Longitudinal Risk for Suicidal Self-Directed Violence Among Veterans With Cancer

JAMA Oncol. 2026 May 28. doi: 10.1001/jamaoncol.2026.1459. Online ahead of print.

ABSTRACT

IMPORTANCE: Patients with cancer experience significantly higher rates of suicidal self-directed violence (SSDV; defined as both fatal and nonfatal suicide attempts) than the general population.

OBJECTIVE: To assess longitudinal risks and methods for SSDV among veterans and identify associated risk factors to improve screening and prevention strategies.

DESIGN, SETTING, AND PARTICIPANTS: This national cohort study of veterans with cancer was conducted between January 2014 and December 2023. Data from established oncology and suicide registries and the Veterans Health Administration (VA) were used. Data were analyzed from January 2025 to February 2026.

EXPOSURE: Diagnosis of invasive solid or hematologic cancer.

MAIN OUTCOMES AND MEASURES: The primary outcome was SSDV and rates per 100 000 person-years. Risk factors for SSDV were estimated as adjusted hazard ratios (aHRs) from multivariable Cox proportional hazards models.

RESULTS: Among 292 271 veterans (mean [SD] age, 69.0 [9.4] years; 7108 female [2%]; 2664 American Indian or Alaska Native [1%]; 1365 Asian [1%]; 62 538 Black or African American [21%]; 13 965 Hispanic or Latino [5%]; 2306 Native Hawaiian or Other Pacific Islander [1%]; 219 205 White [75%]), there were 2400 SSDV events (1%; overall rate, 203 [95% CI, 195-211] per 100 000 person-years). The most common method used was poisoning (eg, opioids; 617 attempts [26%]). Estimated SSDV probabilities were highest for those with central nervous system (CNS), pancreas, head and neck, liver and biliary system, and thyroid cancer types. Veterans with severe frailty (544 [95% CI, 457-648] events per 100 000 person-years), advanced cancer (261 [95% CI, 233-293] events per 100 000 person-years), chronic mental illness (419 [95% CI, 399-439] events per 100 000 person-years), and high pain scores (236 [95% CI, 192-210] events per 100 000 person-years) had high SSDV rates compared with the overall cohort. Younger age (≤45 years; 643 [95% CI, 547-756] events per 100 000 person-years), female sex (369 [95% CI, 306-445] events per 100 000 person-years), American Indian or Alaska Native race (286 [95% CI, 201-407] events per 100 000 person-years), and CNS (394 [95% CI, 311-500] events per 100 000 person-years) and thyroid (359 [95% CI, 290-445] events per 100 000 person-years) cancers had high rates of nonfatal attempts. Increased SSDV hazards (6 months postdiagnosis) occurred among veterans of Asian compared with White race (aHR, 2.55; 95% CI, 1.12-5.76), unmarried veterans (aHR, 1.83; 95% CI, 1.47-2.27), veterans with CNS (aHR, 2.07; 95% CI, 1.13-3.80) or head and neck (aHR, 1.67; 95% CI, 1.13-2.48) compared with lung cancer, and veterans with advanced cancer (aHR, 1.30; 95% CI, 1.00-1.68). Risk for most veterans decreased over time after diagnosis; however, risk remained elevated 5 years postdiagnosis for younger veterans (aged ≤45 vs 46-64 years; aHR, 1.58; 95% CI, 1.29-1.94), unmarried veterans (aHR, 1.48; 95% CI, 1.35-1.62), veterans with CNS vs lung cancer (aHR, 1.63; 95% CI, 1.22-1.27), and veterans with advanced cancer (aHR, 1.30; 95% CI, 1.14-1.50).

CONCLUSIONS AND RELEVANCE: This study found that veterans with cancer were at risk for SSDV that persisted years into survivorship. Previously overlooked high risk, subgroups, such as younger veterans, Asian veterans, or veterans with thyroid cancer, stress the need to systematically track all suicidal behaviors, not just fatal attempts, to inform tailored screening and prevention strategies as a key component of cancer care.

PMID:42207508 | DOI:10.1001/jamaoncol.2026.1459