Table 4.
Effect of vaccine on febrile outbreaks caused by different influenza virus subtype/lineage, based on a two-level random effect zero-inflated negative binomial (RE-ZINB) regression model.
| Subtype/Lineage | Outbreaks | School with Outbreaks | Effect of High Vaccination Coverage (RR) | 95%CI | t | p |
|---|---|---|---|---|---|---|
| A(H3N2) | 152 | 114 | 0.78 | 0.20–3.10 | −0.38 | 0.71 |
| B(Victoria) | 221 | 144 | 0.40 | 0.24–0.67 | −3.8 | 0.002 |
| A(H1N1)pdm09 1 | 64 | 51 | 0.59 | 0.20–1.72 | −1.06 | 0.31 |
| Other 2 | 25 | 23 | - | - | - | - |
| Total 3 | 462 | 283 | 0.50 | 0.34–0.75 | −3.65 | 0.003 |
1 Because the two-level RE-ZINB model was not a convergence model, a two-level random effect zero-inflated Poisson regression (RE-ZIP) model was used to estimate parameters instead. 2 Because there were too few schools with other influenza subtypes/lineages, both the two-level RE-ZINB and RE-ZIP models were not convergence models. 3 There had been outbreaks caused by different subtypes/lineages in 49 schools.