Results from the simulations suggest that routine immunization of an adequately large subset of the population with an effective quadrivalent vaccine can lead to substantial reductions in cases of disease. Furthermore, the indirect benefits associated with herd immunity seem to be just as important as the direct benefits conferred upon vaccinees.

Under base case assumptions, routine immunization of 12 year olds in the US is predicted to reduce the incidence of invasive disease by almost 40%, and vaccine-preventable disease by over 60%. Outbreaks of invasive meningococcal disease from serogroups A, C, Y, and W-135 are all but eliminated. Sensitivity analyses clearly demonstrate the importance of herd immunity and duration of immunity after vaccination in determining the long-term impact of routine vaccination policies.

By providing a framework for the evaluation of strategies to control invasive meningococcal disease which takes into consideration the direct and indirect benefits of immunization, the short and long-term consequences of the disease, and the impact of outbreaks at both the institutional and community level, the simulation developed for these analyses provides a more complete description of the potential impact of meningococcal vaccination than the simpler cohort models used in the past to evaluate the economic efficiency of routine vaccination in the US for meningococcal disease [19, 26, 27]. An important next step using the results of these simulations will be to incorporate cost and resource use data on managing infections, vaccination activities and outbreak control, in order to re-evaluate the cost-effectiveness of instituting routine vaccination of 12 year olds. Given that benefits resulting from herd immunity and outbreak frequency reduction were not considered in the most recent estimate put forth by the CDC [19], one would expect that the cost-effectiveness of routine vaccination of US adolescents with a quadrivalent conjugate vaccine would be considerably lower than their ratio of $121,000 per life year gained.

Modelling the epidemiology of infectious diseases can be complex, but is necessary for reasonable evaluation of infection control strategies [25]. Although still relatively uncommon in epidemiology, discrete event simulations are increasingly used; a recent example being an assessment of smallpox management strategies in the US [35]. Discrete event simulation is a useful technique when modelling interaction between individuals in a population. The simulation is specified and run at the level of the individual, with each person assigned specific characteristics which are carried in the model and updated periodically. This simulation technique also allows the events occurring in individuals to influence outcomes in other individuals, by specifying interaction between them and the consequences of this interaction [24].

As with any model, one of the main challenges rests with limited data availability. Often, available data alone are not sufficient, requiring calibration procedures to be performed in order to ensure that the model best represents what is known about the disease and its prevention. In the simulation described here, calibration was necessary for three key parameters: duration of protection, herd immunity and outbreak risks.

Recently published models in this area have varied greatly in terms of sophistication and scope, ranging from simple deterministic models that ignore the indirect effects of vaccination or incorporate them by brute force [15–17, 19, 22] 16, to more sophisticated dynamic transmission and stochastic mathematical models [18, 20, 21, 23]. The latter have demonstrated the importance of taking the indirect effects of vaccination into consideration. For example, an economic evaluation of use of a conjugate C meningococcal vaccine in the UK using a dynamic age-structured transmission model estimated that cases avoided due to the indirect effects of vaccination were more than three times greater than the direct effect [23], a finding in agreement with our results for the US. Recent models have also highlighted the importance of outbreaks [19, 21], but none have evaluated the effect of vaccination on outbreaks. Although uncommon, outbreaks are a significant public health concern given the serious consequences and the fear generated in affected communities. The simulation described in this paper incorporates real-world data into a framework that allows for evaluation of both the direct and indirect benefits of vaccination, including the effects on outbreaks.

The model does have limitations that should be taken into consideration. Although not a limitation of the model per se, this simulation deals only indirectly with the relationship between carriage and invasive disease because the required data are lacking. We opted to include it anyway given its importance and calibrated the inputs to the observed herd immunity effects of the serogroup C vaccination campaign in the UK [11]. This simplified approach allows for a clear and relatively straightforward testing of assumptions but caution should be exercised in interpreting results as a linear relation is assumed between direct vaccine protection in the community and the extent of herd immunity for the same age group. For example, it is almost surely the case that changing coverage from 10% to 20% versus 70% to 80% will not have the same effect on herd immunity. Furthermore, the strength of the relationship between vaccine protection and coverage was estimated based on the experience of a single large-scale vaccination program [11]. While herd immunity with vaccination seems to have taken place with other vaccination programs, other factors not considered by the simulation may come into play in determining the indirect benefits of vaccination. The effect of vaccination on carriage, and as a consequence on herd immunity, was dealt with conservatively by assuming a shorter duration of protection against carriage than against the disease, and sensitivity analyses evaluated the impact of different assumptions.

By the same token, the incidence of outbreaks is forced, with calibration to existing data. The impact of vaccination on the occurrence of these outbreaks can then be taken into consideration, and sensitivity analyses can be run under alternative assumptions on the frequency and severity of high risk periods. The model does not take into consideration what factors lead to outbreaks, such as the appearance of more invasive strains of meningococcal disease.

Another limitation of the analyses presented here is the assumption of uniform communities of equal size. In reality, the epidemiology of disease may well vary by the size of the community as well as by other community characteristics. The sensitivity analyses run on community size, however, are reassuring in that when calibration procedure account for changes in the size of the community, only a limited impact on model predictions is observed.

Furthermore, uptake rates for both outbreak and routine vaccination may vary from community to community. With adequate data on inter-community migration patterns, as well as the distribution and make-up of communities in a given region or country, the model could handle differing communities and the effects of migration. Acquiring the data required for such analyses, however, would be a considerable undertaking. Furthermore, the computing time required to model large numbers of heterogeneous communities may be a deterrent. Uptake of routine vaccination was also assumed to be immediate. More gradual uptake would delay the time required to reach steady state and achieve maximum benefits.

Finally, the model currently assumes that the overall endemic risk of infection, outbreak frequency, and other epidemiological characteristics of the disease are constant over time. Again, while the model could incorporate longitudinal changes in these, we are not aware of any data that would allow forecasting of these changes over time.

Despite these limitations, the model accurately replicates the current incidence of disease and outbreaks in populations, and provides plausible ranges of estimates of outcomes with routine vaccination of adolescents. The simulation results also highlight the key role herd immunity can play in determining outcomes, and the sensitivity of predicted results to assumptions on the strength of herd immunity, as well as the duration of vaccine effectiveness.