We use six existing sources about past crowd crushes and accident to build a merged data base. By doing so, we show that every source has a partial view of the crowd crushes, with coverage rates between 14% and 59% of our merged data base. Each of the sources contains crowd crushes that are cited by none of the other sources. We then may have a very partial view of past crowd crushes. We examine several biases that can explain under reporting of crowd crushes, notably the less recent ones and the smaller. This partial view affects any statistical study that we can do on the evolution of crowd crushes. However, our data analysis suggests that the number of crowd crushes par capita is not steadily increasing. Crowd crushes may not be a growing tendency in regard to global population. Our analysis suggests it is necessary to continue studying crowd crushes, both globally and in-depth, to gain a more global view of their reasons and their tendencies. We propose to use collaborative projects such as Wikidata to do so.