als (suicide completers versus non-suicide completers, or suicide attempters versus non-suicide attempters). Use of suicidal ideation as an outcome can increase incidence and alleviate the problem to some extent; however, it is unclear whether suicidal ideation is a strong predictor of suicide completion. Using both attempts and completions can confound the analysis since attempters may account for some of the suicides completed within the study period. Because the duration of the prevention studies is frequently too brief to collect sufficient data on the low frequency endpoints of suicide or suicide attempt, proximal measures such as changes in knowledge or attitude are used. Yet the predictive value of these variables is unconfirmed. Statistical approaches (see Appendix A) and proximal endpoints may provide solutions, but a large population base is preferable.

A psychological autopsy is the reconstruction of the events leading up to the death; ascertainment of the circumstances of the death, including suicidal intent; and an in-depth exploration of other significant risk factors for suicide (Beskow et al., 1991; Brent et al., 1988; Brent et al., 1993; Cooper, 1999; Hawton et al., 1998; Kelly and Mann, 1996; Velting et al., 1998). The psychological autopsy is the standard approach to augmenting the information obtained from a death certificate. Information is gathered through a semi-structured interview with key informants, and discrepancies are resolved by re-interviewing informants and through a case conference using “best estimate” procedures (Mitchell, 1982). Among the key issues and risk factors addressed are:

• Circumstances and method of suicide

• Psychopathology

• Family history of psychopathology and suicidal behavior

• Social adjustments and functioning

• Personality characteristics, especially aggression/impulsivity

• Life stressors and supports, including religion

• Characteristics of treatment, especially in the 90 days prior to death.

• Physical health and medical history

• Socioeconomic background and family constellation

• Communication of suicidal intent

• Other record linkage (birth records, child welfare, school, criminal justice)

Most studies have found that the optimal time to conduct this kind of investigation is between 2 and 6 months after the death. Informants’ emo-