delivered face-to-face (n = 141) and then online (n = 27). FC is a 12-week group intervention providing psychoeducation about BPD and teaching coping skills derived from Dialectical Behavior Therapy.

Results

Groups were comparable at baseline, except for certain variables. Controlling for baseline scores, multiple linear regressions showed no significant group differences for post-intervention burden (b=-0.59, p = 0.825) or depression (b = 1.37, p = 0.398). However, the online group improved more in emotion dysregulation (b=-9.05, p = 0.003) but less in coping (b=-0.79, p < 0.000).

Discussion

Findings support the online FC format as comparably effective to the face-to-face version, suggesting its potential to enhance accessibility without compromising outcomes. Replication in randomized designs is warranted.

Highlights

This study supports the online Family Connections as comparably effective to the face-to-face version.

Online Family Connections showed similar improvements in caregiver burden compared to face-to-face groups.

Online format allowed Family Connections to continue during the COVID-19 crisis.

Online Family Connections could improve access across wider geographic regions.

Introduction

Borderline personality disorder (BPD) affects approximately 1.7% of the general population and up to 28% of psychiatric outpatient or inpatient populations [17]. This disorder is characterized by emotional dysregulation, repeated suicidal crises and interpersonal instability. Due to the highly relational nature of these symptoms, BPD has many negative consequences for loved ones, which go beyond the consequences found in other severe somatic or psychiatric disorders [1, 18, 38]. As treatment for BPD is primarily psychotherapeutic, relatives play a major role in the psychosocial support of their loved ones close to an informal psychotherapeutic role [9]. If they feel overwhelmed, it can affect patients. While loved ones want to better understand the disorder and employ effective strategies, they often feel excluded by professionals [24].

Few interventions currently exist to support the families of people living with BPD. The most validated is the Family Connections program ^® [20], with several observational studies [7] and several controlled studies [12, 16]. Based on Dialectical Behavior Therapy (DBT), the psychotherapy with the highest level of evidence in BPD [41], the program consists mainly of explaining BPD to family members and training them in emotional regulation skills. The skills seen in Family Connections focus on those that improve family bonding, i.e. the skills of validation, acceptance and problem management. Validation involves effectively communicating understanding to the relative, acceptance means avoiding denial of the disease, and problem management helps to manage the emotions of the relative, especially during suicidal crisis. The Family Connections program is typically led by a team of two relatives or professionals having received at least 2 days of training. Sessions take place in groups of 10–15 people weekly for two hours over three months.

Unfortunately, this intervention remains poorly available, with unequal access depending on the territory [27]. An online format could be an interesting alternative to the usual in-person version of the program, as it has the advantage of avoiding transport and gathering participants (whether the program leaders or the families) regardless of geographical distance [28]. While this format has rarely been assessed for family interventions, several studies showed that internet interventions facilitate accessibility of psychotherapy [5] and are effective in reducing symptomatology associated with BPD, such as alcohol misuse [35], non-suicidal self-injury [4, 26], suicidal ideation [6, 23], or emotional dysregulation [4]. A meta-analysis of online psychotherapies found no significant differences compared to face-to-face psychotherapies and a large effect size compared to control groups, particularly when the evaluated psychotherapy is CBT [11].

During the COVID-19 crisis, accessibility became more urgent requiring rapid adaptations of psychotherapeutic programs such as Family Connections to online formats [40]. Before the pandemic, only one study had examined an online intervention involving relatives of individuals with BPD. This study reported improvements in both relatives and patients following a combined psychoeducational and psychotherapeutic program [13]. However, no research had evaluated family interventions delivered through teleconsultation platforms before Covid-19, nor directly compared online and face-to-face formats. The COVID-19 context thus created an opportunity to collect data on these questions. One study [15] found that the online version delivered during the pandemic was non-inferior to the face-to-face version offered before it. Even if its conclusions were limited by the small sample size, this study is the only one to investigate the online format of Family Connections.

To confirm this first study, we used data from a naturalistic study comparing Family Connections groups using face-to-face and online formats. As in Guillén’s study, the face-to-face format was delivered in our study before the Covid-19 pandemic, whereas the online format was delivered during the pandemic. Our hypothesis was that online format does not affect the clinical effectiveness of the Family Connections program.

Method

Participants

This study was an extension of a previous study [7] involving a naturalistic sample of family members of people with severe emotional dysregulation in the realm of BPD. The inclusion criterion for the study was participation in Family Connections groups conducted in routine care in 2 French-speaking countries (France and Switzerland) between 2011 and 2021, completion of pre- and post-intervention questionnaires used to assess the clinical evolution, and consent for the use of questionnaire data for research. Participants with incomplete questionnaire data (> 30% missing data) or lack of consent were excluded from the study. In contrast to our previous study [7], the current sample included participants from 2021, a period during which several online groups were conducted due to the Covid-19 pandemic.

Procedure

In this multicenter naturalistic study, we analyzed questionnaires of volunteer participants between 2011 and 2022. Family Connections groups were conducted in several French-speaking centers in Switzerland and France: since 2011 in Fribourg, since 2018 in Geneva, Versailles and Montpellier, and since 2020 in Strasbourg. Participants were recruited for Family Connections based on their motivation to support a family member with BPD; no specific exclusion criteria were applied. The program was promoted by clinicians during patient follow-up, through posters displayed in clinical setting, and via the Family Connections website. Recruitment was validated by a trained relative or by a mental health professional during an interview with the potential participant. This interview ensured that the diagnosis corresponded to BPD and not another diagnosis. Participants completed questionnaires before and after the intervention. These questionnaires were also used by facilitators to assess the effectiveness of their intervention and to identify potential improvements.

Intervention

The Family Connections program was developed by Perry Hoffman and colleagues in the context of the National Education Alliance for Borderline Personality Disorder (NEABPD), a nonprofit organization dedicated to supporting individuals affected by borderline personality disorder (BPD), including patients, their families, and mental health professionals [20]. The program has been specifically dedicated to the relatives and targets psychoeducation, self-care and stress management, coping skills, effective communication, and problem management (Fig. 1). Twelve weekly 2-hour sessions were divided into 6 modules. They were led either solely by healthcare professionals (Fribourg, Geneva), by healthcare professionals accompanied by a trained relative during specific sessions (Montpellier), or by relatives trained with a healthcare professional (Versailles, Strasbourg). The facilitators were trained by an official trainer of the NEA-BPD (Marie-Paule de Valdivia) and used a French translation of the official NEA-BPD Family Connections manual.

Fig. 1.

: Content of the family connections program [20, 21]

Initially, the groups were exclusively face-to-face. However, starting in March 2020, the Covid-related lockdowns prompted a shift to an online format. This format was conducted via a videoconference platform, bringing together participants from different regions of the Francophonie.

Measures

Demographics

Several socio-demographic characteristics were measured: sex, age, proximity between the participant and his/her loved one with BPD, and duration of the disease.

Primary intervention outcome

Burden was selected as the primary outcome, as it represents a fundamental component of close interpersonal relationships and aligns closely with the core objectives of the intervention: Participants’ burden was assessed using the Involvement Evaluation Questionnaire (IEQ). The IEQ is a 31-item self-report scale scored on a 5-point Likert scale. It was developed to evaluate the participants’ experience of burden and the consequences of providing care to people with psychotic disorders [43]. It refers to the four previous weeks. Higher scores indicate a higher burden for the participant. The IEQ has been validated on caregivers of people with borderline personality disorder and show good psychometric properties with Cronbach’s α ranging from 0.70 to 0.85 depending on the subscale considered [30]. Internal consistency in the present sample was good, with a Cronbach’s α of 0.867.

Secondary intervention outcomes

Depression: Depression was assessed with the Center for Epidemiologic Studies-Depression Scale (CES-D). The CES-D comprises 20 items scored by the subject on a four-point Likert Scale. It has been widely used to assess depressive symptoms in community and population-based epidemiological studies [39]. The first validation study showed good internal consistency coefficients, between 0.85 and 0.90 in clinical and non-clinical samples [34]. The factor structure of the French translation was validated using Confirmatory Factor Analysis in 2011 [31]. Internal consistency in the present sample was good, with a Cronbach’s α of 0.898.

Coping: Coping strategies of participants were assessed by the Family Coping Questionnaire (FCQ) [29]. It comprises 31 items about the past 4 weeks, scored on a 5-point Likert scale from 1 = always to 5 = never. The better the score is, the better the coping strategies are. Seven subscales can be distinguished (information gathering, positive communication, social involvement, coercion, avoidance, resignation, and patient’s social involvement). It was first validated in Italian on caregivers of psychotic patients, with a Cronbach’s alpha coefficient ranging from 0.68 to 0.83 [29]. It has been validated in French since then [32]. Internal consistency in the present sample was correct, with a Cronbach’s α of 0.657.

Emotion regulation: Emotional regulation was assessed with the French version of The Difficulties in Emotional Regulation Scale (DERS-F). The DERS was first developed to identify difficulties in emotion regulation in 6 domains: lack of emotional clarity, non-acceptance of emotional responses, difficulties engaging in goals directed behaviors orientation, impulse control difficulties, lack of emotional awareness, and limited access to emotion regulation strategies. It comprises 36 items answered by the subject on a 5-point Likert-Scale. A higher score relates to higher difficulties in emotion regulation. The DERS has been validated on two non-clinical samples with Internal consistency of the original version reaching 0.93 (Cronbach’sα) [14]. The French version of the DERS has shown a high congruence with the original version of 0.98 (Tucker’s phi) with an internal consistency of 0.92 (Cronbach’sα) [8]. Internal consistency in the present sample was excellent, with a Cronbach’s α of 0.915.

Statistical analysis

All the analyses were conducted using R and R studio [33]. Part of the present data were used elsewhere [7]. For quantitative variables, given the low number of missing data (mean = 0.9%), after removal of aberrant values, we performed mean imputation if less than 30% of the items of a scale were missing. For baseline comparisons, we performed Student t-tests with or without Welsh approximation (for quantitative variables) and Chi-Squared tests or Fisher’s exact test (for qualitative variables), depending on whether the validity conditions were met. For the main analyses, we performed multiple linear regressions with post-group scores as dependent variables, and with group (considered as a binary variable, taking 0 for “in person” and 1 for “online”) and baseline scores as independent variables. Within-group effect sizes were calculated using Cohen’s d for paired samples. Between-group effect sizes were computed as standardized adjusted mean differences derived from the models (β_group divided by the residual standard deviation). According to conventional benchmarks, effect sizes of 0.2, 0.5, and 0.8 were interpreted as small, moderate, and large effects, respectively.

Given that this study was exploratory and not confirmatory, we did not calculate sample size, alpha or beta a priori, and we did not correct our analyses for multiple comparisons [3]. Thus, the statistical significance threshold was set at < 0.05.

Ethical aspects

Participants were informed about the context and interest of the study. Participation was voluntary and refusal did not affect the participants’ involvement in the program. Every participants provide their consent to this research. The study protocol was developed following the specific ethical local/national guidelines of each center in accordance with the declaration of Helsinki (for the Réseau Fribourgeois de Santé Mentale: Comité éthique cantonal 016-REP-CER-FR; for Genève: the Ethics Committee of the Republic and Canton of Geneva; for the French centers: the Ethics Committee of Paris-Saclay University, CER-Paris-Saclay-2023- 046).

Results

Between 2011 and 2021, 282 people participated in Family Connections groups, but 114 did not meet the study’s inclusion criteria either due to missing data (n = 98) or absence of consent (n = 16). The population included in this study included 168 participants (66.7% women, n = 112; mean age = 52.5, standard deviation = 10.7). The majority were parents of BPD patients (78%, n = 131), followed by partners (11.3%, n = 19), sisters/siblings (6.5%, n = 11), children (1.8%, n = 3), and others (2.4%, n = 4). The relative (79.1% women, n = 133; mean age = 25.4, standard deviation = 8.2) lived with the participants in 52.9% of cases. The disorder had been diagnosed on average 6.3 years before the relatives took part in the groups (standard deviation = 5.4).

141 participants followed a “face-to-face” group, and 27 followed an “online” group. Sociodemographic characteristics can be found in Table 1. Overall, the two groups were comparable on almost all demographic variables, except for age of the relative (the mean age being younger in the online group: 25.96 vs. 22.28, t = 2.161, p = 0.032). In terms of baseline psychometric scores, no differences were found on the IEQ total scores (t=-0.401, p = 0.690), but CESD total, FCQ mean, and DERS total scores were respectively higher (for CESD : t=-2.837, p = 0.005) and lower (for FCQ : t = 15.349, p < 0.000; for DERS : t = 2.238, p = 0.027) in the online group.

Table 1.

Sociodemographics characteristics

Variables	Face-to-face group (n = 141)	Online group (n = 27)	Differences Test (p value)
Qualitative variables, n (%)
Sex of the participant (female)	91 (64.54%)	21 (77.78%)	ChiS = 1.787 (0.181)
Sex of the relative (female)	115 (81.56%)	18 (66.67%)	ChiS = 3.047 (0.081)
Relationship with the close-one			Fisher (0.746)
Son/daughter	109 (77.3%)	22 (81.48%)
Partner	17 (12.06%)	2 (7.41%)
Brother/sister	8 (5.67%)	3 (11.11%)
Parents	3 (2.13%)	0 (0%)
Other	4 (2.84%)	0 (0%)
Living with the relative			ChiS = 0.027 (0.870)
Yes	75 (53.19%)	14 (51.85%)
Not	65 (46.1%)	13 (48.15%)
Missing data	1 (0.71%)	0 (0%)
Quantitative variables, Mean (SD)
Age of the participant (years)	52.84 (10.65)	51.07 (11.46)	t = 0.778 (0.437)
Age of the relative (years)	25.96 (8.43)	22.28 (6.14)	t = 2.161 (0.032)
Duration of the disorder (years)	6.26 (5.26)	6.26 (5.47)	t=-0.002 (0.999)

Psychometric scores can be found in Table 2. In the online group, pre- to post-intervention comparisons showed moderate effect sizes for IEQ (d=-0.58), CES-D (d=-0.61), and DERS (d=-0.55), and a small effect size for FCQ (d = 0.14). In contrast, the face-to-face group showed small effect sizes for DERS (d=-0.34), IEQ (d=-0.47), and CES-D (d=-0.37), but moderate for FCQ (d = 0.54). Multiple linear regressions were conducted to assess the effect of group membership (“online” vs. “face-to-face”) on post-group psychometric scores, controlling for baseline scores. The group variable was significantly negatively associated with post-treatment DERS (b=-9.05, p = 0.003) and FCQ (b=-0.79, p < 0.000) scores, indicating that participants in the online group showed greater improvement in emotion regulation, but lesser improvement in terms of coping, compared to the face-to-face group. However, no significant differences between groups were observed for post-group IEQ (b=-0.59, p = 0.825) and CES-D scores (b = 1.37, p = 0.398). The very large, adjusted effect size observed for FCQ (d = − 2.37) likely reflects the relatively small residual variance of FCQ scores after adjusting for baseline values in the regression model, which inflates standardized effect size estimates derived from the model.

Table 2.

Pre- and post-treatment psychometric scores

Variables	Face-to-face group		Online group		Group differences Test (p-value)
	Pre	Post	Pre	Post
Burden (IEQ)	39.18 (17.70)	31.97 (15.49)	40.24 (11.33)	31.93 (14.16)	b = -0.59 (0.825) d=-0.05
Depression (CES-D)	20.54 (11.26)	16.99 (9.16)	27.17 (10.38)	21.52 (9.19)	b = 1.37 (0.398) d = 0.18
Coping (FCQ)	3.57 (0.44)	3.77 (0.42)	2.17 (0.41)	2.24 (0.32)	b = -0.79 (0.000) d=-2.37
Emotional regulation (DERS)	82.06 (22.65)	76.74 (20.25)	71.70 (18.38)	61.02 (18.57)	b = -9.05 (0.003) d=-0.64

Abbreviations: CES-D = Center for Epidemiologic Studies Depression Scale ; DERS = Difficulties in Emotion Regulation Scale ; FCQ = Family Coping Questionnaire ; IEQ = Involvement Evaluation Questionnaire

Discussion

Our study aimed to compare the effectiveness of the online Family Connections format with the face-to-face format. As few evidence-based family interventions are available in French-speaking countries, the implementation of Family Connections is particularly important. A previous study we conducted showed that these groups led to statistically significant changes [7], but did not include enough participants in the online format to make a comparison between the two groups. In contrast to this previous study, we included data from 2021, during which several online groups were conducted due to the Covid-19 pandemic.

Our results did not reveal significant differences in effectiveness between the two types of groups on the primary outcome, i.e. the sense of burden of the families assessed by the IEQ, nor on depression assessed by the CES-D, which was a secondary outcome. These results are consistent with the literature. The only study on Family Connections comparing online and face-to-face format [15] found no significant difference in burden, depression, anxiety, stress, level of empowerment, family functioning, and quality of life. Studies on other psychotherapeutic formats generally find a good effectiveness for online psychotherapy, although most studies focus on patients [4, 6, 19, 23, 26, 35–37] and only a minority on their families [13, 25].

While telemedicine interventions have gained in popularity during the Covid-19 pandemic [44], the absence of significant differences between online and face-to-face learning found in our study and in the literature should encourage their development. Many regions still lack access to structured psychotherapeutic programs, notably for Family Connections [27]. In French-speaking countries, some regions are sparsely populated complicating the implementation of specialized psychotherapeutic programs. Online formats can easily overcome these geographical barriers. Reservations regarding their use shouldn’t be overestimated at a time when around 90% of the population has an internet connection in Europe [42] and over 70% own a smartphone [2]. Some studies even suggest that the virtual aspect can be less intimidating and thus increase the therapeutic alliance [22].

In the secondary outcomes of our study, we found that emotional dysregulation assessed by the DERS improved significantly more in the online Family Connections compared to the face-to-face format, while the FCQ improved more in the face-to-face format. For the DERS, a possible hypothesis is that as distance learning participants were more familiar with digital tools, they may have been more likely to watch YouTube videos on the DBT skills recommended during the groups [10]. For the FCQ, another hypothesis is the impact of Covid on family coping. Nevertheless, it is difficult to conclude on these secondary outcomes given that the two groups were significantly different on the DERS and the FCQ before the intervention.

Our study includes several limitations that should be noted. This naturalistic study concerned Family Connections groups conducted in routine practice with a face-to-face format before the Covid-19 pandemic and an online format from the onset of the Covid-19 pandemic, so some biases could not be controlled. First, participants in both groups had significant differences before starting the group. Second, if the face-to-face group served as a control group, we couldn’t randomize participants to equally distribute them between groups. Thirdly, as no assessments were conducted at a distance from the intervention, we lacked access to long-term follow-up to verify the maintenance of clinical progress. Nevertheless, this study gives preliminary data to consider possible implementation of online Family Connections and further randomized controlled studies.

Conclusions

In conclusion, our study provides arguments for using an online format of Family Connections without fear of losing efficiency compared to the face-to-face format. The reluctance to use it is not confirmed by our study, which is consistent with the small existing literature. Given the high number of internet connections, the online format has the advantage of making this intervention more accessible by removing organizational constraints. However, our study has several limitations that require replication in randomized trials.

Author contributions

A. D., S. C., M. S. designed the research protocol. N. P., M. S. developed the intervention program and therapy materials for patients. N. P., M. S., A. D., and S. S. recruited participants and conducted therapy sessions together. S. C. provided support for therapy. (A) D., S. C. and M. (B) conducted the statistical analyses and wrote the first draft of the manuscript, which was then reviewed by N. P. and M.S. All authors approved the final version of the manuscript.

Funding

Open access funding provided by Hospital Center of Versailles, Geneva University Hospitals, Strasbourg University Hospitals. The authors received no specific funding for this work.

Data availability

The datasets supporting the findings of this study are held by the Hospital Center of Versailles. Due to institutional restrictions and confidentiality requirements, they are not publicly available. Data may, however, be obtained upon reasonable request and with permission from the Hospital Center of Versailles and the other contributing centers.

Declarations

Ethical approval and consent to participate

The study protocol was developed following the specific ethical local/national guidelines of each center in accordance with the Declaration of Helsinki (for the Réseau Fribourgeois de Santé Mentale: Comité éthique cantonal 016-REP-CER-FR; for Genève: the Ethics Committee of the Republic and Canton of Geneva; for the French centers: the Ethics Committee of Paris-Saclay University, CER-Paris-Saclay-2023- 046). Every participants provide their consent to this research. Participants were individually informed in writing that their data could be used anonymously to evaluate the program and that they were allowed to refuse to participate.

Consent for publication

All of the material is owned by the authors and/or no permissions are required.

Competing interests

The authors declare no competing interests.