Since publishing our systematic review of the effects of influenza vaccines in healthy children (Feb 26, p 773),1 we have completed our review of the published and unpublished safety evidence for these vaccines.2
We found only one safety study of inactivated vaccine in 35 children aged 12-28 months done nearly 30 years ago with an influenza B vaccine;3 all other safety studies of inactivated vaccine we found were in children aged 3 years or more. Three studies reported safety data of live vaccine in children aged 22 months or younger (combined denominator 827).4-6 Seven further studies included children younger than 2 years.7-13 Only one trial of live vaccine had serious adverse events as an outcome (measured up to 3 months after vaccine administration).8 The lack of reported trial safety data for inactivated vaccines in younger children is particularly surprising given that this vaccine is now recommended for healthy children aged 6 months and older in USA and Canada.1
We wrote to 15 first or corresponding authors or research group leaders of the 31 studies (30 randomised controlled trials [RCTs] and one cohort study) included in our review to enquire about any unpublished data. Some authors had published more than one study, and email addresses for two authors of four studies (one RCT and three cohort studies) could not be found. We received 12 replies (80% response rate), accounting for 27 (87%) of the studies included in our review.
One retrieved paper, by Bergen and colleagues,1,4 reported that more than 2500 medical adverse events had occurred in vaccine recipients and more than 1300 in placebo recipients, but fewer than half-ie, only those that were significantly associated with increased or decreased risk in vaccine recipients-were specified by diagnosis (table). We wrote to the corresponding author, Steven Black, asking for the data on the non-significant events to include in our meta-analysis. Black responded saying that he had the data we requested but would require clearance from the vaccine manufacturer, MedImmune, to send it. He later responded that MedImmune did not want to share the data.
(Table data missing)
Comparison of total medical adverse events recorded and total medical adverse events (by diagnosis) reported in study by Bergen et al 14
We requested a contact at MedImmune and wrote to Robert Walker explaining the need to include non-significant outcomes in the meta-analysis of a systematic review. Walker declined to let us have the data. Walker was contacted once again at a later date after two authors of other studies on live vaccine safety (Belshe and Gruber) informed us that all safety data from their studies2,4-7,15,16 was controlled by MedImmune. Walker responded in the same manner to this new request, stating that “MedImmune does not provide study data to outside parties”.
Although so far we have found no evidence to suggest a serious harmful effect of any of the vaccines in our review, we are concerned by our findings of limited clinical trial evidence for inactivated vaccines. In addition, the withholding of safety data for live attenuated vaccines makes it impossible to present a complete evidence base of their safety. Although a frequent practice, lack of reporting of non-significant outcomes raises the real possibility that our review may present a biased picture.17,18 Heterogeneity of outcome definition and reporting are additional problems in vaccine trials.19
An incomplete or fragmented evidence base could hinder identification of rare and serious adverse events. Given their rarity, these may not be identifiable from single studies but may require a complete data set, as was the case for intussusception and rotavirus vaccines.20
We believe all unpublished trial safety data should be readily accessible to both the regulatory bodies and the scientific community on request. Our evidence gives rise to a concern that lack of access to unreported data prevents published data being put into context and hinders full and independent review. This cannot be good for public confidence in these vaccines.
TJ owned shares in GlaxoSmithKline and received consultancy fees from Sanofi Synthelabo and Roche. All other authors have no conflict of interest.