Podcast: Play in new window

Subscribe: RSS

Date: August 23, 2026

Guest Skeptic: Dr. Roya Caloia is a former EM program director, longtime faculty at Henry Ford Genesys and current interim Associate Medical Director. She is also the Chairperson of the National Clinical Governance Board for US Acute Care Solutions

Case: A 62-year-old woman with diabetes and chronic lung disease presents to the emergency department (ED) with five days of fever, cough, myalgias and progressively worsening dyspnea. Her respiratory rate is 34 breaths per minute, oxygen saturation is 82% on room air, and chest imaging shows bilateral air-space disease. Molecular testing is positive for influenza A.

Despite supplemental oxygen, she requires high-flow nasal oxygen at 50 L/min with an FiO₂ of 0.60. Following initial fluids, her blood pressure remains low, and norepinephrine is started. She has received no more than one dose of oseltamivir before arrival. The ICU team is contacted, and the question is whether she should receive enteral oseltamivir for five days, ten days, or no influenza antiviral.

Background: Influenza season has a way of turning the ED waiting room into a coughing convention. Most patients feel miserable but recover with time and supportive care. The concern is the smaller group at higher risk of complications, or those already sick enough to require hospital admission. That is usually when someone asks the annual question: Should we give Tamiflu? 

Centers for Disease Control and Prevention (CDC) says:

  • “For hospitalized patients with suspected or confirmed influenza, initiation of antiviral treatment with oral or enterically administered oseltamivir is recommended as soon as possible. Antiviral treatment might be effective in reducing morbidity and mortality in hospitalized influenza patients, especially adults, even if treatment is started more than 48 hours after onset of illness.”

For those outside the United States, the World Health Organization (WHO) guideline specifically addresses severely ill and critically ill patients. The WHO guideline recommends oseltamivir for patients with severe influenza and even suggests considering extending treatment from 5 to 10 days in critically ill patients.

Oseltamivir, better known by the brand name Tamiflu, is an oral neuraminidase inhibitor active against influenza A and B. It blocks an enzyme the virus uses to release newly formed viral particles from infected cells and spread to other cells. Put another way, oseltamivir tries to stop the flu from going viral (dad joke). The usual adult treatment is 75 mg twice daily for five days, with dose adjustment for renal impairment. The regulatory indication focuses on acute, uncomplicated influenza when treatment starts within 48 hours, but clinical guidelines recommend starting it as soon as possible in hospitalized patients, even when they present later in the illness.

The challenge is that a biologically plausible mechanism does not automatically produce a patient-oriented outcome (POO). In otherwise uncomplicated influenza, oseltamivir appears to shorten symptoms by about 17 hours on average. That could mean feeling better Tuesday evening instead of Wednesday morning. It has not convincingly reduced hospital admission in outpatient trials, and it does increase nausea and vomiting. Headache and neuropsychiatric events have also been part of the safety discussion, although the causal relationship for the latter remains uncertain.

Tamiflu has also been at the centre of one of medicine’s biggest data-transparency battles [1,2].  Much of the phase III evidence supporting oseltamivir had not been published, and independent Cochrane reviewers spent years trying to obtain the full clinical study reports from Roche, the manufacturer. The BMJ’s Open Data campaign helped keep the pressure on until the reports were eventually released. When the Cochrane team examined the more complete record, the benefits appeared smaller and the harms more apparent [3]. This became a cautionary tale about publication bias and why evidence-based medicine requires access to all the evidence, not just the parts selected for publication.

This is not SGEM’s first rodeo when it comes to talking about oseltamivir. SGEM#98: Don’t Stand So Close to Me reviewed neuraminidase inhibitors and concluded that the small reduction in symptom duration needed to be weighed against adverse effects. SGEM#312: Oseltamivir is like Bad Medicine looked at primary-care patients with influenza-like illness and remained unconvinced that routine treatment offered a favourable benefit-to-harm ratio. SGEM#397: Give a Little Bit…of Oseltamivir examined hospitalized children and highlighted how much uncertainty remained because the evidence was observational. Finally, SGEM#409: Same as It Ever Was found no good evidence that routine oseltamivir prevented hospitalization among influenza-positive outpatients.

For emergency physicians, the more interesting question is not whether an otherwise healthy outpatient gets off the couch 17 hours earlier. It is whether oseltamivir changes outcomes that matter once influenza becomes severe. Much of the rationale for treating hospitalized patients has historically come from observational evidence, while clinical guidelines have continued to recommend early treatment. That is exactly the kind of gap where guideline momentum, biological plausibility, and patient-oriented evidence need to meet under the SGEM skeptical microscope.

Clinical Question: In critically ill patients aged 12 years or older with laboratory-confirmed influenza who require respiratory or cardiovascular organ support, does treatment with oseltamivir improve 90-day survival?

Reference: Murthy et al. Oseltamivir for Critically Ill Patients with Influenza: A Randomised Trial. SSRN: https://ssrn.com/abstract=7172531 or http://dx.doi.org/10.2139/ssrn.7172531

  • Population: Hospitalized patients aged 12 years or older with laboratory-confirmed influenza who were critically ill and receiving at least one of the following: High-flow nasal oxygen at ≥30 L/min and FiO₂ ≥0.40, non-invasive ventilation, invasive mechanical ventilation, or vasopressor or inotrope support
    • Exclusions: Patients who were expected to die within 24 hours and for whom there was no commitment to full support, had received sustained organ support for more than 48 hours, had already received more than one dose of oseltamivir, were planned to receive another influenza antiviral or were considered by the treating clinician not to be suitable for participation.
  • Intervention: Oseltamivir 75 mg enterally twice daily (with renal dose adjustment) assigned for either 5 or 10 days.
  • Comparison: No influenza antiviral treatment. However, some sites did not offer the no-antiviral option because of a lack of local clinical equipoise. Five-day oseltamivir had to remain an available option.
  • Outcome:
    • Primary Outcome: All-cause mortality at 90 days.
    • Secondary Outcomes: Survival time to 90 days, hospital and intensive care unit (ICU) length of stay (LOS), cardiovascular- and overall organ-support-free days, progression to invasive ventilation, ECMO or death, and serious adverse events.
  • Type of Study: A multicentre, international, open-label, multifactorial, Bayesian, response-adaptive platform randomized controlled trial.

Authors’ Conclusions: “Treatment with oseltamivir is ineffective and highly likely to increase 90-day mortality in critically ill patients with influenza.”

Quality Checklist for Randomized Clinical Trials: (Yes, No or Unsure)

  1. The study population included or focused on those in the emergency department. No
  2. The patients were adequately randomized. Yes
  3. The randomization process was concealed. Yes
  4. The patients were analyzed in the groups to which they were randomized. Yes
  5. The study patients were recruited consecutively (i.e. no selection bias). Unsure
  6. The patients in both groups were similar with respect to prognostic factors. No 
  7. All participants (patients, clinicians, outcome assessors) were unaware of group allocation. No
  8. All groups were treated equally except for the intervention. Unsure
  9. Follow-up was complete (i.e. at least 80% for both groups). Yes
  10. All patient-important outcomes were considered. Yes
  11. The treatment effect was large enough and precise enough to be clinically significant. The point estimate is clinically important, but the magnitude is imprecise. 
  12. Who funded the trial. The platform had numerous predominantly governmental, academic and philanthropic funders. No oseltamivir manufacturer was named as a funder. The investigators stated that funders had no role in study design, analysis or reporting.
  13. Did the authors declare conflicts of interest? We could not find the documents online

Results: The analysis included 442 participants: 162 assigned to five days of oseltamivir, 156 to ten days and 124 to no antiviral. They were recruited at 139 sites across 18 countries; only 79 sites offered the no-antiviral option. The Median age was 60 years in each oseltamivir group and 63 years in the no-antiviral group. Approximately 38% to 45% were female, and the median symptom duration before randomization was five days. Approximately 89% had influenza A. At randomization, 33% were receiving high-flow nasal oxygen, 15% non-invasive ventilation and 47% invasive ventilation. Almost half were receiving vasopressors in at least one treatment group. A total of 42% had received one dose of oseltamivir before randomization.

Key Result: In critically ill patients with laboratory-confirmed influenza requiring organ support, neither five nor ten days of oseltamivir improved survival compared with no antiviral treatment, and both treatment strategies raised a concerning signal of increased mortality.

  • Primary Outcome: Mortality by 90 days
    • Control 14%, Five-day 20%, and Ten-day 19%
    • Five days: OR 2.13; 95% credible interval 1.03–4.52
    • Ten days: OR 2.17; 95% credible interval 1.05–4.64
    • The posterior probabilities of harm were 98.0% and 98.2%, respectively. The probability of futility was 99.5% for each duration.
  • Secondary Outcomes: 

  1. Variable Site-Level Clinical Equipoise: Only 79 of 139 sites offered the no-antiviral arm. Sites that believed withholding oseltamivir was unacceptable could opt out, meaning the entire oseltamivir population was not drawn from the same site population as the controls. Differences in local practice, case mix, thresholds for organ support and background mortality could therefore become entangled with treatment assignment. The Bayesian model adjusted for site and eligibility, and the co-eligible-site sensitivity analysis produced a similar direction of effect, but the effect size was much smaller and the credible intervals were very wide. Randomization within a site was protected, but randomization does not automatically eliminate site-level selection or structural imbalance.
  1. Baseline Imbalance: The oseltamivir groups had more invasive ventilation, vasopressor use and higher APACHE II and cardiovascular SOFA scores. These factors strongly predict mortality. The investigators appropriately adjusted for severity and showed that analyses with different adjustment strategies generally pointed in the same direction. Nevertheless, when a moderate-sized randomized trial has relatively few deaths, covariate imbalance can materially change the treatment estimate. The substantial difference between the crude pooled risk increase of about 5.8 percentage points and the adjusted increase of 9.7 percentage points demonstrates how dependent the magnitude is on modelling.

  1. Open Label

    Open-Label & Unmeasured Co-Interventions: Patients and clinicians knew whether oseltamivir was being administered. Although 90-day mortality is objective, open-label assignment can influence antibiotic use, corticosteroid use, antifungal investigations, ventilation and weaning decisions, ECMO referral, goals-of-care discussions, and discharge practices. The investigators specifically acknowledge that they did not collect post-randomization concomitant medication use. Without those data, equal treatment apart from oseltamivir cannot be confirmed. Blinding is intended not only to prevent biased outcome assessment but also to prevent differential co-interventions and performance bias.

  1. Incomplete Separation: A total of 42% of participants had already received one oseltamivir dose before randomization, with rates of 50%, 41% and 32% across the five-day, ten-day and no-antiviral groups. After randomization, 9% of controls received at least one dose of oseltamivir. The assigned ten-day course also had a median actual duration of nine days, and therapy stopped at discharge. The intention-to-treat estimate therefore evaluates assignment to an oseltamivir strategy rather than a clean biological contrast between drug exposure and no exposure. Treatment contamination often dilutes differences toward the null, but unequal pre-randomization exposure and open-label crossover make the direction of bias less certain here.
  1. Adaptive Trial: This was not a simple three-arm trial. It involved response-adaptive allocation, hierarchical borrowing between treatment durations and severity states, adjustment using information from the broader platform, and Bayesian stopping thresholds. The comparison-specific statistical analysis plan was written after the inferiority trigger and DSMB recommendation to stop enrollment, although it was posted before the final analysis was conducted. The extensive sensitivity analyses are reassuring, but some estimates changed meaningfully: when the adjustment used only the confirmed influenza population, the credible intervals crossed 1.0. The individual primary credible intervals were also wide, and the adjusted risk-difference interval extended from essentially no effect to a very large increase in mortality. The direction of the result is reasonably consistent; the precise magnitude and certainty that oseltamivir caused the excess deaths are less secure.

Comment on Authors’ Conclusion Compared to SGEM Conclusion: We generally agree with the authors’ conclusions. 

SGEM Bottom Line: Oseltamivir should not be routinely administered to critically ill adult patients with confirmed seasonal influenza who are already receiving high-flow oxygen, non-invasive or invasive ventilation, or vasopressor support.

Case Resolution: The patient has confirmed influenza A, is approximately five days into symptoms and requires both high-flow nasal oxygen and vasopressor support. She closely resembles the trial population and has received no more than one prior dose of oseltamivir. You discuss the new randomized evidence with the ICU and infectious-diseases teams. Given the absence of demonstrated benefit and the concerning mortality signal, the team decides not to start a routine five- or ten-day course of oseltamivir. She is admitted to the ICU and receives respiratory and hemodynamic support, appropriate investigation and treatment for bacterial coinfection, venous thromboembolism prophylaxis, and other standard critical-care therapies.

Dr. Roya Caloia

Clinical Application: This RCT did not demonstrate a patient-oriented outcome (POO) of benefit from either five or ten days of therapy and identified a concerning signal for increased 90-day mortality. However, the exact magnitude of harm remains uncertain because this is an open-label adaptive platform trial with baseline imbalance, wide credible intervals and no peer review. For a critically ill patient who closely matches this trial, I would not initiate or continue oseltamivir automatically. I would discuss the evidence with the critical care team, infectious diseases consultants, pharmacists and the patient or substitute decision-maker. Because current CDC and WHO recommendations still advise oseltamivir for hospitalized or severe influenza, the disagreement between this RCT and existing guidelines should be explicitly documented and addressed through local multidisciplinary decision-making.

What Do I Tell the Patient?  You have severe influenza and need ICU-level support for your breathing and blood pressure. Until now, guidelines have usually recommended a medication called oseltamivir for patients this sick. A new trial looked specifically at people like you. It did not show any benefit and raised a serious concern about harm. Because you are already critically ill and several days into the infection, I do not think we should give it automatically. We will discuss this with the ICU and infectious-diseases teams, continue all the supportive treatments you need and aggressively look for treatable complications such as bacterial pneumonia.

Keener Kontest: Last episode’s winner was Dr. Steven Stelts. He knew the classic definition of massive transfusion is ≥10 units of red blood cells (RBCs) within 24 hours. 

Other FOAMed:

  • NEJM Voices: Oseltamivir Flops in the Critically Ill
  • PulmCrit: Is oseltamivir assassinating ICU patients?
  • SGEM#409: Same as it Ever Was – Tamiflu for Influenza?
  • SGEM #397: Give a Little Bit…of Oseltamivir to Pediatric Patients Admitted with Influenza
  • SGEM#312: Oseltamivir is like Bad Medicine – for Influenza
  • SGEM#98: Don’t Stand So Close to Me (You have the flu)

Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.

References:

  1. Payne D. Tamiflu: the battle for secret drug data. BMJ. 2012 Oct 29;345:e7303. doi: 10.1136/bmj.e7303. PMID: 23109482.
  2. Godlee F. Open letter to Roche about oseltamivir trial data. BMJ. 2012 Oct 29;345:e7305. doi: 10.1136/bmj.e7305. PMID: 23109484.
  3. Abbasi K. Editor’s Choice. The missing data that cost $20bn. BMJ 2014;348 doi: https://doi.org/10.1136/bmj.g2695