Randomization. Randomization lists were prepared by a biostatistician (B.J.C.). Eligible study participants were randomly allocated to the TIV group or placebo group in the ratio 3:2 using a random number generator (R software). A block‐randomization sequence was generated with randomly permuted block sizes of 5, 10, and 15. More households were allocated to the TIV group to enhance the acceptability of the study to participants.
Blinding. Blinding of households and study nurses was achieved by identical repackaging of the TIV and placebo into numbered syringes by a trained nurse not involved in vaccine administration. A research assistant who had no access to the randomization list allocated unique numbers to participating households based on their order of attendance, and these were subsequently matched to vaccine packages. Allocation of TIV or placebo was concealed from participating households, study nurses, and laboratory staff and was only revealed to investigators after completion of follow‐up.
Sublingual cialis
Laboratory methods. All serum specimens were tested for antibody responses to the vaccine strains A/Brisbane/59/2007(H1N1), A/Brisbane/10/2007(H3N2), and B/Florida/4/2006‐like (Yamagata‐lineage) by hemagglutination inhibition (HAI) and for antibody responses to A/California/04/2009(H1N1) by viral microneutralization (VN) using standard methods. We chose to use VN tests rather than HAI tests for pandemic A/H1N1 after preliminary studies that showed that VN was more sensitive than HAI for the detection of antibody responses in pandemic A/H1N1 infections. The serum samples were tested in serial doubling dilutions from an initial dilution of 1/10. Nose and throat swab samples collected during home visits were tested by RT‐PCR for influenza A and B viruses. Additional technical details of the laboratory procedures are given in the Appendix, which appears only in the online version of the journal.
Statistical analysis. Rates of adverse reaction events experienced by subjects within 4 days after administration of TIV or placebo were compared with use of Fisher’s exact tests. To assess vaccine immunogenicity, pre‐ and post‐vaccination titers and ratios of pre‐ to post‐vaccination titers were compared using Wilcoxon signed‐rank tests. The proportions of study subjects with antibody titer 1:40 post‐vaccination were compared between the intervention group and the control group using χ2 tests.
Post‐season antibody titers were compared with mid‐season titers, which were in turn compared with post‐vaccination antibody titers (or baseline titers in household contacts), to determine serologic evidence of infection during the summer and winter influenza seasons, respectively. Rates of influenza infection determined by serological testing, RT‐PCR, and clinical illness were compared by χ2 tests and Fisher’s exact tests, and 95% confidence intervals (CIs) were obtained using the exact binomial method or the Wald approximation, where appropriate. In an analysis that was not specified in our study protocol (because the study was designed before the pandemic), multivariable logistic regression models were used to study risk of laboratory‐confirmed pandemic influenza infection adjusting for age, sex, receipt of TIV, laboratory‐confirmed seasonal influenza infection, and date of study completion. Infection with a specific influenza strain can lead to increases in antibody titers to other heterologous strains (ie, cross‐reactions), and we identified 15 individuals with 4‐fold or greater increases in antibody titers to >1 influenza strain during either the winter or summer seasons. We adjusted for cross‐reactions by classifying the most likely virus infection during a season based on RT‐PCR confirmation, where available, or otherwise by assuming that the infecting virus was that for which the geometric antibody titer increase was greatest.
All analyses of study outcomes were performed under the principle of intention‐to‐treat. We used multiple imputation with 10 imputations to account for a small amount of missing data. Statistical analyses were conducted in R, version 2.8.1 (R Development Core Team).