Borderline Personality Disorder (BPD) is a prevalent [1
] and complex mental illness associated with high utilization of mental health services [2
] and serious impairments in psychosocial and occupational functioning [4
], leading to high societal costs [7
]. BPD can be effectively treated with specific psychotherapeutic methods. However, only few BPD patients receive these evidence-based treatments [8
One way to provide more treatment facilities and to close the gap between demand and supply in the treatment of BPD is the implementation of internet interventions. In the past years, numerous eHealth interventions were successfully adapted for a broad range of mental disorders like depression, anxiety disorders, or post-traumatic stress disorder [9
]. For BPD, research on internet interventions is also promising, but still in its infancy since most studies are still in early stages. One randomized controlled trial (N = 80) on an internet-based psychoeducation-tool for women with BPD [12
] and several pilot studies on smartphone applications [13
] showed to be efficient in reducing BPD or specifically targeted symptoms. These studies have also shown that patients perceived the apps as user-friendly and helpful [13
]. A recent meta-analysis which included data from twelve studies of ten smartphone applications targeting BPD symptoms found no significant effect on reduction in BPD symptoms of those apps compared to face-to-face therapy or waitlist and concludes that more research is needed to determine if and how such apps could be implemented in providing mental health care for BPD [17
The intervention used for the present study, priovi, is an internet intervention, based on schema therapy (ST) [18
], which was specifically designed for BPD patients. A feasibility study of priovi in conjunction with individual face-to-face ST showed improvement in BPD symptoms in the participating patients. Also, qualitative data revealed that priovi was positively received by both patients and therapists [15
]. A randomized controlled trial comparing priovi plus care as usual to care as usual alone is currently ongoing [19
Priovi was developed by GAIA, a research and development enterprise in the field of e-health tools for mental disorders, together with clinical experts in BPD and BPD patients. It was designed to specifically address the needs and frequent mood shifts of BPD patients. As it is based on ST, priovi tries to establish a good therapeutic alliance with the patient by using a validating, soothing, and friendly tone. [20
One of the reasons why there are so few studies on internet interventions of BPD are concerns that BPD patients need interpersonal contact and a stable therapeutic alliance in treatment and that these needs cannot be met by an e-health tool [20
The therapeutic relationship is widely accepted as an important predictor of therapy outcome [21
]. Also in patients with BPD, the therapeutic alliance has been found to be an important common factor predicting treatment outcome [23
]. For internet-based psychotherapy, a meta-analysis found similarly strong associations between alliance and treatment outcome as in face-to-face psychotherapy [24
]. It should be noted that most of the studies included in that meta-analysis targeted guided internet interventions, where an internet-based program is combined with regular support by a therapist and focused the alliance between patient and supporting therapist. Some studies also assessed the therapeutic relationship of patients with the online intervention itself and suggest that patients are in fact able to form a therapeutic alliance with a technology-based intervention [25
]. One study implies, though, that this alliance may be different from the alliance to the therapist. Specifically, the affective bond seems to be lacking in comparison to the therapist, while there was no difference between the patients’ alliances with the online program and the therapists regarding the agreement on tasks and goals of the therapy [25
]. Another study found patients could establish a stable working alliance with an avatar in an online intervention for insomnia, but some patients missed having a real therapist [26
Until now we are only aware of quantitative studies to measure the alliance in internet interventions. However, to clarify the role of the alliance in internet interventions also qualitative interview studies are needed to understand how patients perceive this ‘technological therapeutic relationship’ and in what ways it differs from and is similar to a human relationship.
When it comes to BPD, we have no information (neither from quantitative nor from qualitative studies) to what extent patients can form a therapeutic alliance with an internet intervention and what the characteristics of such a relationship might be. This information will further our understanding of internet interventions in BPD. This information is crucial to improve and implement internet interventions for patients with BPD.
Thus, the present study aims to address the following research questions:
To what extent are patients able to form a therapeutic relationship with priovi?
What are the characteristics of this relationship?
How does it differ from the working alliance with the human therapist
We investigated the relationship between patients with BPD and the online program priovi on a quantitative and qualitative level and compared it to the relationship with the therapist providing face-to-face ST. To our knowledge, this is the first study to explore the working alliance with an internet intervention for BPD patients. Results show that patients were able to form a relationship with priovi but that it significantly differs from a human relationship. The quantitative results indicate that patients were more in agreement with their therapist about what tasks should be accomplished in therapy than with priovi during the second half of treatment. The qualitative results also indicate significant differences between the two forms of working alliance. In short, patients appreciated that priovi was always available and provided psychoeducational content and cognitive exercises, that helped them to progress in therapy. Although all patients felt validated by priovi and most could identify with priovi to a degree and some even felt it was almost like a relationship with a real person, the relationship with their therapist was perceived as being emotional on a deeper level and more individually suited to their needs.
These findings are in line with earlier qualitative pilot studies on how BPD patients experience online interventions, which report good acceptance, satisfaction, and usability of online tools [13
] and good WAI ratings for technological interventions [25
The WAI ratings of this study are comparable to those of outpatients in the study of Munder et al. [33
]. A study with insomnia patients about the therapeutic relationship to an avatar found that the WAI subscale goal was rated lower for the avatar while task results were comparable to a human relationship [26
]. Ratings of the WAIpriovi
for the task subscale in our sample (M = 3.54, SD = 0.86, after three months of treatment) are comparable to those of a study by Heim et al. [27
] (M = 3.24, SD = 0.79, after three weeks of treatment), although ratings for human therapists were significantly higher at two measurement states in our sample.
Berger et al. [27
] report a mean WAI task score of 3.8 (SD = 0.58) and 3.23 (SD = 0.69) for their tailored (to the specific anxiety symptoms of the patient) and untailored conditions respectively; scores for the WAIpriovi
in this study are in between those results, although priovi is also highly tailored for BPD patients.
A narrative review by Berger [29
] also finds alliance ratings for internet interventions to be roughly equivalent to those measured in face-to-face therapy, irrespective of communication format, diagnostic group, or amount of contact to the therapist. It also points out that the unique characteristics of the alliance in different therapeutic formats need further study.
An alliance-outcome association for the task and goal subscales was found by Penedo et al. [28
]; Heim and colleagues [26
] report that results of the WAI subscales were not associated with outcome in their trial, however, the affective bond with the avatar and whether or not patients missed a human therapist were. The authors noted that a human therapist was increasingly missed as treatment progressed, which they attributed to the use of later stages in the app: similar to priovi, it delivers mostly psychoeducation in the beginning and more specific content in the form of a (restricted) dialogue later on. The authors conclude that the increased desire for a human therapist is probably due to the restricted form of dialogue the app can offer, which would be in line with one of the main findings of this study, “priovi is less flexible”.
It is also interesting to note that we found significant differences between the WAItherapist and WAIpriovi only in the two last measurement states; examination of the mean task scores indicates that WAIpriovi scores are higher during the first two assessments and decrease somewhat after that whereas the WAItherapist scores seem more stable over time. This might be attributable to priovi being especially valuable during the phase of treatment where psychoeducation and learning of new skills are most prominent. Several patients reported this during the interviews and also mentioned having used priovi less later in treatment when they were already able to cope better on their own, and thus its relevance would also decrease over time. However, some patients also reported greater difficulty using priovi during phase two, as the program shifts from mainly delivering psychoeducation to exercises, which may also contribute to less usage.
Patients clearly stated that although priovi evokes emotions, these emotions are not as intensive as with their therapist. While the emotional connection is often regarded as a main healing factor in a human therapeutic alliance, for some patients priovi’s lower emotional impact was perceived as an advantage because they could concentrate better and were able to work more easily with priovi on exercises and psychoeducation than in personal sessions. During face-to-face therapy, these patients were usually so preoccupied with keeping up with the interaction and their thoughts and feelings about it, that they had a hard time processing information and concentrating on any exercise. This is an important aspect for the treatment of BPD patients with high emotional dysregulation and interpersonal problems indicating that teaching psychoeducational and cognitive content might be especially suitable to be done by an internet intervention. This could enhance and speed up treatment effects and disburden therapists if a combination of online treatment and face-to-face psychotherapy is offered.
Priovi’s most prominent advantages, which were mentioned by nearly all participants, are its availability and its unwaveringly validating style of communication. Priovi offers information and comfort 24/7. It never gets frustrated, tired, or angry. Patients can also take as much time as they need to complete exercises or read explanations whereas therapy sessions are always time-limited. Being able to repeat psychoeducation and exercises in between sessions was perceived positively by patients. They felt enabled to learn more quickly, sometimes to improve faster and to feel more self-efficient. This level of stability and availability is not possible to attain for a human therapist, and thus, might be a huge advantage of online interventions
On the other side, Priovi’s most clear disadvantage is its inability to react to patients more flexibly and individually. This was observed by all patients in our sample. Patients found priovi to be less able to respond to their needs in specific situations or when they were in difficult emotional states. Patients felt the need to talk personally about their issues with their therapist. They couldn’t ask priovi specific questions spontaneously and felt priovi was providing a general model of BPD and ST but that they needed their therapist to create their individual model of the disorder and understand their personal and unique issues. This is of course an inherent limitation of all technological interventions in this field.
Interestingly, one patient who initially reported a very favorable view of priovi, reported a rupture in her relationship to priovi as technological errors in the system arose (e.g., text messages stopped coming) which changed her attitude to the program completely. Patterns like this (idealization/devaluation) can also be observed in many human relationships with BPD patients; however, online interventions are not able to repair these ruptures. This implies that eHealth interventions are good additions to traditional therapy but probably not very well-suited to be used as stand-alone tools in many cases, especially in severe and complex BPD patients. This is in line with existing data on this topic [37
]. However, if no traditional therapy is available, priovi could still offer some psychoeducation and comfort, as it has been demonstrated that internet-based psychoeducation can reduce BPD symptoms and improve social functioning [12
]. Moreover, internet interventions have been shown to be suitable low-threshold interventions, also for patients with a high symptom burden [38
]. Priovi seems to be well suited for cognitive exercises and delivering psychoeducation but cannot meet individual needs or address personal issues.
While the therapeutic relationship has for the longest time been considered an underlying, facilitating factor for change, regardless of the psychotherapeutic method used, there have recently been some researchers questioning this view. Schweiger et al. [39
] suggest that the therapeutic relationship may be viewed as an emergent phenomenon, resulting not only from “chemistry” between therapist and patient but is composed by different aspects interacting with each other such as professionalism, emotional bond, plausible models, explanations and suitable techniques, or experience of change early in treatment. Given our results, it might be that specific ingredients of the therapeutic alliance can be better addressed by a human therapist (e.g., emotional bond, corrective experience of a healing relationship or a new way to deal with emotions and needs) while others might be sufficiently or even better addressed by online interventions (e.g., explanation of plausible models, psychoeducational content or explanation of specific techniques).
We did not measure alliance-outcome associations. Literature shows mixed results in this regard. Patterns of findings suggest that it is rather the agreement on goals and tasks than the personal bond that is important for outcome in internet interventions [40
]. Thus, the therapeutic alliance in internet interventions might operate differently, and finally, one cannot clearly state that it is as important as in face-to-face psychotherapy. Heim et al. [26
] suggest that the use of an avatar may facilitate the formation of a relationship, which could also be true for priovi, as it has a protagonist, Pia, to whom patients may connect on a more personal level than to text-only interventions.
This study has several limitations. Most prominently, the small sample size is a major limiting factor. Furthermore, the low number of male participants (two for quantitative results, none for qualitative results) restricts what these results can tell us about online tools for male BPD patients. The WAIpriovi is not a validated, reliable psychometric instrument, but was modified from the WAI-SR (since validated measures were lacking at the time the study was designed) as a first exploration of the working alliance in an online intervention for BPD.
Moreover, in this study priovi was offered in conjunction with weekly face-to-face sessions with ST. We cannot be sure if patients developed a relationship to priovi per se or rather to the concept of ST which it symbolizes. It would probably be reasonable to assume that both priovi as an interactive “face” of ST and ST as a concept are involved in this kind of relationship and are hard to separate.
Possible applications of internet interventions like priovi may be low-threshold delivery of psychoeducation, cognitive content and validation as well as using it in addition to traditional face-to-face therapy to speed up learning and acquisition of new skills along with having a reliable source of comfort outside of therapy sessions.
Future research should include larger sample sizes and more male participants as well as a reliable and valid instrument to capture alliance in Internet Interventions such as the newly developed Working Alliance Inventory for Internet Intervention (WAI-I) [41
]. Also, more qualitative research to better understand the specific characteristics of the therapeutic relationship with the online intervention and with human therapists is urgently needed.