{"paper_id":"10b4eae3-b032-4fc1-96c3-96aa821af89a","body_text":"Metformin at the time of Covid-19 infection and risk of Long Covid: A Target Trial Emulation Study | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 26 November 2025 V1 Latest version Share on Metformin at the time of Covid-19 infection and risk of Long Covid: A Target Trial Emulation Study Authors : Carolyn Bramante 0000-0001-5858-2080 [email protected] , John B Buse 0000-0002-9204-7177 , Jared D. Huling , John Buse B , Christopher Lindsell , Thomas Stewart 0000-0002-5138-0758 , Russell L. Rothman , … Show All … , David Sahner , Sarah E. Dunsmore , Eric Topol , Talia D. Wiggen , Steve Makkar , Andrew Toler , Taylor Estepp , and Steven Johnson Show Fewer Authors Info & Affiliations https://doi.org/10.22541/au.176418398.81170401/v1 4395 views 379 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Background : Our objective was to evaluate metformin prescribed at the time of SARS-CoV-2 infection on the risk of developing Long Covid (LC) in electronic health record data. Methods : We conducted a new user analysis of metformin prescribed within 6 days of documented infection with severe acute respiratory coronavirus syndrome 2 (SARS-CoV- 2) versus experimental control: prescription for fluvoxamine, fluticasone, ivermectin, or montelukast. Inclusion criteria: a clinic visit in the 0- 6 months and the 6-12 months before infection. Exclusion criteria: metformin or control within 12 months. Primary outcome: LC or death (LC/D), to address death as a competing risk, among patients prescribed drug within Days 0-6 of infection. LC was defined by diagnosis code or computable phenotype. We used entropy balancing to estimate the average treatment effect with a weighted log linear model. Results : After weighting, there were 248 in the metformin and control groups; the average age was 53 (16); 16% were Black; and 16% were Hispanic. In the primary analysis, 10/248 (4.0%) in the metformin group developed LC/D vs. 21/248 (8.5%) in the control group, adjusted risk ratio (aRR) 0.47 (95% CI 0.25 to 0.89). For prescriptions on Days 0-1 relative to infection, aRR was 0.39 (95% CI 0.12-1.24); for prescriptions on Days 0-14 the aRR was 0.75 (95% CI 0.52-1.08). Conclusions : In this observational analysis, metformin prescribed within a week of documented SARS-CoV-2 infection was associated with a 53% lower risk of LC over 6 months than comparator medications. Any risk reduction between 75% to 11% is highly compatible with our data. This analysis of electronic health record diagnoses is important for the reproducibility of clinical trial results that ascertained the same outcome but via participant-report. Metformin at the time of Covid-19 infection and risk of Long Covid: A Target Trial Emulation Study Carolyn Bramante, MD, MPH 1 ; Til Stürmer, MD, PhD 2 ; Jared D. Huling, PhD 3 ; John B Buse, MD, PhD 4 ; Christopher Lindsell, PhD 5 ; Thomas Stewart, PhD 6 ; Russell L. Rothman, MD, MPP, 7 ; David Sahner, MD 8,11 , Sarah E. Dunsmore, PhD 8 , Eric Topol, MD, 9 Talia D. Wiggen, MPH 10 ; Steve Makkar, PhD 8,11 , Andrew Toler, MS 8, 11 , Taylor Estepp, PhD 8,11 , and Steven G. Johnson, PhD 10 on behalf of the N3C Consortium. 1. Division of General Internal Medicine, Department of Medicine, University of MN Medical School 2. Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina at Chapel Hill 3. Division of Biostatistics and Health Data Science, University of Minnesota School of Public Health 4. Division of Endocrinology, Department of Medicine, University of North Carolina School of Medicine, Chapel Hill, NC 5. Division of Biostatistics and Bioinformatics, Duke University School of Medicine 6. Division of Biostatistics, University of Virginia School of Data Science 7. Vanderbilt Institute for Medicine and Public Health, Nashville, TN 8. National Center for Advancing Translational Science (NCATS) 9. Scripps Research Translational Institute, La Jolla, CA 10. Institute for Health Informatics, University of Minnesota, Minneapolis, MN 11. NCATS contractor Axle Informatics Data Availability: The data in the N3C is freely available to anyone with an approved data use agreement. The code and codesets are available in GitHub, https://github.com/UMN-IHI/phastr-metformin. Disclosures: Drs. Bramante, Dunsmore, Stewart, Rothman, and Lindsell are on the executive committee for the ACTIV-6 trial. Dr. Bramante is the principal investigator of the COVID-OUT trial. Acknowledgements : We gratefully acknowledge the contributions from Axle Informatics: Breezy Synoground. The analyses were conducted with data or tools accessed through the NCATS N3C Data Enclave https://covid.cd2h.org and N3C Attribution & Publication Policy v 1.2-2020-08-25b supported by NCATS Contract No. 75N95023D00001, Axle Informatics Subcontract: NCATS-P00438-B. This research was possible because of the patients whose information is included within the data and the organizations (https://ncats.nih.gov/n3c/resources/data-contribution/data-transfer-agreement-signatories) and scientists who have contributed to the on-going development of this community resource [https://doi.org/10.1093/jamia/ocaa196]. Disclaimer The N3C Publication committee confirmed that this manuscript MSID:2076.771is in accordance with N3C data use and attribution policies; however, this content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the N3C program. Authorship was determined using the ICMJE recommendations. Abstract Background : Our objective was to evaluate metformin prescribed at the time of SARS-CoV-2 infection on the risk of developing Long Covid (LC) in electronic health record data. Methods : We conducted a new user analysis of metformin prescribed within 6 days of documented infection with severe acute respiratory coronavirus syndrome 2 (SARS-CoV- 2) versus experimental control: prescription for fluvoxamine, fluticasone, ivermectin, or montelukast. Inclusion criteria: a clinic visit in the 0- 6 months and the 6-12 months before infection. Exclusion criteria: metformin or control within 12 months. Primary outcome: LC or death (LC/D), to address death as a competing risk, among patients prescribed drug within Days 0-6 of infection. LC was defined by diagnosis code or computable phenotype. We used entropy balancing to estimate the average treatment effect with a weighted log linear model. Results : After weighting, there were 248 in the metformin and control groups; the average age was 53 (16); 16% were Black; and 16% were Hispanic. In the primary analysis, 10/248 (4.0%) in the metformin group developed LC/D vs. 21/248 (8.5%) in the control group, adjusted risk ratio (aRR) 0.47 (95% CI 0.25 to 0.89). For prescriptions on Days 0-1 relative to infection, aRR was 0.39 (95% CI 0.12-1.24); for prescriptions on Days 0-14 the aRR was 0.75 (95% CI 0.52-1.08). Conclusions : In this observational analysis, metformin prescribed within a week of documented SARS-CoV-2 infection was associated with a 53% lower risk of LC over 6 months than comparator medications. Any risk reduction between 75% to 11% is highly compatible with our data. This analysis of electronic health record diagnoses is important for the reproducibility of clinical trial results that ascertained the same outcome but via participant-report. Words: 3,264 Background The severe acute respiratory coronavirus 2 (SARS-CoV-2) continues to infect adults and children worldwide. 1-3 Long Covid (LC) symptoms often include functional limitations and impairment. 4-9 Even mild limitations after SARS-CoV-2 infection could have broad societal and environmental impacts. 5,7,10 LC may be a greater predictor than severe acute Covid-19 for having post-Covid economic hardship. 11 The potential for LC to increase economic disparities highlights the urgent need to understand whether medications at the time of SARS-CoV-2 infection affect the risk of developing LC. Metformin has inhibited SARS-CoV-2 in several in vitro studies. 12-14 In a randomized clinical trial (RCT) evaluating metformin vs. placebo in over 1,000 outpatient adults (COVID-OUT), LC occurred in 6.2% of metformin vs 10.3% of placebo subjects by day 300; 15,16 and 14% of metformin vs 23% of the placebo group had detectable nasal viral load on day 10. 17 However, only 22% of the trial participants were enrolled during the Omicron era, enrollment was restricted by age (30-85 years) and body mass index (BMI) >=25kg/m 2 , and the LC outcome was diagnosis by a clinician. Thus, our objective was to use electronic health record (EHR) data to understand the impact of new metformin use at the time of SARS-CoV-2 infection on the risk of developing LC in a broader sample to improve generalizability beyond patients who consent to being in a clinical trial. When using EHR data, it may not be possible to determine the exact reason a medication was prescribed, and thus whether the medication was taken during the timeframe of interest (i.e. during acute infection). To this end, we restricted the analysis to patients without an indication for chronic metformin use who were newly prescribed this medication at the time of acute Covid-19. When considering why metformin might be prescribed at the time of SARS-CoV-2 infection before the RCT results, several mechanistic papers on metformin and Covid-19 were published early in 2020, including a February 2020 letter in Lancet suggesting metformin as a safe host-directed immunomodulator. 18-22 Observational analyses of prevalent metformin use and associations with less severe outcomes during acute COVID-19 were published in early 2020. 19,23-25 The Comprehensive Diabetes Center at the University of Alabama at Birmingham published a manuscript in July 2020 reporting an association between metformin and lower mortality from Covid-19 (OR 0.33; 95%CI 0.13-0.84; p=0.0210). 26 These manuscripts may have led to prescriptions for metformin for acute infection. Previous observational analyses that assessed LC as an outcome have not used an active comparator, or ‘experimental control’. 27-29 An experimental control reduces unmeasured confounding, such as likelihood of engagement with healthcare, immortal time bias, and the placebo effect, more effectively than an untreated control. 30-32 The placebo effect is influential on both subjective and objective laboratory outcomes. 33,34 If an experimental control is not used, the placebo effect is only experienced in the exposure group and not the control group, biasing away from the null. We emulated a placebo-controlled trial by analyzing new-use of metformin versus new-use of an experimental control. Design Target trial emulation (TTE) in adults with documented SARS-CoV-2 infection in the National COVID Cohort Collaborative (N3C) database. The N3C is a centralized and harmonized database of de-identified electronic health record (EHR) data from over 21 million individuals from every state in the US (detail in the Supplement). 35,36 The target trials emulated are the COVID-OUT and ACTIV-6 trials, both randomized, blinded, decentralized trials of metformin versus exact-matching placebo to assess whether metformin during acute infection prevents LC. 15,16 TTE is a methodological approach for the design of observational studies that aims to align the structure and intent of the observational study to that of the idealized RCT, minimizing discrepancies in their findings by ensuring they aim to answer the same question. 37 By explicitly specifying eligibility criteria, treatment initiation, follow-up, outcome assessment, and target estimand as an RCT would, TTE can reduce common biases and pitfalls in observational studies such as immortal time bias and confounding. 38 This alignment improves the validity of real-world evidence and enhances the comparability between observational and trial results, fostering more reliable insights into treatment effects. 39 The follow-up in this TTE closely emulates the target trials, in which participants did not receive therapy at the outset of symptoms or immediately upon a positive test. In those decentralized trials, the follow-up period started on the day the study drug was delivered, so participants who were hospitalized between consent and receiving drug in the mail were not eligible (Table 1 and Figure 1). The index dates for inclusion in the TTE ranged from March 2020 – September 2023. The date of the first documented SARS-CoV-2 infection, as defined by the N3C (positive PCR or Covid-19 diagnosis code) 40 , is the index date. 41 Day 1 of follow-up was the prescription of metformin or control. Experimental Control and Intention to Treat We emulated placebo control with medications that have been used for acute SARS-CoV-2 infection but do not have consistent clinical trial data to support an effect on preventing Long Covid-19: fluvoxamine, ivermectin, montelukast, and fluticasone (inhaled and intra-nasal). 16,42,43 The primary exposure of interest was a new metformin prescription on day 0 to 6 of documented SARS-CoV-2 infection. The experimental control was new prescription of one of the control medications on day 0 to 6 of documented infection. Individuals subsequently prescribed metformin or a control remained in the cohort of their first prescription, in keeping with intention to treat in a clinical trial. Analysis Population Documented SARS-CoV-2 infection as defined in the N3C, 40 and evidence of 2 outpatient visits prior to the index date: one in the 0 to 6 months prior and one in the 6 to 12 months prior, to increase the likelihood that comorbidities and pre-existing prescriptions were captured. Per new-user methodology, those with a prescription for metformin or a control in the 12 months prior to the index were excluded. 37 To mimic prospectively starting an intervention, we excluded patients with a documented clinical indication for metformin to minimize the potential of prevalent metformin use in the comparator group, and to enrich the sample for individuals starting the intervention at the time of the indication of interest (SARS-CoV-2 infection). Metformin is the first line pharmacologic agent for type 2 diabetes and prediabetes. Diabetes and diabetes treatments are strongly associated with acute Covid-19 severity and LC, so excluding those with diabetes from both the exposure and control groups is important because measures to address confounding would be insufficient to address imbalance between metformin and experimental controls. 44,45 We also excluded polycystic ovary syndrome, anti-psychotic induced weight gain, and gestational diabetes (other indications for which metformin is often prescribed). Patients with indications for the controls were not excluded because no prospective data suggest that the controls prevent the development of LC, and the control agents are not first-line treatments for their most common indications. 46,47 Individuals prescribed a control at the time of documented infection are more likely to have been prescribed and thus taken the medication for Covid-19 rather than a chronic indication. The most common indications for the control agents (asthma and allergic rhinitis) include symptoms common during acute Covid-19. Therefore, if an individual were coincidentally prescribed fluticasone for allergic rhinitis at the time of SARS-CoV-2 infection, they might obtain and start that prescription right away. In contrast, patients with type 2 diabetes and prediabetes do not frequently have symptoms that are common during acute Covid-19 and therefore may be less likely to start metformin for a chronic indication during acute illness. 48 Key to emulating a clinical trial is finding individuals who initiate the treatment of interest during the timeframe for the indication of interest (acute infection). 37 Contraindications for metformin or controls were excluded (stage 4 or 5 chronic kidney disease, Figure 2). 49 Data from sites which had never reported an International Statistical Classification of Diseases Tenth Revision ( ICD-10 ) code U09.9 for post-Covid-19 condition were excluded to minimize ascertainment error. Individuals who died or developed LC before treatment were excluded, as were individuals with a hospitalization from day -3 to 1 relative to the positive SARS-CoV-2 test because they would be unlikely to have started metformin or an experimental control. This emulates the COVID-OUT and ACTIV-6 trials, as both decentralized trials excluded those hospitalized at the time study drug was delivered. 16,50 Outcome Incidence of LC or Death (LC/D) within 180 days of the metformin or control prescription. Because the endpoint is death or LC, death is not a competing risk. LC was defined as an ICD-10 diagnosis code for LC (U09.9) or a computable phenotype for LC that is based on symptoms and conditions. 40,51,52 The computable phenotype is based on new diagnoses, such as post-viral fatigue or new-onset asthma; or worsening of pre-infection diagnoses. The worsening is calculated by looking at equal windows of time before and after infection and the number of clinic visits for that condition in the pre-infectious window compared to the post-infectious window. Statistical Analysis To balance covariates and adjust for confounding between cohorts we used R weightit package (version 0.14.2) to perform entropy balancing (EB), allowing an estimation of the average treatment effect among the treated (ATT) 53-55 using a weighted log linear model. EB works similarly to inverse propensity score weighting: weights are directly estimated by minimizing a measure of entropy divergence of the control weights relative to base (e.g., uniform) weights in the population, subject to exact balance of covariate distribution moments in the treated and control groups. Although EB weights are constructed via an optimization criterion, they implicitly estimate the propensity score. 55 We weighted on demographics (age, gender, race/ethnicity), BMI, prior medications and conditions that have associations with LC, and laboratory values including hemoglobin A1c (Table 2 and Supplement Tables 3 and 4). An indicator for missingness of BMI was included in the weighting. We did not add an indicator for missingness for other variables that most EHR records had (age, sex) because missingness was low. We weighted by N3C site to adjust for site differences in data practices. Love plots were generated to visualize the balance achieved among cohorts on the covariates (Supplement Figure 1). Because fewer than 10% of sites in the N3C have reliable data on vaccination status, this important variable was not included. The R svyglm package (version 4.0) was used to fit a weighted log linear model to estimate the risk ratio of LC or death between patients exposed to metformin versus the controls. Stratified analyses of interest were viral variant epoch and those with new use of nirmatrelvir-ritonavir. 56 A sample-size calculation was not performed because we had no reason to not analyze all individuals who remained after applying the inclusion and exclusion criteria. Sensitivity analyses by Time to Treatment A time to treatment of 0 to 6 days from documented infection was prospectively chosen as the primary analytic sample to emulate the exclusion of individuals with >= 7 days of symptoms in the target trials. In the COVID-OUT trial, initiation of study drug within 4 days of symptom onset, as opposed to >= 4 days from symptom onset, showed a smaller hazard ratio for metformin. 15,16 To examine whether the duration of time between documented infection and prescription influenced the results in this TTE, we assessed two additional treatment windows (day 0 to +1 and day 0 to +14). We did not use mutually exclusive timeframes because of concern for sample size, but the windows would be expected to contain a progressively larger number of patients who initiated therapy later, with the greatest number of “late starters” in the 0-14-day group. We examined only the day 0-6 and 0-14 treatment windows within subgroups because the day 0-1 window subgroup was not large enough to balance. Sensitivity analyses by Experimental Controls (Excluding Fluvoxamine; Excluding Ivermectin) The research team became aware that some individuals seeking ivermectin for Covid infection may be less likely to seek care for LC symptoms or to allow symptoms to be ascribed to LC. 16 Ascertainment bias could lead to misclassification of the outcome, so subgroup analysis excluded ivermectin from the control cohort. Because of emerging data about selective serotonin reuptake inhibitors (SSRIs), 57 we excluded fluvoxamine from the control cohort in a sensitivity analysis as well. Results After applying the inclusion and exclusion criteria, the primary sample was n=248 in the metformin cohort and n=9,412 in the control cohort (Figure 2). Missingness was frequent for BMI (47% in the metformin group and 28% in the control group), and few patients in the metformin group had a documented BMI of <25kg/m 2 . Before weighting, the metformin group was older, had a higher mean BMI and mean HbA1c; and a larger portion was male (Table 2). After weighting, these variables were well balanced, and the median standardized mean difference (SMD) was 0 (Table 2). After weighting, 16% of the metformin and control groups were Black; 16% of both groups were Hispanic; the mean BMI in the metformin group and control groups were, respectively, 35.9 (SD 9.27) and 35.9 (SD 11.05); and 50% of the metformin cohort and 52% of the control cohort were infected during the Omicron variant epoch. The modal day of prescription was day 0 for both metformin (38%) and the controls (61%). Most of the metformin (72%) and control (84%) prescriptions were on days 0 to 2 relative to documented infection. A minority of the metformin (10%) and control (5%) prescriptions were on day 5 or later (Table 2). The primary analytic sample was those receiving a metformin or comparator prescription within day 0 to 6 relative to infection. The unweighted overall incidence of LC/D was 7.0%, with 1.7% deriving from death and 5.5% from LC. The numerators and denominators are not given due to the N3C data governance rules for counts in a cell of a table (at least 10). (Supplemental Table 2). The weighted overall incidence of LC/D in the day 0 to 6 cohort was 6.3%: 4.0% in the metformin cohort and 8.5% in the experimental control cohort, with absolute risk reduction of 4.5% for metformin versus control (Supplemental Table 2). In the primary analysis, the adjusted risk ratio (aRR) for LC/D was 0.47 (95% confidence interval [CI] 0.25 to 0.89, p=0.02) for metformin versus control. The aRR was 0.37 (95% CI 0.12-1.24) for a time to treatment window of 0 - 1 days relative to infection; and 0.75 (95% CI 0.52 - 1.08) for 0 -14 days (Figure 3). In the day 0-6 treatment window, the aRR for the Omicron era was 0.46 (95% CI 0.14-1.50); the aRR for the pre-Omicron eras combined was 0.41 (95% CI 0.18-0.92). The subgroup of those with a prescription of nirmatrelvir-ritonavir was too small to balance; in the subgroup excluding those with new use of nirmatrelvir-ritonavir, the aRR was 0.48 (95% CI 0.25-0.60). The results were consistent across sensitivity analyses (Figure 4 and Supplemental Figures 2-8). Discussion We conducted a target trial analysis of new metformin use versus an experimental control at the time of SARS-CoV-2 infection to emulate the COVID-OUT and ACTIV-6 randomized clinical trials of metformin vs. placebo for outpatient treatment of SARS-CoV-2 infection. In this TTE, most prescriptions occurred within two days of the infection, suggesting successful identification of individuals who were seeking metformin or a control medication for their SARS-CoV-2 infection. Metformin prescription within the week following documented SARS-CoV-2 infection was associated with a 53% reduction in LC/D over 6 months in this primarily Omicron-infected sample. These results are also consistent with a recent sequential target trial analysis that assessed metformin started within 90 days of infection, hazard ratio 0.36 (95% CI 0.32-0.41). 58 In the COVID-OUT trial, LC was ascertained by asking participants if a medical provider had diagnosed them with LC, and then obtaining medical records to confirm. 16 The overall incidence of LC in COVID-OUT and this TTE is similar, 8.3% over 10 months and 6.3% over 6 months, respectively. These proportions are also similar to the 10% identified as having LC in the RECOVER prospective acute cohort. 59 Emerging data suggest that the risk of LC has decreased throughout the pandemic, which may explain the numerically lower incidence of LC in the TTE analysis. 60 The absolute risk reduction with metformin was similar in this TTE (4.4%) and the COVID-OUT trial (4.1%). In COVID-OUT, the incidence of LC was 6.3% (95% CI 4.2 to 8.2) in the metformin group and 10.4% (7.8 to 12.9) in the placebo group. 16 Unlike the COVID-OUT study, which limited enrollment to those with obesity or overweight, this TTE included a sample that is more heavily weighted toward the Omicron era, and is less restrictive with respect to age, comorbidity, or BMI, although the number of patients with documented BMI under 25kg/m 2 in the metformin group was small. A smaller point estimate with earlier evidence of medication use was also similar between the current analysis and COVID-OUT, in which participants who started metformin in <4 days of symptom onset had a 63% lower likelihood of receiving a diagnosis of LC over the subsequent 10 months. 16 A larger effect when therapy is initiated earlier in infection is consistent with an anti-viral mechanism against SARS-CoV-2. Metformin lowered the viral load in two RCT’s, and in vitro studies found metformin lowered SARS-CoV-2 RNA while increasing cell viability. 12-14,17 Two prospective cohorts also suggest lower incidence of LC in those with prevalent metformin use. 61,62 Small number bias may influence the results in this TTE. We considered other approaches to defining the treatment window to increase the sample size such as allowing the window to start on day -1. However, we were concerned that individuals who received a prescription before day 0 may have been receiving that prescription for another indication and thus would not be starting it while acutely ill with COVID-19. The risk ratio in the subgroup excluding those with a nirmatrelvir-ritonavir prescription was slightly higher, generating the hypothesis that those receiving nirmatrelvir-ritonavir prescription in addition to a new metformin prescription had a lower risk ratio than the sample as a whole. Limitations: Observational analyses may be limited by missing data and misclassification of exposures, outcomes and confounders. For example, the use of nirmatrelvir-ritonavir may not be captured in the EHR because individuals were able to obtain nirmatrelvir-ritonavir directly from pharmacies without seeing their physician. A significant limitation is the lack of vaccination data and residual unmeasured confounding. However, the use of an active comparator should distribute unmeasured confounders more evenly between the metformin and comparator groups than comparing metformin starters to non-starters. In observational data we do not know when symptoms started nor if prescriptions were filled and taken. Because of home testing, the exact number and date of SARS-CoV-2 infections may be missing. There may have been undiagnosed diabetes or prediabetes in the metformin group given the elevated mean hemoglobin A1C in that group, but prevalent metformin use in the metformin group is less of a concern than prevalent metformin use in the comparator group. Several other factors could affect outcome ascertainment, such as vague symptoms, access to health care, and pre-existing conditions, but challenges in defining and identifying LC are not unique to this analysis. Conclusions : In a target trial emulation, metformin prescribed within a week of SARS-CoV-2 infection resulted in a 53% lower risk of LC or death in a primarily Omicron-infected population from a nationwide deidentified EHR database. Anything between an 11% to 75% reduction in risk is highly compatible with our data. This result is consistent with both randomized trials that were emulated and other emerging observational analyses. While observational data alone should be interpreted with caution because there may be missingness in exposure, outcome, and confounder variables, the replication of results in clinical trials is important for informing treatment guidelines. Observational data improves generalizability beyond individuals who are eligible for and consent for clinical trials. Eligibility Criteria Documented SARS-CoV-2 Enrolled within 10 days of documented infection Enrolled within 7 days of symptoms Age 30 – 85 No current metformin use Documented SARS-CoV-2 Enrolled within 3 days of documented infection Symptoms not required, but if present, <7 days Age 30-85 No current metformin use Body Mass Index <25kg/m 2 was an exclusion Known prior Covid-19 infection was an exclusion Documented SARS-CoV-2 Prescription within 6 days (primary analysis population), secondary analyses with prescriptions within 1 day & 14 days Age > 18 years No metformin or comparator prescription within 12 months No minimum body mass index The infection analyzed is the first documented infection Intervention Metformin immediate release or exact-matching placebo tablets Metformin immediate release or exact-matching placebo tablets Any metformin prescription or active comparator (medications with no convincing evidence of favorable impact on acute COVID-19 outcomes: fluvoxamine, ivermectin, fluticasone, montelukast) Treatment assignment Randomization Randomization Prescription for metformin or comparator placed; propensity scores to balance measured covariates across treatment cohorts Considerations for medication acquisition Delivery of medication takes 2-3 days on average Delivery of medication takes 1 day on average Getting a medication prescription then obtaining it from a pharmacy likely takes 0 to 3 days on average Post -randomization exclusions Hospitalized at the time of medication delivery Delivery failure Death before day 1 Hospitalized at the time of medication delivery Delivery failure Death before day 1 Hospitalized between -3 days to +1 days of infection because this would preclude the ability to fill a prescription Death on or before day 1 Day 1 First day of a medication dose First day of a medication dose Day prescription placed Follow-up Day 1 to 180 Day 1 to 300 Day 1 to 180 Table 1: Individuals who experienced death before the prescription could be started were excluded from the trial emulation to mimic both target trials (COVID-OUT and ACTIV-6), as they were both decentralized trials that entailed medication delivery to the home. Those who experienced hospitalization between -3 to 1 days were also excluded to mimic the target trial post randomization exclusions that were necessary because the decentralized design created time between Randomization and the Intervention in the RCTs. Figure 1. This is a visual overview of practical constraints that exist when conducting a decentralized clinical trial (DCT), and an electronic health record (EHR) analysis. Because eligibility assessments, consent, randomization, and starting study drug do not happen at the same time in DCT’s, there may be post-randomization exclusions. Figure 2: Cohort Attrition for the metformin and active comparator (control) cohorts. Days 0 to 6 were chosen a priori as the window that likely most closely matches most participants in the target trials, the lack of symptom data in the medical record makes it impossible to match exactly. The number who died before the treatment window was too small to show while maintaining N3C data privacy rules, so that number is included with the number excluded from sites that have never reported a U09.9 code (those sites are excluded because of potential misclassification of the outcome). Metformin (n=248) Control (n=9,412) SMD Metformin (n=248) Control (n=248.1) SMD Age, mean (SD) 53.31(16.48) 45.28 (20.66) 0.430 53.31(16.45) 52.74 (17.25) 0.034 Age Category < 21 < 10 1,288 (0.14) 0.383 < 10 < 10 0.000 21 to 45 < 70 3,454 (0.37) 0.200 68 (0.27) 68 (0.27) 0.000 46 to 65 116 (0.47) 2,964 (0.31) 0.317 116 (0.47) 116 (0.47) 0.000 >= 66 years 56 (0.23) 1,702 (0.18) 0.112 56 (0.23) 56 (0.23) 0.000 Female 113 (0.46) 6,310 (0.67) 0.444 113 (0.46) 113 (0.46) 0.000 American Indian or Alaska Native < 10 42 (0.00) 0.007 < 10 < 10 0.000 Asian < 10 236 (0.03) 0.063 < 10 < 10 0.000 Black or African American 40 (0.16) 1,638 (0.17) 0.034 40 (0.16) 40 (0.16) 0.000 Native Hawaiian, or Pacific Islander 0 (0.00) 24 (0.00) 0.072 0 (0.00) < 10 0.009 Hispanic or Latino Any Race 39 (0.16) 1,251 (0.13) 0.069 39 (0.16) 39 (0.16) 0.000 White non-Hispanic 153 (0.62) 5,807 (0.62) 0 153 (0.62) 153 (0.62) 0.000 Number of office visits before infection, mean (SD) 0-6 months before infection 6.28 (9.56) 7.87 (9.89) 0.163 6.28 (9.54) 7.38 (10.94) 0.107 6-12 months before infection 5.19 (7.95) 7.34 (9.17) 0.251 5.19 (7.94) 6.38 (8.52) 0.144 Body Mass Index (BMI) n(%) BMI, mean (SD) 35.91 (9.30) 29.64 (8.26) 0.712 35.91 (9.27) 35.90 (11.05) 0.000 18.5 to 24.9 < 10 1,651 (0.18) 0.483 < 10 14 (0.06) 0.120 25.0 to 29.9 < 29 1,849 (0.20) 0.233 < 29 29 (0.12) 0.007 >=30.0 94 (0.38) 2,907 (0.31) 0.148 94 (0.38) 88 (0.35) 0.055 Missing 117 (0.47) 2,654 (0.28) 0.399 117 (0.47) 117 (0.47) 0.000 Medications prior to infection* Serotonin reuptake inhibitors 18 (0.07) 1,197 (0.13) 0.183 18 (0.07) 18 (0.07) 0.000 Angiotensin converting enzyme inhibitors 21 (0.08) 592 (0.06) 0.083 21 (0.08) 21 (0.08) 0.000 Angiotensin receptor blockers 18 (0.07) 552 (0.06) 0.056 18 (0.07) 18 (0.07) 0.000 Statins 18 (0.07) 1,065 (0.11) 0.140 18 (0.07) 18 (0.07) 0.000 Anticoagulant 14 (0.06) 492 (0.05) 0.018 14 (0.06) 14 (0.06) 0.000 Aspirin 11 (0.04) 414 (0.04) 0.002 11 (0.04) 11 (0.04) 0.000 Steroids 61 (0.25) 2,776 (0.29) 0.110 61 (0.25) 61 (0.25) 0.000 GLP-1 RA < 10 44 (0.01) 0.140 < 10 < 10 0.000 Outpatient insulin 14 (0.06) 73 (0.01) 0.279 14 (0.06) 14 (0.06) 0.000 SGLT-2 inhibitor < 10 12 (0.00) 0.054 < 10 < 10 0.001 Clinical conditions prior to infection Coronary artery disease 15 (0.06) 564 (0.06) 0.002 15 (0.06) 15 (0.06) 0.000 Cancer 12 (0.05) 660 (0.07) 0.092 12 (0.05) 12 (0.05) 0.000 Chronic kidney disease, Stage 1-3 < 10 341 (0.04) 0.126 < 10 < 10 0.000 Heart Failure 10 (0.04) 342 (0.04) 0.021 10 (0.04) 10 (0.04) 0.000 Hypertension 71 (0.29) 2,850 (0.30) 0.036 71 (0.29) 71 (0.29) 0.000 Metabolic liver disease 10 (0.04) 315 (0.03) 0.036 < 10 < 10 0.000 Insomnia < 10 308 (0.03) 0.003 < 10 < 10 0.000 Depression 21 (0.08) 1,295 (0.14) 0.169 21 (0.08) 21 (0.08) 0.000 Asthma 23 (0.09) 1,970 (0.21) 0.330 23 (0.09) 23 (0.09) 0.000 COPD < 10 511 (0.05) 0.087 < 10 < 10 0.000 Last Hemoglobin A1c, mean (SD) 7.09 (2.39) 5.39 (0.52) 0.983 7.09 (2.37) 7.09 (2.76) 0.000 Creatinine, mean (SD) 0.92 (0.57) 0.93 (0.72) 0.011 0.92 (0.57) 0.92 (0.41) 0.000 COVID Variant Epoch (not used for weighting) Ancestral (Mar – Sept 2020) 21 (0.08) 391 (0.04) 0.178 21 (0.08) 21 (0.08) 0.001 Alpha (Oct 2020 – May 2021) 70 (0.28) 1,651 (0.18) 0.256 70 (0.28) 49 (0.20) 0.203 Delta (June 2021 – Nov 2021) 34 (0.14) 1,691 (0.18) 0.117 34 (0.14) 49 (0.20) 0.162 Omicron (starting Dec 2021) 123 (0.50) 5,664 (0.60) 0.214 123 (0.50) 129 (0.52) 0.051 Days from infection to treatment (not used for weighting; >5 days not shown) Comparator 0 < 10 5,751 (0.61) 1.557 < 10 127 (0.51) 1.257 1 to 3 < 10 2,165 (0.23) 0.613 < 10 67 (0.27) 0.703 3 to 5 < 10 1,018 (0.11) 0.438 < 10 35 (0.14) 0.521 Metformin 0 94 (0.38) 0 (0.00) 1.105 94 (0.38) 0.0 (0.00) 1.105 1 to 3 85 (0.34) 0 (0.00) 1.021 85 (0.34) 0.0 (0.00) 1.021 3 to 5 43 (0.17) < 10 0.647 43 (0.17) < 10 0.625 Abbreviations: SMD = standard mean difference, absolute; COPD = chronic obstructive pulmonary disease. *Not shown: other diabetes medications, which were all <10 before weighting for both cohorts. The full list of all variables used in weighting is in supplemental tables 3 and 4. Figure 3. Cumulative incidence curves for the outcome of Long Covid/Death in the 180 days after a metformin prescription (blue) or control prescription (red). Panel A is the cohort who received a prescription on days 0 to 1 from infection; Panel B is 0 to 6 from infection; and Panel C is 0 to 14 from infection. The increase in outcomes at day 90 reflects the computable phenotype, which cannot be computed until 90 days after infection. Per data use requirements in the National Covid Cohort Collaborative (N3C), the number at risk is not shown for the 0 to 1 cohort, nor for the starting cohort sizes for the 0 to 6 and 0 to 14, due to small cell sizes or the ability to calculate small cell sizes. Figure 4: This is a forest plot showing metformin use versus control use and the development of Long Covid/Death by day 180. The black squares represent risk ratios (RR), and the lines represent 95% confidence intervals. The number of outcomes, denominators, and exact percentages are sometimes omitted so that cell sizes <10 are not able to be calculated, in accordance with the data use requirements in the National Covid Cohort Collaborative (N3C). The Pre-Omicron Era is larger than the Omicron era in both the metformin and control cohorts, but the denominator for the day 0-6 sample is not shown because of cell sizes <10. The subgroup of those with new nirmatrelvir-ritonavir use was too small to analyze, so we present only the subgroup who did not also receive a prescription for nirmatrelvir-ritonavir. IRB The N3C data transfer to NCATS is performed under a Johns Hopkins University Reliance Protocol # IRB00249128 or individual site agreements with NIH. The N3C Data Enclave is managed under the authority of the NIH; information can be found at https://ncats.nih.gov/n3c/resources. Funding: This research was supported by the Intramural/Extramural research program of the National Center for Advancing Translational Sciences (NCATS), N3C Public Health Study Initiative, PHASTR Project: [RP-C06B65]; and K23 DK124654 from the National Institute of Diabetes and Digestive and Kidney Diseases of the NIH. Dr. Buse was funded by the NCATS UM1TR004406. Dr. Johnson, Dr. Bramante, Dr. Huling and Ms. Wiggen were funded by the National Heart, Lung and Blood Institute OT2HL16184701. Dr. Stürmer received investigator-initiated research funding and support as Principal Investigator (R01 AG056479) from the National Institute on Aging (NIA), as Director of Comparative Effectiveness Research (CER), NC TraCS Institute, UNC Clinical and Translational Science Award (UM1TR004406), and as co-Director of the Human Studies Consultation Core, NC Diabetes Research Center (P30DK124723). References: 1. Davis HE, McCorkell L, Vogel JM, Topol EJ. Long COVID: major findings, mechanisms and recommendations. Nat Rev Microbiol 2023: 1-14. 2. II R-GL. Summer COVID bump intensifies in L.A. and California, fueled by FLiRT variants. LA Times. 2024. 3. CDC. Centers for Disease Control and Prevention. COVID Data Tracker: U.S. Department of Health and Human Services, CDC. . Atlanta, GA, 2024. 4. Davis HE, McCorkell L, Vogel JM, Topol EJ. Long COVID: major findings, mechanisms and recommendations. Nature Reviews Microbiology 2023; 21 (3): 133-46. 5. Al-Aly Z, Rosen CJ. Long Covid and Impaired Cognition - More Evidence and More Work to Do. N Engl J Med 2024; 390 (9): 858-60. 6. Brode WM. Long COVID: The Enduring Pandemic. Annals of Internal Medicine 2024; 177 (7): 978-9. 7. Hampshire A, Azor A, Atchison C, et al. Cognition and Memory after Covid-19 in a Large Community Sample. N Engl J Med 2024; 390 (9): 806-18. 8. Leng A, Shah M, Ahmad SA, et al. Pathogenesis Underlying Neurological Manifestations of Long COVID Syndrome and Potential Therapeutics. Cells 2023; 12 (5). 9. Wong AC, Devason AS, Umana IC, et al. Serotonin reduction in post-acute sequelae of viral infection. Cell 2023; 186 (22): 4851-67.e20. 10. WHO. Post COVID-19 Condition (Long Covid). 2022. https://www.who.int/europe/news-room/fact-sheets/item/post-covid-19-condition. 11. Hair NL, Urban C. Association of Severe COVID-19 and Persistent COVID-19 Symptoms With Economic Hardship Among US Families. JAMA Network Open 2023; 6 (12): e2347318-e. 12. Gordon DE, Jang GM, Bouhaddou M, et al. A SARS-CoV-2 protein interaction map reveals targets for drug repurposing. Nature 2020; 583 (7816): 459-68. 13. Parthasarathy H, Tandel D, Siddiqui AH, Harshan KH. Metformin suppresses SARS-CoV-2 in cell culture. Virus Res 2022; 323 : 199010. 14. Ventura-López C, Cervantes-Luevano K, Aguirre-Sánchez JS, et al. Treatment with metformin glycinate reduces SARS-CoV-2 viral load: An in vitro model and randomized, double-blind, Phase IIb clinical trial. Biomed Pharmacother 2022; 152 : 113223. 15. Bramante CT, Huling JD, Tignanelli CJ, et al. Randomized Trial of Metformin, Ivermectin, and Fluvoxamine for Covid-19. N Engl J Med 2022; 387 (7): 599-610. 16. Bramante CT, Buse JB, Liebovitz DM, et al. 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. The Lancet Infectious Diseases 2023. 17. Bramante CT, Beckman KB, Mehta T, et al. Favorable Antiviral Effect of Metformin on Severe Acute Respiratory Syndrome Coronavirus 2 Viral Load in a Randomized, Placebo-Controlled Clinical Trial of Coronavirus Disease 2019. Clin Infect Dis 2024. 18. Chen Y, Yang D, Cheng B, et al. Clinical Characteristics and Outcomes of Patients With Diabetes and COVID-19 in Association With Glucose-Lowering Medication. Diabetes Care 2020; 43 (7): 1399-407. 19. Nafakhi H, Alareedh M, Al-Buthabhak K, Shaghee F, Nafakhi A, Kasim S. Predictors of adverse in-hospital outcome and recovery in patients with diabetes mellitus and COVID-19 pneumonia in Iraq. Diabetes Metab Syndr 2021; 15 (1): 33-8. 20. Xian H, Liu Y, Rundberg Nilsson A, et al. Metformin inhibition of mitochondrial ATP and DNA synthesis abrogates NLRP3 inflammasome activation and pulmonary inflammation. Immunity 2021; 54 (7): 1463-77.e11. 21. Scheen AJ. Metformin and COVID-19: From cellular mechanisms to reduced mortality. Diabetes Metab 2020; 46 (6): 423-6. 22. Zumla A, Hui DS, Azhar EI, Memish ZA, Maeurer M. Reducing mortality from 2019-nCoV: host-directed therapies should be an option. Lancet 2020; 395 (10224): e35-e6. 23. Crouse AB, Grimes T, Li P, Might M, Ovalle F, Shalev A. Metformin Use Is Associated With Reduced Mortality in a Diverse Population With COVID-19 and Diabetes. Frontiers in Endocrinology 2021; 11 (1081). 24. Cariou B, Hadjadj S, Wargny M, et al. Phenotypic characteristics and prognosis of inpatients with COVID-19 and diabetes: the CORONADO study. Diabetologia 2020; 63 (8): 1500-15. 25. Hariyanto TI, Kurniawan A. Metformin use is associated with reduced mortality rate from coronavirus disease 2019 (COVID-19) infection. Obes Med 2020; 19 : 100290. 26. Crouse AB, Grimes T, Li P, Might M, Ovalle F, Shalev A. METFORMIN USE IS ASSOCIATED WITH REDUCED MORTALITY IN A DIVERSE POPULATION WITH COVID-19 AND DIABETES. medRxiv 2020: 2020.07.29.20164020. 27. Ioannou GN, Berry K, Rajeevan N, et al. Effectiveness of Nirmatrelvir–Ritonavir Against the Development of Post–COVID-19 Conditions Among U.S. Veterans. Annals of Internal Medicine 2023; 176 (11): 1486-97. 28. Wong CKH, Lau JJ, Au ICH, et al. Optimal timing of nirmatrelvir/ritonavir treatment after COVID-19 symptom onset or diagnosis: target trial emulation. Nature Communications 2023; 14 (1): 8377. 29. Wan EYF, Yan VKC, Mok AHY, et al. Effectiveness of Molnupiravir and Nirmatrelvir–Ritonavir in Hospitalized Patients With COVID-19. Annals of Internal Medicine 2023; 176 (4): 505-14. 30. Lund JL, Richardson DB, Stürmer T. The active comparator, new user study design in pharmacoepidemiology: historical foundations and contemporary application. Curr Epidemiol Rep 2015; 2 (4): 221-8. 31. Laursen DRT, Hansen C, Paludan-Müller AS, Hróbjartsson A. Active placebo versus standard placebo control interventions in pharmacological randomised trials. Cochrane Database Syst Rev 2020; 2020 (7). 32. Tang H, Lu Y, Donahoo WT, et al. Glucagon-Like Peptide-1 Receptor Agonists and Risk for Suicidal Ideation and Behaviors in U.S. Older Adults With Type 2 Diabetes. Annals of Internal Medicine 2024; 177 (8): 1004-15. 33. Vollert J, Cook NR, Kaptchuk TJ, Sehra ST, Tobias DK, Hall KT. Assessment of Placebo Response in Objective and Subjective Outcome Measures in Rheumatoid Arthritis Clinical Trials. JAMA network open 2020; 3 (9): e2013196. 34. Zunhammer M, Spisák T, Wager TD, Bingel U. Meta-analysis of neural systems underlying placebo analgesia from individual participant fMRI data. Nat Commun 2021; 12 (1): 1391. 35. Pfaff ER, Girvin AT, Bennett TD, et al. Identifying who has long COVID in the USA: a machine learning approach using N3C data. The Lancet Digital Health 2022; 4 (7): e532-e41. 36. Haendel MA, Chute CG, Bennett TD, et al. The National COVID Cohort Collaborative (N3C): Rationale, design, infrastructure, and deployment. Journal of the American Medical Informatics Association 2021; 28 (3): 427-43. 37. Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. Am J Epidemiol 2016; 183 (8): 758-64. 38. Hernán MA, Sauer BC, Hernández-Díaz S, Platt R, Shrier I. Specifying a target trial prevents immortal time bias and other self-inflicted injuries in observational analyses. J Clin Epidemiol 2016; 79 : 70-5. 39. Hubbard RA, Gatsonis CA, Hogan JW, Hunter DJ, Normand ST, Troxel AB. ”Target Trial Emulation” for Observational Studies - Potential and Pitfalls. N Engl J Med 2024; 391 (21): 1975-7. 40. Pfaff ER, Girvin AT, Bennett TD, et al. Identifying who has long COVID in the USA: a machine learning approach using N3C data. Lancet Digit Health 2022; 4 (7): e532-e41. 41. Bennett TD, Moffitt RA, Hajagos JG, et al. Clinical Characterization and Prediction of Clinical Severity of SARS-CoV-2 Infection Among US Adults Using Data From the US National COVID Cohort Collaborative. JAMA network open 2021; 4 (7): e2116901-e. 42. Vaira LA, Hopkins C, Petrocelli M, et al. Efficacy of corticosteroid therapy in the treatment of long- lasting olfactory disorders in COVID-19 patients. Rhinology 2021; 59 (1): 21-5. 43. Bisgaard H, Flores-Nunez A, Goh A, et al. Study of montelukast for the treatment of respiratory symptoms of post-respiratory syncytial virus bronchiolitis in children. Am J Respir Crit Care Med 2008; 178 (8): 854-60. 44. Wong R. Diabetes and Long Covid in the N3C. . In preparation . 45. Harding Jea. Diabetes as a Risk Factor for Long-COVID-19—A Scoping Review. 2024. 46. Seidman MD, Gurgel RK, Lin SY, et al. Clinical Practice Guideline: Allergic Rhinitis. Otolaryngology–Head and Neck Surgery 2015; 152 (1_suppl): S1-S43. 47. Dykewicz MS, Wallace DV, Amrol DJ, et al. Rhinitis 2020: A practice parameter update. Journal of Allergy and Clinical Immunology 2020; 146 (4): 721-67. 48. George CM, Brujin LL, Will K, Howard-Thompson A. Management of Blood Glucose with Noninsulin Therapies in Type 2 Diabetes. Am Fam Physician 2015; 92 (1): 27-34. 49. Zhu Y, Hubbard RA, Chubak J, Roy J, Mitra N. Core concepts in pharmacoepidemiology: Violations of the positivity assumption in the causal analysis of observational data: Consequences and statistical approaches. Pharmacoepidemiol Drug Saf 2021; 30 (11): 1471-85. 50. McCarthy MW, Naggie S, Boulware DR, et al. Effect of Fluvoxamine vs Placebo on Time to Sustained Recovery in Outpatients With Mild to Moderate COVID-19: A Randomized Clinical Trial. Jama 2023; 329 (4): 296-305. 51. Ely EW, Brown Lisa M, Fineberg Harvey V. Long Covid Defined. New England Journal of Medicine ; 0 (0). 52. Pfaff ER, Madlock-Brown C, Baratta JM, et al. Coding long COVID: characterizing a new disease through an ICD-10 lens. BMC Med 2023; 21 (1): 58. 53. N G. WeightIt: Weighting for Covariate Balance in Observational Studies. R package version 0.14.2.9003. 2023; https://github.com/ngreifer/WeightIt , https://ngreifer.github.io/WeightIt/. 54. Hainmueller J. Entropy Balancing for Causal Effects: A Multivariate Reweighting Method to Produce Balanced Samples in Observational Studies. Political Analysis 2012; 20 (1): 25-46. 55. Zhao Q, Percival D. Entropy Balancing is Doubly Robust. 2017; 5 (1). 56. Hadley E, Yoo YJ, Patel S, et al. Insights from an N3C RECOVER EHR-based cohort study characterizing SARS-CoV-2 reinfections and Long COVID. Communications Medicine 2024; 4 (1): 129. 57. Sidky H, Hansen KA, Girvin AT, et al. Assessing the effect of selective serotonin reuptake inhibitors in the prevention of post-acute sequelae of COVID-19. Comput Struct Biotechnol J 2024; 24 : 115-25. 58. Chaichana U, Man KK, Ju C, Makaronidis J, Wei L. Effect of metformin on the risk of post-COVID-19 condition among individuals with overweight and obese: a population-based retrospective cohort study. Clinical Infectious Diseases 2025; In press . 59. Thaweethai T, Jolley SE, Karlson EW, et al. Development of a Definition of Postacute Sequelae of SARS-CoV-2 Infection. JAMA 2023; 329 (22): 1934-46. 60. Xie Y, Choi T, Al-Aly Z. Postacute Sequelae of SARS-CoV-2 Infection in the Pre-Delta, Delta, and Omicron Eras. New England Journal of Medicine 2024; 391 (6): 515-25. 61. Rigo S, Barbic F, Khalaf K, et al. The Long-COVID autonomic syndrome in hospitalized patients: A one-year prospective cohort study. European Journal of Internal Medicine 2024; 120 : 38-45. 62. Mateu L, Tebe C, Loste C, et al. Determinants of the onset and prognosis of the post-COVID-19 condition: a 2-year prospective observational cohort study. Lancet Reg Health Eur 2023; 33 : 100724. Information & Authors Information Version history V1 Version 1 26 November 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords coronavirus economic reason immune responses inflammation social science virus classification Authors Affiliations Carolyn Bramante 0000-0001-5858-2080 [email protected] University of Minnesota Medical School View all articles by this author John B Buse 0000-0002-9204-7177 The University of North Carolina at Chapel Hill Gillings School of Global Public Health View all articles by this author Jared D. Huling University of Minnesota Twin Cities Division of Biostatistics View all articles by this author John Buse B The University of North Carolina at Chapel Hill School of Medicine View all articles by this author Christopher Lindsell Duke University Department of Biostatistics and Bioinformatics View all articles by this author Thomas Stewart 0000-0002-5138-0758 University of Virginia View all articles by this author Russell L. Rothman Vanderbilt Institute for Medicine and Public Health View all articles by this author David Sahner National Center for Advancing Translational Sciences View all articles by this author Sarah E. Dunsmore National Center for Advancing Translational Sciences View all articles by this author Eric Topol The Scripps Research Institute Skaggs Graduate School of Chemical and Biological Sciences View all articles by this author Talia D. Wiggen University of Minnesota Twin Cities Institute for Health Informatics View all articles by this author Steve Makkar National Center for Advancing Translational Sciences View all articles by this author Andrew Toler National Center for Advancing Translational Sciences View all articles by this author Taylor Estepp National Center for Advancing Translational Sciences View all articles by this author Steven Johnson University of Minnesota Twin Cities Institute for Health Informatics View all articles by this author Metrics & Citations Metrics Article Usage 4395 views 379 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Carolyn Bramante, John B Buse, Jared D. Huling, et al. Metformin at the time of Covid-19 infection and risk of Long Covid: A Target Trial Emulation Study. Authorea . 26 November 2025. DOI: https://doi.org/10.22541/au.176418398.81170401/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(\".js__slcInclude\").on(\"change\", function(e){ if ($(this).val() == 'refworks') $('#direct').prop(\"checked\", false); $('#direct').prop(\"disabled\", ($(this).val() == 'refworks')); }); Cited by Razaz A. Felemban, Talha S. Mayet, Zeyad O. Khaimi, Yazeed H. Aljabri, Mohammed A. Hadi, Abdulrahman F. Alsulami, Alaa Althubaiti, Duaa Babaer, The length of hospital stay in COVID-19 patients using pre-hospital metformin at a Saudi tertiary hospital: a retrospective cohort study, BMC Infectious Diseases, 26 , 1, (2026). https://doi.org/10.1186/s12879-026-13170-4 Crossref Loading... View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {\"doi\":\"10.22541/au.176418398.81170401/v1\",\"type\":\"Article\"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob(\"bG9jYXRpb24=\"),_bnb=atob(\"b3JpZ2lu\"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(\" \")); $.get(\"/resource/lodash?t=\"+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML=\"window.__CF$cv$params={r:'a008232fbb2f8650',t:'MTc3OTU4MjA1Ng=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);\";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();","source_license":"CC-BY-4.0","license_restricted":false}