Background: The four-component serogroup B meningococcal 4CMenB vaccine (Bexsero, GSK) has been routinely given to all infants in the United Kingdom at 2, 4 and 12 months of age since September 2015. After three years, Public Health England (PHE) reported a 75% [95% confidence interval: 64%; 81%] reduction in the incidence of serogroup B invasive meningococcal disease (IMD) in age groups eligible to be fully vaccinated. In contrast, vaccine effectiveness (VE) evaluated in the same immunization program applying the screening method was not statistically significant. We re-analyzed the data using an incidence model and obtained more precise VE estimates.
Methods: Aggregate data stratified by age, year and doses received were provided by PHE: serogroup B IMD case counts for the entire population of England (years 2011-2018) and 4CMenB vaccine uptake in infants. We combined uptake with national population estimates to obtain counts of vaccinated and unvaccinated person-time by age and time. We re-estimated VE comparing incidence rates in vaccinated and non-vaccinated using a Bayesian Poisson model for case counts with person-time data as an offset. The model was adjusted for age, time and number of doses received.
Results: The incidence model showed that cases decreased until 2013-2014, followed by an increasing trend that continued in the non-vaccinated population during the immunization program. VE in fully vaccinated (three doses) was 80.1% [95% Bayesian credible interval (BCI): 70.3%; 86.7%]. After a single dose, VE was 33.5% [12.4%; 49.7%]95%BCI, and, after two doses, 78.7% [71.5%; 84.5%]95%BCI. We estimated that vaccination has averted 312 cases [252; 368]95%BCI between 2015 and 2018. VE was in line with the previously reported incidence reduction.
Conclusions: Our estimates of VE had higher precision than previous estimates based on the screening method, that were not statistically significant. When disease incidence is low and vaccine uptake is high, the screening method applied to cases exclusively from the population eligible to vaccination may not be enough precise and produce misleading point-estimates. Precise and accurate VE estimates are fundamental for informing public health decision making. VE assessment can be enhanced using models that leverage data on subjects not eligible for vaccination.