Controlling for Known Risk Factors for Suicide in Multivariate Models

Abstract

Dear Editor: This letter is in response to the article by Lemstra et al1 published in the September 2009 issue. We would like to commend the authors for tackling such an important issue using the Canadian Community Health Survey: Mental Health and Weil-Being (CCHS 1.2) and a social determinants of health perspective; however, we have significant concerns about the manner in which the analysis was conducted which cast doubt on the authors' interpretation of their findings. It is well known that previous history of suicide attempt, psychiatric disorders, substance abuse, and access to lethal means, in addition to the risk factors noted by the authors (physical and social environment, cultural factors, childhood adversity, alcohol abuse, and poverty) are important risk factors for suicide in Aboriginal people in Canada.25 However, none of these key variables were controlled for in the analysis, despite the fact that CCHS 1.2 (2002) investigated Canadians' mental health and well-being and collected information on prevalence of mental disorders and substance use, both for the preceding 12 months as well as for the lifetime. Because of their strong, previously documented relation to suicide, it is possible that these factors could explain the variability found by the authors. By not controlling for these risk factors, the authors may have erroneously reached the conclusion reported. Further, this study used suicidal ideation as the outcome of interest; however, we assume that the purpose of the study was to promote better understanding of the issue of suicide, not just suicidal ideation. It is well appreciated in the literature that most people reporting suicidal ideation do not die by suicide. The history of suicide research is bedevilled by the use of proxy measures such as suicidal ideation, which do not and cannot accurately reflect the complex interplay amongst the different variables associated with suicide death. Therefore, policies and programs such as the ones suggested by the authors in this study on suicidal ideation may be advanced or applied at substantial direct and opportunity costs as suicide prevention strategies but lead to little or no substantive change in the primary outcome, suicide. It is our contention that suicide researchers must address these important methodological and conceptual issues. Indeed, we would go so far as to suggest that research that neglects these issues should not be reported, as publications arising from such work, rather than improving our understanding of suicide, may actually hinder it. Stan Kutcher, MD Magdalena Szumilas, MSc Halifax, Nova Scotia