Metformin on the Presence of COVID-19 Symptoms Over 6 Months: The ACTIV-6 Randomized Clinical Trial

Carolyn T. Bramante, Thomas G. Stewart, David R. Boulware, Matthew W. McCarthy, Yue Gao, Russell L. Rothman, Ahmad Mourad, Florence Thicklin, Jonathan B. Cohen, Idania T. Garcia del Sol, Nirav Shah, Manisha Mehta, Orlando Quintero, Jake Scott, Adit A. Ginde, Mario Castro, Dushyantha Jayaweera, Mark Sulkowski, Nina Gentile, Kathleen M. McTigue, G. Michael Felker, Sean P. Collins, Sarah E. Dunsmore, Stacey J. Adam, Christopher J Lindsell, Adrian F. Hernandez, Susanna Naggie, the Accelerating Covid-19 Therapeutic Interventions and Vaccines (activ)-6 Study Group and InvestigatorsmedRxiv 2025 · DOI

TL;DR

The ACTIV-6 randomized clinical trial investigated whether a 14-day course of metformin could prevent Long COVID (PASC) in low-risk outpatient adults. While the study did not meet its strict primary statistical threshold for efficacy (PPE of 0.83 vs. a goal of 0.975), secondary outcomes—including clinician-diagnosed Long COVID—showed a strong numerical trend toward protection, suggesting metformin remains a candidate for further study in preventing post-viral sequelae.

What Problem Does This Paper Address?

Post-acute sequelae of SARS-CoV-2 (PASC), commonly known as Long COVID, affects approximately 7% of adults in the US, potentially ranking as the sixth most common chronic disease. While previous studies (like the COVID-OUT trial) suggested metformin might reduce Long COVID risk in higher-risk populations, its effectiveness in a broader, "low-risk" population—those with high levels of prior immunity from vaccines or previous infections—remained unproven.

The authors sought to determine if immediate-release metformin, administered during the acute phase of infection, could prevent symptoms from persisting or emerging 180 days later.

Methods

The study was part of the ACTIV-6 platform, a decentralized, double-blind, placebo-controlled trial.

  • Participants: 2,983 outpatient adults (≥30 years) with mild-to-moderate COVID-19. Notably, 83% of the cohort had prior immunity (vaccination or previous infection).
  • Intervention: A 14-day titrated dose of metformin (up to 1,500 mg daily) versus a matching placebo.
  • Statistical Approach: The researchers used a Bayesian multivariate model (specifically a first-order Markov model) to estimate treatment effects at days 90, 120, and 180.
  • Efficacy Threshold: They defined success as a Posterior Probability of Efficacy (PPE) greater than 0.975. This Bayesian approach allowed them to incorporate outcome information from participants who contributed at least one follow-up survey while adjusting for covariates like age, BMI, and vaccination status.

Datasets

As a clinical trial, the "dataset" consists of the participant cohort data collected between September 2023 and May 2024 across 90 US sites.

  • Cohort Size: 2,983 participants in the modified intention-to-treat (mITT) sample.
  • Demographics: 63% female, 47% Hispanic/Latino, median age 47.
  • Context: Most participants were enrolled during the dominance of the JN.1 variant, which is characterized by higher infectivity but lower morbidity compared to earlier strains.

Key Insights and Ideas

  • The "Floor Effect" of Modern Variants: The overall incidence of Long COVID symptoms at day 180 was remarkably low (only 2.6%). This low baseline makes it statistically difficult to prove a significant reduction, even if a drug is effective.
  • Secondary Signal vs. Primary Goal: While the primary endpoint (patient-reported symptoms) missed the mark, the risk of a clinician diagnosis of Long COVID was halved in the metformin group (Relative Risk 0.495, PPE 0.96). This suggests that while "any symptom" might be too broad a metric, "clinically significant Long COVID" might be more responsive to treatment.
  • Biological Mechanism: The paper highlights that SARS-CoV-2 activates the mTORC1 pathway, which damages the gut's epithelial barrier. Metformin is a known mTOR inhibitor and promotes beneficial gut bacteria (SCFA-producers), providing a plausible biological reason for its potential protective effect against long-term systemic inflammation.
  • Safety in Low-Risk Groups: This was the first phase 3 trial to randomize adults without chronic diseases to metformin. It proved exceptionally safe, with no cases of lactic acidosis and fewer episodes of low blood sugar than the placebo group.

Key Results

  • Primary Outcome (Day 180 PASCD): The covariate-adjusted risk difference was -0.008, resulting in a PPE of 0.83. This failed to meet the pre-specified 0.975 threshold.
  • Clinician Diagnosis (Day 180): Metformin showed a more pronounced effect here, with a risk difference of -0.007 and a PPE of 0.96. The adjusted risk ratio was 0.79 (95% CrI 0.474 to 1.230).
  • Symptom Burden (Day 120): The PPE for reducing symptom burden reached 0.98, indicating a high probability of benefit at the four-month mark.
  • Mortality: There were zero deaths in the study, reflecting the lower severity of the JN.1 variant and the high level of community immunity.

Limitations and Future Work

  • Missing Data: Approximately 20% of follow-up surveys were missing at the 90, 120, and 180-day marks.
  • Enrollment Timing: Participants were enrolled up to 10 days after testing positive. Since metformin has antiviral properties, starting the treatment earlier in the infection cycle might yield stronger results.
  • Intermittent Symptoms: The survey captured symptoms on the specific day of the survey. Because Long COVID symptoms can be "relapsing-remitting," some cases may have been missed.
  • Future Direction: The authors suggest that while metformin didn't meet the primary efficacy bar in this specific low-risk, high-immunity context, the consistent direction of benefit across secondary outcomes warrants continued investigation into its role as a low-cost, safe intervention for post-viral syndromes.

Summary updated

Abstract

ClinicalTrials.gov ( NCT04885530 ).
BibTeX
@misc{bramante2025metformin,
  author = {Carolyn T. Bramante and Thomas G. Stewart and David R. Boulware and Matthew W. McCarthy and Yue Gao and Russell L. Rothman and Ahmad Mourad and Florence Thicklin and Jonathan B. Cohen and Idania T. Garcia del Sol and Nirav Shah and Manisha Mehta and Orlando Quintero and Jake Scott and Adit A. Ginde and Mario Castro and Dushyantha Jayaweera and Mark Sulkowski and Nina Gentile and Kathleen M. McTigue and G. Michael Felker and Sean P. Collins and Sarah E. Dunsmore and Stacey J. Adam and Christopher J Lindsell and Adrian F. Hernandez and Susanna Naggie and the Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV)-6 Study Group and Investigators},
  title = {{Metformin on the Presence of COVID-19 Symptoms Over 6 Months: The ACTIV-6 Randomized Clinical Trial}},
  year = {2025},
  doi = {10.1101/2025.08.08.25333305},
  howpublished = {medRxiv},
}
Figure