Authors' reply

McPhedran & Baker point out an unsolved problem of Australian suicide research. There are concerns about the quality of mortality data and statistics based upon them. Therefore, they urge researchers to approach Australian firearms data with caution. The authors cite a letter to the Medical Journal of Australia (De Leo, 2007) which highlighted inconsistencies in Australian mortality data since the year 2001 and called for standardised certification procedures of deaths according to ICD-10 and for other improvements of death registries. However, in Austria autopsies are performed when there is any uncertainty regarding the cause of death. The autopsy rate is high, with a mean rate of 29% in 1991-2000 (Waldhoer et al, 2003). If the cause of death is not clear, an additional investigation by Statistics Austria takes place. Statistics Austria registers deaths as suicide if that is the most probable cause of death. The International Classification of Diseases (ICD-8, -9, -10) has been applied for many years and there are no signs of a decrease in the data quality of Statistics Austria. The work reported by Kapusta et al (2007) is based on these data.

Furthermore, De Leo (2007) realistically states that some underreporting is ubiquitous and has to be tolerated in suicide statistics. On the other hand, underreporting of firearm deaths seems less probable than underreporting of, for example, deaths due to poisonings (with longer survival periods), which tend to be classified as disease-related deaths.

We agree with McPhedran & Baker that Australian firearm laws should be reevaluated on the basis of more reliable data, but as long as sufficient evidence is not available, theoretical assumptions that Australian firearm laws had no life-saving effects remain speculative. This applies also to Europe where independent scientific evaluations of firearm law are still rare.