Pilot Study of Blended-Format Interpersonal Psychotherapy for Major Depressive Disorder

Internet-delivered psychotherapy may increase access to care, empower patients, reduce therapists’ work time, and improve the quality and cost-effectiveness of care (1). Internet-delivered cognitive-behavioral therapy (CBT) has been proven effective (2) in clinical populations (3). In the Netherlands, most mental health care organizations offer Internet-delivered CBT to patients who are interested in this format. Internet access is seldom a barrier: as of 2019, 97% of the adult Dutch population had Internet access (4). To our knowledge, although Internet-delivered interpersonal psychotherapy (IPT) is in development, it has not yet been studied in clinical populations.

For clinical populations, Internet-delivered treatment has mostly been delivered through an integrated format of online psychotherapy in combination with face-to-face (either in person or via videoconferencing) sessions with a therapist (blended format). A blended format combines the advantages of face-to-face therapy with those of active participation in an interactive online program. This approach has been supported by a recent study on blended treatment (5), in which it was demonstrated that adding an in-person component to enable the development of a therapeutic alliance was associated with higher motivation to initiate and sustain engagement with Internet-delivered care. When in-person sessions are replaced by videoconferencing sessions, geographical distances can be overcome, as can social distancing restrictions caused by the COVID-19 pandemic (6).

In the current study, Internet-delivered IPT for patients with depression was developed and tested to examine whether the blended format would be feasible in specialized mental health care to serve the growing group of patients who are motivated to engage with an online program. We hypothesized that it would be feasible to deliver IPT in a blended format to patients receiving specialized mental health services, and we expected the blended IPT to result in clinical improvement from baseline to posttreatment.

Methods

Participants and Procedure

This study was designed as an uncontrolled pilot study with repeated measures. The study (registration: 2018.58) was approved by the Medical Ethics Committee of Vrije Universiteit (VU) Medical Center, Amsterdam. All participants provided written informed consent.

Participants were recruited from September 2019 to August 2020 at an outpatient academic clinic for mood disorders within GGZ inGeest Mental Health Care in Amsterdam. Participant inclusion criteria consisted of having a diagnosis of major depressive disorder, being ages 18–64, having Dutch language proficiency, and having access to a computer or tablet with an Internet connection. Exclusion criteria consisted of having a severe psychiatric disorder or disability (current or past psychosis, current or past bipolar disorder, acute suicidality requiring emergency intervention, intellectual disabilities). We aimed to include 20 participants.

The researcher (A.Y.S.) informed the clinic staff about the study during multidisciplinary treatment team meetings and regularly contacted the team to remind the clinicians to recruit patients for the study. Patients expressing interest in participating were contacted by a researcher from VU and screened over the telephone to determine whether they met all the inclusion criteria or any of the exclusion criteria. After providing written informed consent, participants were enrolled in the study and were asked to complete an online assessment for baseline measurements.

Intervention: Blended Face-to-Face and Internet-Delivered IPT

Our blended-format IPT was modeled after the brief, six-session version of IPT called interpersonal counseling, which was developed by Weissman and colleagues for use in primary care settings (7). Therefore, our intervention included six online sessions and followed the structure of interpersonal counseling. We developed the first version of our Internet-delivered IPT (an unguided self-help format for use in clinical populations) (8) and adapted it for a blended-format approach. In our blended approach, we intended to alternate the delivery of six in-person and six Internet-delivered IPT sessions. This alternating approach was also used by Kooistra et al. (3) for CBT and was demonstrated to be feasible and effective. We offered the option of having one to four extra in-person sessions in case more time with the therapist was needed for optimal patient benefit. At the start of the therapy, patients were informed about its duration and the option of having extra in-person sessions if needed. See Table 1 for the content of the online sessions.

The blended IPT began with a face-to-face session in which the therapist explored the patient’s current symptoms and problems and introduced the online platform and blended format. The next session was online, with the patient participating at home. In this online session, psychoeducation about depression was given and the concept of the link between life events and depression was introduced. Patients then performed writing exercises in which they described their depressive symptoms and problems and filled out a timeline of the depressive episodes and related life events. The duration of an online session depended on how long it took a patient to read the material and to perform the writing exercises. We did not measure the time spent on the Internet module in this study. After the patient completed the online session and its writing exercises, the therapist received a notification and provided comments (via asynchronous chat) on the patient’s work. In the next face-to-face session, the therapist and patient briefly reviewed the online session and discussed points of confusion; in addition, the therapist used conventional IPT strategies that fit the treatment phase. Techniques such as exploration, encouragement of affect, and clarification were used. Sessions could be repeated if necessary.

Below, we give some examples of the content of online exercises, which were performed by patients at home. In the first online session, patients completed an interpersonal inventory by filling out a sociogram and describing important relationships. Possible problem areas were introduced in the second online session, during which patients chose the problem area that seemed to best fit their needs. This choice was further elaborated during the following face-to-face session with the therapist. During the treatment phase, patients could access only the sessions related to their chosen problem area. Exercises differed by problem area. For example, in sessions to address grief, patients were asked to describe the relationship with their loved one, the circumstances under which their loved one had died, and how they had handled their emotions and with whom they had shared their feelings. In sessions for role disputes, communication techniques were described and illustrated with short videos. Patients were asked which communication styles they recognized and which options for improvement they saw. Throughout the entire module, fictionalized patient vignettes and examples of completed writing exercises were included so that patients could learn from others. Because IPT is an affect-focused treatment, patients were encouraged to describe the emotions they experienced in specific situations and to learn to link mood states with depression, via an exercise to reflect on the mood changes indicated on weekly administrations of the nine-item Patient Health Questionnaire (PHQ-9). During the face-to-face sessions, as well as in the online feedback, therapists could use the technique of encouragement of affect.

The Internet-delivered components of our blended IPT intervention were made available as an online treatment protocol on the Minddistrict platform; online IPT formats may be developed on other electronic health platforms in the future. The Internet-delivered IPT intervention was developed in 2019 by VU on the initiative of the sixth author (T.D.), in collaboration with the first (D.v.S.) and third authors (E.D.) and in accordance with the guidance of Columbia University Professor Myrna Weissman.

Therapists

Therapists from the mood disorder department of GGZ inGeest Mental Health Care who were interested in the study were trained in the blended IPT during two 2.5-hr face-to-face workshops. The first session focused on IPT in general, and the second session focused on how to work with the Internet-delivered IPT and blended format. In addition, a treatment manual was made available, in which the content and aim of each session were described. Various scenarios and suggestions for online feedback were included. The therapists were instructed to deliver 45-minute face-to-face sessions. The therapists attended supervision sessions with a certified IPT supervisor (D.v.S.) for 1 hr every 2 weeks for 4 months. The training workshops were recorded as video sessions, and some therapists who started to participate at a later stage viewed the videos for self-study.

Participating therapists consisted of a psychologist, a nurse specialist, three residents in psychiatry, and a psychiatrist. The nurse specialist and psychiatrist were experienced IPT therapists, but the others were not familiar with IPT.

Measurements

All questionnaires were completed online. Data on sociodemographic characteristics (sex, age, marital status, education, work status, and previous mental health care) were collected at baseline.

The primary outcome measure was the feasibility of the intervention. The intervention was considered to be feasible when more than 60% of the participants had completed more than 50% of the online sessions. Therapists reported the number of sessions that were completed to the research assistant. In addition, postintervention (at 16 weeks) satisfaction was measured with the Client Satisfaction Questionnaire–8 (CSQ-8) (9). This questionnaire includes eight items on a 4-point scale. The total score ranges from 8 to 32, and a higher score indicates greater satisfaction. In addition, qualitative interviews were conducted. We expected to get good feedback by asking five patients who completed all the online sessions (almost 25% of the total sample) to participate in a more in-depth interview about their experiences and satisfaction with the intervention. The usability of the online sessions of the blended IPT was measured with the System Usability Scale (SUS). A score of 68 on the SUS (range 0–100, where higher scores indicate greater usability) is often used as a cutoff point to distinguish above-average from below-average usability scores (10, 11). The SUS has been used in several studies to measure usability of Internet-delivered CBT (12). Therapists’ feedback on the blended IPT was collected through e-mail and individual videoconferencing sessions.

Finally, to assess the initial effects of the intervention on depressive symptoms, the PHQ-9 scores at baseline and postintervention (16 weeks) were used. PHQ-9 scores range from 0 to 27, with higher scores indicating more severe depression. The optimal cutoff score for a probable diagnosis of major depressive disorder is 10 (13). Recent research (14) has shown that the PHQ-9 is also a feasible and reliable screener for depression when delivered via Web-based technology.

Analysis

Descriptive statistics were used to analyze demographic variables, dropout rates, feasibility, usability (assessed with the SUS), and patient satisfaction (assessed with the CSQ-8). Depressive symptom severity, as assessed with the PHQ-9, was investigated with a paired-samples t test, with a significance level of p<0.05, before and after the treatment to analyze the within-group treatment effect. Treatment response was defined as a 50% reduction in PHQ-9 score, and remission was defined as a PHQ-9 score of <5 (15). To obtain Hedges’ g, we standardized the difference in depression symptom scores from baseline to postintervention by its standard deviation and then multiplied the result by the finite sample size correction factor J (16). The standard deviation of the difference could be derived from the value of the paired t test statistic, the difference itself, and the number of patients. We chose Hedges’ g because it allows for small sample size bias correction. A Shapiro-Wilk test was performed on SUS, CSQ-8, and PHQ-9 difference sum scores to test for the normality assumption. All statistical analyses were performed with SPSS, version 26.0; Hedges’ g was calculated by using an online effect size calculator (https://effect-size-calculator.herokuapp.com).

Thematic content analysis, the standard methodology for analyzing qualitative structured interview data, was used (17). Interviews were recorded, anonymized, and transcribed verbatim by an independent research assistant. A different independent research assistant manually conducted the content analysis of the transcripts. Data sampling was carried out according to a supervised, guarded protocol to ensure data quality.

Results

In total, 26 eligible patients were identified; five refused to participate in the study. Twenty-one patients started the blended IPT. Baseline characteristics of 20 of the patients are presented in Table 2. One patient’s baseline demographic data became corrupted because of technical problems when the questionnaires were uploaded.

Four patients did not fill out the postintervention questionnaires; therefore, the outcomes of usability and satisfaction were based on 17 questionnaires. For the symptom severity outcome, analyses were based on all 21 participants, because we were able to extract missing data from the treatment platform.

Feasibility

Of the 21 patients who started the blended IPT, 19 (90%, 95% CI=70%–99%) fully completed the online portion of the intervention by working through all six sessions, one patient completed two sessions, and another completed five sessions. Of these two latter patients, one preferred standard CBT therapy, and the other dropped out because his therapist left for another job and the patient did not want to continue the intervention with a new therapist. The original protocol called for the patients to alternate in-person and online sessions, but for about half the patients, the in-person sessions were replaced by videoconferencing sessions because of COVID-19–related restrictions. Despite COVID-19, the in-person or videoconferencing completion rate was high (19 of 21 participants completed treatment).

Usability and Patient Satisfaction

The mean±SD score (N=17) for usability, as measured with the SUS, was 66.0±12.4. The mean score (N=17) for patient satisfaction with the treatment as a whole, as measured with the CSQ-8, was 25.12 ±3.55. Although the CSQ-8 has no cutoff scores, the score in our study can be interpreted as positive. Five patients were purposively selected to participate in a qualitative interview by telephone (two men, ages 30 and 49, and three women, ages 35–58). All five had completed five to six online and six to 10 face-to-face sessions. These participants rated the overall experience with the treatment as favorable (average rating of eight out of 10; range 1–10, with a higher score indicating more satisfaction). All five participants indicated that the online platform and modules were easy to use, despite some temporary problems with the platform. Sufficient access to technology at home, especially for videoconferencing, was seen as important. Participants reported that they liked taking an active role in their treatment by completing a large part of the IPT individually online, and they liked the encouragement to make their own choices. The coherence of the online sessions, combined with the in-person or video conversations with the therapist, was mentioned as most valuable. All five interviewees had former experience with psychotherapy and mentioned that the blended IPT distinguished itself from conventional therapies in that the online sessions made the setup and overview of the therapy well structured and clear. In addition, the blended IPT was perceived by some patients as more intense than other therapies, because the content of online exercises could compel confrontation.

Therapist Feedback

Except for some technical problems with the treatment platform, which were solved during the study, the therapists were generally satisfied with Internet-delivered IPT. They noticed that the exercises urged patients to actively work on their recovery. Therapists with no former experience with IPT found the online protocol helpful in guiding the treatment process. The two experienced IPT therapists reported that the online activities and insights gained by the patients (e.g., in describing their relationships and communication styles) exceeded their expectations.

Symptom Severity

The intention-to-treat analysis (N=21) showed a significant decrease in PHQ-9 score, from 17.48±5.41 at baseline to 11.90±6.45 postintervention (t=4.86, df=20, p=0.001). In this same group of 21 participants, the treatment response rate (50% reduction in PHQ-9 score) was 33% (95% CI=15%–57%, N=7), and the remission rate (PHQ-9 score <5) was 19% (95% CI=6%–42%, N=4). Hedges’ g (pre- to posttreatment) was 0.90 (95% CI=0.44–1.41), representing a large effect size.

Discussion

This study was, to our knowledge, the first to examine the feasibility and initial effects of a blended face-to-face and Internet-delivered version of IPT for depression in routine specialized mental health care. Results showed that Internet-delivered IPT was feasible in specialized mental health care, with >60% of the participants completing >50% of the online sessions. The completion rate of 90% (95% CI=70%–99%) among patients who started the blended-format IPT seems to be comparable to two other studies on blended-format CBT. Kooistra et al. (3) found a completion rate of 75%, and Thase et al. (18) found a completion rate of 82%. For guided Internet-delivered CBT, without face-to-face contact and with only written feedback, an average completion rate of 68% was found in a meta-analysis (19).

A concern about Internet-delivered IPT may be that some human elements of the therapy, especially the affect-focused aspect, may be lost when patients work on their own (e.g., when describing their role dispute). In that sense, IPT may differ from CBT, in which cognitive and behavioral strategies may be more easily translated into Internet-delivered formats. However, both IPT and CBT have a clear structure that can be described explicitly on the Internet. Patient autonomy is promoted: patients are activated and urged to think about their situation and possible solutions. In developing the Internet-delivered IPT, extra attention was given to addressing affects in online exercises, as described earlier. For some patients, this approach worked surprisingly well. Of course, this was not the case for all patients. Some patients needed extra guidance and support to work through the exercises. All patients had the option to attend additional face-to-face sessions to elicit affects and improve interpersonal skills.

Overall, patients felt satisfied with the intervention as a whole (mean CSQ-8 score=25.12±3.55). In a meta-analysis (20) of 24 trials with mostly guided Internet-delivered CBT, median satisfaction for Internet-delivered CBT was rated as positive to very positive by 86% (range 60%–100%) of the patients. Thus, our findings seem comparable with those of previous studies, although, to our knowledge, until now no direct comparisons have been available for patient satisfaction with blended versus guided Internet-delivered interventions.

The SUS score for the Internet-delivered IPT was 66.0±12.4; this score was somewhat below the cutoff value for average usability of 68.0. In a study of Internet-delivered CBT, the mean score on the SUS was 67.9±16.3 (12). During our study, there were some technical problems with the online program. These temporary glitches could have added to the relatively low score on the SUS.

We found significant improvement in depressive symptoms, from 17.48±5.41 on the PHQ-9 at baseline to 11.90±6.45 postintervention, with a Hedges’ g of 0.90 (95% CI=0.44–1.41) indicating a large effect size. This result lies within the range of effects found in Internet-delivered CBT for depression and anxiety (21). Our treatment response (33%, 95% CI=15%–57%) and remission (19%, 95% CI=6%–42%) rates appeared somewhat lower than those found in the literature. Several factors may have accounted for these findings, but considering the small sample size and accompanying broad CIs, no firm conclusions can be drawn. In a recent meta-analysis (22), mean response and remission rates of psychotherapy for depression were 41% and about 30%, respectively.

This study was limited by its design as an uncontrolled pilot study with a small sample size. The results were favorable regarding completion rate and effect size, but the latter must be interpreted with caution because of the small sample size. The results from the qualitative interviews and satisfaction measures may have been positively biased because they were based on data from those who fully completed the treatment. In this small sample, there was no reliable way to handle missing data.

Conclusions

Blended-format IPT, which integrated face-to-face and Internet-delivered sessions, for depression in a clinical sample showed favorable results regarding feasibility and patient satisfaction. For patients who are motivated to engage with an online program, this format may increase access to care and may improve care quality because of the platform’s clear setup and active patient participation in the online sessions. More extensive research should be done to test Internet-delivered IPT in a randomized controlled trial, with an adequate sample size, to assess cost-effectiveness and predictors of treatment outcome.