Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

VA Health Systems Research

Go to the VA ORD website
Go to the QUERI website

CDA 09-216 – HSR Study

CDA 09-216
Assessing Medications as Interventions to Prevent Suicide in the VHA
Eric G. Smith, MD PhD MPH
VA Bedford HealthCare System, Bedford, MA
Bedford, MA
Funding Period: August 2010 - July 2015
Suicide prevention is a key priority of the Veterans Health Administration (VHA). Studies have suggested that the mood stabilizer lithium may prevent suicide in patients with mood disorders, but most studies are observational, lack thorough control for confounding, and do not involve Veterans.

Primary: To use advanced comparative effectiveness techniques such as high-dimensional propensity scores to assess whether lithium is associated with reduced risks of suicide for VHA patients. Lithium was compared to valproate, the most popular anticonvulsant mood stabilizer. Secondary: Assess the rates and consequences of discontinuation/nonadherence and the association of lithium with nonsuicide mortality.

A high-dimensional propensity score of over 900 covariates from the VHA's detailed clinical databases was defined and used to 1:1 match new users of lithium and valproate. A matched sample of 42,388 patients (21,194 patients/treatment) was derived that retained 99% of lithium new users, and balanced lithium and valproate new users extremely closely (standardized difference < 0.018) on all covariates. Covariates included age, sex, other demographics, recent psychiatric and nonpsychiatric hospitalizations, recent psychiatric and nonpsychiatric clinic visits, recent and current psychiatric and nonpsychiatric medications, and some nontraditional risk factors (e.g., discharges against medical advice). No significance difference was observed in suicide risk over 365 days in the primary, intent-to-treat analysis (OR 1.22, p=0.32), but higher risks were observed in patients discontinuing lithium than valproate (significant over 180 days at OR 2.72, p=0.015). Such discontinuation-associated risks may reflect either risks produced by lithium discontinuation, confounding biasing against lithium, or both. When individuals with bipolar disorder versus other mood or psychotic disorders were compared, lithium was associated with increased risks of suicide death over 365 days (HR 1.50, p=0.026), but this increased risk was attributable entirely to risk observed after discontinuation, not during active treatment. Among patients without bipolar disorder, lithium was associated with nonsignificant but nontrivial protective effects in both intent-to-treat analysis (HR 0.77, p=0.26) and active treatment (HR 0.43, p=0.11). An analysis of nonsuicide mortality risks also found increased risk upon lithium discontinuation after 180 days (HR 1.54, p=0.045), similar to the suicide mortality analyses. This suggests that post-discontinuation risks are not simply the result of confounding based on practices concerning suicide risk. Intent-to-treat analyses showed significant protective association of lithium with nonsuicide mortality over 90 days (HR 0.67, p=0.003), but not longer (when risks observed after lithium discontinuation reduced the overall protective association between lithium initiation and nonsuicide mortality). Significantly reduced nonsuicide mortality risks were observed among patients receiving active lithium treatment over all time periods examined (e.g., HR[365d] 0.62, p=0.002).

Not yet available.

High. While confounding cannot be excluded from nonrandomized designs, the observation of emergent risks upon the discontinuation of lithium, especially in the nonsuicide mortality analysis, do not appear to be easily explained completely by confounding. The nonsignificant findings of reduced risks of suicide in patients without bipolar disorder receiving active lithium treatment also is not easily explained by confounding (although random error cannot be excluded). This study may immediately impact clinical care by emphasizing the need to reduce the rate of lithium discontinuation and enhance the care received after lithium discontinuation. This study may also promote rethinking of suicide prevention strategies by suggesting that any benefits that lithium may have against suicide may be more easily observed in patients without bipolar disorder than in patients with bipolar disorder. The nonsuicide mortality findings suggest that additional attention needs to be paid to the potential general health impacts of psychiatric medications. This attention should include evaluating the possibility that some psychiatric medications may appreciably reduce mortality risks. This study helped inform the design of the VHA's Collaborative Studies Program trial evaluating lithium's impact on suicidal behaviors. My CDA research also undoubtedly played a major role in my being invited to the White House to participate in a roundtable discussion of suicide prevention leaders.

External Links for this Project

Dimensions for VA

Dimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.

Learn more about Dimensions for VA.

VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address.
    Search Dimensions for this project


Journal Articles

  1. Smith EG, Henry AD, Zhang J, Hooven F, Banks SM. Antidepressant adequacy and work status among Medicaid enrollees with disabilities: a restriction-based, propensity score-adjusted analysis. Community mental health journal. 2009 Oct 1; 45(5):333-40. [view]
  2. Smith EG, Deligiannidis KM, Ulbricht CM, Landolin CS, Patel JK, Rothschild AJ. Antidepressant augmentation using the N-methyl-D-aspartate antagonist memantine: a randomized, double-blind, placebo-controlled trial. The Journal of clinical psychiatry. 2013 Oct 1; 74(10):966-73. [view]
  3. Stankiewicz BW, Smith EG, Herz L. Brief CBT and Suicide Risk: Ruling Out Nonspecific Effects of Individual Therapy. The American journal of psychiatry. 2015 Oct 1; 172(10):1022. [view]
  4. Del Vecchio N, Elwy AR, Smith E, Bottonari KA, Eisen SV. Enhancing self-report assessment of PTSD: development of an item bank. Journal of traumatic stress. 2011 Apr 1; 24(2):191-9. [view]
  5. Smith EG, Austin KL, Kim HM, Eisen SV, Kilbourne AM, Miller DR, Zivin K, Hannemann C, Sauer BC, Valenstein M. Mortality associated with lithium and valproate treatment of US Veterans Health Administration patients with mental disorders. The British Journal of Psychiatry; The Journal of Mental Science. 2015 Jul 1; 207(1):55-63. [view]
  6. Gören JL, Rose AJ, Engle RL, Smith EG, Christopher ML, Rickles NM, Semla TP, McCullough MB. Organizational Characteristics of Veterans Affairs Clinics With High and Low Utilization of Clozapine. Psychiatric services (Washington, D.C.). 2016 Nov 1; 67(11):1189-1196. [view]
  7. Miller CJ, Li M, Penfold RB, Lee AF, Smith EG, Osser DN, Bajor L, Bauer MS. Patterns of initiation of second generation antipsychotics for bipolar disorder: a month-by-month analysis of provider behavior. BMC psychiatry. 2014 Nov 30; 14(1):339. [view]
  8. Smith EG, Austin KL, Kim HM, Miller DR, Sauer BC, Valenstein M. Suicide death over the first year of lithium versus valproate treatment in cohorts with and without bipolar disorder. Journal of psychiatric research. 2022 Mar 1; 147:349-356. [view]
  9. Smith EG, Kim HM, Ganoczy D, Stano C, Pfeiffer PN, Valenstein M. Suicide risk assessment received prior to suicide death by Veterans Health Administration patients with a history of depression. The Journal of clinical psychiatry. 2013 Mar 1; 74(3):226-32. [view]
  10. Smith EG, Austin KL, Kim HM, Miller DR, Eisen SV, Christiansen CL, Kilbourne AM, Sauer BC, McCarthy JF, Valenstein M. Suicide risk in Veterans Health Administration patients with mental health diagnoses initiating lithium or valproate: a historical prospective cohort study. BMC psychiatry. 2014 Dec 17; 14:357. [view]
  11. Smith EG. The ACCE method: an approach for obtaining quantitative or qualitative estimates of residual confounding that includes unmeasured confounding. F1000 research. 2014 Aug 11; 3(3):187. [view]
  12. Smith EG. Treatment of veterans with depression who die from suicide: timing and quality of care at last VHA visit. The Journal of clinical psychiatry. 2011 Feb 1; 72(1):622-9. [view]
  13. Smith EG, Zhao S, Rosen AK. Using the patient safety indicators to detect potential safety events among US veterans with psychotic disorders: clinical and research implications. International journal for quality in health care : journal of the International Society for Quality in Health Care. 2012 Aug 1; 24(4):321-9. [view]
  14. Kim HM, Smith EG, Stano CM, Ganoczy D, Zivin K, Walters H, Valenstein M. Validation of key behaviourally based mental health diagnoses in administrative data: suicide attempt, alcohol abuse, illicit drug abuse and tobacco use. BMC health services research. 2012 Jan 23; 12(1):18. [view]
Journal Other

  1. Smith EG. Additional effect size measures helpful in understanding lithium and valproate trial results. The American journal of psychiatry. 2012 Jan 1; 169(1):97-8; author reply 99. [view]

DRA: Mental, Cognitive and Behavioral Disorders, Health Systems
DRE: Prevention, Treatment - Comparative Effectiveness, Treatment - Observational
Keywords: Pharmaceuticals, PTSD
MeSH Terms: none

Questions about the HSR website? Email the Web Team

Any health information on this website is strictly for informational purposes and is not intended as medical advice. It should not be used to diagnose or treat any condition.