Fig. 2
A Incidence rate of hospital admission due to non-measles infectious diseases by age, gender and by a selection of time points relative to measles admission, with 95% confidence intervals*, in Children in Ho Chi Minh City from 2005 to 2015. B Incidence rate ratio of hospital admission due to non-measles infectious diseases post vs. pre-measles by time after measles, with 95% confidence intervalsδ, in children in Ho Chi Minh City from 2005 to 2015. C Incidence rate difference of hospital admission due to non-measles infectious diseases post vs. pre-measles by a selection of time points after measles, with 95% confidence intervals*, in children in Ho Chi Minh City from 2005 to 2015. δWe report the IRR until 6 years after measles admission. All hospital admissions occurred within 6 years after measles admission, and there was hardly any follow-up beyond 6.5 years after measles admission (see Fig. 1). Thus the estimate beyond this time point is based on extrapolation and very sensitive to the specified knot locations of the spline functions in the Poisson regression model. *In our data no children older than 10 years were admitted to hospital due to infectious diseases. Therefore, we only display the fitted curves of incidence rate of hospital admission by age up to 10 years. Also in our data, six infants were diagnosed with measles 1 day after birth and 68 infants had measles within 1 month after birth. Therefore, the incidence rate curves at 2, 4 and 6 years post-MeV starting at age 2, 4 and 6 are not only extrapolation. Of course, children that had measles shortly after birth do not contribute to the pre-measles data. But that is not a problem, because other children serve as the reference group for the pre-measles period