AAS: Experts say suicide research needs a reboot

ATLANTA– Progress has stalled in understanding the predictors and prevention of suicide, and it’s time for researchers to step up their game, experts agreed at the annual conference of the American Association of Suicidology.

“In the past couple of decades we’ve learned a fair amount about suicidal behavior. However, I think progress has been fairly slow – some might even say a little stagnant – in our pushing things forward and improving our understanding,” Matthew K. Nock, Ph.D., said in the meeting’s opening plenary talk.

He cited a soon-to-be-published meta-analysis led by his post-doctoral fellow Joseph C. Franklin, Ph.D., which evaluated all of the studies of predictors of suicide attempts and completed suicides published during the last 5 decades. The eye opening finding: The predictive odds ratios for the standard risk factors have remained essentially the same – namely, weak – for the past 50 years.

“In general, we’re not getting better in our ability to predict suicidal behavior – and that’s a serious problem for us. We still have enormous gaps in our understanding and in our ability to predict and prevent these outcomes,” declared Dr. Nock, professor of psychology at Harvard University, Boston.

The necessity for a fresh approach to suicide and suicide risk also was emphasized by E. David Klonsky, Ph.D., in his Edwin Shneidman Award Lecture.

“Despite what seems like a very large body of knowledge, suicide rates in the U.S. have increased for numerous consecutive years, and the same is true worldwide,” observed Dr. Klonsky, a psychologist at the University of British Columbia, Vancouver.

“What’s really hard to wrap our heads around is that we’re still only at a 1960s level in our ability to predict suicide. And the main reason for that is our risk factors don’t tell us what we think they do,” he continued.

This was first demonstrated in a 1999 study by Dr. Ronald C. Kessler of Harvard Medical School and coworkers (Arch. Gen. Psychiatry 1999;56:617-26).

They showed that the widely accepted suicide risk factors -- including any mood or anxiety disorder or substance disorders -- are strong predictors of suicidal ideation, but not significant predictors of who will transition from ideation to suicidal action. This finding has subsequently been confirmed by Dr. Nock and others in both adults and adolescents in a massive World Health Organization-sponsored project. Yet to date the concept hasn’t really sunk in broadly in the mental health and medical fields, according to Dr. Klonsky.

In his plenary talk, Dr. Nock focused on four key gaps in the current understanding of how to predict and prevent suicide and outlined how he and others are addressing these needs:

The need for objective markers of suicidal risk: Historically, nearly all patient assessments have relied upon self-report and cross-sectional surveys. That has an obvious limitation, since people are often motivated to conceal their thoughts of suicide. For example, one study found that 78% of patients who died by suicide while in a psychiatric hospital denied suicidal thoughts or intent in their last assessment.

The emerging emphasis is on creating brief computerized tests of memory and reaction time to gain a window into people’s implicit cognitions. Dr. Nock and colleagues have developed one such test, the Implicit Association Test. They had patients who presented to a psychiatric emergency department take the 5-minute word association test and demonstrated that those who scored high for implicit associations between death and suicide were six-fold more likely to make a suicide attempt in the next 6 months (Psychol. Sci. 2010;21:511-7). These findings have since been confirmed by a Canadian group (Psychol. Assess. 2013;25:714-21). The test is available online (www.ImplicitMentalHealth.com) with expert feedback provided as a public education tool and as a means for Dr. Nock and coinvestigators to gather large quantities of data.

Other objective tests for suicide risk that measure physiologic and neural responses to suicide-related stimuli include the Suicide Stroop and Affect Misattribution Procedure.

The need for better predictors of the transition from ideation to attempt: There are a few early leads on such predictors from the WHO dataset and other large studies. These include disorders characterized by aggression, agitation, and/or anxiety, such as conduct disorder, bipolar disorder, and a history of physical or sexual abuse. In a large study in the U.S. Army, the number-one predictor is intermittent explosive disorder.

The need for methods of combining risk factor data: Nearly all studies of suicide risk factors have utilized bivariate analysis -- that is, they examine risk based upon the presence or absence of an individual risk factor, such as a personal history of a mental disorder. But in a study led by Guilherme Borges, Sc.D., of the National Institute of Psychiatry in Mexico City, a group including Dr. Nock showed using National Comorbidity Survey Replication data that by simply together individual risk factors to create a 0-11 scale it became possible to identify a high-risk subgroup consisting of 13.7% of survey participants. This subgroup accounted for 67% of all suicide attempts within the next 12 months (Psychol. Med. 2006;36:1747-57).

The investigators have gone on to validate this approach in more than 108,000 subjects in 21 countries participating in the World Health Organization mental health project (J. Clin. Psychiatry 2010;71:1617-28).

Simple addition of suicidality risk factors, while a big step forward in risk assessment, is still a relatively crude predictive tool. More recently, Dr. Kessler, collaborating with Dr. Nock and others, has developed a much more sophisticated actuarial risk algorithm and applied it to more than 54,000 U.S. Army soldiers hospitalized for psychiatric disorders. They found that subjects who scored in the top 5% in terms of predicted suicide risk accounted for 53% of all suicides that occurred within the next 12 months. The suicide rate in this highest-risk group was massive: 3,624 per 100,000 per year as compared to a background rate of 18.5/100,000/year in the Army overall.

Moreover, nearly one-half of soldiers with a risk score in the top 5% had a 12-month composite adverse outcome, defined as another suicide attempt, death by suicide, accidental death, or psychiatric rehospitalization (JAMA Psychiatry 2015;72:49-57).

The need for data on imminent risk: Dr. Nock called this the biggest unmet need in suicidology; it’s what clinicians and family members desperately want but don’t have. At present there is “approximately zero data” on how to predict suicidal behavior in the hours, days, or weeks before it occurs, Dr. Nock said. Indeed, Dr. Franklin’s meta-analysis showed that in the past 50 years more than three-quarters of studies examining suicide risk have looked at risk a year or more in the future. Only 2% of studies have looked at risk during the window of the next month or so.

Numerous groups are now looking at real-time patient monitoring using cell phones and smart watches as a means of developing short-term risk predictors. These tools enable investigators to monitor changes in mood, thoughts, behavior, and physiology in large populations in order to see what leads up to a suicide attempt. Dr. Nock’s group is collaborating with information scientists at Massachusetts Intitute of Technology on such projects.

This technology also shows promise for therapeutic intervention. Dr. Franklin and coworkers have developed a brief, game-like mobile app to administer what he calls Therapeutic Evaluative Conditioning. In three soon-to-be-published randomized controlled trials, he has shown that this simple intervention – essentially, playing a game on a cell phone – resulted in reductions of 42%-49% in self-cutting and other nonsuicidal self-injury, 21%-64% reductions in suicidal planning, and 20%-57% decreases in suicidal behaviors, according to Dr. Nock.

Dr. Nock’s research is funded chiefly by the National Institute of Mental Health, the World Health Organization, and the Department of Defense; he reported having no financial conflicts. Dr. Klonsky’s research is largely supported by the American Foundation for Suicide Prevention.

[email protected]

ATLANTA– Progress has stalled in understanding the predictors and prevention of suicide, and it’s time for researchers to step up their game, experts agreed at the annual conference of the American Association of Suicidology.

“In the past couple of decades we’ve learned a fair amount about suicidal behavior. However, I think progress has been fairly slow – some might even say a little stagnant – in our pushing things forward and improving our understanding,” Matthew K. Nock, Ph.D., said in the meeting’s opening plenary talk.

He cited a soon-to-be-published meta-analysis led by his post-doctoral fellow Joseph C. Franklin, Ph.D., which evaluated all of the studies of predictors of suicide attempts and completed suicides published during the last 5 decades. The eye opening finding: The predictive odds ratios for the standard risk factors have remained essentially the same – namely, weak – for the past 50 years.

“In general, we’re not getting better in our ability to predict suicidal behavior – and that’s a serious problem for us. We still have enormous gaps in our understanding and in our ability to predict and prevent these outcomes,” declared Dr. Nock, professor of psychology at Harvard University, Boston.

The necessity for a fresh approach to suicide and suicide risk also was emphasized by E. David Klonsky, Ph.D., in his Edwin Shneidman Award Lecture.

“Despite what seems like a very large body of knowledge, suicide rates in the U.S. have increased for numerous consecutive years, and the same is true worldwide,” observed Dr. Klonsky, a psychologist at the University of British Columbia, Vancouver.

“What’s really hard to wrap our heads around is that we’re still only at a 1960s level in our ability to predict suicide. And the main reason for that is our risk factors don’t tell us what we think they do,” he continued.

This was first demonstrated in a 1999 study by Dr. Ronald C. Kessler of Harvard Medical School and coworkers (Arch. Gen. Psychiatry 1999;56:617-26).

They showed that the widely accepted suicide risk factors -- including any mood or anxiety disorder or substance disorders -- are strong predictors of suicidal ideation, but not significant predictors of who will transition from ideation to suicidal action. This finding has subsequently been confirmed by Dr. Nock and others in both adults and adolescents in a massive World Health Organization-sponsored project. Yet to date the concept hasn’t really sunk in broadly in the mental health and medical fields, according to Dr. Klonsky.

In his plenary talk, Dr. Nock focused on four key gaps in the current understanding of how to predict and prevent suicide and outlined how he and others are addressing these needs:

The need for objective markers of suicidal risk: Historically, nearly all patient assessments have relied upon self-report and cross-sectional surveys. That has an obvious limitation, since people are often motivated to conceal their thoughts of suicide. For example, one study found that 78% of patients who died by suicide while in a psychiatric hospital denied suicidal thoughts or intent in their last assessment.

The emerging emphasis is on creating brief computerized tests of memory and reaction time to gain a window into people’s implicit cognitions. Dr. Nock and colleagues have developed one such test, the Implicit Association Test. They had patients who presented to a psychiatric emergency department take the 5-minute word association test and demonstrated that those who scored high for implicit associations between death and suicide were six-fold more likely to make a suicide attempt in the next 6 months (Psychol. Sci. 2010;21:511-7). These findings have since been confirmed by a Canadian group (Psychol. Assess. 2013;25:714-21). The test is available online (www.ImplicitMentalHealth.com) with expert feedback provided as a public education tool and as a means for Dr. Nock and coinvestigators to gather large quantities of data.

Other objective tests for suicide risk that measure physiologic and neural responses to suicide-related stimuli include the Suicide Stroop and Affect Misattribution Procedure.

The need for better predictors of the transition from ideation to attempt: There are a few early leads on such predictors from the WHO dataset and other large studies. These include disorders characterized by aggression, agitation, and/or anxiety, such as conduct disorder, bipolar disorder, and a history of physical or sexual abuse. In a large study in the U.S. Army, the number-one predictor is intermittent explosive disorder.

The need for methods of combining risk factor data: Nearly all studies of suicide risk factors have utilized bivariate analysis -- that is, they examine risk based upon the presence or absence of an individual risk factor, such as a personal history of a mental disorder. But in a study led by Guilherme Borges, Sc.D., of the National Institute of Psychiatry in Mexico City, a group including Dr. Nock showed using National Comorbidity Survey Replication data that by simply together individual risk factors to create a 0-11 scale it became possible to identify a high-risk subgroup consisting of 13.7% of survey participants. This subgroup accounted for 67% of all suicide attempts within the next 12 months (Psychol. Med. 2006;36:1747-57).

The investigators have gone on to validate this approach in more than 108,000 subjects in 21 countries participating in the World Health Organization mental health project (J. Clin. Psychiatry 2010;71:1617-28).

Simple addition of suicidality risk factors, while a big step forward in risk assessment, is still a relatively crude predictive tool. More recently, Dr. Kessler, collaborating with Dr. Nock and others, has developed a much more sophisticated actuarial risk algorithm and applied it to more than 54,000 U.S. Army soldiers hospitalized for psychiatric disorders. They found that subjects who scored in the top 5% in terms of predicted suicide risk accounted for 53% of all suicides that occurred within the next 12 months. The suicide rate in this highest-risk group was massive: 3,624 per 100,000 per year as compared to a background rate of 18.5/100,000/year in the Army overall.

Moreover, nearly one-half of soldiers with a risk score in the top 5% had a 12-month composite adverse outcome, defined as another suicide attempt, death by suicide, accidental death, or psychiatric rehospitalization (JAMA Psychiatry 2015;72:49-57).

The need for data on imminent risk: Dr. Nock called this the biggest unmet need in suicidology; it’s what clinicians and family members desperately want but don’t have. At present there is “approximately zero data” on how to predict suicidal behavior in the hours, days, or weeks before it occurs, Dr. Nock said. Indeed, Dr. Franklin’s meta-analysis showed that in the past 50 years more than three-quarters of studies examining suicide risk have looked at risk a year or more in the future. Only 2% of studies have looked at risk during the window of the next month or so.

Numerous groups are now looking at real-time patient monitoring using cell phones and smart watches as a means of developing short-term risk predictors. These tools enable investigators to monitor changes in mood, thoughts, behavior, and physiology in large populations in order to see what leads up to a suicide attempt. Dr. Nock’s group is collaborating with information scientists at Massachusetts Intitute of Technology on such projects.

This technology also shows promise for therapeutic intervention. Dr. Franklin and coworkers have developed a brief, game-like mobile app to administer what he calls Therapeutic Evaluative Conditioning. In three soon-to-be-published randomized controlled trials, he has shown that this simple intervention – essentially, playing a game on a cell phone – resulted in reductions of 42%-49% in self-cutting and other nonsuicidal self-injury, 21%-64% reductions in suicidal planning, and 20%-57% decreases in suicidal behaviors, according to Dr. Nock.

Dr. Nock’s research is funded chiefly by the National Institute of Mental Health, the World Health Organization, and the Department of Defense; he reported having no financial conflicts. Dr. Klonsky’s research is largely supported by the American Foundation for Suicide Prevention.

[email protected]

ATLANTA– Progress has stalled in understanding the predictors and prevention of suicide, and it’s time for researchers to step up their game, experts agreed at the annual conference of the American Association of Suicidology.

“In the past couple of decades we’ve learned a fair amount about suicidal behavior. However, I think progress has been fairly slow – some might even say a little stagnant – in our pushing things forward and improving our understanding,” Matthew K. Nock, Ph.D., said in the meeting’s opening plenary talk.

He cited a soon-to-be-published meta-analysis led by his post-doctoral fellow Joseph C. Franklin, Ph.D., which evaluated all of the studies of predictors of suicide attempts and completed suicides published during the last 5 decades. The eye opening finding: The predictive odds ratios for the standard risk factors have remained essentially the same – namely, weak – for the past 50 years.

“In general, we’re not getting better in our ability to predict suicidal behavior – and that’s a serious problem for us. We still have enormous gaps in our understanding and in our ability to predict and prevent these outcomes,” declared Dr. Nock, professor of psychology at Harvard University, Boston.

The necessity for a fresh approach to suicide and suicide risk also was emphasized by E. David Klonsky, Ph.D., in his Edwin Shneidman Award Lecture.

“Despite what seems like a very large body of knowledge, suicide rates in the U.S. have increased for numerous consecutive years, and the same is true worldwide,” observed Dr. Klonsky, a psychologist at the University of British Columbia, Vancouver.

“What’s really hard to wrap our heads around is that we’re still only at a 1960s level in our ability to predict suicide. And the main reason for that is our risk factors don’t tell us what we think they do,” he continued.

This was first demonstrated in a 1999 study by Dr. Ronald C. Kessler of Harvard Medical School and coworkers (Arch. Gen. Psychiatry 1999;56:617-26).

They showed that the widely accepted suicide risk factors -- including any mood or anxiety disorder or substance disorders -- are strong predictors of suicidal ideation, but not significant predictors of who will transition from ideation to suicidal action. This finding has subsequently been confirmed by Dr. Nock and others in both adults and adolescents in a massive World Health Organization-sponsored project. Yet to date the concept hasn’t really sunk in broadly in the mental health and medical fields, according to Dr. Klonsky.

In his plenary talk, Dr. Nock focused on four key gaps in the current understanding of how to predict and prevent suicide and outlined how he and others are addressing these needs:

The need for objective markers of suicidal risk: Historically, nearly all patient assessments have relied upon self-report and cross-sectional surveys. That has an obvious limitation, since people are often motivated to conceal their thoughts of suicide. For example, one study found that 78% of patients who died by suicide while in a psychiatric hospital denied suicidal thoughts or intent in their last assessment.

The emerging emphasis is on creating brief computerized tests of memory and reaction time to gain a window into people’s implicit cognitions. Dr. Nock and colleagues have developed one such test, the Implicit Association Test. They had patients who presented to a psychiatric emergency department take the 5-minute word association test and demonstrated that those who scored high for implicit associations between death and suicide were six-fold more likely to make a suicide attempt in the next 6 months (Psychol. Sci. 2010;21:511-7). These findings have since been confirmed by a Canadian group (Psychol. Assess. 2013;25:714-21). The test is available online (www.ImplicitMentalHealth.com) with expert feedback provided as a public education tool and as a means for Dr. Nock and coinvestigators to gather large quantities of data.

Other objective tests for suicide risk that measure physiologic and neural responses to suicide-related stimuli include the Suicide Stroop and Affect Misattribution Procedure.

The need for better predictors of the transition from ideation to attempt: There are a few early leads on such predictors from the WHO dataset and other large studies. These include disorders characterized by aggression, agitation, and/or anxiety, such as conduct disorder, bipolar disorder, and a history of physical or sexual abuse. In a large study in the U.S. Army, the number-one predictor is intermittent explosive disorder.

The need for methods of combining risk factor data: Nearly all studies of suicide risk factors have utilized bivariate analysis -- that is, they examine risk based upon the presence or absence of an individual risk factor, such as a personal history of a mental disorder. But in a study led by Guilherme Borges, Sc.D., of the National Institute of Psychiatry in Mexico City, a group including Dr. Nock showed using National Comorbidity Survey Replication data that by simply together individual risk factors to create a 0-11 scale it became possible to identify a high-risk subgroup consisting of 13.7% of survey participants. This subgroup accounted for 67% of all suicide attempts within the next 12 months (Psychol. Med. 2006;36:1747-57).

The investigators have gone on to validate this approach in more than 108,000 subjects in 21 countries participating in the World Health Organization mental health project (J. Clin. Psychiatry 2010;71:1617-28).

Simple addition of suicidality risk factors, while a big step forward in risk assessment, is still a relatively crude predictive tool. More recently, Dr. Kessler, collaborating with Dr. Nock and others, has developed a much more sophisticated actuarial risk algorithm and applied it to more than 54,000 U.S. Army soldiers hospitalized for psychiatric disorders. They found that subjects who scored in the top 5% in terms of predicted suicide risk accounted for 53% of all suicides that occurred within the next 12 months. The suicide rate in this highest-risk group was massive: 3,624 per 100,000 per year as compared to a background rate of 18.5/100,000/year in the Army overall.

Moreover, nearly one-half of soldiers with a risk score in the top 5% had a 12-month composite adverse outcome, defined as another suicide attempt, death by suicide, accidental death, or psychiatric rehospitalization (JAMA Psychiatry 2015;72:49-57).

The need for data on imminent risk: Dr. Nock called this the biggest unmet need in suicidology; it’s what clinicians and family members desperately want but don’t have. At present there is “approximately zero data” on how to predict suicidal behavior in the hours, days, or weeks before it occurs, Dr. Nock said. Indeed, Dr. Franklin’s meta-analysis showed that in the past 50 years more than three-quarters of studies examining suicide risk have looked at risk a year or more in the future. Only 2% of studies have looked at risk during the window of the next month or so.

Numerous groups are now looking at real-time patient monitoring using cell phones and smart watches as a means of developing short-term risk predictors. These tools enable investigators to monitor changes in mood, thoughts, behavior, and physiology in large populations in order to see what leads up to a suicide attempt. Dr. Nock’s group is collaborating with information scientists at Massachusetts Intitute of Technology on such projects.

This technology also shows promise for therapeutic intervention. Dr. Franklin and coworkers have developed a brief, game-like mobile app to administer what he calls Therapeutic Evaluative Conditioning. In three soon-to-be-published randomized controlled trials, he has shown that this simple intervention – essentially, playing a game on a cell phone – resulted in reductions of 42%-49% in self-cutting and other nonsuicidal self-injury, 21%-64% reductions in suicidal planning, and 20%-57% decreases in suicidal behaviors, according to Dr. Nock.

Dr. Nock’s research is funded chiefly by the National Institute of Mental Health, the World Health Organization, and the Department of Defense; he reported having no financial conflicts. Dr. Klonsky’s research is largely supported by the American Foundation for Suicide Prevention.

[email protected]