Most household models of disease transmission assume static household distributions. Although this is a reasonable simplification for assessing vaccination strategies at a single point in time or over the course of an outbreak, it has considerable drawbacks for assessing long term vaccination policies or for predicting future changes in immunity. We demonstrate that household models that include births, deaths and movement between households can show dramatically different patterns of infection and immunity to static population models. When immunity is assumed to be life-long, the pattern of births by household size is the key driver of infection, suggesting that the influx of susceptibles has most impact on infection risk in the household. In a comparison of 12 countries, we show that both the crude birth rate and the mean household size affect the risk of infection in households.