Clinical effectiveness of an online supervised group physical and mental health rehabilitation programme for adults with post-covid-19 condition (REGAIN study): multicentre randomised controlled trial

Trial design and setting

The REGAIN trial was a pragmatic, multicentre, parallel group, superiority randomised controlled trial with embedded process evaluation, recruiting throughout England and Wales. Each participant identification centre was granted NHS Trust site specific approval. The NHS Digital “Digi-Trials” service was approved to identify and invite potential participants in accordance with Regulation 3(4) of the Health Service (Control of Patient Information, COPI) Regulations 2002, requiring NHS Digital to share confidential patient information with organisations entitled to process this under COPI for the purposes of covid-19 research.36 The trial protocol and details of the intervention’s development have been published previously78 and are reported in accordance with the template for intervention description and replication (TIDieR)9 (see supplementary material).

Participants and procedures

Participants were adults (26-86 years) who had been discharged from hospital three or more months previously after hospital admission with covid-19 and who had ongoing substantial (as defined by participants) covid-19 related physical and/or mental health sequelae. In the absence of agreed diagnostic criteria or clinical coding for post-covid-19 condition, participants were asked to self-report any substantial lasting effects that they attributed to their hospital admission with covid-19. This was confirmed during an eligibility telephone call with the clinical trial team before study enrolment. Exclusion criteria were contraindication to exercise training; severe mental health problems preventing engagement with study procedures; previous randomisation in the present study; already engaged, or planning to engage, in an alternative NHS rehabilitation programme in the next three months; or a household member had been randomised into the REGAIN trial previously.

Randomisation and masking

After completion of the online consent and baseline questionnaires (to ensure allocation concealment), participants were randomly allocated (1:1.03) to the REGAIN intervention or to usual care by a centralised computer generated randomisation sequence using a bespoke web based system, administered independently by Warwick Clinical Trials Unit. We used a minimisation algorithm, stratified by age (<65 years v ≥65 years), level of hospital care (intensive care unit (ICU) or high dependency unit (HDU) v ward), and case level mental health symptomatology (impact of event scale-6 (IES-6) post-traumatic stress disorder score ≥11/24, or hospital anxiety and depression scale (HADS) anxiety subscore ≥11/21, or HADS depression subscore ≥11/21; compared with IES-6 post-traumatic stress disorder score <11/24, or HADS anxiety subscore <11/21, or HADS depression subscore <11/21) (see supplementary material). Participants and practitioners delivering REGAIN could not be masked to group allocation. Follow-up outcome assessments were completed by participants online, or, in a small number of cases, over the telephone by a member of the trial team, blind to treatment allocation.

Procedures

Potential participants were contacted by post, either locally through secondary care NHS Trusts or, for England and Wales, through an NHS Digital “Digi-Trials” mailout. Self-referral to the trial was also possible. Those with persistent physical or mental health sequelae, or both (estimated at 10% of people discharged from hospital after covid-19)10 were invited to register their interest by completing a brief online questionnaire for suitability. On confirmation of suitability, the clinical research team contacted participants by telephone to complete further eligibility checks. For those who self-referred, eligibility was confirmed through their general practitioner. Participants then completed an online consent form and baseline outcomes questionnaire before randomisation. We informed general practitioners in writing of participants whose baseline (or follow-up) scores met any of our predefined criteria for case level mental health (see supplementary material). These people continued in the trial. Trial and intervention materials were translated into the five most spoken non-English languages in the UK (Bengali, Gujarati, Urdu, Punjabi, and Mandarin) and a non-English speaking pathway developed to allow access to the trial.

Clinical exercise physiologists or physiotherapists trained in the REGAIN intervention and supported by health psychologists delivered interventions exclusively online from a central trial hub. The hub was based at Atrium Health, a non-profit rehabilitation centre, subcontracted to University Hospitals of Coventry and Warwickshire NHS Trust. Intervention staff included NHS and Atrium Health employees, with some delivering both intervention and usual care treatments. Interventions were informed by a rapid review of existing literature relating to rehabilitation programmes for people affected by chronic obstructive pulmonary disease, chronic fatigue syndrome, and the 2003 severe acute respiratory distress syndrome pandemic. Details on co-development of the intervention are provided elsewhere.8

Interventions

Participants in the usual care group received best practice usual care, consisting of a 30 minute, online, one-to-one consultation with a trained practitioner. A trial booklet was provided that incorporated components of the NHS England “your covid recovery” programme,11 information and advice that is freely available online. During the one-to-one consultation, practitioners used the NHS England your covid recovery programme as a template to discuss hospital admission with covid-19, resulting physical and mental health sequalae, and other relevant medical history with participants. The consultation covered generic advice as to how participants might facilitate recovery and undertake self-directed physical activity. A structured physical activity plan was not provided, and no specific psychological techniques were used.

The REGAIN intervention comprised an eight week, online, home based, supervised, group rehabilitation programme (see supplementary figure S1), supported by a workbook for participants (https://wrap.warwick.ac.uk). Participants received a one hour, online, one-to-one consultation with a REGAIN practitioner, which provided an opportunity to discuss hospital admission with covid-19 and sequelae, medical history, and practical ways in which physical and mental health recovery could be supported. Participants subsequently enrolled on weekly practitioner led live (ie, synchronous) online group exercise sessions and six live online group psychological support sessions (one hour each) delivered through Zoom using the Beam platform (https://www.beamfeelgood.com). Individualised, equipment-free exercise sessions performed at home in online groups under the supervision of a REGAIN practitioner aimed to improve cardiovascular fitness, strength, balance, and fatigue while restoring confidence in completing activities of daily living. Semistructured facilitated psychological support sessions were designed to enhance psychological capability and increase knowledge and understanding of covid-19 and its impact on daily living, while giving participants the opportunity to share their own experiences with the group (≤12 participants). Topics of discussion, supported by short introductory videos, included motivation, fear avoidance, activity pacing, managing emotions and set-backs, sleep and fatigue, and stress and anxiety management. Finally, a library of prerecorded, on-demand physical activity videos was made available for participants to access independently online as required. Sessions ranged in duration and intensity from simple breathing exercises, Pilates, and yoga to light seated activity and upright moderate to high intensity exercise.

Outcomes

Outcomes were measured at three, six, and 12 months. The primary outcome was health related quality of life using the patient reported outcomes measurement information system (PROMIS) 29+2 Profile v2.1 at three months post-randomisation.12 This measure is one of a portfolio of outcomes from the US National Institutes of Health. The system is reliable, generic, and validated for online use, generating a single overall preference based score (absolute score rather than effect size)—the PROMIS preference (PROPr) score (range −0.022 to 1.0, where 0 indicates a health state equivalent to death and 1.0 indicates perfect health).12 The overall score is generated from seven subscores for depression, fatigue, sleep disturbance, pain interference, physical function, social roles or activities, and cognitive function. As with other preference based measures such as the EuroQol 5 dimension 5 level (EQ-5D-5L) instrument, a difference of 0.03 to 0.05 is considered to be clinically important.13

For analysis purposes, we rescaled raw data from the seven PROMIS subscores to standardised T scores with a mean of 50 and a standard deviation (SD) of 10. Therefore, a person with a T score of 40 is 1 SD below the mean. Higher T scores represent more of the concept being measured. For negatively worded concepts such as pain interference, a T score of 60 is 1 SD worse than the mean, whereas for positively worded concepts such as physical function, a T score of 60 is 1 SD better than the mean. For anxiety, depression, fatigue, pain interference, and sleep disturbance, higher scores indicate more severe symptoms. For physical function and social participation, higher scores indicate better health outcomes. A change in T score of between 2.0 and 6.0 is considered clinically important in the PROMIS subscores and subscales.14 A further two PROMIS subscales independently measured anxiety and pain intensity. All subscores and subscales were rated over the preceding seven days, apart from physical function and social roles or activities, which do not have a specified timeframe.

Our secondary outcomes were dyspnoea (PROMIS dyspnoea severity short form), cognitive function (PROMIS Neuro-QoL short-form v2.0-cognitive function), quality of life (EQ-5D-5L),15 physical activity (international physical activity questionnaire short-form, IPAQ),16 severity of post-traumatic stress disorder (impact of events scale-revised, IES-R),17 anxiety and depression (hospital anxiety and depression scale, HADS),18 general health (self-report of current overall health compared with baseline), and mortality.

Adverse events and serious adverse events were recorded in both trial arms, in line with the principles of good clinical practice.19 Additionally, participants in the intervention group were asked to report any events before each exercise session through a confidential, secure online poll. The trial team routinely reviewed responses and contacted participants for further information as required. As the presentation of post-covid-19 condition and chronic fatigue syndrome/myalgic encephalomyelitis overlaps,3 we prospectively monitored for post-exertional symptom exacerbation20 in the intervention arm during each contact of participants with the intervention team.

Full adherence to the REGAIN intervention was defined as attendance at the initial one-to-one session along with completion of four or more of six support sessions and five or more of eight exercise sessions. Partial adherence was defined as attendance at the initial one-to-one session and completion of fewer than four of six support sessions and fewer than five of eight exercise sessions. To assess fidelity to the intervention, we reviewed a randomly selected 5% subsample of video recorded one-to-one sessions and group exercise and support sessions against predetermined checklist criteria. Results of this and the rest of our process evaluation will be reported elsewhere.

Sample size

The sample size was calculated based on identifying a small to moderate standardised mean difference of 0.3. No data exist on which to base a sample size estimation, nor normative data for the PROPr health related quality of life score in a covid-19 population. Also, there is no indication of what a worthwhile benefit might be from the intervention. We inflated the size of the intervention group to compensate for any clustering effect owing to the delivery of the group intervention. We assumed, based on our experience with other rehabilitation interventions, that groups would comprise a maximum of eight participants. Assuming an intracluster coefficient of 0.01, 90% power, and type I error rate of 5%, with a 10% loss to follow-up, we determined that 535 participants would be required. This equated to 272 participants in the intervention arm (up to 34 intervention groups) and 263 participants in the usual care arm (allocation intervention to usual care ratio of 1.03:1).21 To compensate for the slightly higher than anticipated loss to follow-up in the observed data, we recruited a total of 585 participants.

Statistical analysis

Statistical analysis followed a predefined plan.22 Our primary analysis was done on an intention-to-treat basis, which included all participants randomly assigned to a treatment group. For the primary outcome (PROPr score) we performed a partially nested heteroscedastic model23 to compare health related quality of life at three months between the REGAIN intervention group and usual care group, producing unadjusted and adjusted estimates. Adjustments were made using age, level of hospital care, level of mental health disorder, baseline PROPr score, and therapist effect as a random effect. Deaths were included with a score of zero. The only ordinal categorical outcome was the overall health score. This outcome was fitted using linear regression models (for unadjusted and adjusted variables). We checked normality assumptions and used the Mann-Whitney test to test the treatment effect (unadjusted). To accommodate for non-adherence, we did a complier average causal effect analysis based on a single equation instrumental variable regression model. The complier average causal effect estimates the treatment effect in people randomly assigned to the intervention who actually received it by comparing participants who fully or partially adhered in the intervention group with participants in the control group who would have been classed as adherent had they been allocated to the intervention group.

We performed unadjusted analyses of subgroups defined according to age (<65 years v ≥65 years), level of hospital care (critical care v standard ward), HADS depression and HADS anxiety score (<11 v ≥11), severity of post-traumatic stress disorder (impact of events scale-6 (six item subscale of the impact of events scale-6-revised) score (<11 v ≥11), ethnicity (white v non-white), wave of pandemic (first, second, third, or fourth), and method of recruitment (NHS Digital v NHS Trusts or self-referral). In a sensitivity analysis, we used the multiple imputation by chained equations procedure,24 imputing the primary outcome. To aid interpretation, we report the number needed to treat based on the responses on the global health transition question. These responses are presented as a number needed to treat to be “much better” and at least “somewhat better.” All analyses were conducted using Stata version 17 and R version 4.

Study monitoring

The data monitoring and trial steering committees reviewed the progress of the trial and safety periodically (see supplementary material).

Patient and public involvement

The concept for the trial and grant funding application was driven by our patient partner working group early in the covid-19 pandemic. Patient representatives were involved as co-applicants in the grant funding application. On receipt of the award, our patient and stakeholder working group were integral to the rapid design, co-creation, and pilot testing of the REGAIN intervention and trial processes.8 Subsequently, patient partners participated as members of the trial management group and trial steering committee.