Short-Term Magnesium Therapy Alleviates Moderate Stress in Patients with FM: A Randomized Double-Blind Clinical Trial, 2022, Nicolas Macian et al

Interesting to note that the FM patients had normal levels of Mg in the serum and in erythrocytes at baseline. However, it's noted that magnesium can be released from the bones to maintain Mg blood levels.

 

I had to laugh at that. Stress-related trouble. Despite that, they had to screen people with FM in order to weed out those who did not score highly enough on the stress measuring survey.

The study is to see if this magnesium supplement reduces reported stress.

The study failed to show a difference for this branded magnesium supplement. The DASS-42 score was a measure of how much stress the participants felt. Both the treatment and the placebo arms started with scores of just over 30 and both ended up wth scores of just over 21.

Therefore, it seems pretty clear that the supplement did not reduce stress levels. Up to this point, the study was done well, with blinding.

But, this is a trial of a commercial product, so there was a post-hoc analysis.



And that chart in a. looks good, doesn't it? That's for patients that were rated as having mild or moderate stress versus the placebo. The dark bar is the baseline, and the light bar is at day 28. Patients with mild or moderate stress treated with magnesium had a much bigger reduction in stress than the patients with mild or moderate stress treated with the placebo.

But, have a look at chart b. Look at the effect size difference for all patients and for severely stressed patients. There's basically no difference between treatment and placebo. So, whatever is going on in the mildly/moderately stressed patients, there can't be very many of them. Also, the improvement in the mild/moderate patients is basically the same improvement that was seen in both treated and placebo patients, except for the placebo mild/moderate patients. We can't say that the magnesium treatment had a useful effect, it's just that something odd happened in what was surely a small number of mild/moderate patients in the placebo arm - on average, they didn't improve, unlike basically everyone else.

 

I've tried to find out how many patients were in the post-hoc mild/moderate groups. I don't think the paper tells us that anywhere.
Table 2 gives the characteristics of the mild/moderate and severely stressed groups, but the numbers are for the patients in both treatment arms together.

I mean, there were 37 patients in the magnesium treatment and 38 patients in the placebo treatment. So the 16 mild/moderate patients were spread over the magnesium treatment and the placebo treatment. Maybe there was only a handful of mild/moderate patients in the placebo group.

 

The only secondary outcome that the authors claim showed any difference was in this same small strange mild/moderate stress category.

And, nope, that's not going to fly with me. Knowing that the total sample, spread between the treatment and the placebo was only 16, and knowing that that MM placebo group had a completely different response to the stress survey than the SS placebo group and the total placebo group, this result could easily be chance. And the before and after differences aren't very big.

I think it's irresponsible of the authors to not show that the sum of the mildly/moderately people in the magnesium treatment arm and in the placebo arm was only 16. If they are claiming a result, they absolutely should have given the n. I reckon they had to find something to keep the trial funder happy when it turned out there was no result above placebo, so they came up with the post hoc analysis.

 

I think this study shows that
* not all people with FM score sufficiently highly on a stress survey to qualify as stressed. So, maybe FM doesn't equal "stress-trouble"

* people with FM can have normal blood magnesium, indeed in this trial most did (edit - and their blood levels marginally decreased after a month on the supplement, it didn't increase)

* the branded magnesium taken for a month had no impact on reported stress levels

* the branded magnesium taken for a month had no impact on reported pain levels.
Yeah, it's actually worse than I first thought. They had a whole range of pain measures, and cherry picked the one single finding from the small aberrant post-hoc sample.


Have a look at the Brief Pain Inventory - Pain severity row.
Have a look at the All patient placebo column - there was an improvement from 5.7 to 5.1.
Now look at the mM Mg treatment column - there was an improvement from 5.7 to 5.1

Even in this cherry picked example, it wasn't that the mildly/moderately stressed people responded to the magnesium more than the average placebo effect. The difference was because the people in the small placebo treatment subset reported getting worse.

Have a look at the total sample for the Fibromyalgia Impact questionnaire. People on the magnesium improved by 4 points, people on the placebo improved by 9 points. The difference at day 28 was significant (I'm not sure if they compared the change or the absolute values). But, whatever, a significant result - the placebo was better at reducing the impact of fibromyalgia.

 

An interesting thought @bobbler. Yes, lactose does seem like an odd choice for a placebo to me. Against the idea that it is that that is causing the mild/moderate stress placebo treatment group to worsen
1. the amount (two tablets) is pretty low

2. the people in the severe stress placebo treatment didn't worsen, they improved as much as those given the magnesium treatment. Likewise the placebo group (all of them, both mild/moderately stressed and severely stressed) on average didn't worsen, and in fact improved as much as those given the magnesium treatment. The confidence limits around the total placebo group were tight, suggesting there were just a few outliers affecting the result in the mild/moderately stressed post-hoc group.

 

In this magnesium study, the unit of magnesium in the tablet was mg, as in a total of 100 mg in two tablets (50 mg x2), or 0.1 g. A standard '0' size capsule can hold about 0.5g of powder, so at most the daily lactose dose was 1 g.

In the lactose experiment you refer to, they talk about 'lactose loading' and a dose of 25 g of lactose, delivered in a solution that participants drank, not pills. Therefore, I think it's really unlikely that the quantity of lactose in the placebo of the magnesium study had any effect; it was just too small.

Yes, I think chance events in a very small cohort probably explain the odd result. There is another possibility - that the blinding was somehow broken in some or all of the people given the placebo and in the mild/moderate stress category. The product being tested is a branded one, with money to be made if the product can be promoted as having scientific evidence to support its use. We have no evidence to suggest that blinding was broken, deliberately or otherwise, of course, other than what the authors did when there was no positive result in the primary outcome. They did a post hoc analysis, they did fail to disclose the low number of participants in the post hoc group of interest, and they did present the results as a success of the magnesium treatment rather than an anomalous result for the placebo treatment. They also cherry picked a positive result for a secondary outcome, and presented it as supporting the use of the magnesium product without noting the substantial evidence in their trial against that conclusion, including a significant negative result.

 

The question is, how can we somehow make the authors accountable for their selective use of the data and try to stop papers like these being used to promote products? There's so many of these problematic papers, and we have so little energy. We review papers a bit like other people do the morning crossword - it's sort of a fun mental exercise. But, what then?

There seems to be some sort of forum for discussion of papers embedded in this journal. But, without knowing how influential the journal is and how the comment facility works, it's hard to know if it's worth using limited energy to set out the concerns. Not having any standing in a research institution is also a problem, one that makes providing comments on some papers impossible.

 

I found that Chronomag is produced by Alliospharma in France. The researchers are also from France.