Annals of Internal Medicine CLINICAL GUIDELINE Assessment and Management of Patients at Risk for Suicide: Synopsis of the 2019 U.S. Department of Veterans Affairs and U.S. Department of Defense Clinical Practice Guidelines James Sall, PhD, FNP-BC; Lisa Brenner, PhD, ABPP; Amy M. Millikan Bell, MD, MPH; and Michael J. Colston, MD Description: In May 2019, the U.S. Department of Veterans Af- Recommendations: This synopsis, which includes 3 clinical fairs (VA) and U.S. Department of Defense (DoD) approved an practice algorithms, summarizes the key recommendations of update to the 2013 joint clinical practice guideline for assessing the guideline related to screening and evaluation, risk man- and managing patients who are at risk for suicide. This guideline agement and treatment, and other management methods. provides health care providers with a framework by which to screen Risk management and treatment recommendations address for, evaluate, treat, and manage the individual needs and prefer- both pharmacologic and nonpharmacologic approaches for ences of VA and DoD patients who may be at risk for suicide. patients with suicidal ideation and behavior. Other manage- ment methods address lethal means safety (such as restricting Methods: In January 2018, the VA/DoD Evidence-Based Prac- access to firearms, poisons, and medications and installing tice Work Group convened to develop a joint VA/DoD guideline barriers to prevent jumping from lethal heights) and popula- including clinical stakeholders and conforming to the National tion health strategies. Academy of Medicine's tenets for trustworthy clinical practice guidelines. The guideline panel drafted key questions, systemat- ically searched and evaluated the literature through April 2018, created algorithms, and advanced 22 recommendations in ac- Ann Intern Med. 2019;171:343-353. doi:10.7326/M19-0687 Annals.org cordance with the GRADE (Grading of Recommendations As- For author affiliations, see end of text. sessment, Development and Evaluation) system. This article was published at Annals.org on 27 August 2019. uicide is a public health problem, with worsening GUIDELINE DEVELOPMENT PROCESS Strends in recent decades. Nationwide, suicide rates The recommendations were developed according increased 25% from 1999 to 2016 (1). All states report- to a process established by the Evidence-Based Practice ing to the National Violent Death Reporting System, ex- Work Group that adheres to the standards described for cept Nevada, indicated an increase in suicide rates dur- trustworthy guidelines published by the Institute of Med- ing this period, ranging from 6% to 58% (2). During that icine in 2011 (6–8). Members of the guideline project same time, the U.S. Department of Defense (DoD) team completed conflict-of-interest disclosures for rela- active component suicide rate increased from 10.7 tionships in the previous 2 years. Web-based surveillance to 21.5 suicide-related deaths per 100 000 service (for example, by ProPublica) also was used to screen for members (3). potential conflicts of interest among project team mem- Suicide rates have been particularly high among bers. The Evidence-Based Practice Work Group selected army personnel, the service members who engaged in 3 guideline panel cochairs—1 from the VA and 2 from the DoD. The cochairs selected a multidisciplinary panel the most ground combat during the recent conflicts in of practicing clinician stakeholders, including primary Iraq and Afghanistan. In 2016, suicide occurred in 26.7, care physicians, psychologists, psychiatrists, pharma- 31.6, and 20.6 per 100 000 U.S. Army, Army National cists, nurse practitioners, social workers, and nurses, to Guard, and Army Reserve members, respectively. Ac- develop the guideline. cording to DoD Suicide Event Report data, 127 army The Evidence-Based Practice Work Group con- soldiers and 150 national guard or reserve members tracted with the Lewin Group, a third party with exper- took their own lives in 2016 (3). Among these suicides, tise in developing clinical practice guidelines, to facili- personally owned firearms were the most common tate meetings and to help draft key questions using the method used (58.9% of all suicide deaths) (4). PICOTS (population, intervention, comparator, out- Each day, 20 veterans die by suicide (4). Compared comes, timing of outcomes measurement, and setting) with age- and sex-matched civilian cohorts, veterans format. The guideline panel developed 12 key ques- have a 21% higher suicide rate (5). Differences also ex- ist between veterans who do and those who do not use Veterans Health Administration (VHA) services: An 8% See also: increase in suicides was observed among veterans who used VHA services versus 35.5% among those who did Related article ............................334 not. Of note, rates among female veterans who do not Editorial comments ....................372, 374 use VHA services have increased by 81.6%; however, Web-Only rates among female veterans who do use VHA services CME/MOC activity decreased by 2.6% (5). Annals.org Annals of Internal Medicine • Vol. 171 No. 5 • 3 September 2019 343 Downloaded from https://annals.org by VA Central Office of Acquisition Operations user on 09/03/2019 CLINICAL GUIDELINE Assessment and Management of Patients at Risk for Suicide Figure 1. Algorithm A: Identification of risk for suicide. 123 Person presents with Person identified to be at Person presents in context warning signs (may have high risk for suicide via of routine suicide risk suicidal ideation or recent predictive analytics screening self-directed violence) 4 Screen for current suicide risk: Ask the person direct questions about recent thoughts of suicide 6 5 Continue routine Are the screening No management of care and results positive?* presenting concerns Build protective factors Yes 8 If local procedures exist 7 for either completing Are safety concerns secondary suicide risk such that immediate No screening or conducting management is a comprehensive suicide required? risk evaluation, follow those procedures Yes 910Continue to Continue to Algorithm C, Algorithm B Step 19 (Figure 2) (Figure 3) * Continue to Step 7 if screening results are negative but additional evidence (e.g., collateral) suggests the need for continued screening or evaluation. tions to guide the evidence review. ECRI Institute per- tors as part of a comprehensive evaluation of suicide formed a systematic search of the peer-reviewed liter- risk. It also found support for cognitive behavioral ther- ature beginning with the end date of the literature apy (CBT)–based interventions focused on suicide pre- review from the previous version of the guideline— vention for patients with a recent history of self-directed November 2011—through April 2018. The search iden- violence, to reduce future incidents of self-directed vi- tified 70 studies relevant to the key questions, including olence. Despite a preponderance on the national sui- randomized trials, systematic reviews, and meta- cide prevention stage of strategies for community- analyses of fair or better quality. The search methods based intervention, evidence for the benefits of such and results are detailed in the full guideline (available interventions is lacking. The full guideline report pro- at www.healthquality.va.gov). All members of the vides complete recommendations, rationale, and sup- guideline panel participated in the evidence review porting evidence (www.healthquality.va.gov). and development of the recommendations in accor- dance with the GRADE (Grading of Recommendations Screening and Evaluation Assessment, Development and Evaluation) method Within clinical settings, suicide prevention includes (9–11). screening for and evaluation of suicide risk. In screen- ing patients for such risk, the question often arises as to RECOMMENDATIONS whether screening itself might cause patients to think The guideline's goal is to reduce the incidence of about suicide, subsequently increasing their risk. Al- suicide through screening for and evaluation of suicidal though we found no studies that identified risks or risk, as well as to provide quality care to patients iden- harms associated with screening patients for suicide, tified as having an elevated risk. The guideline panel screening is not problem-free. Current screening tools developed 3 algorithms to highlight current best prac- tend to have an unacceptably high false-positive pre- tices (Figures 1 to 3) and formulated 22 evidence- diction rate (that is, many persons determined to be “at based recommendations organized into 3 categories risk” never have clinically significant suicidal thoughts (Appendix Table, available at Annals.org). In general, or behavior) and a low degree of accuracy for identify- the panel found strong support for assessing risk fac- ing true cases (that is, a substantial portion of persons 344 Annals of Internal Medicine • Vol. 171 No. 5 • 3 September 2019 Annals.org Downloaded from https://annals.org by VA Central Office of Acquisition Operations user on 09/03/2019 Assessment and Management of Patients at Risk for Suicide CLINICAL GUIDELINE who die by suicide are not identified by the screening analysis of ketamine trials, 55% of patients who re- tools) (15, 16). ceived ketamine reported no suicidal ideation after 24 Several studies identified in the search support the hours and 60% reported that they were no longer hav- use of Patient Health Questionnaire-9 (PHQ-9) item 9 as ing suicidal ideations at 7 days (20). a universal screening instrument to identify suicide risk Lithium may reduce the risk for suicide in patients (16, 17). Louzon and colleagues (16) evaluated 447 245 with unipolar depression or bipolar disorder. Several patients who received PHQ-9 assessments across VHA cohort studies and systematic reviews found that lith- care settings and found that higher levels of suicidal ium maintenance therapy was associated with fewer ideation, as identified by responses to item 9, were as- suicidal behaviors and deaths (12–29). Clozapine may sociated with an increased risk for death by suicide. reduce suicidal behaviors in patients with schizophre- Likewise, Simon and colleagues (17) examined the re- nia or schizoaffective disorder (30, 31).