Heterogeneity of case definitions used in vaccine effectiveness studies--and its impact on meta-analysis.

In the literature, different definitions of clinical illness cases and observation period are used to assess influenza vaccine effectiveness, usually without addressing their actual specificity and sensitivity. These properties, however, have large implications on influenza vaccine effectiveness estimates in single trials and meta-analyses of trials. Re-arranging 30 trials (in 17 publications) with a total of 29,265 subjects according to specificity and sensitivity of their case definitions and observation periods, resulted in large differences between vaccine effectiveness estimates for various levels of specificity and sensitivity. For highly specific definitions the combined vaccine effectiveness estimate was 54%, but for highly sensible definitions only 11%. Such findings call for a cautious interpretation of combined vaccine effectiveness estimates. Internationally agreed case and observation period definitions would be desirable to facilitate future meta-analyses.