Authors' reply to Drs J Atashili and L Kalilani

We were perhaps overly critical of the meta-analysis of Atashili et al. [1]. We did not mean to imply that their work was inaccurately conducted, rather that pooling such different trial designs is methodologically inappropriate. Case–control studies cannot be interpreted as providing evidence of influenza incidence as the cases are chosen because of their confirmed diagnosis. Further, the outcomes in Table 2 in their study do not match the pooled data in Fig. 1, which is simply a reporting error. Given the small number of studies available, the fact that pooled estimates are similar is not surprising. We maintain that only randomized trials that are adequately powered can provide evidence of effectiveness.