Changes in statistical methods affected the validity of official suicide rates

Abstract Objective This study investigates whether changes in registration and coding practices influenced official suicide rates in Norway from 1988 to 2002. Study Design and Setting A Poisson regression model was used to evaluate rates of suicide and potentially competing underlying causes of death. Setting in Norway 1988–2002. Results From 1988 to 1994, suicide mortality decreased significantly, by 23.7%. Simultaneously, rates of causes of death potentially masking suicide decreased or remained fairly stable. From 1994 to 2002, however, there were no significant changes in suicide rates but accidental poisoning, which may mask suicide, increased significantly by 32.4%. Also, “ill-defined causes” of death increased by 16.7%, indicating poorer data quality. Conclusion This study suggests that the decreasing suicide rate in 1988–94 reflects a real change. However, the general quality of mortality statistics has deteriorated since the late 1990s, making it difficult to assess developments since 1994. Such variations in the reliability of official suicide statistics complicate international comparisons. However, shifts in the death rate because of “ill-defined” causes could serve as a warning that data quality is not consistent over time.