Trial Report Immediate and Long-Term Effects of Hyperbaric Oxygenation in Patients with Long COVID Syndrome, 2023, Lindenmann et al

[Hutan]

Immediate and Long-Term Effects of Hyperbaric Oxygenation in Patients with Long COVID-19 Syndrome Using SF-36 Survey and VAS Score: A Clinical Pilot Study

Spoiler: Authors

Lindenmann, Joerg; Porubsky, Christian; Okresa, Lucija; Klemen, Huberta; Mykoliuk, Iurii; Roj, Andrej; Koutp, Amir; Kink, Eveline; Iberer, Florian; Kovacs, Gabor; Krause, Robert; Smolle, Josef; Smolle-Juettner, Freyja Maria

Abstract
(1) Background: Long COVID syndrome (LCS) is a heterogeneous long-standing condition following COVID-19 infection. Treatment options are limited to symptomatic measures, and no specific medication has been established. Hyperbaric oxygenation (HBO) has been found to have a positive impact on the treatment of COVID-19 infection. This study evaluates both the feasibility and outcome of supportive HBO in patients with LCS.

(2) Methods: Within 17 months, 70 patients with proven LCS were prospectively included. Each patient underwent a cycle of 10 subsequent HBO treatment sessions administered for 75 min at 2.2 atmospheres. Evaluation of the patients was performed before the first and after the last HBO session and 3 months afterwards. Statistical evaluation was based on an intention-to-treat analysis using Fisher’s exact test and Student’s t-test for paired samples.

(3) Results: In total, 59 patients (33 females, 26 males; mean age: 43.9 years; range: 23–74 years; median: 45.0) were evaluable. After HBO, a statistically significant improvement of physical functioning (p < 0.001), physical role (p = 0.01), energy (p < 0.001), emotional well-being (p < 0.001), social functioning (p < 0.001), pain (p = 0.01) and reduced limitation of activities (p < 0.001) was confirmed.

(4) Conclusions: Physical functioning and both the physical and emotional role improved significantly and sustainably, suggesting HBO as a promising supportive therapeutic tool for the treatment of LCS.

Web | DOI | PMC | PDF | Journal of Clinical Medicine

 

[Hutan]

I'm surprised we didn't have this one, but I couldn't find it.

No sham treatment, so unblinded, with subjective outcomes , SF-36 and VAS

 

[Hutan]

HBO treatment was carried out on an outpatient basis in a large walk-in, drive-in hyperbaric chamber.

HBO sessions were carried five times a week with a weekend break (two series of five compressions were performed). Each session lasted 75 min for a scheduled total time of 12 h and 30 min per patient. HBO was administered at a pressure of 2.2 atmospheres using medical oxygen. During compression the patients breathed 100% oxygen.

100% oxygen, 2.2 atmospheres, 75 mins - seems pretty full on.

Outcomes

Patient evaluation was carried out at three defined time points: immediately before HBO, immediately after the 10th HBO session and after 3 months.

The SF-36 survey is a widely used standardized questionnaire consisting of 36 self-reported items that are grouped into 8 dimensions [18]. These eight dimensions (physical functioning; physical role; energy; emotional role; emotional well-being; social functioning; pain; general perception of health) were used as main outcome measures in our exploratory analysis. In addition, the SF-36 questionnaire comprises a list of 38 everyday activities, where patients have to judge whether they were limited in performing these activities.

The question proposed for the visual analogue scale was “How would you rank your present health and fitness by using this scale?”, where the scale extended across a distance of 100 mm. The mark set by the patient was measured with a ruler and recorded as mm from 0, where 0 indicated the worst case and 100 the best case

 

[Hutan]

Results

After the dropout of 11 patients (9 due to non-compliance, 1 because of barotrauma to the middle ear and 1 because of an anxiety attack during treatment) 59 patients (33 females, 26 males; mean age: 43.9 years; range: 23–74 years; median: 45.0) were evaluable for the study and had the planned number of HBO sessions.

Thirty-seven patients wished to continue the HBO treatment beyond the 10th session due to subjective improvement of symptoms. They had a varying number of further HBO sessions and were excluded from the statistical evaluation beyond this time point. Out of the 22 remaining patients, 18 entered the final evaluation at 3 months, whilst 4 declined to show up for the investigation

So the 37 out of 59 people who decided to have more sessions were excluded from followup. That seems like it could bias things a lot. I'm not sure in what way, perhaps people who weren't yet feeling a benefit wanted to keep going to see if there was one?

Out of the initial 70 patients who started the treatment, only 18 did the 3 month evaluation.

I'm wondering if the arrangements meant that the patients were resting more than usual aside from the treatment. It was 5 days a week for two weeks, possibly the patients had to stay near the HBO facility and would not be doing the usual housework and cooking? Anyway, with only 18 out of the 70 patients making it to the 3 month evaluation and the biased exclusions, along with no sham treatment, I can't see how this trial can be reliable evidence of a benefit.

The other thing is that it is likely that some people with Long Covid were recovering naturally over this time.

 

[Hutan]

That's the results for the 59 people completing the 10 sessions. There are some odd things there. The General Perception of Health hardly changes - it goes fro 39.5 to 42.9, whereas the VAS score for the question 'How would you rank your present health and fitness?' changed from 5.85 to 3.79.

(There seems to be a problem in the paper, the Methods section says the scale for the VAS was 0 to 100, with 100 being the best possible. Whereas in Table 1, the score appears to be out of 10, and the score decreases, but I doubt people were rating their health and fitness worse.)

Overall, mean changes were fairly small - consistent with a placebo effect and possibly some natural improvement.

 

[Jonathan Edwards]

Was there even a control group?
If not these p values are meaningless. The null hypothesis is that time did not pass.

 

[Hutan]

Was there even a control group?

No sham treatment, so unblinded, with subjective outcomes , SF-36 and VAS

 

[EndME]

Treatment here seems comparable to a similar placebo-controlled trial which had a negative result, but where both groups improved: Ten sessions of hyperbaric oxygen versus sham treatment in patients with long covid HOT-LoCO. Unfortunately the outcome measures in that trial are different ones, so not perfectly comparable, but the other trial included a lot of objective outcome measures as well.

 

[EndME]

At also seems like the procotol of this study had not been registered anywhere.

 

[Jonathan Edwards]

There was no sham, but was there even a control?
If not, blinding and stuff doesn't even matter. No meaningful null causal hypothesis can be tested.

 

[EndME]

RAND-36 used in the placebo-controlled study and SF-36 used in this study are the same questions but are sometimes evaluated differently, depending on the domain. If one had the raw data one could probably see whether the results here are in line with the results of placebo-controlled study. Most domains in the questionnaires get scored equivalently as far as Google tells me. For example physical functioning and physical role. At a first glance the results of the this non-placebo-controlled study seem to be in-line with the results of the placebo-group in the other study. For example in this study there is a 8.5 point improvement in the physical function score if you look at all patients (table 1) (and for the small group of patiens the results are comparable and things only get better if you select down to 18 patients), whilst the placebo-group in the other study had a 8.59 improvement.

The placebo-controlled study also included a VAS, however essentially the opposite one where you rate yourself how good you feel, not had bad you feel. So the scales and the results are not comparable. However, given that the other study already gives us unusually strong evidence about nonspecific improvement because it was randomized, double-blinded and placebo-controlled I think the results of this study carry close to no weight.

 

[EndME]

(There seems to be a problem in the paper, the Methods section says the scale for the VAS was 0 to 100, with 100 being the best possible. Whereas in Table 1, the score areas to be out of 10, and the score decreases, but I doubt people were rating their health and fitness worse.)

This seems to be based on a ruler and then measuring the distance with a ruler. The worse scores are consistent because people are rating the severity of their illness.

The description in the paper is a bit inconsistent, first it says "How would you rank your present health and fitness by using this scale?” (where better health would presumably score better) but then it says "This numeric score describes the patients’ general perception of the severity of the disease on a 10-piece scale, with 10 indicating the highest severity". I think the second one is the one they used which is consistent with them later writing "mean pre-therapeutic VAS score describing the patients´ general perception of the severity of their actual disease was 5.85 +/− 2.01 on a 10-piece scale, with 10 indicating the highest severity."

I must admit that I find measuring severity of illness like this less intuitive than measuring my health.

 

[EndME]

So the 37 out of 59 people who decided to have more sessions were excluded from followup. That seems like it could bias things a lot. I'm not sure in what way, perhaps people who weren't yet feeling a benefit wanted to keep going to see if there was one?

I agree it's hard to understand where the bias is coming from but it doesn't seem unlikely that those 18 are simply in the group of people that report the largest improvements at 3 months. I also don't understand why in table 3 they don't include the "Post-HBO" data for the 18 participants which would give more information of whether there is a "sustained effect" vs simply a "selection effect" of choosing your 18 pariticipants.

 

[Hutan]

For example physical functioning and physical role. At a first glance the results of the this non-placebo-controlled study seem to be in-line with the results of the placebo-group in the other study. For example in this study there is a 8.5 point improvement in the physical function score if you look at all patients (table 1) (and for the small group of patiens the results are comparable and things only get better if you select down to 18 patients), whilst the placebo-group in the other study had a 8.59 improvement.

That is helpful EndME.

So, in the open label study, there was an 8.5 point improvement in physical function. In the placebo group in the blinded study, there was essentially the same reported improvement.

Imagine you are in an open label study that you really think will help - that is going to encourage you to report an improvement. Now imagine you are in a blinded study and know that there is a chance you are in the placebo arm. You absolutely don't want to be that person who is on the placebo and reports a great improvement. Any feeling of being better is going to be met with scrutiny 'But am I really better?'.

I reckon there are quite different dynamics going on between the two scenarios. It's far more likely that improvements will be overstated in the open label but ineffective treatment scenario than in the blinded trial placebo arm scenario. So, the fact that the reported improvements in this open label trial are around the same as the reported improvements with a placebo is pretty good evidence that there were no real benefits.