The findings in this report are subject to at least two limitations.
First, the use of news media reports for mortality surveillance incurs the risk of low sensitivity and specificity. LexisNexis does not include all newspapers and does not include most (e.g., local) television news reports. In this assessment, a LexisNexis search identified only 59.8% of decedents, compared with 87.8% of decedents identified on the two choking-game awareness websites. Even when all newspapers in an area are examined, their sensitivity for unintentional injury surveillance has ranged from 59% for drowning deaths8 to 96% for deaths from fires9 and has been reported as low as 13% for homicides.10 Further, this approach cannot be used to assess or characterize nonfatal injuries resulting from the choking game.2,6,7 Additionally, newspaper reports might attribute deaths to causes or intents that differ from those recorded on death certificates.8 In the design used in this study, information from news media reports could not be subjected to independent verification.
Second, news media reports usually did not provide information on characteristics such as race/ethnicity, education, income, or the role of influence by peers or the media/Internet; therefore, analysis of these characteristics was not possible.