Identifying suicidal intent in patients is difficult; in behavioral health, valid and reliable objective measurements and physiologic testing in the diagnostic process are lacking. The first indication of error in assessing depressive symptoms and suicide risk may be an injury or death resulting from a suicide attempt. Potential failures in the diagnostic process were identified were identified in events reported from May 1, 2015, through April 30, 2018, through the Pennsylvania Patient Safety Reporting System (PA-PSRS) database. A review of PA-PSRS data found that about 20% of suicide-attempt event reports mentioned the patient had been assessed as being at low risk of suicide. A complete assessment explores not just the patient's first statements of suicidal ideation but delves deeper into other plans that the patient may have been withholding and the evolution of those plans over time. In addition, information gathering for patient assessment appeared to be inconsistent or incomplete in several event reports. Improving information gathering at all stages of the patient's crisis and ensuring that relevant data is communicated throughout the continuum of care can contribute to a more accurate diagnostic process.
The diagnostic process is impacted by key cognitive and psychological factors on the part of both the diagnostician and the patient, especially during the stages of history, assessment, and monitoring/follow-up. However, accurate and timely identification and treatment of behavioral health disorders may be more challenging than in general medicine, and failures in the process could be more difficult to identify.
The diagnostic process in behavioral health relies on identifying symptoms from patient reports and clinician observations without access to objective laboratory tests or biomarkers.1 This lack of valid and reliable confirmation makes it difficult to identify errors in the diagnostic process before they cause harm. The first indication of error in uncovering depressive symptoms and suicide risk may be an injury or death resulting from a suicide attempt. Information found in events reported through PA-PSRS related to suicide risk assessment provides an opportunity to identify diagnostic errors or failures in the behavioral health diagnostic process.
Suicide remains a significant patient safety issue, with the number of cases rising annually and now ranking as the 10th leading cause of death in the United States and 11th in Pennsylvania.2 Most of the people who die by suicide received services from a healthcare professional within the year prior to death, but providers often do not recognize the risk. Just as concerning, for each suicide there are an additional 25 attempts per year nationwide.2
Of the 9.8 million adults in the United States who reported having had serious thoughts of suicide in 2016, 2.8 million reported having made a suicide plan and 1.3 million attempted suicide. Of the adults who attempted suicide, 77% reported having made plans prior.3 A methodical suicide risk assessment presents the opportunity to understand and refine the diagnostic process and reduce the likelihood of diagnostic error resulting in unforeseen suicide attempts. Not just in behavioral health, but in medicine generally, chances are high that patients will experience at least one meaningful diagnostic error in their lifetime.4
Gathering accurate information is vital, particularly when assessing suicidal intent. Some patients may openly share their intent to attempt suicide, even if they are uncertain or the interviewer is ineffective.5 However, a skilled, detailed discussion and review of past documentation is often needed to gain an accurate portrait of a patient's true suicidal intent. Consideration should be given not just to a patient's stated intent, but also to reflected and withheld intent.
Reflected intent is the "quality and quantity of the patient's suicidal thoughts, desires, plans, and extent of action taken to complete the plans, which reflect how much the patient truly wants to commit suicide."5 Patients may withhold intent if they are unwilling to share information or consciously or unconsciously minimize their intent. Recognizing the potential pitfalls of inadequate exploration of reflected and withheld intent when assessing suicide risk can help improve the diagnostic process.
The assessment stage is another source of potential error when diagnosing depression and assessing suicide risk. Psychiatric diagnosis often relies on clinical evaluation against a prototype of symptom clusters, despite the availability of diagnostic criteria in the Diagnostic and Statistical Manual of Mental Disorders (DSM; current version is DSM-5). Although a multidimensional approach is used to assess the patient against the prototype, a subjective determination is then made whether the patient meets the diagnostic requirements.1
The provider is also tasked with weighing and deciding on each criterion with the available information, which may be incomplete. The individualized weighing of diagnostic criteria by clinicians based on subjective data poses several potential failure points to reaching an accurate diagnosis and clinical picture.1 Furthermore, false negatives, as well as misinterpretation or misinformation resulting from underlying disease process during the assessment of suicide risk, greatly increase the likelihood that the patient will be moved to an unsafe setting where attempting suicide is more probable and more likely to result in completion.5
The author reviewed the PA-PSRS database events reported by Pennsylvania healthcare facilities as "Suicide Attempt – Injury" that resulted in an unanticipated injury requiring additional healthcare services or events reported as "Suicide – Death" that occurred during a three-year period from May 1, 2015, through April 30, 2018. The author manually reviewed the total resulting set of 70 event report narratives to identify reports describing the results of the latest completed suicide risk assessment. The events were also grouped by method of injury and reported level of harm resulting from the suicidal actions.
The author conducted a review of the literature to identify the prevalence of and strategies to reduce the occurrence of failures in the diagnostic process when assessing risk for suicidal actions.
Of the 70 events, 81.4% (n = 57) were reported as "Suicide Attempt – Injury" and 18.6% (n = 13) were reported as "Suicide – Death".
In 18.6% (n = 13 of 70) of event report narratives, it was mentioned that the patient either denied suicidal ideation or had been assessed as being of low risk for suicidal behaviors prior to the event.
Information gathering appeared to be inconsistent or incomplete in several event reports.
Of method of injury among event report narratives that indicated a low-risk assessment, 38.5% (n = 5 of 13) involved self-mutilation (e.g., cutting), 15.4% (n = 2) involved a jump from height, and one each involved hanging, gunshot, intentional overdose, car traffic, and train tracks. One narrative did not mention the method of injury.
In 38.5% (n = 5 of 13) of the events involving patients assessed as low risk, the injury occurred after the patient was discharged from the reporting facility, indicating that a diagnosis of suicidal intent may have been missed during the latest assessment. Of the 13 events, 69.2% (n = 9) resulted in an unanticipated injury that required additional healthcare services, and the remaining events resulted in the death of the patient.
The review of 70 event reports revealed that nearly one-fifth specifically mentioned the patient was assessed as being at low risk for suicide due to lack of stated intent or result of assessment. In more than one-third of those instances, the patient was then discharged from the facility to a lower level of care where the suicide attempt occurred.
Accurately assessing risk for impulsive actions and suicidal intent is especially important prior to care transitions, oftentimes because the patient's ability to execute suicide plans increases as their psychotic or depressive symptoms improve. The patients involved in those events may have had a level of suicidal intent or risk of impulsive actions but did not state their intent to the clinician during the most recent assessment. This may indicate that a greater weight is given to a patient's stated intent during the assessment process than other underlying risks, or plans may have been overlooked.
In many areas of medicine, symptom-based diagnostic systems have been replaced with objective measurements and physiologic testing of physical data. This conversion has been slow within the behavioral health field. Although the DSM gives behavioral health providers a definition of symptoms (criteria) for the diagnosis of various conditions, it does little more to guide the diagnostic process.6 Limited information is available on valid biomarkers and other differential testing for behavioral health disorders. Standardized screening and assessment tools for suicide risk suffer from the same limitations. As such, behavioral health is anchored in determining diagnoses and providing treatment based on symptoms, which often have multiple potential causes. Unfortunately, treatment based on symptoms alone often fails to effectively address underlying issues completely.6
These issues are perpetuated by the limits of identifying diagnostic errors and adjusting the process accordingly. Consensus of experts, a common method for determining error in behavioral health, is problematic. To recognize an error, agreement must be reached about which diagnostic prototype fits a patient's symptoms and clinical presentation. In addition, most behavioral health diagnostic errors are discovered long after the error has been made.1 For example, a patient in whom schizophrenia or depression has been diagnosed may emerge as having bipolar disorder, but not until several years later, revealing the initial misdiagnosis.
When assessing a patient's risk for suicidal actions, it is critical to gather information directly from the patient about his or her intent to die by suicide. However, the information gathered at that time may not be truthful or complete.5 A patient may not share accurate information for a variety of reasons. Although most of these reasons are from conscious decisions, there may also be instances when a patient implements unconscious defense mechanisms, such as denial, rationalization, or intellectualization.5 Conscious reasons that patients may choose to misrepresent suicidal intent identified by The Training Institute for Suicide Assessment & Clinical Interviewing7 include the following:
- The suicide attempt may be a result of impulsivity without extensive suicidal ideation.5
- Patients may purposely choose to not relay suicidal ideation or may withhold the method of choice because they do not want to be stopped from completing the plan.5
- Patients may believe that suicide is a sign of weakness or is an immoral act.5
- Patients may be worried that they will be perceived as crazy or will be shamed.5
- Patients may fear that they will be hospitalized if suicidal ideation is shared.5
- Patients may believe that their suicidal ideation cannot be helped.5
- Patients may have trouble recognizing or describing emotional pain.8
Additionally, information gathering may be inconsistent throughout the diagnostic process. During the initial intervention, the patient may be emotionally charged—with either positive or negative feelings—and be more willing to share the truth. It is extremely important that as much data is gathered at that time as possible and equally important the information is passed along to subsequent healthcare providers.5 This relayed information can often greatly affect the level of suicide risk identified. Past documentation and information from prior providers can be weighed with current evaluations to assess suicidal intent accurately.
Although stated intent is important, withheld and reflected intent are vital pieces often not weighed as heavily during the diagnostic process. Patients may relay their suicidal intent in stages, based on how the provider responds. This is also true of information the patient may have shared, but reevaluated based on a perceived negative reaction with a previous provider during either the current episode or prior episodes.5 When patients with strong suicidal intent are asked about a plan, they often choose not to share their method of choice even when asked directly and might reveal abandoned methods first.9 For example, a patient may share a plan to shoot herself, but hold back on her true method of choice, such as to overdose on drugs, for fear the pills will be removed as an option. If that shared method is addressed by the provider, and the gun is removed, there may be a false belief that the risk of suicide is sufficiently mitigated, resulting in the patient's true suicidal intent being grossly underestimated.5
Overlooking reflected intent may lead to erroneous assumptions about a patient's potential for suicidal actions. As Shea states, "Those patients most likely to die by suicide are those patients least likely to relate their intention to do so."9
Motivational theory supports the importance of reflected intent. When looking at the motivation to do something that is hard to do but beneficial, such as substance-abuse counseling, the extent of the patient's goal-directed thinking and subsequent actions may be much better indicators of intent to proceed than stated intent. Although motivational theory is typically used to discuss initiating difficult-to-do actions for positive change, it can be equally effective for initiating a difficult-to-do action that is harmful, such as suicide. Therefore, exploring the "amount of time a patient spends thinking, planning, and practicing a suicide attempt"5 by examining the patient's history, previous clinical interactions, and statements made outside of crisis periods, may be a more accurate indicator of imminent danger than a patient's current words.5,7
Behavioral health patients are especially vulnerable to diagnostic errors related to clinical reasoning, including perceptions, failed critical thinking processes, and cognitive biases. There is a tendency for clinicians to be more judgmental with behavioral health patients and blame them for their own illnesses, potentially skewing data to fit preconceived expectations.10 This is also true for patients with substance abuse disorders, which are common comorbidities with depression and suicidal ideation. The provider's cognitive constructs can affect how criteria are selected and weighed when assessing for suicide risk.11 These cognitive errors are pervasive and predictable and require providers to take action. Croskerry provides strategies to help counteract these cognitive errors, including the following:10
- Increase awareness of cognitive biases by describing attitudes towards and among individuals with mental health disorders.
- Force consideration of alternative etiologies for symptoms so all possibilities are explored.
- Provide opportunities to step back from a problem and reflect on the thinking process.
- Use cognitive aids such as algorithms, mnemonic devices, and handheld technology, to improve the accuracy of judgments and decrease reliance on memory.
- Develop simulation opportunities to observe triggers of cognitive bias and their consequences.
- Construct training materials to compare biased and debiased approaches.
- Develop strategies to avoid anticipated bias when discussing sensitive issues such as suicidal intent.
- Make information available in a concise, clear, well-organized format.
- Minimize time pressures and provide adequate time for complete history gathering and assessment.
- Communicate clear accountability for decisions and establish a process for follow-up.
- Provide feedback as soon as possible so that errors are quickly recognized and corrected.
Because of the reliance on history taking and assessment during interviews, behavioral health providers are in a unique position to contribute to the study and improvement of the diagnostic process in all areas, not just behavioral health diagnoses and suicide risk assessment. Described or observed behaviors and responses of patients, family members, and providers may contribute to missed or incorrect diagnoses.1,10 Understanding the source and reason for those behaviors and reactions can contribute to decreased errors during the history and assessment stages of the diagnostic process.
Diagnostic reasoning, acknowledging and learning from mistakes, communicating with patients and families, and using motivational techniques to gather information are processes with strong roots in behavioral healthcare that can help improve the diagnostic process.
This study is limited by the type and number of reports collected, which depend on the degree to which facility reporting is accurate and complete. Other limitations include scant detail in some of the PA-PSRS reports and potential misinterpretation of information in the narrative descriptions.
Suicide attempts that do not result in an unanticipated injury requiring additional healthcare services might be reported as Infrastructure Failures, which are not reported to the Pennsylvania Patient Safety Authority through PA-PSRS. Events reported as "Self-mutilation" or "Ingestion" and not as "Suicide Attempt" were not included in the review, as well as potentially applicable events reported as "Other/miscellaneous."
The sample size of event report narratives reviewed is too small for statistical analysis.
The diagnostic process for behavioral health, specifically in the assessment of suicide risk, is different from most other areas of medicine. This difference presents both unique challenges and opportunities. Review of PA-PSRS event reports indicates a potential overreliance on stated intent during this assessment process. Recognizing how suicidal intent can be more thoroughly assessed and discovered, then communicated to all providers, is an important step in reducing the harm from associated failures of the diagnostic process.
- Phillips J. Detecting diagnostic error in psychiatry. Diagnosis (Berl). 2014 Jan 01;1(1):75-8. Also available: http://dx.doi.org/10.1515/dx-2013-0032. PMID: 29539971.
- Suicide statistics. [internet]. New York (NY): American Foundation for Suicide Prevention (AFSP); [accessed 2018 Apr 30]. [4 p]. Available: https://afsp.org/about-suicide/suicide-statistics/.
- Center for Behavioral Health Statistics and Quality. Results from the 2016 National Survey on Drug Use and Health: detailed tables. Rockville (MD): Substance Abuse and Mental Health Services Administration; 2017. Also available: https://www.samhsa.gov/data/sites/default/files/NSDUH-DetTabs-2016/NSDUH-DetTabs-2016.pdf.
- National Academies of Sciences, Engineering, and Medicine. Improving diagnosis in health care. Washington (DC): The National Academies Press; 2015. Also available: https://www.nap.edu/read/21794/chapter/1.
- Shea SC. Suicide assessment. Psychiatr Times. 2009;26(12):17.
- Insel TR. The NIMH Research Domain Criteria (RDoC) Project: precision medicine for psychiatry. Am J Psychiatry. 2014 Apr;171(4):395-7. Also available: http://dx.doi.org/10.1176/appi.ajp.2014.14020138. PMID: 24687194.
- Clinical interviewing tips. [internet]. Training Institute for Suicide Assessment & Clinical Interviewing (TISA); [accessed 2018 Apr 30]. [12 p]. Available: https://suicideassessment.com/useful-clinical-and-suicide-prevention-links/interviewing-tips/.
- Mays D. Structured assessment methods may improve suicide prevention. Psychiatr Ann. 2004 May;34(5):366-72. Also available: https://doi.org/10.3928/0048-5713-20040501-17.
- Shea SC. Uncovering a patient's hidden method of choice for suicide: Insights from the chronological assessment of suicide events (CASE approach). Psychiatr Ann. 2017 Aug;47(8):421-7. Also available: doi: 10.3928/00485713-20170708-01.
- Croskerry P. The importance of cognitive errors in diagnosis and strategies to minimize them. Acad Med. 2003 Aug;78(8):775-80. PMID: 12915363.
- Martin MS, Hynes K, Hatcher S, Colman I. Diagnostic error in correctional mental health: prevalence, causes, and consequences. J Correct Health Care. 2016 Apr;22(2):109-17. Also available: http://dx.doi.org/10.1177/1078345816634327. PMID: 26984134.