This study demonstrated the feasibility of an online self-directed psychosocial program for women with MBC, with high uptake rates, acceptable retention, and positive qualitative feedback. While the recruitment target was not met, this pilot has generated estimates for future recruitment timeframes. Although early indicators are promising that FMW-A may improve cancer-specific distress, fear of progression, global QOL, and role functioning, this must be balanced against the mixed findings arising in other quality-of-life domains, particularly mental QOL, and the increases in general distress. Given the small sample size, the potential impact of baseline distress on program engagement and subsequent escalations of some symptoms, revisions to the program, and replication of findings are warranted via a larger study before efficacy conclusions are drawn.

Preliminary indicators of feasibility were obtained: uptake was high in the enrolment period at 63.6% — notably 10% higher than the documented pooled uptake rate of MBC intervention RCTs [8] and 20% higher than the average uptake rate for group therapy RCTs. This uptake rate is similar to individual therapy RCTs for MBC [8]; while one might have predicted uptake to be superior in overcoming accessibility barriers, these modalities reach arguably different populations with different needs and preferences [38]; thus, comparable uptake rates are a positive outcome. While the uptake rate was nearly identical for those approached via clinicians vs social media, the recruitment rate per month varied dramatically (1 per month via clinicians vs 12 per month via social media). This was consistent with other recent trials utilising social media advertising to augment recruitment efforts for their digital psycho-oncology trials [39, 40] and highlights the importance of this recruitment avenue. Limitingly, those who were recruited via social media self-screened for eligibility as part of sign-up. Future research may consider obtaining consent from this recruitment group to contact their treatment team to confirm eligibility and medical information.

While uptake was high, twenty eligible individuals declined to participate. In many instances, reasons for declining were unable to be documented but commonly cited reasons for declining related to already coping well and/or having adequate support. Potential other reasons might relate to those arising among participants with lower engagement levels — e.g. higher distress and not wishing to be exposed to content. Alternatively, people with advanced cancer experience time toxicity (i.e. losing time as a result of cancer treatments), causing them to prioritise how they spend their time) [41].

Retention indicators of overall trial feasibility met acceptability benchmarks (attrition rate of 31.4% post-intervention) and were identical to the pooled attrition rate documented in a systematic review of MBC RCTs [8]. This was further demonstrated via qualitative feedback, where participants found the program acceptable, practical, and in demand. Due to the accessibility of the program during Covid-19, there may have been artificially increased demand for the intervention at the time of this trial, though it could equally be argued that people were ‘zoomed out’ during the pandemic and were less inclined towards digital modalities than at other timepoints. Participants also provided positive feedback about the program serving as a prompt to access more intensive services, consistent with findings of increased supportive care service utilisation in the FMW trial [6]. However, it should be noted that limited conclusions can be made regarding the qualitative feedback within the current study, due to the small sample size of self-selected participants who provided this feedback that are not necessarily a representative sample of the MBC population. Furthermore, retention after randomisation differed across conditions (77.8% for control versus 58.9% for intervention). Thus, while the overall trial’s retention rate met a priori feasibility benchmarks, the intervention group’s rate was 10% below and thus sub-optimal and may reflect the increased time commitment, higher distress experienced, and higher digital health literacy required. However, qualitative data indicated that the program was not viewed as too large of a time commitment. Although, this may be biased, given participants are self-selected for the interview and are likely to be high program engagers. Alternatively, while participants were provided with a PDF tutorial for how to use the program, this may not have been sufficient for those with low computer literacy skills, as indicated to be a potential barrier in the qualitative feedback.

Contrary to the literature supporting that low-intensity interventions have higher adherence than face-to-face psychological interventions [8, 42], programme engagement in the current study was lower than expected at a median of 3/6 available modules; on average completing one module less than the original FMW [6]. While the a priori benchmark for engagement was not met, there is limited evidence to suggest what ‘dose’ of digital self-guided interventions is required to derive benefit [43]. Interestingly, the bi-modal pattern of program engagement noted in this study (see Fig. 3), whereby participants tended to complete either very little or nearly all of the program, is common in digital interventions for cancer populations [20, 44], however, seemingly amplified in the metastatic setting. Indeed, these findings may reflect the individual needs of participants, given they were informed to use the program as much or as little as they like and to choose modules, or content within modules, that was relevant to them. Alternatively, this may reflect the nature of cancer generally, given this is a typical finding in digital mental health [2, 6], or reflect the nature of MBC specifically, given the program is targeting those with a chronic condition, whose illness demands fluctuate Future research may consider removing the tunnelling/sequential release of modules, to enable immediate full access. A common theme raised in the qualitative analyses was participants’ preference to dip in and out of modules. This fits the adult learning framework proposed by Cornelius and Gordon, whereby there are three types of adult learners¾ universalists (who study all available materials), butterflies (who dip in and out of materials), and changelings (who change their study approach over time) [45]. Thus, the participants in this study may fall into the category of butterflies. While this pattern of engagement results in lower overall usage rates, immediate access to all modules may increase usage rates and cater to all three categories of learners. While outside the scope of the current feasibility study, future research may include a longer follow-up period to accommodate people who may have an increasing number of clinical events and needs for which they might seek information via the program.

Furthermore, while digital intervention engagement is considered important for intervention effectiveness in the broad adult population [46], there is mixed research regarding the association between increased engagement and improved outcomes within the curatively treated cancer [44] and survivorship populations [47,48,49]. Therefore, it is critical to understand how those in the advanced/metastatic setting engage with digital interventions and how this engagement is related to psychological outcomes. Future research may (a) aim to increase engagement of those in the advanced/metastatic setting to determine its impact on the efficacy of digital interventions, (b) explore engagement correlates (e.g. demographic characteristics, baseline outcomes), (c) qualitatively evaluate engagement further with a larger sample of participants (i.e. facilitators, barriers, self-reported engagement versus objective indicators of engagement), or (d) modify the intervention to suit participants current level of engagement, through the development of a micro intervention. Some promising signals of efficacy were obtained. While cancer-specific distress, fear of progression, global QOL, and role functioning did not change across time among control participants, FMW-A participants mean scores showed trends towards improving. The changes in fear of progression in particular for FMW-A participants demonstrated an overall shift from the ‘high’ to the ‘moderate’ range, according to Sarkar and colleagues’ cut-off score of 34 [50]. However, the less promising findings for FMW-A participants in unmet needs, general distress, and mental and social QOL from baseline to post-intervention were contrary to expectations and potentially concerning. These aligned with qualitative findings, where one participant discussed how some information in FMW-A may cause emotional distress or discomfort. However, a body of literature shows that exposure to potentially emotional stimuli or situations previously avoided can lead to short-term increases in distress, particularly among those more vulnerable to distress, prior to habituation occurring [51,52,53]. Given that distress is negatively associated with QOL in the breast cancer population, the increased general distress of FMW-A participants may be explained by these exposure principles and explain the decreased QOL scores in turn. In addition, the efficacy of the intervention may have been impacted by the lower engagement observed. Future studies need to include a follow-up period longer than 6 weeks to determine if general distress habituates, and as a result, QOL starts to increase. Conversely, this finding could reflect the impact of (dis)engagement and ongoing avoidance of content.

That said, to address the potential that distress was caused by the program, a number of changes could be made for future testing: removing module-tunnelling and unlocking all modules at once (enabling freedom of choice regarding which modules to access), introducing an ‘accordion’ function that opens up further information within modules when clicked (to enable freedom of choice regarding which information within a module participants access), improved worksheet functionality and consequently improved user experience, and the separation of research surveys from the therapeutic program to clearly delineate content and reduce overwhelm. In addition, further studies are underway regarding the addition of human support to the program, aiming to increase engagement, and consequently, outcomes.

Another solution is a modification of FMW-A into a blended format, where online-based therapies are used in adjunct to face-to-face therapy or as part of a stepped care model, whereby baseline distress is screened prior to participation to determine the suitability of participants for a self-directed protocol. Indeed, systematic reviews in the mental health population support that online interventions with guidance are more effective in improving engagement and psychological outcomes, such as anxiety and depression [54, 55]. A recent systematic review found that health professionals preferred blended therapy over online only [56], particularly for those with more severe psychological distress who may be at higher risk of dropout and may benefit from the therapeutic rapport and more active monitoring. Previous blended therapy trials in cancer populations have shown effectiveness in improving fear of cancer recurrence [57] and psychological distress [58]. However, these have focused only on those in post-treatment survivorship rather than metastatic populations and have not measured the impact on QOL. FMW-A could provide important data for a blended model.

In conclusion, while this study was limited by a small sample size, this is the first randomised trial of an online self-directed intervention for women with MBC, demonstrating the feasibility and acceptability of the program and avenues for improvement. Preliminary signals of efficacy were found but must be balanced against the mixed findings and potential for escalations in distress among those with high baseline symptoms who are less likely to engage with the program. Further research is warranted to revise the program and replicate findings in a larger study and to explore avenues for improving program engagement.