Although not formally powered to test for change, the feasibility data indicate larger reductions in CFQ-11 in the intervention than the wait-list group, and the wait-list group when using the intervention experienced similar reductions in their scores (Table 3).

There were indications of benefit in a patient group with often intractable symptoms. We provide preliminary evidence that warrants further evaluation of this intervention. Based on this feasibility study, we believe that the self-help intervention in isolation could help support a reduction of fatigue in some people.

RCTs are usually costly to run and failure to recruit is often a barrier. Collaboration with Research for the Future (RfTF) helped us to meet and achieve our two-stage recruitment targets, with 82% stage 2 participants recruited via this route.

The technique employed is an established manual therapeutic approach based on several tenets of osteopathic manipulative medicine such as improving the spinal biomechanics to reduce dysautonomia and to aid blood flow and lymphatic drainage of muscles and the central nervous system through manual soft tissue, cranial and spinal manipulative techniques [7,8,9,10,11].

The remote methodology, with centralized research support, enabled the study to run in the highly constrained research environment post-COVID-19 UK lockdown. This also spared people with fatigue from further attending clinic appointments and reduced the risk of infection transmission.

Adopting a wait-list methodology meant that all participants who consented to participate would at some point either immediately or after 12 weeks receive the self-help intervention. This may have helped drive study recruitment, as it avoided half the participants from receiving ‘treatment as usual’. All participants had baseline data collected prior to randomization, reducing any bias in reporting, once allocation was known.

Importantly, the groups were balanced in terms of gender and recruitment route, which were controlled for within the randomization process.

Based on participant feedback, we updated the self-help component. Changes included providing a full rationale for using the technique and re-recording the online video guidance with indexed search functionality for easier access to specific areas of support. We recognize the need to convey this to users to motivate them to perform the self-help aspect. Future work will utilize this updated guidance.

The 3-month and 6-month follow-up dates were calculated by adding 90 and 180 days to the consent date; this allowed for some of the delays in utilizing the pack that occurred due to (i) difference between baseline data collection and randomization date (9 days) plus (ii) time for participants to receive the intervention pack (3 days; 1 day for collection of pack from post room and 2 days for Royal Mail delivery. This brings the time that participants can experience use of the intervention closer to the 12 weeks (84 days) intended.

Between 0 and 3 months, retention rates in the study were good, with 79% being on the border between Amber (50–79%) and Green (80–100%) in the Red Amber Green (RAG) ratings often seen in feasibility study protocols [20]. These are purely arbitrary boundaries that were not pre-published for this study per se, but represent when modifications should be made to improve study retention.

It is unusual to see a higher follow-up response rates in ‘control’ compared to ‘intervention’ arms, 88.2% vs. 69.4%, respectively. The reason for this is likely to be twofold: firstly, completing the follow-up survey for those in the wait-list group triggered dispatch of their equipment packs, allowing them to begin using the intervention; secondly, those in the intervention group who had not adhered to the intervention may not have wanted to provide follow-up data.

Interestingly, at 6 months, there was a reverse in this trend, with a larger proportion of non-completers in the wait-list arm than the initial intervention arm, resulting in similar 6-month follow-up rates overall, 61.2% vs. 56.9%, respectively.

Our results suggest that there was an interest within the long COVID community for this type of intervention to be delivered remotely. It remains to be seen as to whether the decline in COVID infection and its potential decrease in severity will lead to a reduction in the number of new long COVID cases arising and whether a similar recruitment could be achieved in future studies.

Further research is warranted, powered to test the effect size of the intervention for an efficacy RCT. Using baseline data from all 100 participants in the current feasibility study, an estimate of the upper 80% confidence interval for the standard deviation of the CFQ is 5.5 units. Assuming an attrition rate of 30% between baseline and the primary endpoint (at 3 months), we calculate that 230 participants would be required in each group (460 in total) to detect a two-point reduction in CFQ scores with 90% power. This could either be as a stand-alone intervention or in conjunction with delivered in-person sessions that focus on lymphatic drainage though specific massage and cranial osteopathic techniques. Introducing a third arm would increase the total sample size to 690. There would also need to be an adjustment for making multiple comparisons.

For individuals who are housebound or not able to attend the hospital because of mobility issues, this is an ideal approach, and so fills a gap in existing care provision.

Our team currently believes that many of those with long COVID after 12–18 months will develop ‘long’ long COVID and are likely to be diagnosed with ME/CFS in the future. This is similar to the progression seen following glandular fever, to post Epstein–Barr Virus (EBV) syndrome, then to ME [21].

Spontaneous remission in long COVID is higher than in ME [22] (evidenced by three participants (5.9%) returning intervention packs after 3 months in the wait-list arm of the trial). Future trials could address this by considering the inclusion of two types of patients, namely participants with long COVID and those with diagnosed ME/CFS as a distinct other group.

A point to take into account is that the wait-list control group was given access to the intervention after 12 weeks, providing an opportunity to experience the same use of the intervention, although some authors suggest that this might contribute to a nocebo effect where waiting to access a treatment limits participants’ self-healing behaviors [23].

As a feasibility RCT, we aimed to compare recruitment and retention of those using a self-help intervention to reduce long COVID-related fatigue. While the two-stage approach worked remotely, it is uncertain if the same enthusiasm for the study and its methodology would remain now that there are no further COVID lockdowns.

The wait-list strategy was employed so as to give all participants access to the treatment. This is in keeping with good osteopathic research and medical research practice. The higher attrition rate in the intervention group is thought to be related to the time required each day to undertake the self-help intervention as supported by the participant interviews.

We accept that a wait-list control has the potential for exaggerated effect sizes/bias in intervention effect estimates and difficulties in attributing improvements to the intervention itself (vs. natural recovery). However, this was not the primary focus of our study. Participants in the wait-list group were happy enough to wait their turn for the intervention, but there was an obvious drop in numbers in the intervention group after the 3-month follow-up. We cannot say whether this was simply fatigue or whether it was due to the intervention itself.

It is possible that many of those who were lost to follow-up either did not comply with the intervention or experienced side effects. Without carrying out qualitative interviews with all participants, this is impossible to ascertain.