June 8, 2023

expert reaction to study on Metformin after testing positive for SARS-CoV-2 and risk of developing long COVID

A study published in The Lancet Infectious Diseases looks at the use of Metformin (a diabetes drug) in the treatment of COVID-19, and incidence of post-COVID-19 condition over 10 months.

Dr Frances Williams, Professor of Genetic Epidemiology at King’s College London:

“The study by Bramante et al published in the Lancet is a well designed intervention in overweight people with COVID in which a well known, safe drug called metformin was started during the first week of symptoms to see if it influenced  the development of long-lasting symptoms, so-called long COVID.

“This was not an easy study to conduct but the design was well considered and robust. The metformin group had reduced long covid subsequently – 35 cases developing compared to 58 in the placebo group. This represents an impressive-sounding 40% reduction in risk of long COVID. But one important aspect is the number of people exposed to the drug. In total, 564 people were exposed to the drug metformin to prevent 23 hypothetical cases. This means 24 people would need to take metformin to prevent one case of long covid.

“While metformin has very good safety data, being widely used and safe in pregnancy and renal failure, this is a considerable drug burden in the context of a poorly understood condition. We still lack knowledge of long COVID – its definition remains a challenge –  and what it means for people long term. Ethnic minority groups were under-represented in the study and it will be of importance to see if this treatment reduces symptoms enough to allow people to return to work.”

Dr Adam Jacobs, Senior Director, Biostatistical Science, Premier Research, said:

“The paper has some good points to it, though it’s not perfect.

“The most important point in its favour is that it’s a double-blind randomised trial. We’ve seen a lot of observational research around covid, which is always hard to interpret, but the design of this study immediately makes it more believable than any of the observational studies. Another good point was that it included non-hospitalised patients, and since most people with covid are not hospitalised, this makes the results more generalisable than any studies limited to hospitalised patients.

“The follow-up time was long enough to determine whether patients had long covid: at least 180 days, and up to 300 days (different patients had different participation times).

“Definition of long covid was based on a medical diagnosis, not self report.

“There were, however, some weaknesses.

“The most important weakness is that long covid was not the primary objective of the trial, it was one of the secondary objectives. The authors state in the paper that the trial also met its primary objective and showed metformin to be superior to placebo in acute treatment of covid. However, this does not appear to be accurate. I looked up the paper describing the primary results (Reference 6, Bramante et al 2022), which reports that metformin failed to show a significant difference from placebo on the primary outcome measure of a composite of hypoxia, emergency department visit, hospitalisation, or death at 14 days. Although metformin was significantly superior to placebo on the secondary outcome measure of emergency department, hospitalisation, or death (ie the same as the primary endpoint but ignoring hypoxia, and which is claimed in the current paper to be the primary endpoint), results of secondary analyses are never entirely trustworthy when the study has failed to meet its primary objective. It is also strange that the authors describe the primary outcome differently in this paper and the paper reporting the results of acute covid treatment.

“The study also failed to analyse all patients. 1421 patients were randomised to treatment, but only 1126 were analysed. It does not appear that the authors made any attempts to take account of the missing data, and it cannot be ruled out that the experience of patients who were randomised but not analysed would have been different.

“There was a small sex imbalance between the groups: 54.1% of the metformin patients were female, compared with 58.2% of the placebo patients. They do not appear to have adjusted for sex in the analysis. Since other research has shown that women are at greater risk for long covid than men, it is possible that this sex imbalance may have made metformin look better than it really is.

“As the authors themselves note, the study is not completely generalisable, as it was limited only to overweight patients. This may be particularly important, as subgroup analyses showed a greater effect in obese patients compared with non-obese (but still overweight) patients. The authors also note that a further limitation was that no systematic attempt was made to diagnose long covid: long covid diagnoses were dependent on patients seeking out healthcare for any long covid symptoms. However, because of the double-blind and randomised nature of the trial, it seems unlikely that this would have introduced any bias in the treatment comparison. It may, however, mean that the absolute risks of long covid reported in this paper underestimate the true risks of long covid in this population.

“It is worth noting that this study did not investigate only metformin: it was a factorial trial that also investigated ivermectin and fluvoxamine. Neither of those other 2 drugs had any significant effects on either acute treatment or long covid.

“The main results of this study showed that metformin reduced the risk of long covid from 10.4% to 6.3%, a relative decrease of 41%. This is certainly a clinically significant effect, and was also statistically significant. In a subgroup analysis of patients who received metformin within 3 days of symptom onset, the effect was still more impressive, reducing the risk from 11.8% to 4.6%, a relative risk reduction of 63%. Although subgroup analyses should generally be treated with caution, it seems biologically plausible that starting treatment earlier during an acute infection would have a greater effect, so this subgroup analysis seems believable.

“Finally, this study recruited patients mostly in the pre-omicron period. Although the study did include some omicron patients (about 23%), and the relative risk reduction shown in those patients (55%) was broadly in line with the overall patient population, the small sample size of omicron patients means that it is hard to be sure that the results still apply now that omicron is the dominant strain, although I can think of no particular reason to think that they wouldn’t.

“Overall, despite these weaknesses, I think this study still makes a useful contribution to the literature, though I would like to see these results confirmed in a further randomised trial.”

Prof Ewan Pearson, Professor of Diabetic Medicine, University of Dundee, said:

“Metformin is a drug that has been used to treat type 2 diabetes for over 60 years. It has a long established safety record and is effective at reducing diabetes related complications and reducing risk of cardiovascular disease in people with Type 2 diabetes. 

“In this study published in Lancet Infectious Diseases, a placebo controlled randomised controlled study established that, compared to placebo,  the use of metformin reduced risk of long-covid (as diagnosed by a health care professional) by 41%.   This was consistent with an earlier reported endpoint from this study, where metformin reduced risk of emergency department visits, hospitalisation and death by 42%.  Overall the study outcomes are convincing, although it should noted that the number of people developing long covid was relatively small (n=93) and the study population were largely those with private healthcare and were predominantly White.  As metformin reduces risk of developing diabetes it is notable that the rate of development of diabetes was not reported; with symptoms of diabetes potentially overlapping with those of long-covid.  

“Metformin was identified from the French Lilac flower as a potential therapeutic agent for treatment of type 2 diabetes.  Metformin has multiple complex modes of action with its effects on glucose lowering being mediated via the liver and the gut.  Colleagues in Dundee, and others, have also shown that metformin has an anti-inflammatory effect, this with putative antiviral mechanisms may mediate the benefits for COVID outcomes seen in this study.  Beyond the use of metformin to treat diabetes,  its potential to reduce risk of cancer, to treat multiple sclerosis and to slow ageing are under investigation. Metformin has also been used for many years to treat people without diabetes, but with polycystic ovarian syndrome.   This exciting study adds yet another therapeutic indication for metformin – as a safe effective treatment to improve outcomes in COVID-19 infection.”

Dr Stephen M Lawrence, Associate Clinical Professor, University of Warwick, said:

“This randomised control trial accurately reflects the science that people with obesity have a high risk of developing type 2 diabetes. More than 90% of people with type 2 diabetes are either overweight or obese. It therefore stands to reason that their metabolic risk, which includes their tendency to hyperglycaemia, is going to be raised compared with those of normal weight. 

“The researchers have suggested that the observed benefit is attributed to Metformin interfering with replication of the Covid virus.

“However, an alternative explanation for the positive outcome following two weeks of treatment with Metformin may be that it is addressing a tendency to hyperglycaemia which is considered a risk factor for adverse outcomes following infection with Covid in people with established type two diabetes. Poor blood glucose level control is associated with deleterious outcomes following Covid infection. 

“The authors indicate that treatment with Metformin is safe and affordable. However, we need to also acknowledge well-established research, citing the importance of monitoring, kidney function, and the frequency of gastrointestinal side-effects in people taking this medication.

“This study demonstrates useful information supporting the principle of early intervention for people with obesity, who have yet to develop diabetes.”

‘Outpatient treatment of COVID-19 and incidence of post-COVID-19 condition over 10 months (COVID-OUT): a multicentre, randomised, quadruple-blind, parallel-group, phase 3 trial’ by Carolyn T Bramante et al. was published in The Lancet Infectious Diseases at 23:30 UK time on Thursday 8^th June.

DOI: https://doi.org/10.1016/S1473-3099(23)00299-2

Declared interests

Dr Adam Jacobs: “No conflicts to declare.”

Prof Ewan Pearson: No conflicts in relation to this study. Has received honoraria for lectures from Novo Nordisk, Lilly and Illumina.

Dr Stephen M Lawrence: “No conflicts of interest.”