Does Oseltamivir Really Reduce Complications of Influenza?

TO THE EDITOR—Public health agencies have already hailed Hernán and Lipsitch’s recent reanalysis of Roche data on the effects of oseltamivir (Tamiflu) [1] as confirmation [2] of the problematic Roche meta-analysis by Kaiser et al [3]. However, on the basis of the methodology used in their study, we are left unconvinced of claims that Roche’s clinical trials demonstrated that oseltamivir reduces complications of influenza.

First, can complications even be meta-analyzed? Oseltamivir’s US product label states, “Serious bacterial infections may begin with influenza-like symptoms or may coexist with or occur as complications during the course of influenza. Tamiflu has not been shown to prevent such complications” [4]. This Food and Drug Administration (FDA) statement in part reflects the fact that trials did not use standardized definitions of secondary complications (such as pneumonia) [3] and confirmatory testing was not done in all patients. We presume that the FDA had unrestricted access to the entire manufacturer dataset, not just the 11 trials included in Hernán and Lipsitch’s study.

Second, even if complications are meta-analyzed, what outcomes should be selected? Hernán and Lipsitch’s study only included pneumonia, bronchitis, and other lower respiratory tract infections in their combined outcome of complications. This is at odds with the a priori–specified definition of complications in the original oseltamivir clinical study reports. The prespecified outcome also included sinusitis and otitis media. This appears to be an example of selective reporting, or “cherry picking,” by which outcomes reported are chosen post-hoc, as was also done in the Kaiser meta-analysis [3].

We are also concerned that not all of the relevant trials were included in Hernán and Lipsitch’s meta-analysis, which they selected from a so-called “complete list” provided by Roche. In January 2011, we documented the existence of 3 trials not on Roche’s list: 1 trial of oseltamivir treatment that reported on complications of influenza in adults (ML16369) [5] and 2 pediatric trials (WP15979 and WP16137) [6]. We believe that independent analyses must independently reconstruct complete trial lists, not rely on manufacturers.

Finally, it was unclear whether any cross-checks were performed of trial details and summary measures against published studies or whether the (limited) trial data provided by Roche were simply accepted at face value by the authors. In particular, it is unclear whether inconsistencies were found and how they were resolved. This is important because we have found some inconsistencies between the more detailed clinical study reports and published articles [5].

We firmly believe that multiple, independent, thorough analyses of oseltamivir are a good thing. In 2009, such a prospect seemed possible after Roche promised to release full study reports to physicians and scientists [7]. Sadly, Roche has denied us these data, despite our repeated requests. Even more alarming is that Roche neither provided these data to Hernán and Lipsitch, nor did Hernán and Lipsitch request it (personal correspondence, Miguel Hernán, June 2011).