1. by Niket Thakkar, and Kevin A. McCarthy
Comment on “Long-term measles-induced immunomodulation increases overall childhood infectious disease mortality”
by Niket Thakkar, and Kevin A. McCarthy
Science
Volume 365(6449):eaax5552
July 12, 2019
Copyright © 2019, American Association for the Advancement of Science

2. Fig. 1 Measles in Iceland.
Measles in Iceland. (A) Monthly measles incidence per 1000 population and 1- to 9-year-old all-cause mortality from 1904 to 1974 can be used to test the effect of measles importation on mortality. (B and C) Although we find no statistically significant mortality increase in the years after measles outbreaks, applying the method of Mina et al. offers weak support (R2 = 0.21) for a 47-month immunomodulation duration after measles infection.
Niket Thakkar, and Kevin A. McCarthy Science 2019;365:eaax5552
Copyright © 2019, American Association for the Advancement of Science

3. Fig. 2 Simulation study.
Simulation study. (A) Synthetic measles trace M(t) (black) is created by summing Gaussian functions with standard deviation of 2 months and means separated by P = 60 months. Simultaneously, y(t) (red) is a line with m = –0.1 per month and b = 10. Both traces are distorted by white noise with standard deviation 0.05 and 0.5, respectively. Applying the regression procedure yields an optimal duration d* = P = 60, in agreement with our theoretical analysis. (B) On data created to mimic the UK vaccine introduction, R2 reaches 0.7 while maintaining the periodicity dependence of d*. (C) We also find d* = P when the UK’s M(t) is used and y(t) is an endpoint-constrained random walk (red) with the UK’s mortality variance (dotted black). Using 10,000 samples of y(t), we find a d* distribution that encompasses the value reported in (1). The corresponding optimal R2 values have a 95% CI of (0.28, 0.71), demonstrating the potential for spurious correlation.
Niket Thakkar, and Kevin A. McCarthy Science 2019;365:eaax5552
Copyright © 2019, American Association for the Advancement of Science