Sixteen studies from 14 different research reports. Seven of the studies consisted of multiple comparison groups with a total of 23 specific interventions. 10 studies were randomised controlled trials, 4 were pre-post intervention designs (no control group) and 2 controlled trials. The total number of subjects included in these trials = 70 835.
The Q test for heterogeneity was 95.29 (n = 16), p < 0.05. The overall average pooled odds ratio was 2.08 (95% CI: 1.80, 2.41) suggesting that influenza immunisation interventions do increase coverage rates (this value should be interpreted with caution due to the significant heterogeneity present).
The results were also analysed separately by intervention type. The heterogeneity present in these models was not reported and so these values should be interpreted with caution:
Patient focused intervention (n=6): OR=1.85 (95% CI 1.25, 2.75).
Provider focused interventions (n=4): OR = 2.06(1.70, 2.48).
Mixed interventions (n=6): OR 2.50 (1.75, 3.58).
Only 3 of the studies did not find a statistically significant odds ratios suggesting a benefit of the intervention.
The association of the immunisation rate of the comparison (control or pre-intervention) group with post-intervention coverage was analysed to investigate the existence of a ceiling effect. Studies were divided into 2 groups (8 studies in each group), low group with immunisation < 50%and high group with immunisation > = 50%. Effect size for high group was 1.58 (95% CI: 1.34, 1.86) and for the low group was 3.35 (2.56, 4.38)suggesting that the effect of the intervention is greater in those studies in which the control group has a vaccine uptake of < 50%. Difference in log ORs is statistically significant (p < 0.05).
Average random effect sizes were compared by study design, 10 RCTS compared to 6 other designs. The average OR for the RCT studies was lower (1.96, 95% CI: 1.54, 2.49) than for the other study designs (2.31, 95% CI:1.73, 3.08), difference was not statistically significant (p > 0.05).