User login
Suicide is not a trivial matter – it upends families, robs partners of a loved one, prevents children from having a parent, and can destroy a parent’s most cherished being. It is not surprising that societies have repeatedly made it a goal to study and reduce suicide within their populations.
The suicide rate in the United States is trending upward, from about 10 per 100,000 in 2000 to about 15 per 100,000 in more recent reports. The increasing suicide rates have been accompanied by increasing distress among many strata of society. From a public health level, analysts are not just witnessing increasing suicide rates, but a shocking rise in all “deaths of despair,”1 among which suicide can be considered the ultimate example.
On an individual level, many know someone who has died of suicide or suffered from a serious suicide attempt. From the public health level to the individual level, advocacy has called for various interventions in the field of psychiatry to remedy this tragic problem.
Psychiatrists have been firsthand witnesses to this increasing demand for suicide interventions. When in residency, the norm was to perform a suicide risk assessment at the time of admission to the hospital and again at the time of discharge. As the years passed, the new normal within psychiatric hospitals has shifted to asking about suicidality on a daily basis.
In what seems to us like an escalating arms race, the emerging standard of care at many facilities is now not only for daily suicide risk assessments by each psychiatrist, but also to require nurses to ask about suicidality during every 8-hour shift – in addition to documented inquiries about suicidality by other allied staff on the psychiatric unit. As a result, it is not uncommon for a patient hospitalized at an academic center to receive more than half a dozen suicide risk assessments in a day (first by the medical student, at least once – often more than once – by the resident, again by the attending psychiatrist, then the social worker and three nurses in 24 hours).
One of the concerns about such an approach is the lack of logic inherent to many risk assessment tools and symptom scales. Many of us are familiar with the Patient Health Questionnaire (PHQ-9) to assess depression.2 The PHQ-9 asks to consider “over the last 2 weeks, how often have you ...” in relation to nine symptoms associated with depression. It has always defied reason to perform a PHQ-9 every day and expect the answers to change from “nearly every day” to “not at all,” considering only 1 day has passed since the last time the patient has answered the questions. Yet daily, or near daily, PHQ-9 scores are a frequently used tool of tracking symptom improvement in response to treatments, such as electroconvulsive therapy, performed multiple times a week.
One can argue that the patient’s perspective on how symptomatic he or she has been over the past 2 weeks may change rapidly with alleviation of a depressed mood. However, the PHQ-9 is both reported to be, and often regarded as, an objective score. If one wishes to utilize it as such, the defense of its use should not be that it is a subjective report with just as much utility as “Rate your depression on a scale of 0-27.”
Similarly, many suicide scales were intended to assess thoughts of suicide in the past month3 or have been re-tooled to address this particular concern by asking “since the last contact.”4 It is baffling to see a chart with many dozens of suicide risk assessments with at times widely differing answers, yet all measuring thoughts of suicide in the past month. Is one to expect the answer to “How many times have you had these thoughts [of suicide ideation]? (1) Less than once a week (2) Once a week ...” to change between 8 a.m. and noon? Furthermore, for the purpose of assessing acute risk of suicidality in the immediate future, to only consider symptoms since the last contact – or past 2 weeks, past month, etc. – is of unclear significance.
Provider liability
Another concern is the liability placed on providers. A common problem encountered in the inpatient setting is insurance companies refusing to reimburse a hospital stay for depressed patients denying suicidality.
Any provider in the position of caring for such a patient must ask: What is the likelihood of someone providing a false negative – a false denial of suicidality? Is the likelihood of a suicidal person denying suicidality different if asked 5 or 10 or more times in a day? There are innumerable instances where a patient at a very high risk of self-harm has denied suicidality, been discharged from the hospital, and suffered terrible consequences. Ethically, the psychiatrist aware of this risk is no more at ease discharging these patients, whether it is one suicide risk scale or a dozen that suggests a patient is at low risk.
Alternatively, it may feel untenable from a medicolegal perspective for a psychiatrist to discharge a patient denying suicidality when the chart includes over a dozen previously documented elevated suicide risk assessments in the past 72 hours. By placing the job of suicide risk assessment in the hands of providers of varying levels of training and responsibility, a situation is created in which the seasoned psychiatrist who would otherwise be comfortable discharging a patient feels unable to do so because every other note-writer in the record – from the triage nurse to the medical assistant to the sitter in the emergency department – has recorded the patient as high risk for suicide. When put in such a position, the thought often occurs that systems of care, rather than individual providers, are protected most by ever escalating requirements for suicide risk documentation. To make a clinical decision contrary to the body of suicide risk documentation puts the provider at risk of being scapegoated by the system of care, which can point to its illogical and ineffective, though profusely documented, suicide prevention protocols.
Limitations of risk assessments
Considering the ongoing rise in the use of suicide risk assessments, one would expect that the evidence for their efficacy was robust and well established. Yet a thorough review of suicide risk assessments funded by the MacArthur Foundation, which examined decades of research, came to disheartening conclusions: “predictive ability has not improved over the past 50 years”; “no risk factor category or subcategory is substantially stronger than any other”; and “predicting solely according to base rates may be comparable to prediction with current risk factors.”5
Those findings were consistent with the conclusions of many other studies, which have summarized the utility of suicide risk assessments as follows: “occurrence of suicide is too low to identify those individuals who are likely to die by suicide”;6 “suicide prediction models produce accurate overall classification models, but their accuracy of predicting a future event is near zero”;7 “risk stratification is too inaccurate to be clinically useful and might even be harmful”;8 “suicide risk prediction [lacks] any items or information that to a useful degree permit the identification of persons who will complete suicide”;9 “existing suicide prediction tools have little current clinical value”;10 “our current preoccupation with risk assessment has ... created a mythology with no evidence to support it.”11 And that’s to cite just a few.
Sadly, we have known about the limitations of suicide risk assessments for many decades. In 1983 a large VA prospective study, which aimed to identify veterans who will die by suicide, examined 4,800 patients with a wide range of instruments and measures.12 This study concluded that “discriminant analysis was clearly inadequate in correctly classifying the subjects. For an event as rare as suicide, our predictive tools and guides are simply not equal to the task.” The authors described the feelings of many in stating “courts and public opinion expect physicians to be able to pick out the particular persons who will later commit suicide. Although we may reconstruct causal chains and motives, we do not possess the tools to predict suicides.”
Yet, even several decades prior, in 1954, Dr. Albert Rosen performed an elegant statistical analysis and predicted that, considering the low base rate of suicide, suicide risk assessments are “of no practical value, for it would be impossible to treat the prodigious number of false positives.”13 It seems that we continue to be unable to accept Dr. Rosen’s premonition despite decades of confirmatory evidence.
“Quantity over quality”
Regardless of those sobering reports,
One can reasonably argue that the periodic performance of a suicide risk assessment may have clinical utility in reminding us of modifiable risk factors such as intoxication, social isolation, and access to lethal means. One can also reasonably argue that these risk assessments may provide useful education to patients and their families on epidemiological risk factors such as gender, age, and marital status. But our pursuit of serial suicide risk assessments throughout the day is encouraging providers to focus on a particular risk factor that changes from moment to moment and has particularly low validity, that being self-reported suicidality.
Reported suicidality is one of the few risk factors that can change from shift to shift. But 80% of people who die by suicide had not previously expressed suicidality, and 98.3% of people who have endorsed suicidality do not die by suicide.14 While the former statistic may improve with increased assessment, the later will likely worsen.
Suicide is not a trivial matter. We admire those that study it and advocate for better interventions. We have compassion for those who have suffered the loss of a loved one to suicide. Our patients have died as a result of the human limitations surrounding suicide prevention. Recognizing the weight of suicide and making an effort to avoid minimizing its immense consequences drive our desire to be honest with ourselves, our patients and their families, and society. That includes the unfortunate truth regarding the current state of the evidence and our ability to enact change.
It is our concern that the rising fascination with repeated suicide risk assessment is misguided in its current form and serves the purpose of appeasing administrators more than reflecting a scientific understanding of the literature. More sadly, we are concerned that this “quantity-over-quality” approach is yet another barrier to practicing what may be one of the few interventions with any hope of meaningfully impacting a patient’s risk of suicide in the clinical setting – spending time connecting with our patients.
Dr. Badre is a clinical and forensic psychiatrist in San Diego. He holds teaching positions at the University of California, San Diego, and the University of San Diego. He teaches medical education, psychopharmacology, ethics in psychiatry, and correctional care. Dr. Badre can be reached at his website, BadreMD.com. Dr. Compton is a member of the psychiatry faculty at University of California, San Diego. His background includes medical education, mental health advocacy, work with underserved populations, and brain cancer research. Dr. Badre and Dr. Compton have no conflicts of interest.
References
1. Joint Economic Committee. (2019). Long Term Trends in Deaths of Despair. SCP Report 4-19.
2. Kroenke K and Spitzer RL. The PHQ-9: A new depression diagnostic and severity measure. Psychiatr Ann. 2013;32(9):509-15. doi: 10.3928/0048-5713-20020901-06.
3. Columbia-Suicide Severity Rating Scale (C-SSRS) Full Lifetime/Recent.
4. Columbia-Suicide Severity Rating Scale (C-SSRS) Full Since Last Contact.
5. Franklin JC et al. Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychol Bull. 2017 Feb;143(2):187-232. doi: 10.1037/bul0000084.
6. Beautrais AL. Further suicidal behavior among medically serious suicide attempters. Suicide Life Threat Behav. 2004 Spring;34(1):1-11. doi: 10.1521/suli.34.1.1.27772.
7. Belsher BE. Prediction models for suicide attempts and deaths: A systematic review and simulation. JAMA Psychiatry. 2019 Jun 1;76(6):642-651. doi: 10.1001/jamapsychiatry.2019.0174.
8. Carter G et al. Royal Australian and New Zealand College of Psychiatrists clinical practice guideline for the management of deliberate self-harm. Aust N Z J Psychiatry. 2016 Oct;50(10):939-1000. doi: 10.1177/0004867416661039.
9. Fosse R et al. Predictors of suicide in the patient population admitted to a locked-door psychiatric acute ward. PLoS One. 2017 Mar 16;12(3):e0173958. doi: 10.1371/journal.pone.0173958.
10. Kessler RC et al. Suicide prediction models: A critical review of recent research with recommendations for the way forward. Mol Psychiatry. 2020 Jan;25(1):168-79. doi: 10.1038/s41380-019-0531-0.
11. Mulder R. Problems with suicide risk assessment. Aust N Z J Psychiatry. 2011 Aug;45(8):605-7. doi: 10.3109/00048674.2011.594786.
12. Pokorny AD. Prediction of suicide in psychiatric patients: Report of a prospective study. Arch Gen Psychiatry. 1983 Mar;40(3):249-57. doi: 10.1001/archpsyc.1983.01790030019002.
13. Rosen A. Detection of suicidal patients: An example of some limitations in the prediction of infrequent events. J Consult Psychol. 1954 Dec;18(6):397-403. doi: 10.1037/h0058579.
14. McHugh CM et al. (2019). Association between suicidal ideation and suicide: Meta-analyses of odds ratios, sensitivity, specificity and positive predictive value. BJPsych Open. 2019 Mar;5(2):e18. doi: 10.1192/bjo.2018.88.
Suicide is not a trivial matter – it upends families, robs partners of a loved one, prevents children from having a parent, and can destroy a parent’s most cherished being. It is not surprising that societies have repeatedly made it a goal to study and reduce suicide within their populations.
The suicide rate in the United States is trending upward, from about 10 per 100,000 in 2000 to about 15 per 100,000 in more recent reports. The increasing suicide rates have been accompanied by increasing distress among many strata of society. From a public health level, analysts are not just witnessing increasing suicide rates, but a shocking rise in all “deaths of despair,”1 among which suicide can be considered the ultimate example.
On an individual level, many know someone who has died of suicide or suffered from a serious suicide attempt. From the public health level to the individual level, advocacy has called for various interventions in the field of psychiatry to remedy this tragic problem.
Psychiatrists have been firsthand witnesses to this increasing demand for suicide interventions. When in residency, the norm was to perform a suicide risk assessment at the time of admission to the hospital and again at the time of discharge. As the years passed, the new normal within psychiatric hospitals has shifted to asking about suicidality on a daily basis.
In what seems to us like an escalating arms race, the emerging standard of care at many facilities is now not only for daily suicide risk assessments by each psychiatrist, but also to require nurses to ask about suicidality during every 8-hour shift – in addition to documented inquiries about suicidality by other allied staff on the psychiatric unit. As a result, it is not uncommon for a patient hospitalized at an academic center to receive more than half a dozen suicide risk assessments in a day (first by the medical student, at least once – often more than once – by the resident, again by the attending psychiatrist, then the social worker and three nurses in 24 hours).
One of the concerns about such an approach is the lack of logic inherent to many risk assessment tools and symptom scales. Many of us are familiar with the Patient Health Questionnaire (PHQ-9) to assess depression.2 The PHQ-9 asks to consider “over the last 2 weeks, how often have you ...” in relation to nine symptoms associated with depression. It has always defied reason to perform a PHQ-9 every day and expect the answers to change from “nearly every day” to “not at all,” considering only 1 day has passed since the last time the patient has answered the questions. Yet daily, or near daily, PHQ-9 scores are a frequently used tool of tracking symptom improvement in response to treatments, such as electroconvulsive therapy, performed multiple times a week.
One can argue that the patient’s perspective on how symptomatic he or she has been over the past 2 weeks may change rapidly with alleviation of a depressed mood. However, the PHQ-9 is both reported to be, and often regarded as, an objective score. If one wishes to utilize it as such, the defense of its use should not be that it is a subjective report with just as much utility as “Rate your depression on a scale of 0-27.”
Similarly, many suicide scales were intended to assess thoughts of suicide in the past month3 or have been re-tooled to address this particular concern by asking “since the last contact.”4 It is baffling to see a chart with many dozens of suicide risk assessments with at times widely differing answers, yet all measuring thoughts of suicide in the past month. Is one to expect the answer to “How many times have you had these thoughts [of suicide ideation]? (1) Less than once a week (2) Once a week ...” to change between 8 a.m. and noon? Furthermore, for the purpose of assessing acute risk of suicidality in the immediate future, to only consider symptoms since the last contact – or past 2 weeks, past month, etc. – is of unclear significance.
Provider liability
Another concern is the liability placed on providers. A common problem encountered in the inpatient setting is insurance companies refusing to reimburse a hospital stay for depressed patients denying suicidality.
Any provider in the position of caring for such a patient must ask: What is the likelihood of someone providing a false negative – a false denial of suicidality? Is the likelihood of a suicidal person denying suicidality different if asked 5 or 10 or more times in a day? There are innumerable instances where a patient at a very high risk of self-harm has denied suicidality, been discharged from the hospital, and suffered terrible consequences. Ethically, the psychiatrist aware of this risk is no more at ease discharging these patients, whether it is one suicide risk scale or a dozen that suggests a patient is at low risk.
Alternatively, it may feel untenable from a medicolegal perspective for a psychiatrist to discharge a patient denying suicidality when the chart includes over a dozen previously documented elevated suicide risk assessments in the past 72 hours. By placing the job of suicide risk assessment in the hands of providers of varying levels of training and responsibility, a situation is created in which the seasoned psychiatrist who would otherwise be comfortable discharging a patient feels unable to do so because every other note-writer in the record – from the triage nurse to the medical assistant to the sitter in the emergency department – has recorded the patient as high risk for suicide. When put in such a position, the thought often occurs that systems of care, rather than individual providers, are protected most by ever escalating requirements for suicide risk documentation. To make a clinical decision contrary to the body of suicide risk documentation puts the provider at risk of being scapegoated by the system of care, which can point to its illogical and ineffective, though profusely documented, suicide prevention protocols.
Limitations of risk assessments
Considering the ongoing rise in the use of suicide risk assessments, one would expect that the evidence for their efficacy was robust and well established. Yet a thorough review of suicide risk assessments funded by the MacArthur Foundation, which examined decades of research, came to disheartening conclusions: “predictive ability has not improved over the past 50 years”; “no risk factor category or subcategory is substantially stronger than any other”; and “predicting solely according to base rates may be comparable to prediction with current risk factors.”5
Those findings were consistent with the conclusions of many other studies, which have summarized the utility of suicide risk assessments as follows: “occurrence of suicide is too low to identify those individuals who are likely to die by suicide”;6 “suicide prediction models produce accurate overall classification models, but their accuracy of predicting a future event is near zero”;7 “risk stratification is too inaccurate to be clinically useful and might even be harmful”;8 “suicide risk prediction [lacks] any items or information that to a useful degree permit the identification of persons who will complete suicide”;9 “existing suicide prediction tools have little current clinical value”;10 “our current preoccupation with risk assessment has ... created a mythology with no evidence to support it.”11 And that’s to cite just a few.
Sadly, we have known about the limitations of suicide risk assessments for many decades. In 1983 a large VA prospective study, which aimed to identify veterans who will die by suicide, examined 4,800 patients with a wide range of instruments and measures.12 This study concluded that “discriminant analysis was clearly inadequate in correctly classifying the subjects. For an event as rare as suicide, our predictive tools and guides are simply not equal to the task.” The authors described the feelings of many in stating “courts and public opinion expect physicians to be able to pick out the particular persons who will later commit suicide. Although we may reconstruct causal chains and motives, we do not possess the tools to predict suicides.”
Yet, even several decades prior, in 1954, Dr. Albert Rosen performed an elegant statistical analysis and predicted that, considering the low base rate of suicide, suicide risk assessments are “of no practical value, for it would be impossible to treat the prodigious number of false positives.”13 It seems that we continue to be unable to accept Dr. Rosen’s premonition despite decades of confirmatory evidence.
“Quantity over quality”
Regardless of those sobering reports,
One can reasonably argue that the periodic performance of a suicide risk assessment may have clinical utility in reminding us of modifiable risk factors such as intoxication, social isolation, and access to lethal means. One can also reasonably argue that these risk assessments may provide useful education to patients and their families on epidemiological risk factors such as gender, age, and marital status. But our pursuit of serial suicide risk assessments throughout the day is encouraging providers to focus on a particular risk factor that changes from moment to moment and has particularly low validity, that being self-reported suicidality.
Reported suicidality is one of the few risk factors that can change from shift to shift. But 80% of people who die by suicide had not previously expressed suicidality, and 98.3% of people who have endorsed suicidality do not die by suicide.14 While the former statistic may improve with increased assessment, the later will likely worsen.
Suicide is not a trivial matter. We admire those that study it and advocate for better interventions. We have compassion for those who have suffered the loss of a loved one to suicide. Our patients have died as a result of the human limitations surrounding suicide prevention. Recognizing the weight of suicide and making an effort to avoid minimizing its immense consequences drive our desire to be honest with ourselves, our patients and their families, and society. That includes the unfortunate truth regarding the current state of the evidence and our ability to enact change.
It is our concern that the rising fascination with repeated suicide risk assessment is misguided in its current form and serves the purpose of appeasing administrators more than reflecting a scientific understanding of the literature. More sadly, we are concerned that this “quantity-over-quality” approach is yet another barrier to practicing what may be one of the few interventions with any hope of meaningfully impacting a patient’s risk of suicide in the clinical setting – spending time connecting with our patients.
Dr. Badre is a clinical and forensic psychiatrist in San Diego. He holds teaching positions at the University of California, San Diego, and the University of San Diego. He teaches medical education, psychopharmacology, ethics in psychiatry, and correctional care. Dr. Badre can be reached at his website, BadreMD.com. Dr. Compton is a member of the psychiatry faculty at University of California, San Diego. His background includes medical education, mental health advocacy, work with underserved populations, and brain cancer research. Dr. Badre and Dr. Compton have no conflicts of interest.
References
1. Joint Economic Committee. (2019). Long Term Trends in Deaths of Despair. SCP Report 4-19.
2. Kroenke K and Spitzer RL. The PHQ-9: A new depression diagnostic and severity measure. Psychiatr Ann. 2013;32(9):509-15. doi: 10.3928/0048-5713-20020901-06.
3. Columbia-Suicide Severity Rating Scale (C-SSRS) Full Lifetime/Recent.
4. Columbia-Suicide Severity Rating Scale (C-SSRS) Full Since Last Contact.
5. Franklin JC et al. Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychol Bull. 2017 Feb;143(2):187-232. doi: 10.1037/bul0000084.
6. Beautrais AL. Further suicidal behavior among medically serious suicide attempters. Suicide Life Threat Behav. 2004 Spring;34(1):1-11. doi: 10.1521/suli.34.1.1.27772.
7. Belsher BE. Prediction models for suicide attempts and deaths: A systematic review and simulation. JAMA Psychiatry. 2019 Jun 1;76(6):642-651. doi: 10.1001/jamapsychiatry.2019.0174.
8. Carter G et al. Royal Australian and New Zealand College of Psychiatrists clinical practice guideline for the management of deliberate self-harm. Aust N Z J Psychiatry. 2016 Oct;50(10):939-1000. doi: 10.1177/0004867416661039.
9. Fosse R et al. Predictors of suicide in the patient population admitted to a locked-door psychiatric acute ward. PLoS One. 2017 Mar 16;12(3):e0173958. doi: 10.1371/journal.pone.0173958.
10. Kessler RC et al. Suicide prediction models: A critical review of recent research with recommendations for the way forward. Mol Psychiatry. 2020 Jan;25(1):168-79. doi: 10.1038/s41380-019-0531-0.
11. Mulder R. Problems with suicide risk assessment. Aust N Z J Psychiatry. 2011 Aug;45(8):605-7. doi: 10.3109/00048674.2011.594786.
12. Pokorny AD. Prediction of suicide in psychiatric patients: Report of a prospective study. Arch Gen Psychiatry. 1983 Mar;40(3):249-57. doi: 10.1001/archpsyc.1983.01790030019002.
13. Rosen A. Detection of suicidal patients: An example of some limitations in the prediction of infrequent events. J Consult Psychol. 1954 Dec;18(6):397-403. doi: 10.1037/h0058579.
14. McHugh CM et al. (2019). Association between suicidal ideation and suicide: Meta-analyses of odds ratios, sensitivity, specificity and positive predictive value. BJPsych Open. 2019 Mar;5(2):e18. doi: 10.1192/bjo.2018.88.
Suicide is not a trivial matter – it upends families, robs partners of a loved one, prevents children from having a parent, and can destroy a parent’s most cherished being. It is not surprising that societies have repeatedly made it a goal to study and reduce suicide within their populations.
The suicide rate in the United States is trending upward, from about 10 per 100,000 in 2000 to about 15 per 100,000 in more recent reports. The increasing suicide rates have been accompanied by increasing distress among many strata of society. From a public health level, analysts are not just witnessing increasing suicide rates, but a shocking rise in all “deaths of despair,”1 among which suicide can be considered the ultimate example.
On an individual level, many know someone who has died of suicide or suffered from a serious suicide attempt. From the public health level to the individual level, advocacy has called for various interventions in the field of psychiatry to remedy this tragic problem.
Psychiatrists have been firsthand witnesses to this increasing demand for suicide interventions. When in residency, the norm was to perform a suicide risk assessment at the time of admission to the hospital and again at the time of discharge. As the years passed, the new normal within psychiatric hospitals has shifted to asking about suicidality on a daily basis.
In what seems to us like an escalating arms race, the emerging standard of care at many facilities is now not only for daily suicide risk assessments by each psychiatrist, but also to require nurses to ask about suicidality during every 8-hour shift – in addition to documented inquiries about suicidality by other allied staff on the psychiatric unit. As a result, it is not uncommon for a patient hospitalized at an academic center to receive more than half a dozen suicide risk assessments in a day (first by the medical student, at least once – often more than once – by the resident, again by the attending psychiatrist, then the social worker and three nurses in 24 hours).
One of the concerns about such an approach is the lack of logic inherent to many risk assessment tools and symptom scales. Many of us are familiar with the Patient Health Questionnaire (PHQ-9) to assess depression.2 The PHQ-9 asks to consider “over the last 2 weeks, how often have you ...” in relation to nine symptoms associated with depression. It has always defied reason to perform a PHQ-9 every day and expect the answers to change from “nearly every day” to “not at all,” considering only 1 day has passed since the last time the patient has answered the questions. Yet daily, or near daily, PHQ-9 scores are a frequently used tool of tracking symptom improvement in response to treatments, such as electroconvulsive therapy, performed multiple times a week.
One can argue that the patient’s perspective on how symptomatic he or she has been over the past 2 weeks may change rapidly with alleviation of a depressed mood. However, the PHQ-9 is both reported to be, and often regarded as, an objective score. If one wishes to utilize it as such, the defense of its use should not be that it is a subjective report with just as much utility as “Rate your depression on a scale of 0-27.”
Similarly, many suicide scales were intended to assess thoughts of suicide in the past month3 or have been re-tooled to address this particular concern by asking “since the last contact.”4 It is baffling to see a chart with many dozens of suicide risk assessments with at times widely differing answers, yet all measuring thoughts of suicide in the past month. Is one to expect the answer to “How many times have you had these thoughts [of suicide ideation]? (1) Less than once a week (2) Once a week ...” to change between 8 a.m. and noon? Furthermore, for the purpose of assessing acute risk of suicidality in the immediate future, to only consider symptoms since the last contact – or past 2 weeks, past month, etc. – is of unclear significance.
Provider liability
Another concern is the liability placed on providers. A common problem encountered in the inpatient setting is insurance companies refusing to reimburse a hospital stay for depressed patients denying suicidality.
Any provider in the position of caring for such a patient must ask: What is the likelihood of someone providing a false negative – a false denial of suicidality? Is the likelihood of a suicidal person denying suicidality different if asked 5 or 10 or more times in a day? There are innumerable instances where a patient at a very high risk of self-harm has denied suicidality, been discharged from the hospital, and suffered terrible consequences. Ethically, the psychiatrist aware of this risk is no more at ease discharging these patients, whether it is one suicide risk scale or a dozen that suggests a patient is at low risk.
Alternatively, it may feel untenable from a medicolegal perspective for a psychiatrist to discharge a patient denying suicidality when the chart includes over a dozen previously documented elevated suicide risk assessments in the past 72 hours. By placing the job of suicide risk assessment in the hands of providers of varying levels of training and responsibility, a situation is created in which the seasoned psychiatrist who would otherwise be comfortable discharging a patient feels unable to do so because every other note-writer in the record – from the triage nurse to the medical assistant to the sitter in the emergency department – has recorded the patient as high risk for suicide. When put in such a position, the thought often occurs that systems of care, rather than individual providers, are protected most by ever escalating requirements for suicide risk documentation. To make a clinical decision contrary to the body of suicide risk documentation puts the provider at risk of being scapegoated by the system of care, which can point to its illogical and ineffective, though profusely documented, suicide prevention protocols.
Limitations of risk assessments
Considering the ongoing rise in the use of suicide risk assessments, one would expect that the evidence for their efficacy was robust and well established. Yet a thorough review of suicide risk assessments funded by the MacArthur Foundation, which examined decades of research, came to disheartening conclusions: “predictive ability has not improved over the past 50 years”; “no risk factor category or subcategory is substantially stronger than any other”; and “predicting solely according to base rates may be comparable to prediction with current risk factors.”5
Those findings were consistent with the conclusions of many other studies, which have summarized the utility of suicide risk assessments as follows: “occurrence of suicide is too low to identify those individuals who are likely to die by suicide”;6 “suicide prediction models produce accurate overall classification models, but their accuracy of predicting a future event is near zero”;7 “risk stratification is too inaccurate to be clinically useful and might even be harmful”;8 “suicide risk prediction [lacks] any items or information that to a useful degree permit the identification of persons who will complete suicide”;9 “existing suicide prediction tools have little current clinical value”;10 “our current preoccupation with risk assessment has ... created a mythology with no evidence to support it.”11 And that’s to cite just a few.
Sadly, we have known about the limitations of suicide risk assessments for many decades. In 1983 a large VA prospective study, which aimed to identify veterans who will die by suicide, examined 4,800 patients with a wide range of instruments and measures.12 This study concluded that “discriminant analysis was clearly inadequate in correctly classifying the subjects. For an event as rare as suicide, our predictive tools and guides are simply not equal to the task.” The authors described the feelings of many in stating “courts and public opinion expect physicians to be able to pick out the particular persons who will later commit suicide. Although we may reconstruct causal chains and motives, we do not possess the tools to predict suicides.”
Yet, even several decades prior, in 1954, Dr. Albert Rosen performed an elegant statistical analysis and predicted that, considering the low base rate of suicide, suicide risk assessments are “of no practical value, for it would be impossible to treat the prodigious number of false positives.”13 It seems that we continue to be unable to accept Dr. Rosen’s premonition despite decades of confirmatory evidence.
“Quantity over quality”
Regardless of those sobering reports,
One can reasonably argue that the periodic performance of a suicide risk assessment may have clinical utility in reminding us of modifiable risk factors such as intoxication, social isolation, and access to lethal means. One can also reasonably argue that these risk assessments may provide useful education to patients and their families on epidemiological risk factors such as gender, age, and marital status. But our pursuit of serial suicide risk assessments throughout the day is encouraging providers to focus on a particular risk factor that changes from moment to moment and has particularly low validity, that being self-reported suicidality.
Reported suicidality is one of the few risk factors that can change from shift to shift. But 80% of people who die by suicide had not previously expressed suicidality, and 98.3% of people who have endorsed suicidality do not die by suicide.14 While the former statistic may improve with increased assessment, the later will likely worsen.
Suicide is not a trivial matter. We admire those that study it and advocate for better interventions. We have compassion for those who have suffered the loss of a loved one to suicide. Our patients have died as a result of the human limitations surrounding suicide prevention. Recognizing the weight of suicide and making an effort to avoid minimizing its immense consequences drive our desire to be honest with ourselves, our patients and their families, and society. That includes the unfortunate truth regarding the current state of the evidence and our ability to enact change.
It is our concern that the rising fascination with repeated suicide risk assessment is misguided in its current form and serves the purpose of appeasing administrators more than reflecting a scientific understanding of the literature. More sadly, we are concerned that this “quantity-over-quality” approach is yet another barrier to practicing what may be one of the few interventions with any hope of meaningfully impacting a patient’s risk of suicide in the clinical setting – spending time connecting with our patients.
Dr. Badre is a clinical and forensic psychiatrist in San Diego. He holds teaching positions at the University of California, San Diego, and the University of San Diego. He teaches medical education, psychopharmacology, ethics in psychiatry, and correctional care. Dr. Badre can be reached at his website, BadreMD.com. Dr. Compton is a member of the psychiatry faculty at University of California, San Diego. His background includes medical education, mental health advocacy, work with underserved populations, and brain cancer research. Dr. Badre and Dr. Compton have no conflicts of interest.
References
1. Joint Economic Committee. (2019). Long Term Trends in Deaths of Despair. SCP Report 4-19.
2. Kroenke K and Spitzer RL. The PHQ-9: A new depression diagnostic and severity measure. Psychiatr Ann. 2013;32(9):509-15. doi: 10.3928/0048-5713-20020901-06.
3. Columbia-Suicide Severity Rating Scale (C-SSRS) Full Lifetime/Recent.
4. Columbia-Suicide Severity Rating Scale (C-SSRS) Full Since Last Contact.
5. Franklin JC et al. Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychol Bull. 2017 Feb;143(2):187-232. doi: 10.1037/bul0000084.
6. Beautrais AL. Further suicidal behavior among medically serious suicide attempters. Suicide Life Threat Behav. 2004 Spring;34(1):1-11. doi: 10.1521/suli.34.1.1.27772.
7. Belsher BE. Prediction models for suicide attempts and deaths: A systematic review and simulation. JAMA Psychiatry. 2019 Jun 1;76(6):642-651. doi: 10.1001/jamapsychiatry.2019.0174.
8. Carter G et al. Royal Australian and New Zealand College of Psychiatrists clinical practice guideline for the management of deliberate self-harm. Aust N Z J Psychiatry. 2016 Oct;50(10):939-1000. doi: 10.1177/0004867416661039.
9. Fosse R et al. Predictors of suicide in the patient population admitted to a locked-door psychiatric acute ward. PLoS One. 2017 Mar 16;12(3):e0173958. doi: 10.1371/journal.pone.0173958.
10. Kessler RC et al. Suicide prediction models: A critical review of recent research with recommendations for the way forward. Mol Psychiatry. 2020 Jan;25(1):168-79. doi: 10.1038/s41380-019-0531-0.
11. Mulder R. Problems with suicide risk assessment. Aust N Z J Psychiatry. 2011 Aug;45(8):605-7. doi: 10.3109/00048674.2011.594786.
12. Pokorny AD. Prediction of suicide in psychiatric patients: Report of a prospective study. Arch Gen Psychiatry. 1983 Mar;40(3):249-57. doi: 10.1001/archpsyc.1983.01790030019002.
13. Rosen A. Detection of suicidal patients: An example of some limitations in the prediction of infrequent events. J Consult Psychol. 1954 Dec;18(6):397-403. doi: 10.1037/h0058579.
14. McHugh CM et al. (2019). Association between suicidal ideation and suicide: Meta-analyses of odds ratios, sensitivity, specificity and positive predictive value. BJPsych Open. 2019 Mar;5(2):e18. doi: 10.1192/bjo.2018.88.