Eighteen studies (n=5,000) were included in the review.
Systemic vaccine reactions (11 trials).
The proportion of systemic reactions varied considerably between trial arms (range: 0 to 31.6%). No significant differences between AA-LIV and IIV were found in any of the 11 studies, and the pooled OR was 0.96 (95% CI: 0.74, 1.24). Between-trial heterogeneity did not occur. The funnel plot gave no indication of publication bias.
Systemic haemagglutination inhibition (HI) antibody response (8 trials).
For influenza A-H3N2, data from 7 of the 8 trials showed that in all but one comparison, IIV-induced antibodies exceeded the protective HI-threshold after vaccination in approximately twice as many patients as AA-LIV did; the pooled OR was 0.17 (95% CI: 0.08, 0.33) favouring IIV. Heterogeneity between trials was large (p<0.001), apparently caused by between-trial differences in the pre-vaccination state. To eliminate heterogeneity caused through vaccination status, a subgroup analysis of seronegative participants was performed. This yielded a pooled OR of 0.07 (95% CI: 0.04, 0.15), with no significant heterogeneity (p=0.919).
For influenza A-H1N1, data from 10 trials found the percentages of patients exceeding the protective threshold for AA-LIV and IIV were closer than for A-H3N2, but the pooled OR was 0.41 (95% CI: 0.26, 0.67), which significantly favoured IIV. To eliminate heterogeneity caused through vaccination status, a subgroup analysis of seronegative participants was performed. This yielded a pooled OR virtually identical to that of A-H3N2, namely 0.09 (95% CI: 0.02, 0.33), with no significant heterogeneity (p=0.200).
For influenza B, data from 3 trials favoured IIV (83.3% versus 20.5%) and the pooled OR was 0.06, (95% CI: 0.02, 0.19). Between-trial heterogeneity was not found.
Local IgA antibody response.
Data from 4 trials on A-H3N2 found a stronger IgA induction by AA-LIV than IIV. Approximately half (52.4%) of the AA-LIV recipients, and a quarter (23.9%) of the IIV recipients responded to influenza A-H3N2 (pooled OR 5.54, 95% CI: 2.48, 12.36). For A-H1N1, data from 4 trials showed that 52% of the AA-LIV recipients and 25.3% of the IIV recipients responded (pooled OR 4.61, 95% CI: 1.34, 15.85). For influenza B, data from 2 trials found that an IgA response occurred less frequently in both vaccine groups (AA-LIV, 20.7%; IIV, 7.4%), with an insignificant pooled OR of 2.46 (95% CI: 0.44, 13.75).
Culture-positive illness and vaccine efficacy.
For A-H3N2, 6 trials found an efficacy of 68.0% for AA-LIV and 78.2% for IIV; the pooled OR was 1.50 (95% CI: 0.80, 2.82). For A-H1N1, the result was 72.3% for AA-LIV and 72.5% for IIV, with a pooled OR of 1.03 (95% CI: 0.58, 1.82). For influenza B, the data from one trial yielded a vaccine efficacy of 100.0% and a OR of 1.00.
The robustness and sensitivity analysis showed these results to be stable. The funnel plot analysis did not indicate publication bias. Age did not apparently affect the OR values. Trials with a close match between vaccine strains and those with a mismatch showed similar OR-values, suggesting that neither vaccine offered an advantage in the case of antigenic drift.
Challenge success was greater with the experimental design (culture-positive illness in placebo group: 36.4 to 45.8%) than with the epidemic approach (1.8 to 6.8%; only in one trial was it greater, i.e. 25.6%). There were no significant differences between the OR values from either design. Vaccine efficacy was high for both vaccine types (50 to 100%), with the exception of two A-H1N1 trials performed in children and young adults previously unprimed for that subtype (16.7 to 49.4%).