No abstract

There is a lack of psychological autopsy studies assessing the influence of axis I disorders on axis II disorders as risk factors for suicide. Therefore, we investigated the association between personality disorders, axis I disorders, and suicide. Psychiatric disorders were evaluated by a semi-structured interview including the Structured Clinical Interview for DSM-IV Axis I (SCID-I) and Personality Disorders (SCID-II) in 163 completed suicides (mean age 49.6 +/- 19.3 years; 64.4% men) and by personal interview in 396 population-based control persons (mean age 51.6 +/- 17.0 years; 55.8% men). In both genders, suicides significantly more often had personality disorders of all clusters than controls, also after adjustment for axis I disorders (p < 0.001, each). In addition, alcohol-related disorders, major depression, and co-occurrence of personality disorders of more than one cluster (men: OR = 16.13; women: OR = 20.43) remained independent predictors for suicide in both genders, "pure" cluster B personality disorders only in women and "pure" cluster C personality disorders only in men. In both genders, co-occurrence of personality disorders of more than one cluster contributed to risk of completed suicide after control for axis I psychiatric disorders and has to be considered as an independent risk factor for suicide.

Screening instruments for detection of alcohol consumption, abuse, and dependence for use in psychological autopsy studies with case control design are not validated. Therefore, interrater and test-retest reliability of the Luebeck Alcohol Dependence and Abuse Screening Test (LAST) and the usability of this test for the psychological autopsy method were investigated. Alcohol consumption was evaluated by a semi-structured interview including the Structured Clinical Interview for DSM-IV Axis I (SCID-I) and the LAST in 163 completed suicides (mean age 49.6 +/- 19.3 years; 64.4% men) and by personal interview in 396 population-based controls (mean age 51.6 +/- 17.0 years; 55.8% men). Of the controls, 35 were additionally assessed by interviewing informants; these results were compared with those generated by personal interview. Comparison of LAST scores by personal and informant's interview of controls generated a Spearman correlation coefficient of 0.74 (P < 0.0001). The LAST (7 item-version, cut-off of 2) revealed high sensitivity and specificity for alcohol abuse and dependence, in both controls and suicides. LAST scores were significantly associated with high, frequent, and hazardous alcohol consumption (P < 0.001) in suicides. Our findings provide support for reliability and validity of identifying individuals with alcohol dependence and abuse obtained through the best-estimate method using the LAST. This 7-item questionnaire can be recommended as a useful tool for the psychological autopsy procedure in postmortem research.