Characterizing the relative strength of metapopulation dynamics in measles persistence

Background/Aims

Measles is a highly infectious, vaccine-preventable disease that remains persistent across much of the world despite continued large-scale control efforts. Local dynamics of measles are shaped by births of susceptible individuals and seasonal drivers of transmission, while broader regional dynamics are shaped by population mobility and metapopulation connectivity. Incidence data, useful for tracking the progress of measles control, are frequently reported only at the aggregate level (e.g., states/provinces), and persistence at this scale may arise because of the independent persistence in all local sub-units or because of regional metapopulation dynamics. Disentangling the contribution of local versus regional persistence is critical in identifying effective strategies for control.

Methods/Results

We show, as vaccination reduces local measles prevalence, metapopulation colonization and extinction events dominate regional persistence. This manifests as a change in the seasonal distribution of cases throughout the year; incidence becomes more uniformly distributed through the year even with no change in underlying seasonal drivers. We demonstrate this using: 1) a seasonally forced, stochastic metapopulation compartmental model; 2) historical measles incidence in the USA; and 3) recent subnational-level measles incidence in Nigeria.

We create a novel, scale-free measure to quantify the contribution of metapopulation dynamics to persistence, roundness, defined as the geometric mean of the scaled maximum and minimum monthly incidence. A low value of roundness indicates strong local persistence dominated by the seasonal attractor, while a high value indicates strong metapopulation persistence. We illustrate the properties of this measure using simulations and observed data. In the USA, roundness nearly doubled after the introduction of the measles-containing vaccine. In Nigeria, roundness is well correlated with observed vaccine coverage, with a Pearson’s r of 0.70 (95% CI: 0.57, 0.80), discriminating between northern states dominated by local persistence of measles and southern states where persistence is maintained by regional dynamics.

Conclusion

Identifying how measles persists is important in its control—vaccination strategies may differ if persistence is largely local, or if persistence is driven by connectivity from surrounding areas, even where yearly incidence is comparable. Our approach can be valuable areas where other measures (connectivity, mobility, laboratory confirmed illness) are not widely available.