Online-delivered eye movement desensitization and reprocessing treatment for adults with post-traumatic stress disorder due to multiple traumas: a non-concurrent multiple baseline design

ABSTRACT

Background: No controlled studies incorporating randomization have been conducted to investigate the effectiveness of online eye movement desensitization and reprocessing (EMDR) treatment, despite its use in clinical practice.

Objective: This study evaluated the effect of online EMDR treatment in adults aged 18–65 years with post-traumatic stress disorder (PTSD) resulting from multiple traumas.

Method: A multiple baseline single-case experimental design (n = 21) was employed. Participants were patients with PTSD due to multiple traumas, recruited from a mental healthcare institution in the Netherlands. They were randomly assigned to baseline phases of 2, 3.5, or 5.5 weeks. After this, participants received 10 weekly online EMDR sessions. The primary outcome was the total score on an adapted version of the PTSD Checklist for DSM-5 (PCL-5), which was administered twice a week during baseline and intervention phases, and once 12 weeks after the end of the intervention phase. We performed visual analysis and calculated the improvement rate difference (IRD) for each individual separately to determine whether online EMDR was effective. We also performed a paired t-test and calculated Cohen’s d for pretreatment and post-treatment comparisons, and pretreatment versus follow-up to evaluate effects at the group level.

Results: Visual analysis and IRD scores showed that the treatment was effective for 15 participants, with effect sizes ranging from small to very large. The mean scores on the PCL-5 at group level decreased significantly over time between pretreatment and post-treatment (M_diff= 23.6, Cohen’s d = 1.34, 95% CI 0.74–1.93), as well as between pretreatment and follow-up (M_diff= 27.2, Cohen’s d = 1.62, 95% CI 0.95–2.27).

Conclusion: Most participants showed a reduction in symptoms following the start of the online EMDR. Furthermore, at the group level there was a significant and clinically relevant reduction in symptoms over time. This provides preliminary evidence for the effectiveness of online EMDR treatment.

HIGHLIGHTS

• This study used a single-case experimental design to evaluate the effects of online EMDR on trauma symptoms in 21 patients with PTSD due to multiple traumas.

• Online EMDR was effective in 15 out of 21 participants.

• At the group level, the average reduction in trauma symptoms on the PCL-5 was clinically relevant.

1. Introduction

In 2020, the Netherlands, like the rest of the world, was confronted with coronavirus disease 2019 (COVID-19). The government strongly advised working from home, leading mental health institutions in the country to transition to treating patients through video call programmes. Therapies had to be shifted to online platforms. Currently, online treatments in mental healthcare are still utilized, even in the absence of COVID restrictions. Online treatment in general can have advantages for some patients, owing to increased accessibility, convenience, and flexibility, and being in the familiar environment of one's own home can lead to a more relaxed and comfortable experience. These general benefits of online treatment raise the question of whether evidence-based interventions, such as eye movement desensitization and reprocessing (EMDR), can also be effectively delivered in an online format. EMDR is a guideline-recommended treatment for post-traumatic stress disorder (PTSD) according to the guidelines committee of the International Society for Traumatic Stress Studies (ISTSS, 2019). However, these guidelines do not provide information on the effectiveness of online EMDR.

Various studies have demonstrated the clinical effectiveness of online treatment of PTSD using cognitive–behavioural therapy (CBT), another guideline therapy for PTSD. A systematic review in 2019 (Lewis et al., 2019) showed the positive effects of online CBT treatment for PTSD. In a non-inferiority trial, online-delivered CBT for PTSD was not inferior to in-person CBT (Acierno et al., 2017). In addition, another study found no difference in clinical effectiveness between online CBT and in-person CBT (Morland et al., 2020).

Less evidence is available for the effectiveness of online EMDR treatments for PTSD. Online EMDR was as effective as in-person EMDR in 160 people with post COVID related trauma (Russo et al., 2024), but this study did not focus specifically on people with PTSD. A relevant line of experimental research has examined the effects of eye movements and other dual-attention tasks in analogue paradigms. These studies show that recalling an aversive autobiographical memory while simultaneously performing eye movements reduces the vividness and emotionality of that memory, consistent with the working-memory account of EMDR (Houben et al., 2020; van Veen et al., 2015). More recent technology-supported paradigms, including computerized and virtual-reality eye-movement tasks, demonstrate that such dual-attention tasks can also be effectively delivered digitally while maintaining their impact on memory vividness and emotionality (IJdema et al., 2023).

Importantly, these experimental findings appear to translate to the clinical context. Studies investigating online EMDR and remote trauma-focused treatments that include EMDR have shown that EMDR can be effectively delivered via digital platforms, with substantial reductions in PTSD symptoms (Bongaerts et al., 2021; Bongaerts et al., 2022; Ellenbroek et al., 2024; Matthijssen & Menses, 2024). In addition, studies examining online EMDR as a stand-alone intervention suggest comparable effectiveness to in-person delivery, although controlled evidence remains limited (Spence et al., 2013; Strelchuk et al., 2023). Together, these findings support the plausibility of delivering the eye-movement component of EMDR in an online format, while also highlighting the need for controlled studies.

In addition, traditional group-based studies only give information about average effects on a group level, but not about effects for individual participants. It is increasingly being recognized that group-based studies calculating average results often yield information that is not applicable to individual patients (Vlaeyen et al., 2020; Zuidersma et al., 2020). The single-case experimental design (SCED) can serve as a complementary study design to obtain information about treatment effects. In SCEDs, participants act as their own control, which comes closer to the ideal counterfactual experiment than studies making comparisons between people (Vandenbroucke, 2006). In a SCED, all participants have a baseline phase as a control condition, and outcomes during the intervention phase are compared with the baseline control phase. SCEDs are distinguished from traditional group-based randomized controlled trials (RCTs) in that in SCED studies, effects are analysed for each participant separately, while group-based RCTs analyse average effects for the whole sample. This eliminates the possibility that treatment outcomes can be explained by differences between individuals [which may even occur with randomization (Saint-Mont, 2015)]. In addition, in SCEDs, each person has a large number of repeated observations, taking into account within-person variability and thereby increasing the reliability and interpretation of the results, compared with studies relying on only one or a few outcome assessments (Senn, 2018).

The current study is a SCED study on the effect of online EMDR treatment in adults aged 18–65 years diagnosed with PTSD resulting from multiple traumas. First, we hypothesize that online EMDR is effective in reducing self-reported PTSD symptoms for most people during treatment (based on analyses at the individual level; results from individual participants). The second hypothesis is that, on average in the whole group, there will be a clinically relevant reduction in PTSD symptoms in the short and long term (based on analyses at the group level; traditional group-aggregated results).

2. Method

2.1. Participants

From October 2022 to November 2023, adults (18–65 years) who were referred to Synaeda, a mental healthcare institution in the Netherlands, and diagnosed with PTSD due to multiple traumas, were offered the opportunity to participate in this study. Patients who were interested in participating in the study were screened by therapists for the inclusion and exclusion criteria. These therapists had received training in the administration of the Mini-International Neuropsychiatric Interview – Simplified for DSM-5 (MINI-S) (Sheehan et al., 1998). Inclusion criteria were a diagnosis of PTSD determined according to the MINI-S and having experienced a minimum of three A-criteria worthy traumatic events. Additional inclusion criteria were access to a laptop or personal computer, a stable internet connection, and the ability to ensure a secure and quiet environment for EMDR sessions. Patients were excluded if there was a severe level of suicidality or if they were using hard drugs. In cases where patients used soft drugs or benzodiazepines, they were excluded unless they agreed to no use on the day before, the day of, and the day after the EMDR session.

2.2. Study design

This controlled study employed a non-concurrent multiple baseline design, a type of SCED suitable for evaluating effects of interventions with long-lasting or irreversible effects. The multiple baseline design addresses threats to internal validity by randomizing participants to different lengths of the baseline phase (Ferron & Sentovich, 2002). Non-concurrent means that not all participants started the study at the same time.

The current study was designed according to the What Works Clearinghouse standards for SCED studies (Kratochwill et al., 2012). Participants were randomly assigned to a baseline phase of 2, 3.5, or 5.5 weeks. These lengths were set to meet the minimum requirement of five measurements in the shortest baseline phase and to prevent the longest baseline phase from resulting in an excessively long waiting time before the start of the intervention. During the baseline phase, no treatment took place. Following this, 10 sessions of online EMDR took place, predominantly once a week, with each session lasting for 60–75 min. The duration of the sessions was not predetermined in this study, but varied depending on therapeutic needs. Throughout both the baseline and intervention phases, severity of PTSD symptoms were assessed twice a week with the Dutch version of the PTSD Checklist for DSM-5 (PCL-5) (Boeschoten et al., 2014; Blevins et al., 2015), which we adapted for assessment twice a week. This adaptation was necessary because SCED methodology requires frequent repeated measurements to reach a sufficient number of data points to reliably evaluate intervention effects. Twelve weeks after the end of the intervention phase, a follow-up measurement with the same adapted version of the PCL-5 was performed. As a result, each participant had a minimum of 26 measurements and a maximum of 33 measurements, including the follow-up measurement.

The study protocol was submitted to the Medical Ethics Review Board of the University Medical Center Groningen (METc UMCG). The board concluded that no approval was needed from an METc since the study was not clinical research with human subjects as meant in the Medical Research Involving Human Subjects Act (WMO). The protocol was developed before analyses were performed, but the study was registered after data collection, on https://osf.io/vx3kj. Deviations from the protocol after registration are described in Appendix 1 in the Supplementary material.

2.3. Randomization

After individuals had provided informed consent, randomization for the length of the baseline phase was performed for the included participants. Simple randomization was achieved by rolling a dice, using the web application available at https://www.random.org/dice/. If a roll resulted in a 1 or 2, the participant was assigned to the 2 week baseline phase. If the roll resulted in a 3 or 4, the participant was assigned to the 3.5 week baseline phase, and if the roll resulted in a 5 or 6, the participant was assigned to the 5.5 week baseline phase.

2.4. Intervention

The EMDR treatment was conducted by nine EMDR therapists, each of whom had completed at least the advanced EMDR training by an EMDR Europe accredited trainer. Each participant had only one therapist throughout the intervention phase. Group supervision sessions with an EMDR consultant in training took place once every 2 months. At the beginning of each treatment, the case conceptualization was presented to the EMDR consultant in training, who was also available for questions in between consultations.

In this study, the online EMDR treatment consisted of 10 sessions of 60–75 min, ideally scheduled weekly within a 10 week period. Treatment was completed earlier when a participant no longer reported intrusive images, and the PCL-5 score was below 31. The participant was then considered an ‘early completer’. If a participant either did not initiate therapy or completed fewer than five out of the 10 treatment sessions, and did not qualify as an early completer, they were categorized as a dropout. Data of dropouts were not included in the analysis.

During the first session, the case conceptualization was established collaboratively with the participant. In consultation between the participant and the therapist, it was decided whether there was sufficient time to proceed with EMDR during this first session. The standard Dutch EMDR protocol (2022) was employed in this study, utilizing the online EMDR platform WeMind (https://moovd.nl). This platform integrates video calling and EMDR treatment, incorporating regular eye movements while also using an algorithm. The WeMind platform uses an adaptive dual-task procedure to optimize working-memory load during EMDR. Participants track a moving visual stimulus (a ball) and respond by clicking when the ball enters a target (a cylinder). Response speed and errors are used as indicators of working-memory load. When responses are fast and accurate, task difficulty is increased by accelerating the stimulus; when responses slow down or errors increase, task difficulty is reduced. This allows continuous within-session calibration of working-memory load, accounting for individual differences and fluctuations over time. The rationale for the within-session calibration procedure is explained in Appendix 2. The exact algorithmic thresholds are proprietary and were not available to the researchers, which limits full replicability.

2.5. Measurements

2.5.1. Mini-International Neuropsychiatric Interview – Simplified for DSM 5 (MINI-S)

The Dutch translation of the MINI-S (Hergueta & Weiller, 2017; Dutch translation: Overbeek & Schruers, 2019) was administered before the start of the study to confirm the diagnosis PTSD made by the referring therapist. The MINI-S is a simplified version of the MINI (Sheehan et al., 1998) and is updated for the DSM-5. It is a brief structured clinical interview focusing on the symptoms of psychiatric disorders. In this study, only the part related to PTSD was administered. The MINI provides a reliable DSM diagnosis (Sheehan et al., 1998). The validity and reliability of the Dutch version of the MINI-S are not available.

2.5.2. PTSD Checklist for DSM-5 (PCL-5)

The PCL-5 is a widely used self-report instrument consisting of 20 items assessing the 20 symptoms of PTSD according to the DSM-5. In this study, the Dutch translation of the PCL-5 was used (Boeschoten et al., 2014). The reliability and validity of the original English version of the PCL-5 have been established (Blevins et al., 2015). This instrument was adapted for administration twice a week by asking participants to report on the past half-week instead of the typical past week.

2.6. Statistical analysis

Statistical analysis was performed by an independent researcher (SK) who was not involved in EMDR treatment.

2.6.1. Hypothesis 1: Effects of online EMDR at the individual level

To test the hypothesis that online EMDR is effective in reducing self-reported PTSD symptoms for most people, visual analysis (Morley, 2018) was conducted for each participant separately to determine effectiveness at the individual level. In this analysis, we compared level, trend, and variability between the baseline (A) and intervention (B) phases, and in addition evaluated immediacy and overlap (Ledford et al., 2018). We used the online application of Bouwmeester and Jongerling (2021) to examine the level, trend, and variability of the PCL-5 scores within both the baseline and intervention phases as well as between the phases. We deemed the intervention to be effective when there was a mean difference between baseline and intervention of at least 10 points (level, based on the fact that a change of around 10 points is seen as an indicator for response) (Weathers et al., 2013), when there was a negative change in the mixed regression slope (trend) and when the data were stable in both phases (variability). Based on criteria set out by Ledford et al. (2018), we considered the data pattern to be stable when 80% of the data points fell within the stability envelope around the median regression slope within a phase. The stability envelope was set at 25% of the median value. In addition, we inspected the immediacy of the effect. We deemed the intervention to be effective on immediacy when the effect became visible within the first 3 weeks.

Overlap was calculated with the ‘improvement rate difference’ (IRD) (Parker et al., 2009). For this purpose, we calculated the improvement rate (IR) for each phase by dividing the number of improved data points by the total number of data points in that phase. As defined by Parker et al. (2009), we defined an improved data point in the baseline phase as one that ties or exceeds any data point in the intervention phase. We defined an improved data point in the intervention phase as any that exceeds all data points in the baseline phase (where ‘exceeds’ means a lower PCL-5 score, in our situation). After calculating the IR for both phases, the IRD was calculated by subtracting the baseline IR from the intervention IR. Benchmarks for the IRD are as follows: IRD < 0.50 is considered a small and questionable effect, IRD between 0.50 and 0.70 is considered a moderate effect, between 0.70 and 0.75 as large, and IRD > 0.75 as very large (Parker et al., 2009). As we anticipated that the effect may occur after one to three EMDR sessions, the IRD is expected to be lower. Hence, we set a minimum threshold for the IRD of 0.3 to consider it demonstrating effectiveness. In addition, we employed Parker’s benchmarks for small, moderate, large, and very large effect.

For proper calculation of the IRD, there should be no trend in the baseline phase. We expected no trend during the baseline phase, because PTSD symptoms are not likely to decline or increase without treatment. However, when there was a clear trend in the baseline phase, which could not be ascribed to outliers or extreme variability, we calculated the tau-U (Morley, 2018), instead of IRD. This test accounts for trends in the baseline (Morley, 2018). We deemed a trend present when all of the following criteria were met: (1) there were at least five measurements in the baseline phase; (2) the value of the slope was greater than 2 or less than −2 (based on regression, suggesting an average increase or decrease of 2 PCL-5 points per assessment); and (3) the trend was statistically significant (p values were obtained from the online application of Bouwmeester and Jongerling, 2021, and manually recalculated when they were below .10, so that they could be presented with three decimals).

Inference criteria for effectiveness at the individual level: When the difference in level between phases was at least 10 points favouring the intervention, along with at least two of the other indices (i.e. trend, variability, immediacy, overlap) demonstrating effect, we considered EMDR to be effective for that individual. Furthermore, if the above-mentioned criteria were met but the difference in level was < 10 points, we considered the effect to be interpreted ‘with caution’; and when there were more than five missing data points we considered the effect to be interpreted as ‘unclear’. When there was a ‘positive effect’ or a ‘positive effect with caution’, we indicated, using the IRD, whether the effect was small (≥ 0.3 and < 0.5), moderate (≥ 0.5 and < 0.7), large (≥ 0.7 and < 0.75), or very large (≥ 0.75).

2.6.2. Hypothesis 2: Effects of online EMDR at the group level

To test the second hypothesis, a paired t-test and Cohen’s d calculation was conducted, comparing the total average group scores of the PCL-5 pretreatment with the total PCL-5 score post-treatment, and the total PCL-5 score at the follow-up measurement 12 weeks after the intervention ended. These are considered the primary outcomes for this hypothesis.

We considered the effect to be clinically relevant if the paired t-test was statistically significant (p < .05) and Cohen’s d was ≥ 0.80. These tests required a minimum sample size of 16 to detect a statistically significant effect with two-tailed p < .05, d = 0.8, with 80% power (Dhand & Khatkar, 2014).

3. Results

3.1. Sample characteristics

Of the 40 individuals interested in participating in the study, 30 met the inclusion criteria. Of these, 26 provided informed consent. Four men and 22 women participated in the study. The mean age was 32.9 (SD = 12.77) years. The type of trauma memories was diverse: physical violence (n = 13), sexual violence (n = 12), traumatic loss (n = 5), armed robbery (n = 2), and medical trauma (n = 2). Twenty of them had multiple trauma memories in one trauma category, and six had trauma memories in multiple categories. Of the 26 participants, two dropped out during the baseline phase owing to difficulties in integrating the treatment into their daily lives and two because of non-completion of the questionnaires, and one participant discontinued the treatment after three sessions owing to a change in treatment preference. These five individuals were not included in the visual and statistical analyses.

Of the 21 remaining participants, five were assigned to the baseline phase of 2 weeks, eight to the baseline phase of 3.5 weeks, and eight to the baseline phase of 5.5 weeks. Of the 21 participants, seven received all 10 sessions, seven were considered early completers and received between five and nine sessions, six participants received eight or nine sessions as a result of unforeseen circumstances such as illness, and one participant did not want to address a new trauma cluster and decided to stop the treatment after eight sessions. For five participants it was possible to start EMDR in the first session after establishing the case conceptualization. The other 16 started the EMDR in the second session. Details for each participant are provided in Appendix 3 (3A–U).

3.2. Hypothesis 1: Effects of online EMDR at the individual level

Visual analysis (see Figure 1 and Appendix 3A–U), combined with the gathered indices [Table 1 (see Appendix 4 for a version without colours)], showed that online EMDR was effective in 15 participants; for six of these participants the effect was considered ‘with caution’. For two participants it was deemed ‘unclear’, according to our predefined criteria. For the other four participants, the intervention was not effective, according to our predefined criteria. The formation of these conclusions and details per participant are outlined in Appendix 3A–U. Furthermore, among the 15 participants showing a positive effect, six exhibited a small effect, five a moderate effect, one a large effect, and five a very large effect, as observed in the IRD (see Table 1).

Figure 1.

Individual scores on the PTSD Checklist for DSM-5 (PCL-5) with mean plotted for both phases. Colours of the borders indicate: red: no effect; orange: unclear (owing to missing data: > 5 missing data points); light green: positive effect (difference in level between A and B ≥ 10, and at least two other indices demonstrate effect), small (IRD between 0.3 and 0.5); medium green: positive effect (difference in level between A and B ≥ 10, and at least two other indices demonstrate effect), moderate (IRD 0.5–0.7); dark green: positive effect (difference in level between A and B ≥ 10, and at least two other indices demonstrate effect), large or very large (IRD > 0.70); striped border: positive effect is viewed with caution (there is a positive effect, but difference in level between A and B < 10).

Figure 1.

Continued

Figure 1.

Continued

Table 1.

Visual analysis of PTSD Checklist for DSM-5 (PCL-5) scores and conclusions.

Note: To see this table without colours, see Appendix 4.

¹Positive effect: decline in level of at least 10 points (moderate green for level) and at least two of the other indices should also suggest an effect, i.e. negative change in trend (moderate green for trend), and/or minimal variability [at least 80% within stability envelope (moderate green for variability)], and/or immediacy within three sessions (moderate green for immediacy) and/or overlap (IRD > 0.3). Positive effect with caution: these criteria are met, but with a difference in level < 10 points (light green for level). Unclear: these criteria are met, but there are five or more missing data points (orange for all indices). No effect: no decline in level and/or fewer than two of the other indices suggest an effect (red for conclusion). Small effect: IRD 0.3–0.5 (light green for overlap); moderate effect: IRD 0.5–0.7 (moderate green for overlap); large effect: IRD 0.7–0.75 (dark green for overlap); very large effect: IRD > 0.75 (dark green for overlap).

A = baseline phase; B = intervention phase; IRD = improvement rate difference.

3.3. Hypothesis 2: Effects of online EMDR at the group level

The mean score of the 21 participants on the PCL-5 was 44.9 (SD = 18.19), 21.2 (SD = 16.52), and 17.7 (SD = 12.58) at pretreatment, post-treatment, and follow-up, respectively. The paired-sampled t-test and Cohen’s d showed a statistically significant and clinically relevant reduction of 23.6 points on the PCL-5 from pretreatment to post-treatment [t(20) = 6.15, p < .001, d = 1.34, 95% confidence interval (CI) 0.74–1.93], and of 27.2 points from pretreatment to follow-up [t(20) = 7.42, p < .001, d = 1.62, 95% CI 0.95–2.27] (Figure 2). Five participants scored below the clinical cut-off of 31 on the PCL-5 at baseline. Therefore, as a sensitivity analysis, we repeated the paired-samples t-test and Cohen’s d for the group without these five participants. This yielded a clinically relevant reduction of 27.9 points on the PCL-5 from pretreatment to post-treatment [t(15) = 6.41, p < .001, d = 1.60, 95% CI 0.84–2.34], and of 32.3 points from pretreatment to follow-up [t(15) = 8.32, p < .001, d = 2.08, 95% CI 1.18–2.95].

Figure 2.

Mean (SE) PTSD Checklist for DSM-5 (PCL-5) group scores at pretreatment, post-treatment, and follow-up.

4. Discussion

This controlled study, adopting a multiple baseline design, evaluated the effect of online EMDR treatment in adults aged 18–65 years diagnosed with PTSD resulting from multiple traumas. A clear effect was observed at individual and group level following the start of the intervention. In accordance with our hypothesis, visual analysis showed an effect on self-reported PTSD symptoms for 15 of the 21 participants after the start of the online EMDR. Of the 15 participants for whom the intervention was effective, effect sizes were small for five participants, moderate for four participants, large for one participant, and very large for five participants, according to the IRD. In line with our second hypothesis, we found a clinically relevant average reduction of 23.6 points on the PCL-5 from pretreatment to post-treatment, which is greater than what is considered clinically relevant (Weathers et al., 2013) and with a large effect size (Cohen’s d = 1.34). Furthermore, this reduction was sustained until 12 weeks after the treatment, with an average reduction of 27.2 (Cohen’s d = 1.62).

This was the first controlled study incorporating randomization on online EMDR treatment in PTSD patients, yet the results are comparable to those of previous, uncontrolled studies on online EMDR for PTSD. This study showed a similar percentage of participants who improved (15 out of 21), compared to the results of Bongaerts et al. (2021). In that study, four out of six participants showed a decrease in PTSD symptoms over time after a fully online intensive trauma treatment. Our results regarding effectiveness over time at a group level are comparable to those of Bongaerts et al. (2022) and Ellenbroek et al. (2024), despite some differences in the intervention delivered. Bongaerts et al. (2022) found a 4 day remotely delivered intensive treatment programme, containing a combination of prolonged exposure, EMDR therapy, physical activities, and psychoeducation, to be effective (d = 1.59) in 73 PTSD patients. Ellenbroek et al. (2024) demonstrated effectiveness (d = 1.15) of a 6 day remote intensive treatment programme, which included prolonged exposure, EMDR, physical activity, and psychoeducation in 26 PTSD patients. In addition, our effect size was also comparable to that of RCTs comparing the effects of in-person EMDR with waiting-list conditions (Hedges’ g = 1.13) (Cuijpers et al., 2020).

The single-case design of this study made it possible to detect variability in the effects of the intervention between participants. In this way, we found that the magnitude of the effect varied substantially between patients, with effects ranging from small to very large according to the IRD, and mean declines ranging between 3.5 (Participant 25) and 34.9 (Participant 6) points on the PCL-5. Furthermore, our results also showed that, for some patients, the effects of online EMDR are not immediate. Of the 15 participants who showed a positive effect of the intervention, six participants started to improve after at least 4 weeks following the start of the intervention. This delayed effect could be explained by the fact that our sample included patients with multiple traumas. Such delayed effects should therefore be taken into account when designing future studies on EMDR treatment for PTSD with multiple traumas.

In our study, five participants had low PCL-5 scores (below the cut-off score of 31). While a PTSD diagnosis can technically exist with subthreshold symptoms, low PCL-5 scores may also indicate limited self-awareness in some patients, resulting in underreporting of symptoms. Another explanation for the low PCL-5 scores could be that we adapted our PCL-5 to reflect a half-week instead of a week. Including patients with low baseline symptom scores in studies focusing on change or treatment effects is likely to underestimate the changes and treatment effects. The results of the present study indeed suggest that this is the case, because effects were smaller for the five participants with scores below 31 at pretreatment.

Overall, this study suggests online EMDR to be an effective treatment for PTSD with multiple traumas in most, but not all, participants. The effect sizes align with those observed in in-person EMDR therapy, endorsing the integration of online EMDR into clinical practice. This approach can mitigate obstacles such as travel time and expenses for patients, and increases accessibility for those in remote areas or those with mobility limitations. In addition to promoting access, online treatments ensure continuity of care in challenging circumstances, such as the COVID-19 pandemic. It remains unclear whether patients for whom online EMDR was not effective would have benefited from EMDR in person. Some differences between online and in-person EMDR should be noted. In the online format, the therapist interacts via a screen, limiting visibility of the patient’s full body and preventing any physical contact. In many in-person sessions, therapists guide eye movements using hand movements or a light bar; however, some also use the online EMDR platform WeMind, as was employed in the present study for the online EMDR treatment. Despite these differences, the core components and procedures of EMDR remain unchanged.

The present study has several strengths. First, this was the first study comparing online EMDR with a predefined control condition incorporating randomization within individual PTSD patients. By randomizing the length of the baseline (control) phase, internal validity is enhanced (Michiels & Onghena, 2019). Secondly, the adoption of the single-case design enabled us to find effects for individual participants and thereby make statements about the number of people for whom the intervention is effective. Thirdly, the inclusion of participants with different types of trauma enhances the overall generalizability of the outcomes of the study. Notably, the study’s strength lies in its inclusion of participants who have experienced multiple traumas, highlighting the effectiveness of the treatment in this population.

This study also has several limitations. Some limitations are related to the internal validity of the study. First, this study adopted a non-concurrent design, meaning that the participants’ baseline phases did not start simultaneously. As a result, the internal validity of the findings is compromised as it is less clear whether effects are due to external events (history) rather than to the intervention itself (Tate et al., 2013). Unfortunately, we did not ask participants systematically to mention any external events that could have influenced their symptoms. Despite this, it should be mentioned that randomization and a priori determination of baseline length also enhanced the internal validity. Secondly, we did not assess treatment adherence. Although EMDR is a highly protocolized intervention, we are less certain whether EMDR was conducted according to protocol. Treatment effects may be underestimated if the intervention was not delivered according to the protocol. In addition, the lack of assessment of treatment adherence makes it unclear whether improvements are due to the active ingredients of the intervention, or to any supportive actions by the treating specialists. This limits the internal validity and replicability of the study. Thirdly, we were not able to blind participants and their clinicians, and the researcher who performed analyses was not blind to randomization status. Therefore, internal validity is threatened by potential placebo/nocebo effects and observer bias. Fourthly, randomization was not balanced (i.e. the ratio of participants randomized to 2, 3.5, and 5.5 weeks of baseline was 5:8:8). As a result, there may be an increased risk of history and maturation. Yet, we believe that this is rather a small effect on internal validity, because the baseline lengths were still random and a priori determined. Despite these limitations threatening internal validity, the internal validity in our study is enhanced through replication (Michiels & Onghena, 2019).

There are some other limitations of this study. First, adverse effects were not systematically assessed. Because adverse effects were not included as a predefined outcome measure, no conclusions can be drawn about the safety of the online EMDR intervention in this sample. This is relevant because adverse effects in psychotherapy are increasingly being recognized as underreported, and recent work has highlighted that adverse effects in EMDR research are rarely monitored systematically (van Schie & van Veen, 2025). Although meta-analytic evidence suggests that psychological interventions for adult PTSD are generally safe, future online EMDR studies should include standardized monitoring of adverse effects, such as symptom deterioration, increased distress, dissociation, suicidality, crisis contacts, and dropout due to worsening (Hoppen et al., 2022). Secondly, the adaptation of the PCL-5 to a twice-weekly assessment schedule may have reduced the sensitivity of our outcome. Thirdly, although participants were intended to follow the same protocol of 10 weekly sessions, external circumstances prevented this from being achieved for all of the participants. Yet, this reflects the realities of clinical practice, where treatments differ and may not always proceed as planned owing to various factors. Fourthly, of the 30 individuals who met the inclusion and exclusion criteria, nine participants were not included in the final results owing to dropout or not having provided informed consent. In addition, the ratio of men to women (4:22) and the relatively low age (M = 32.9 years, SD = 12.77) may have limited the generalizability. Finally, although we tried to follow best practices for open science, after registration of our study we added inference criteria for the other visual analysis indices besides IRD, as described by Ledford et al. (2018).

Further research is needed to strengthen the existing evidence. First, this study should be replicated in a different setting/population to generalize evidence in a broader setting. Future studies should also incorporate the assessment of adverse effects of the online EMDR treatment and adherence to the protocol, and other potential factors influencing effects, including patient characteristics as well as therapist characteristics. In general, we advise researchers in future SCEDs to take into account possible threats to their internal validity, to ask participants to mention any possible external events that may have affected their symptoms, and to take into account possible delayed effects. Generating a large database of multiple repeated single-case studies would help in informing when and for whom online EMDR should be initiated. This would enable more tailored advice for individuals regarding suitable treatment options in the future. To test the hypothesis that there is no difference in effectiveness between online EMDR and in-person EMDR for PTSD, a group study, such as a non-inferiority trial comparing online EMDR with in-person EMDR, is recommended.

5. Conclusion

To the best of our knowledge, this is the first controlled study incorporating randomization to investigate the efficacy of online EMDR in the treatment of PTSD. Online EMDR can be particularly useful to increase accessibility and continuity of care in challenging circumstances, such as during a pandemic. Because we adopted a single-case design study, we were able to evaluate effectiveness for individual participants. We observed a reduction in symptoms in most participants following the start of online EMDR, along with a significant and clinically relevant effect over time at the group level, providing evidence for the effectiveness of online EMDR treatment. Controlled replication studies are needed to provide more evidence for the effectiveness of online EMDR treatment.

Disclosure statement

Suzy J.M.A. Matthijssen is part of the advisory board of WeMind, the platform that was used to deliver the online EMDR (https://moovd.nl). The authors report no competing interests to declare.

Data availability statement

The study protocol and its metadata are accessible at OSF (https://osf.io/vx3kj). For reasons of confidentiality the study data are not publicly available, but will be made available upon reasonable request to the corresponding author.

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20008066.2026.2702141.