In Summary, for those who took the neuraminidase inhibitors:

Benefits in NNT

  • 95% saw no benefit
  • 5.0% were helped by preventing 1 culture-positive influenza case
  • None were helped (preventing an influenza-like illness)
  • 1 in 20 were helped (preventing 1 culture-positive influenza case)

Harms in NNT

  • 14.3% were harmed by nausea
  • 2.7% were harmed by vomiting
  • 1 in 7 were harmed (nausea)
  • 1 in 37 were harmed (vomiting)

View As:

Source: JJefferson T, Jones M, Doshi P,Del Mar C. Neuraminidase inhibitors for preventing and treating influenza in healthy adults: systematic review and meta-analysis. BMJ 2009 Dec 8;339:b5106. doi: 10.1136/bmj.b5106

Jefferson T, Demicheli V, Di Pietrantonj C, Jones M, Rivetti D. Neuraminidase inhibitors for preventing and treating influenza in healthy adults. Cochrane Database of Systematic Reviews 2006, Issue 3. Art. No.: CD001265

Cooper NJ et al. Effectiveness of neuraminidase inhibitors in treatment and prevention of influenza A and B: systematic review and meta-analyses of randomised controlled trials BMJ 2003; 326: 1235-41.

Kaiser L et al. Impact of Oseltamivir treatment on influenza-related lower respiratory tract complications and hospitalizations. Arch Int Med 2003; 163: 1667-72.

Miller FG, Joffe S. Equipoise and the Dilemma of Randomized Controlled Trials. N Engl J Med 2011; 364:476-480.

Efficacy Endpoints: Symptomatic influenza, laboratory-confirmed influenza, time to symptom relief.

Harm Endpoints: Adverse effects including nausea, vomiting, and neuropsychiatric events. Level of evidence for harms is low due to poor reporting of harms in trials.

Narrative: Most trials of neuraminidase inhibitors (NI’s) in the prophylaxis and treatment of influenza are industry-sponsored and small. The Cochrane Review and the systematic review combine the data.

Prophylaxis: When NI's are given to someone during influenza season to prevent influenza, that person's chances of reporting flu symptoms (fever, cough, aches, etc) are the same whether they took an NI or whether they took a placebo. However, there was a small reduction in the chance of an illness during which tests found influenza virus: 8.8% placebo versus 4.0% NI (NNT=20).

Treatment: Influenza is mostly a self-limited respiratory illness that lasts 7 days on average. When one takes a NI for treatment of acute influenza the review finds an average of about 1 day of symptom relief. Unfortunately this only applies to those with proven influenza infection who take the drug within 48 hours of the first symptom. No high-risk groups were identified that might benefit more, though the studies may be underpowered to find such a benefit. There is no convincing data to suggest serious harms from taking NI’s, though there are some reports of increased neuropsychiatric events in children.

A critical question is "Do NI's reduce serious complications of influenza such as pneumonia or death?" While no data address mortality there is one meta-analysis examining pneumonia rates (Kaiser, 2003). It is authored by employees of the industry sponsor (Hoffman-LaRoche Pharmaceuticals) and includes selected trials from a proprietary company database. It is unclear why some are excluded, some data are from non-peer reviewed sources, and most endpoints have no concrete definition or gold standard. The Cochrane review update has excluded this trial as the data may be biased and is unavailable for public scrutiny; they conclude there is currently no good data to support the claim that NI's reduce influenza complications such as pneumonia, hospitalization, or mortality. There is a large amount of unpublished pharmaceutical trial data that may or may not be available in the future to add to this question.

Caveats: Published data is mostly industry-sponsored and therefore may be disproportionately positive; negative data (showing no benefit) may exist and not be available or published. It is also notable and unfortunate that there are no data addressing mortality. It would be helpful to know if hospitalized patients with influenza infections are benefited, harmed, or unaffected by NI’s. Given the 30,000-40,000 in-hospital deaths that occur each year secondary to influenza it would be important to have these data, and the subject pool for randomized trials is certainly available. We worry that such a study will never occur; while we believe evidentiary equipoise exists it is unlikely that clinical equipoise does*. Plainly, the existing data seems equivocal (evidentiary equipoise) but clinical consensus (clinical equipoise) will likely strongly favor the use of NI’s making the ethics of a randomized controlled trial controversial.



There are numerous studies showing a small symptom benefit to NI’s. The benefit seems clinically insignificant which is why many current health policy guidelines do not recommend NI’s for otherwise healthy adults. There is no evidence to support NI's as an agent to reduce morbidy/mortality for any group of adults, healthy or otherwise. Use in the setting of a pandemic remains unstudied, though theoretically may be helpful to prevent the spread of influenza via the drug’s mechanism of action (decreased viral shedding from infected cells).

For more information on recent controversies and opinions please explore www.BMJ.com, where many recent publications have addressed the evidentiary base for NI’s, and also Tom Jefferson's (Acute Respiratory Infections Group, Cochrane Collaboration) blog post.

Author: Joshua Quaas, MD

Published/Updated: January 26, 2010

  1. The Title Bar

    The title bar is color-coded with our overall recommendation.

    • Green: Benefits outweigh risks.
    • Yellow: Unclear risk/benefit profile.
    • Red: Benefits do not outweigh risks.
    • Black: Obvious harms, no clear benefits.
  2. Tip content...