Again from The Atlantic article (which you should just go read the entire thing).
As with vaccines, the scientific evidence for Tamiflu and Relenza is thin at best. In its general-information section, the CDC’s Web site tells readers that antiviral drugs can “make you feel better faster.” True, but not by much. On average, Tamiflu (which accounts for 85 to 90 percent of the flu antiviral-drug market) cuts the duration of flu symptoms by 24hours in otherwise healthy people. In exchange for a slightly shorter bout of illness, as many as one in five people taking Tamiflu will experience nausea and vomiting. About one in five children will have neuropsychiatric side effects, possibly including anxiety and suicidal behavior. In Japan, where Tamiflu is liberally prescribed, the drug may have been responsible for 50 deaths from cardiopulmonary arrest, from 2001 to 2007, according to Rokuro Hama, the chair of the Japan Institute of Pharmacovigilance.
Such side effects might be worth risking if the antivirals prevented serious complications of flu, such as pneumonia, hospitalization, and death. Roche Laboratories, the company licensed to manufacture and market Tamiflu, says its drug does just that. In two September2006 press releases, the company announced, “Tamiflu significantly reduces the risk of death from influenza: New data shows treatment was associated with more than a two third reduction in deaths,” and “Children with influenza [are] 53 percent less likely to contract pneumonia when treated with Tamiflu.” Once again cohort studies (the same kind of potentially biased research that led to the conclusion that flu vaccine cuts mortality by 50 percent) are behind these claims. Tamiflu costs $10 a pill. It is possible that people who take it are more likely to be insured and affluent, or at least middle-class, than those who do not, and a large body of evidence shows that the well-off nearly always fare better than the poor when stricken with an infectious disease, including flu. In both 2003 and 2009, reviews of randomized placebo-controlled studies found that the study populations simply weren’t large enough to answer the question: Does Tamiflu prevent pneumonia?
As late as this August, the company’s own Web site contained the following statement, which was written under the direction of the FDA: “Tamiflu has not been proven to have a positive impact on the potential consequences (such as hospitalizations, mortality, or economic impact) of seasonal, avian, or pandemic influenza.” An FDA spokesperson said recently that the agency is unaware of any data submitted by Roche that would support the claims in the company’s September 2006 news release about the drug’s reducing flu deaths.