A Method for Identifying High-Risk Suicidal Individuals Among Those Receiving Primary or Behavioral Health Care for a Mental Health Problem

Group Health Cooperative, an integrated health system on the west coast, serves approximately half a million consumers. Several recent developments – in service delivery and research methods - make it possible to study the phenomena of medically treated suicidal behavior and suicide death in people who are receiving mental health and primary care treatment. We will describe the routine use of the depression questionnaires (the PHQ-9) and the accurate methods we have developed for rapidly identifying probable suicide attempts from electronic medical records data, as well as the real-time linkages of electronic medical record data regarding self-reported suicidal thoughts and subsequent suicidal behavior.

Our research at GHC concludes that item 9 (‘thoughts that you would be better off dead or of hurting yourself in some way more than half the days’ or ‘nearly every day for past two weeks’) of the PHQ depression questionnaire predicts heightened risk of suicide attempt and suicide death.  Patients who score 2 or 3 on this suicide question are 5.5 times more likely to die by suicide and 9 times more likely to make a suicide attempt than individuals reporting a 0 (not at all). This conclusion is based on analysis from data linking electronic medical record, insurance claims, and death certificate data from 84,418 outpatients with PHQs (n=207,625) to suicide attempts (n=708) and suicide deaths (n=46) between 2007 and 2012. This method - the translation of basic science into prevention models - has allowed us to develop and provide more targeted interventions for high-risk individuals. We will describe the benefits and limitations of this method (improving identification of those who don’t endorse suicide ideation but who then harm themselves), as well as potential implications for public policy development.