IPD cases among children 2-59 months old were identified through the Active Bacterial Core surveillance system during 2001-2009. We used logistic regression to calculate the odds ratio of vaccination (versus no vaccination) among cases (PCV7-type IPD cases) and non-cases (non-PCV7-type IPD cases), controlling for the presence of underlying conditions. Vaccine effectiveness (VE) was calculated as one minus the adjusted odds ratio.
Among 4225 IPD cases reported during 2001-2009, 2680 (63 % ) had serotype information and vaccine history. Effectiveness of ? dose of PCV7 against PCV7-types was 88 % (95 % confidence interval (CI) 78-94 % ) among children with comorbid conditions and 97 % (95 % CI 92-98 % ) among healthy children. Among healthy children, VE was higher in 2001-2003 (97 % , 95 % CI 95-98 % ) compared to 2004-2009 (81 % , 95 % CI 64-90 % ). The annual estimates of VE in 2004-2009 showed great variability and wide confidence intervals due to the small number of PCV7-type cases.
An indirect cohort design using IPD surveillance data confirms the findings of the case-control study and, therefore, appears suitable for estimating PCV7effectiveness. This method would be most useful shortly after vaccine introduction, and less useful in a setting of very high vaccine coverage and fewer vaccine-type cases.