Increasingly, data portals allow researchers to download data-sets generated by others for futher analyses. Open-access data have many benefits. However, researchers should be aware of the need to carefully assess the data before they are used. We illustrate the challenges using the example of the cross-national homicide data generate by the World Health Organisation (WHO) as part of their Global Health Estimates. For many countries these data are statistically modelled on the basis of complex algorithms, and WHO warns against their uncritical use. However, in some publications researchers have downloaded these data for regression models, using the same predictors that had been used by WHO for generating the estimates, meaning that the same information is used on both sides of the equation. In this comment we alert data providers, researchers, and reviewers to the need for caution when using secondary data.