Podcast: Play in new window

Subscribe: RSS

Reference: Huang Y-N, et al. Early Antiviral Therapy in Pediatric Outpatients and Risk of Influenza-Related Hospitalization. Pediatrics. 2026

Date: September 29, 2026

Dr. Gabrielle Jasmine

Guest Skeptic: Dr. Gabrielle Jasmine is a pediatric emergency medicine fellow at Children’s National Hospital in Washington DC. Her research interests include health equity, social determinants of health, and medical education.

Case: A previously healthy 7-year-old boy presents to the emergency department (ED) with fever, cough, rhinorrhea, and decreased energy. His parents tell you that these symptoms have been going on for the past day and a half. Influenza A testing is performed and is positive. On your exam, the boy is mildly tachycardic with the fever but otherwise well appearing. Oxygen saturation is normal. He is not in any respiratory distress, and he is drinking an apple juice.

The parent asks you, “I remember feeling really miserable when I got the flu, and I heard that there are some kids who can get really sick from it. Are there any particular medications that can treat the virus?”

Background:

Influenza is common in children and, for many, is a self-limited illness. The clinical spectrum can range from a nonspecific febrile illness with upper respiratory symptoms to bronchiolitis, croup, gastrointestinal symptoms, and other relatively uncomplicated presentations. However, influenza can also cause serious complications, including pneumonia, myocarditis, myositis, rhabdomyolysis, secondary bacterial infections, and neurologic disease. The 2026–2027 American Academy of Pediatrics (AAP) technical report notes that neurologic complications occur in approximately 8% to 11% of children hospitalized with influenza and can include seizures, encephalitis, and influenza-associated encephalopathy or encephalitis (IAE).

One particularly devastating form is acute necrotizing encephalopathy (ANE), characterized by rapid neurologic deterioration and characteristic bilateral brain lesions, often involving the thalami. Although uncommon, the morbidity and mortality remind us that the clinically important question about influenza treatment is not simply whether we can make a child feel better a little sooner, but whether we can prevent progression to serious disease.

Antiviral medications are recommended by multiple professional organizations for children at higher risk for complications, including the CDC. Treatment can also be considered for otherwise healthy children when started early in the course of illness.

The most familiar of these medications is oseltamivir, or Tamiflu, which is the AAP’s antiviral of choice. It is an oral neuraminidase inhibitor that prevents newly formed influenza virions from being released from infected cells. Randomized trials in children suggest that oseltamivir may shorten symptoms by roughly a day and may decrease the risk of acute otitis media. What has been far less clear is whether it prevents the outcomes that matter even more to patients and clinicians, such as hospitalization, critical illness, or death.

Tamiflu also has a somewhat notorious and controversial history. Much of the original phase III evidence supporting oseltamivir was not initially published, leading Cochrane investigators and The BMJ to spend years seeking access to the complete clinical study reports from the manufacturer, Roche. When the fuller dataset was eventually examined, the benefits appeared more modest and adverse effects such as nausea and vomiting became more apparent. The saga has become a classic example of publication bias, selective reporting, and the importance of data transparency.

The controversy has not gone away. Recently, SGEM#518 reviewed a recent randomized trial of critically ill, predominantly adult patients with influenza requiring respiratory or cardiovascular organ support. In that study, oseltamivir not only did not improve survival, but it also showed a concerning signal of increased mortality.

Clinical Question: In children with laboratory-confirmed influenza presenting for outpatient or emergency care, does starting antiviral therapy within 48 hours of symptom onset lower the likelihood of influenza-related hospitalization compared with delayed or no antiviral therapy?

Reference: Huang Y-N, et al. Early Antiviral Therapy in Pediatric Outpatients and Risk of Influenza-Related Hospitalization. Pediatrics. 2026

  • Population: Children younger than 18 years with laboratory-confirmed influenza diagnosed between January 1, 2020, and October 31, 2023, in northern Taiwan. Influenza was confirmed by rapid influenza diagnostic testing (RIDT), PCR, or viral culture.
    • Excluded: Patients without lab confirmation of influenza or with diagnoses made >14 days after hospital admission or with hospitalization unrelated to influenza
  • Exposure: Early antiviral therapy defined as treatment initiated within 48 hours of symptom onset.
  • Comparison: Antiviral treatment initiated more than 48 hours after symptom onset or no antiviral treatment.
  • Outcome:
    • Co-Primary Outcome: Influenza-related hospitalization or death within 14 days of symptom onset.
    • Secondary Outcomes: Influenza-related severe illness, including: Intensive care unit (ICU) admission and/or mechanical ventilation, acute necrotizing encephalopathy with long-term neurologic sequelae
  • Trial: Multicentre, retrospective, matched case-control observational study.

Authors’ Conclusions: Early antiviral therapy in pediatric outpatients with influenza substantially reduces the risk of subsequent severe illness and hospitalization, supporting prompt initiation of treatment in this population.

Quality Checklist for Observational Study:

  1. Did the study address a clearly focused issue? Yes
  2. Did the authors use an appropriate method to answer their question? Yes
  3. Were the cases selected in an acceptable way? Yes.
  4. Were the controls recruited in an acceptable way? Yes
  5. Was the exposure accurately measured to minimize bias? Unsure
  6. Aside from the exposure, were groups treated equally? No
  7. Have the authors identified all-important confounding factors? Unsure
  8. Was the exposure effect large? Yes
  9. Was the estimate precise? Yes
  10. Do you believe the results? Unsure
  11. Can the results to applied to the local population? Unsure
  12. Do the results fit with other available evidence. Unsure
  13. Conflicts of Interest: No
  14. Funding of the Study: The study was supported by the Taiwan Ministry of Education and Taiwan National Science and Technology Council. The authors reported that the funders had no role in study design, analysis, interpretation, manuscript preparation, or the decision to publish.

Results: They included 1,492 children (354 hospitalized cases and 1,138 non-hospitalized controls). Mean age was 7.1 years (SD, 4.6), with slightly less than half (44%) being girls. Most (87%) were diagnosed with influenza A and treated with oral oseltamivir (56%).

When it came to antiviral therapy: 81% received early therapy, 18% received late antiviral therapy, and 1% did not receive it at all.

Key Results: Early antiviral therapy was associated with lower odds of influenza-related hospitalization.

Early antiviral therapy was associated with lower rates of hospitalization (16.1% vs 55.9%) and influenza-related severe illness (0.7% vs 3.1%) compared to those who received late or no antiviral therapy.

Antiviral therapy was associated with 81% lower odds of hospitalization (aOR, 0.19, 95% CI 0.14-0.27).

Not All Influenza Tests Are Created Equal

There is a bit of diagnostic alphabet soup here. Influenza could be confirmed by rapid influenza diagnostic test (RIDT), PCR, or viral culture, but these tests do not have equivalent sensitivity and may identify different spectra of influenza illness. As Kohn et al. remind us, an imperfect reference standard can misclassify true disease, and using different reference standards can introduce differential verification and spectrum effects. In this study, a false-negative test would do more than mislabel a patient; it could exclude a true influenza case from the cohort altogether. If the choice and performance of the influenza test were associated with illness severity, timing of presentation, antiviral treatment, or subsequent hospitalization, this could produce selection bias or residual confounding in the observed association between early antivirals and hospitalization. Because the study does not establish that diagnostic ascertainment was independent of exposure and outcome, the direction and magnitude of this bias are uncertain.

Prescription ≠ Medication Administration

One potential area for misclassification bias is how the authors defined “early antiviral therapy.” They classified children as receiving early treatment if an antiviral was prescribed within 48 hours of symptom onset. They justify this by stating that, in Taiwan, “an early prescription is considered equivalent to early medication receipt because prescriptions are typically filled immediately at hospital-based or nearby community pharmacies.” There is no reference to support that assumption.

Even if we accept that prescriptions are usually filled immediately, dispensing a medication is not the same thing as taking it. We do not know whether the child actually received the first dose, how many doses were taken, or whether the prescribed course was completed. This is particularly relevant for oseltamivir because gastrointestinal adverse effects are common. A child could take one dose, vomit, and the family could decide not to give any more. That child would still be counted as having received “early antiviral therapy,” despite having minimal actual drug exposure.

That creates the possibility of misclassification of the exposure. If nonadherence occurred similarly among children who were and were not subsequently hospitalized, it would generally tend to dilute a true treatment effect toward the null. But adherence may not be random for example, children who were sicker, vomiting, or unable to tolerate oral medication could also have been more likely to require hospitalization so the direction of the bias is not necessarily predictable. The larger point is that this study really measured early antiviral prescribing, not confirmed early antiviral treatment.

 Which Came First: The Antiviral or the Admission?

 For treatment to prevent hospitalization, the exposure must occur before the outcome. But the authors explicitly included children who were diagnosed with influenza in the ED and hospitalized on that same day. They categorized treatment by whether it was given within 48 hours of symptom onset, rather than whether it was administered before the decision to hospitalize.

That makes the temporal relationship messy. Some children could theoretically be classified as receiving “early antiviral therapy” even though the drug was given during the encounter in which they were already being admitted. Similarly, children receiving late therapy may have received it specifically because they were deteriorating. This time-dependent treatment issue makes causal statements about antivirals preventing hospitalization much harder.

Odds are Not Risk

 The authors report an adjusted OR of 0.19 and translate this into an “81% lower risk of hospitalization.” This language is not correct. An odds ratio of 0.19 means approximately 81% lower odds of hospitalization.

Risk asks, “What proportion of patients had the outcome?”

Odds ask, “How many had the outcome compared with how many did not?”

A quick example:

  • If 20 out of 100 children are hospitalized, the risk is 20/100 = 20%.
  • The odds are 20 hospitalized compared with 80 not hospitalized: 20/80 = 25%

Odds ratios approximate relative risks reasonably well when outcomes are rare. Hospitalization was not particularly rare here: 354 of 1,492 children were hospitalized, and 55.9% of the late/no-treatment comparison group was hospitalized. When outcomes are this common, the OR can make the relative benefit look substantially larger than the corresponding risk ratio.

Because this is a case-control study, the odds ratio is the appropriate measure of association. Rather than saying early antivirals “reduced hospitalization risk by 81%,” the more accurate wording is: “Early antiviral treatment was associated with 81% lower adjusted odds of hospitalization.”

Confounding by Severity

The investigators did a lot of statistical adjustment. They matched on season and age, adjusted for comorbidities, vaccination, co-infection, antibiotics, corticosteroids, and measures of previous healthcare utilization, and performed multiple sensitivity analyses.

But what we really want to know is: How sick was the child when the decision to prescribe or not prescribe the antiviral was made?

We do not get adjustment for oxygen saturation, respiratory distress, dehydration, abnormal vital signs, appearance, or the clinician’s assessment of severity. One striking clue is antibiotic use. Children receiving antibiotics within two days had an adjusted odds ratio for hospitalization of 7.23. Antibiotics almost certainly did not cause that increase in hospitalization. They are probably acting as a marker for children who looked sicker.

The same phenomenon could influence antiviral timing. The authors appropriately calculate E-values and report a main-analysis E-value of 10.56. But an E-value tells us how strong an unmeasured confounder would need to be to explain an association under certain assumptions. It does not magically remove time-related bias, selection bias, misclassification, or inadequately measured severity.

This Isn’t Really Antiviral Versus No Antiviral

The headline question sounds like: Do antivirals prevent hospitalization? That is not really the comparison being made. Of 1,492 children, only 17 received no antiviral medication at all. The comparison group was overwhelmingly composed of children who received antivirals more than 48 hours after symptom onset. Interestingly, the unadjusted hospitalization rate was 23.5% in the untreated group but 58% among those receiving late antivirals.

This suggests that late antiviral treatment may function partly as a marker of worsening disease. A child becomes progressively sicker, comes back for another evaluation, gets started on an antiviral, and subsequently gets admitted. This study provides much stronger evidence of an association between treatment timing and hospitalization than it does of whether antiviral treatment itself prevents hospitalization.

Comment on Authors’ Conclusion Compared to SGEM Conclusion: The authors conclude that early antiviral therapy “substantially reduces” hospitalization and argue that the results support prompt antiviral treatment. We would make that conclusion more cautious.

This is a retrospective case control study demonstrating an association between antiviral therapy initiated within 48 hours and lower odds of hospitalization. However, there are multiple limitations we mentioned that make us question the strength of the conclusion.

SGEM Bottom Line: Antivirals for kids with influenza may be associated with fewer hospitalizations, but observational data can’t cure confounding. Don’t let an impressive association become a prescription for certainty.

Case Resolution:

The child is still within the first 48 hours of illness and engage in shared decision making with the family. You discuss the potential harms and benefits of antiviral therapy based on the child’s age, clinical situation, and existing recommendations. You state that there is some data suggesting earlier treatment may reduce the odds of subsequent hospitalization.

You discharge the well-appearing child with appropriate return precautions and follow-up.

Clinical Application:

This study should inform rather than dictate practice. Early antivirals were associated with fewer hospitalizations, but observational data cannot prove causation, and residual confounding and several different biases remain concerning.

What Do I Tell My Patient?

Some evidence suggests that starting influenza medicine early may shorten the illness, but it can also cause side effects, most commonly nausea and vomiting. At this time, it’s a bit unclear as to whether these medications can reduce the risk of being hospitalized or other bad outcomes. Let’s talk a bit more about what you and your child would prefer.

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

Other SGEM Episodes about Influenza:

  • SGEM#518: What Have You Done for Me Lately – Oseltamivir for Hospitalized 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)