Post-Hospitalisation COVID-19 Rehabilitation (PHOSP-R): A randomised controlled trial of exercise-based rehabilitation, 2025, Daynes et al

Post-Hospitalisation COVID-19 Rehabilitation (PHOSP-R): A randomised controlled trial of exercise-based rehabilitation

Enya Daynes, Rachael A Evans, Neil J Greening, Nicolette C Bishop, Thomas Yates, Daniel Lozano-Rojas, Kimon Ntotsis, Matthew Richardson, Molly M Baldwin, Malik Hamrouni, Emily Hume, Hamish McAuley, George Mills, Dimitrios Megaritis, Matthew Roberts, Charlotte E Bolton, James D Chalmers, Trudie Chalder, Annemarie B Docherty, Omer Elneima, Ewen M Harrison, Victoria C Harris, Ling p Ho, Alex Horsley, Linzy Houchen-Wolloff, Olivia C Leavy, Michael Marks, Krishna Poinasamy, Jennifer K Quint, Betty Raman, Ruth M Saunders, Aarti Shikotra, Amisha Singapuri, Marco Sereno, Sarah Terry, Louise V Wain, William D-C Man, Carlos Echevarria, Ioannis Vogiatzis, Christopher Brightling, Sally J Singh

Objective
Post-COVID syndrome involves prolonged symptoms with multi-system and functional impairment lasting at least 12 weeks after acute COVID-19. We aimed to determine the efficacy of exercise-based rehabilitation interventions, either face-to-face or remote, compared to usual care in individuals experiencing Post-COVID syndrome following a hospitalisation of acute COVID-19.

Design
This single-blind randomised controlled trial compared two COVID exercise-based rehabilitation interventions (face-to-face or remote) to usual care in participants with Post-COVID syndrome following a hospitalisation. The interventions were either a face-to-face or remote eight-week program of individually prescribed exercise and education. The primary outcome was the change in Incremental Shuttle Walking Test (ISWT) following eight weeks of intervention (either face-to-face or remote) compared to usual care. Other secondary outcomes were measured including health related quality of life (HRQoL), and exploratory outcomes included lymphocyte immunotyping.

Results
181 participants (55% male, mean [sd] age 59 [12] years, length of hospital stay 12 [19] days) were randomised. There was an improvement in the ISWT distance following face-to-face rehabilitation (mean 52 [95% CI 19 to 85]m, p=0·002) and remote rehabilitation (mean 34 [95% CI 1 to 66]m, p=0·047) compared to usual care alone. There were no differences between groups for HRQoL of self-reported symptoms. Analysis of immune markers revealed significant increases in naïve and memory CD8+ T cells following face-to-face rehabilitation versus usual care alone (p<0·001, n=31).

Conclusion
Exercise-based rehabilitation improved short-term exercise capacity in Post-COVID syndrome following an acute hospitalisation and showed potential for beneficial immunomodulatory effects.

Link | PDF (European Respiratory Journal) [Open Access]

 

Individuals were excluded if they: had any contraindication to exercise22; experienced symptoms indicative of another medical condition that required further investigation/management (i.e. clinical diagnosis or self-reported severe post-exertional malaise (PEM)/ symptom exacerbation (PESE) rendering the individual bedbound or postural orthostatic tachycardia syndrome); unstable comorbidities; or completion of a rehabilitation programme in the preceding six months.


Usual care
Usual care was offered to all groups and ensured participants could access any treatment that was offered in the management of their post-COVID syndrome. All participants were seen in a specialist outpatient COVID clinic hosted within secondary care, led by a multidisciplinary team of consultants (respiratory, cardiology, neurology, diabetologist, renal and general practitioner), nurses and physiotherapists.

Participants had their care optimised prior to enrolment in the trial through the COVID clinic, frequency of appointments is tailored to the individual’s needs. At the time of this trial conduct, treatment was not standardised but was offered based on an individual and thorough assessment, this could include, but is not limited to, psychological interventions, medical treatment, symptom self-management or breathing. As rehabilitation was not considered usual care at trial set up, participants were excluded from engaging in exercised-based rehabilitation during the trial period. The control group for this study was usual care alone.

 

The face-to-face rehabilitation group improved from 285 [219 to 351] m to 312 [244 to 380] m (p<0·001). The
unadjusted analysis demonstrated a statistically significant difference of 55 [19 to 92] m in favour of the intervention (p<0·001).

The number (%) of participants achieving more than a 35m improvement was 25(56%), 22(50%) and 11(19%) for the face-to-face, remote, and usual care groups respectively.

My comment:
Average distance increased by 27 meters within the treatment group. That’s less than 10 % and hardly clinically significant. For reference, I went from 400 to 600 meters in the 6 min walking test (not the same test) after 5 weeks of rehab 6 months into LC. I’m bedbound with ME/CFS two and a half year later. Edit: I had PEM since day one.

 

It should definitely have been mentioned in the abstract that this is only about non-ME/CFS long COVID, considering how many people with LC have ME/CFS, and how many people will only read the abstract.

 

Hospitalized-Non-PEM/POTS-LC patients, much broader than ME/CFS.

 

The design seems decent. Their conclusions are terrible.

Data only available at reasonable requests, and you have to document ‘relevant’ training! These guys are full of themselves..

 

The walking test is described here.

The source the guide uses, states that:
The minimum clinically important improvement for the ISWT is 47.5 m. In addition, patients were able to distinguish an additional benefit at 78.7 m.

For context, the mean distance was 168.5 m. The increase was 28 % of the baseline.

In this study, the mean increase within the treatment group was 27 m (or 10 % of their baseline). So the results are not clinically significant.

 

Some patients with post-COVID syndrome experience PEM/PESE which can present a challenge. We excluded those under active investigation for, or a diagnosis of severe and debilitating PEM (n=1) resulting in inability to leave the house as this intervention was not deemed appropriate for this patient group. We used extensive PESE monitoring and screening in those included in the trial to ensure symptoms were not worsened by the intervention in any participants6. Measures of fatigue and symptoms improved across all three groups suggesting that the rehabilitation exercise was not harmful in appropriately selected patients The presence of severe PEM/PESE is likely higher in patients under the care of COVID clinics than reported in this trial, which is potentially the result of careful participant selection and identification.

 

Consistent with our primary outcome we found potential improvements in several other physical outcomes above the established MCID when compared to usual care25,26 . Despite this, we did not detect improvements in self-reported health-related quality of life above the usual care group. This study did not demonstrate improvements in the EQ5D utility index between groups. Given the broad range of symptoms in post-COVID syndrome, it is plausible that many participants meet the floor or ceiling of the symptom specific outcome measures, and that generic health-related quality of life tools are insensitive, and therefore specific Post-COVID syndrome health-related quality of life outcome measures maybe more sensitive and are now available27,28 . The REGAIN trial has demonstrated improvements in quality of life, measured by PROMIS-PROPr score, through a synchronised remote intervention in the absence of a measure of exercise capacity/physical function17

 

«The sample size was based on a mean difference between groups of 50m in the primary outcome (ISWT), with a SD of 72m taken from previous cohort studies16. The study was powered at 90% with a type I error of two sided alpha 0.05 and required 44 participants per group, 132 participants in total. The sample size was inflated by 20% to account for attrition to give a final target sample size of 159 participants21»

«140/181 participants were included in the per-protocol analysis, 40/56 (71%) face-to-face, 38/62 (61%) remote, 60/62 (98%) usual care completing 75% of the intervention and the follow-up measures.»

29 % and 39 % drop out from the two treatment groups. 2 % in the CAU group. The treatment groups were smaller than they planned for. I don’t know how to check for that in their analysis.

 

I believe they changed the threshold for clinical significance for the primary outcome measurment!

In the protocol, they state the following:

«The primary objective of this study is to determine whether rehabilitation (either face-to-face or digital) added to usual care increases physical function compared to usual care alone in individuals with ongoing symptoms following COVID-19 hospitalisation.
(…)
The sample size is calculated on the ISWT (primary outcome) with a change of 50 m at 90% power, with a standard deviation of 72 m and a 0.05 type 1 error as previously documented in the literature as the minimum important difference and variance of the ISWT [14, 37].»

37 is this article is the one I mentioned here:

In the paper, they say this:
«In this fully powered randomised controlled trial, we demonstrated that both face-to-face and remote exercise-based rehabilitation significantly improve exercise capacity compared to usual care alone in those previously hospitalised with COVID-19. These between group improvements exceed the established MCID (35m)24 , highlighting improvements of clinical relevance in those with post-COVID syndrome.»

24 is this study om patients with COPD:
«Conclusion The minimum important improvement in the ISWT is 35.5 m using the distributive method.»

This change is significant because they use the 35 m here:
«The number (%) of participants achieving more than a 35m improvement was 25(56%), 22(50%) and 11(19%) for the face-to-face, remote, and usual care groups respectively.»

I’m assuming the numbers would have been worse if they used a higher threshold of >47 m.

Furthermore, the remote rehab group did not achive clinical significance regardless of which threshold they used. And the 95 % CI for the face-to-face crosses the threshold:
«There was an improvement in the ISWT distance following face-to-face rehabilitation (mean 52 [95% CI 19 to 85]m, p=0·002) and remote rehabilitation (mean 34 [95% CI 1 to 66]m, p=0·047) compared to usual care alone.»

 

It's really silly how the quacks are holding on to the distinction over PEM being relevant, when the reality is that it doesn't matter one bit. None of this actually works for anything. It has the exact same efficacy as homeopathy or acupuncture, regardless of what the problem is. PACE made that just as clear, as did many other trials. None of this rehab junk works for this type of health problems.

But somehow they will waste many years on this crap argument. Because wasting time is all they ever had. As long as they prevent significant work from invalidating their stuff, they can simply occupy the space and practice their own version of the tragedy of the commons, content with doing laps around the land they block, never getting anywhere.

 

What does "potential improvements" even mean here? There is no such thing. Either it improves, or it doesn't. Have to go with Yoda here: No. Try not. Do... or do not. There is no try.

What they mean of course is something borderline, the needle slightly moving erratically and sometimes getting close to the absolute minimum level of a blip that they can detect. This corrupt industry has been pretending for decades that this means something, that their putting fingers on scales and making the needle wiggle slightly is significant, they just have to figure out how to make it happen for real.

But always the minimum. Literally always aiming for the bare minimum, and only ever getting close when they manually extend the target as needed. No one would employ freeloaders like this in private industries.

 

Yes, they would. They are good at supporting a narrative, that’s a highly valued trait for many, regardless of the sector.

 

@dave30th has written the latest Trial by Error piece on the Virology Blog on this study:

Trial By Error: Trudie Chalder Is Co-Author on Another Bad Exercise Paper
see https://virology.ws/2025/02/28/trial-by-error-trudie-chalder-is-co-author-on-another-bad-exercise-paper/

[Mod Edit: remove the Facebook referrer/tracker at the end of the link.]

 

Apologies for the mess of cross-references, I’m struggling with formating in my phone!

@dave30th :

I believe there’s an inaccuracy here in the blog. If the face-to-face group only improved by 27 meters, but they improved by 52 meters compared to usual care, the usual care group must have had a decrease of 25 meters compared to their baseline. ‘More than the increase’ should have been ‘more than the decrease’.

I don’t believe they actually mentioned this in the paper, but I might have overlooked it. I can’t find the data for the ISWT, only the mean - is it included in the paper?

Additionally, the study with the MCID for ISWT was on patients with COPD, but that isn’t mentioned in the blog.

@dave30th

I also believe that they swapped to the more favourable MCID for COPD after they had published their protocol, as per my post above. I’m not 100 % certain on this point, though, I might have misunderstood something somewhere.

 

As I understand it, it's used for pulmonary rehabilitation and respiratory status studies. I don't know if the MCID would depend on the specific diagnosis, but maybe so. The other MCID studies were for related conditions. They obviously picked the most favorable. I don't know if it's different than in the protocol, but perhaps using an updated MCID study would be an acceptable change. How they got even that one wrong in presenting their data, however, is a mystery.

 

good catch! you're right, they don't seem to mention the results in this draft of the usual care group, and it's not included in the table. I assume perhaps when the published-published verison comes out, that info will be included. And my phrasing, given that, was inaccurate, so I fixed it.

But I don't understand the statistics here. In the face-to-face group, the absolute improvement was 27 but that turned into a difference of 52 from care as usual. But in the remote group, the absolute improvement was 35 meters, and that turned into a difference of 34 meters from usual care.

 

@dave30th is it acceptable to not mention the change from the protocol, even if the change was warranted?

Huh, I didn’t think to do the calculations for the remote group. That a very weird result.

The p-value is very close to 0.05 for the relative change for the remote group. I’m not saying that has to mean anything..

Are anyone able to get their hands on the data?