The reliability of reported suicide mortality statistics: an experience from Belgium.

To assess the reliability of official Belgian suicide statistics in epidemiological comparisons on national and district level, various mortality indicators for suicide, undetermined deaths, accidental poisoning deaths and 'estimated' suicide in 1968-1972 and 1978-1981 were compared. Results did not reveal important bias on national comparisons by underreporting. But the biased distribution of undetermined deaths among the 43 geographical entities (districts) suggests that the assumption about constancy of underreporting is not always met and has to be checked if possible. Nevertheless, the numerical weight of the underreporting categories was not found heavy enough to bias the geographical suicide pattern in Belgium.