The accuracy of suicide statistics: are true suicide deaths misclassified?

PURPOSE: Official suicide statistics often produce an inaccurate view of suicide populations, since some deaths endorsed as being of uncertain manner are in fact suicides; it is common, therefore, in suicide research, to account for these deaths. We aimed to test the hypothesis that non-suicide death categories contain a large potential reservoir of misclassified suicides.
METHODS: Data on undetermined intent and ill-defined death causes, and official suicide deaths recorded in the district of Tel Aviv for the years 2005 and 2008 were extracted. Based on supplementary data, cases regarded as probable suicides ("suicide probable") were then compared with official suicides ("suicide verdicts") on a number of socio-demographic variables, and also in relation to the mechanism of death.
RESULTS: Suicide rates were 42 % higher than those officially reported after accounting for 75 probable suicides (erroneously certified under other cause-of-death categories). Both death classifications ("suicide probable" and "suicide verdicts") had many similarities, significantly differing only with respect to method used. Logistic regression confirmed that the most powerful discriminator was whether the mechanism of death was considered "less active" or "more active" (p < 0.001). Indeed, deaths among the less active group were 4.9 times as likely to be classified as "suicide probable" than were deaths among the more active group.
CONCLUSIONS: Caution is needed when interpreting local area data on suicide rates, and undetermined and ill-defined deaths should be included in suicide research after excluding cases unlikely to be suicides. Improving suicide case ascertainment, using multiple sources of information, and uniform reporting practices, is advised.