In a study of the heterogeneity in malaria infection rates among children Smith et al.1 fitted several mathematical models to data from community studies in Africa. They concluded that 20% of children receive 80% of infections, that infections last about six months on average, that children who clear infections are not immune to new infections, and that the sensitivity and specificity of microscopy for the detection of malaria parasites are 95.8% and 88.4%, respectively. These findings would have important implications for disease control, but we show here that the statistical analysis is unsound and that the data do not support their conclusions.
@article{arxiv.1105.2559,
title = {Biological and Statistical Heterogeneity in Malaria Transmission},
author = {Brian G. Williams and Christopher Dye},
journal= {arXiv preprint arXiv:1105.2559},
year = {2014}
}
Comments
Updated the text to make the connection between the AIC and the likelihood ratio explicit