1 INTRODUCTION

In 2019, the adjusted suicide rate among Veterans was 52.3% greater than among non-Veteran US adults.¹ Among Veterans, those with recent Veterans Health Administration (VHA) encounters had greater suicide risk than other Veterans.² Suicide has many causes and as part of ongoing Veterans Affairs (VA) suicide surveillance it is important to assess potential suicide risk factors.

Recent studies suggest that new cancer diagnoses are positively associated with suicide risk, particularly in the months immediately following diagnosis.^3-7 Suicide risk corresponding to a cancer diagnosis varies by cancer stage, type, prognosis, and other patient characteristics.^{3-5, 7} Cancer patients within the US from 1973–2014 had 4.44 times the suicide risk of the general population. This risk was higher immediately following diagnosis, among younger patients, and in more recent years.³ Among US adults diagnosed with cancer 2000–2014, odds of suicide were highest in the second month following diagnosis and for distantly metastasized, pancreatic, and lung cancer subtypes.⁴ Furthermore, the suicide standardized mortality ratios corresponding to new cancer diagnoses increased in more recent years, rising from 1.27 in 2000–2007 to 1.58 in 2007–2013.⁷

The Interpersonal Theory of Suicide posits that active suicidal desire is driven by the presence of thwarted belongingness, perceived burdensomeness, and hopelessness that these interpersonal struggles will improve, while the capacity for suicidal behavior is an independent process driven by fearlessness and familiarity with pain.⁸ A cancer diagnosis may represent a life event that triggers both interpersonal struggles, offers little hope for change in the future, and increases a patient's capability for suicide through an increase in pain tolerance and acceptance of death.⁸ Cancer diagnosis and subsequent treatment may diminish social engagement, reciprocal relationships, and perceived belonging, while the physical and financial strain of cancer may increase feelings of burdensomeness. Among cancer patients, physical stressors associated with suicidal behavior include pain, disfigurement, body-image concerns, and dysfunction.^9-12 Financial distress affects approximately 47%–49% of cancer survivors and increases risk for depression and anxiety.^{13, 14} Cancer diagnoses can also stress social support networks, leading to relationship troubles.^{15, 16} Among Veterans diagnosed with cancer, those reporting relationship problems had a five times greater likelihood of suicidal ideation than those without them.¹⁷ Physiological factors, such as cytokines produced by tumors and glucocorticoids used to treat cancer, may also increase depression and anxiety risk.¹⁷ A high prevalence of mental health and substance use disorders among cancer patients may also increase suicide risk among cancer patients.^{18, 19}

On average, the Veteran population is older than the non-Veteran US adult population and has a higher burden of comorbid health conditions, including cancer.²⁰ Veteran VHA users diagnosed with cancer have an elevated prevalence of suicide risk factors, including PTSD symptoms, suicidal ideation, and mental health and substance use disorders.^{17, 21, 22} Understanding associations between new cancer diagnoses and suicide risk, overall and adjusting for mental health disorders, is important for enhancing Veteran suicide prevention efforts. Although cancer has been associated with adverse mental health outcomes among Veterans, we know of no studies that have assessed the relative suicide risk of Veteran VHA users who receive a new, or incident, cancer diagnosis. Furthermore, prior studies of cancer and suicide in the US population have often lacked adjustment for comorbid mental health conditions.^{3, 4, 7}

The Veterans Health Administration's Suicide Prevention Now plan aims to identify and target strategies to immediately prevent Veteran suicide, including the identification of high-risk medical diagnoses. In support of this initiative, the aims of these analyses were to: (1) assess the association between new cancer diagnoses and suicide risk among Veterans engaged in VHA care, adjusting for mental health comorbidities, (2) identify cancer subtypes and post-diagnosis periods of elevated risk for suicide, and (3) determine whether method of suicide among suicide decedents differs for those with versus without an incident cancer diagnosis.

2 METHODS

All analyses were conducted as a part of VA's Office of Mental Health and Suicide Prevention operations and quality improvement activities, and institutional review board was not required.

3 RESULTS

3.2 High suicide risk periods following a new cancer diagnosis

Table 3 presents suicide rates per 100,000 person-years in the months following a new cancer diagnosis. Among 240,410 Veteran VHA users with a new cancer diagnosis 2012–2018, 393 had a subsequent death by suicide in this period. Suicide rates were highest during the first 3 months following diagnosis (rate = 128.3, 95% CI: 100.4, 161.6), accounting for 18.3% of the 393 documented suicides (Table 3 & Figure 1). Relative to those without any cancer diagnosis (rate = 39.4, 95% CI: 38.7, 40.2), rates among cancer patients remained elevated during the 3–6 months (rate = 78.7, 95% CI: 56.2, 107.2) and 6–12 months (rate = 66.6, 95% CI: 50.8, 85.7) following diagnosis (Table 3 and Figure 1).

TABLE 3. Suicide rates following any new cancer diagnosis among Veteran VHA users

Time interval following a new cancer diagnosis	N, Veterans alive at the beginning of the interval	N, suicides in risk perioda	Suicide ratea per 100,000 person-years (95% confidence interval)
0–3 months following cancer diagnosis	240,410	72	128.3 (100.4, 161.6)
3–6 months following cancer diagnosis	215,440	40	78.7 (56.2, 107.2)
6–12 months following cancer diagnosis	197,323	60	66.6 (50.8, 85.7)
12–24 months following cancer diagnosis	168,501	65	44.1 (34.1, 56.3)
24–36 months following cancer diagnosis	127,454	70	62.7 (48.9, 79.2)
36–48 months following cancer diagnosis	95,734	39	47.8 (34.0, 65.3)
48–60 months following cancer diagnosis	67,741	28	50.5 (33.5, 73.0)
60–72 months following cancer diagnosis	43,184	14	43.1 (23.6, 72.4)
72–84 months following cancer diagnosis	21,459	-	-

• ^a Suicide rates based on suicide counts less than 10 are suppressed per CDC reporting guidelines, indicated with a “-.” Italicized rates are based on counts less than 20, which indicate unstable rates.

4 DISCUSSION

Cancer was less common among Veteran VHA users who died by suicide when compared with other Veteran VHA users. However, after accounting death by non-suicidal causes of death and adjusting for important confounders using cox proportional hazards regression, an incident cancer diagnosis was associated with increased suicide risk. Cancer sub-types with increased suicide risk included esophageal, head and neck, and lung cancer. Cancer stages at diagnosis associated with increased suicide risk included stages 3, 4, and unknown. Suicide risk was particularly high in the first 3 months following a new cancer diagnosis, and in this period, deaths were more likely to involve firearms.

Study findings are consistent with the literature indicating positive associations between an incident cancer diagnosis and suicide risk among the general US adult population, with risk highest in the months immediately following diagnosis.^{3, 4, 7, 25} As with this study, prior work has shown that individuals diagnosed at more advanced stages^3-7 and with lung,^3-5 esophageal^5-7 or head and neck^{3, 9, 26} cancer were at elevated risk for suicide. There are several factors which may increase suicide risk within these cancer subtypes.^9-12, First, disruptions to basic functions such as swallowing, talking, and breathing may contribute to reduced quality of life which is associated with higher risk of suicidal behavior in cancer patients.^{10, 11} Second, the poor prognosis corresponding to these subtypes likely leads to feelings of hopelessness.^{10, 11} Third, due to lung cancer's associations with well-known, modifiable behaviors there may be a unique component of self-blame which contributes to poorer psychological adjustment and subsequent elevations in suicide risk.²⁷

In contrast to the current literature, we did not find evidence for an increase in suicide risk (after adjusting for multiple comparisons) corresponding to colorectal,^{3, 4, 6} pancreatic,^{4, 5, 7} prostate,³ bladder,³ gastric,²⁸ or leukemia cancers.³ Limited power within these cancer types may account for some of these observed differences. For cancer sub-types with sufficient power, such as prostate cancer, our findings may also represent differences specific to the Veteran VHA user population.

Among Veteran suicide decedents, those with a cancer diagnosis used firearms more often than those without cancer and were most frequently involved among suicide decedents with a cancer diagnosis in the prior 3 months. Similarly, prior studies found that 93% of VHA Veteran suicide decedents diagnosed with lung cancer died by firearm and US adult suicide decedents with a history of cancer were more likely to die by firearms.^{29, 30} Furthermore, within US adult and Veteran suicide decedent populations, those who used firearms were less likely to have mental health diagnoses and other common suicide risk factors than those who died by other methods.^{31, 32} Together, this may suggest that cancer represents an acute crisis that can trigger suicidal behavior in the absence of prior mental health conditions. Alternatively, an increased propensity to firearm suicide following a cancer diagnosis may indicate that cancer's effect on suicide risk is specific to Veterans who are more likely to use a firearm method of suicide.

Strategies to prevent Veteran suicide among VHA patients diagnosed with cancer may be warranted. Root cause analyses found that a majority of VHA suicide decedents diagnosed with cancer died following a new diagnosis or change in clinical status, suggesting that these may be key intervention points for suicide prevention.³³ Among Veterans diagnosed with cancer, post-traumatic stress disorder (PTSD) is positively associated with suicide risk and combat- and cancer-related PTSD interact to increase the likelihood of chronic pain, a well-known suicide risk factor.^{22, 34-36} Additionally, factors corresponding to increased risk for suicidal-behavior among Veterans diagnosed with head and neck cancer included prior mood disorder diagnoses and mental health and substance use disorder encounters within 90 days of cancer diagnosis. Receipt of palliative care within 90 days of diagnosis was associated with lower suicidal-behavior risk.³⁷ In addition, engaging non-small cell lung cancer diagnosed Veterans with pre-existing mental health conditions in mental health treatment was associated with significant decreases in all-cause and cancer-specific mortality.³⁸ These findings indicate that suicide prevention initiatives within mental health specialties and referral to palliative care may support Veteran oncology patients.

However, awareness of oncology patient's suicide risk is necessary across all specialties. A root cause analysis among VHA oncology patients found that deaths by suicide were preceded by a lack of coordinated care, patient or staff education, or communication.³⁹ These findings correspond well with several promising practices for suicide prevention in oncology settings including (1) routine distress and suicidal risk screening^{40, 41} (2) normalized discussion about suicidal behavior between oncologist and patients,^{40, 42} and (3) interdisciplinary communication and collaboration between oncologists, social workers, mental health care providers, and chaplains.^{18, 42, 43} In addition, documentation of Veteran distress in the electronic health record can facilitate clinician awareness to the social, mental, and physical needs of their cancer patients⁴⁴ and training oncologists on lethal means safety counseling may help promote safety among oncology patients.⁴² We found that suicide risk was highest among late-stage cancer patients who often receive systemic, non-curative intent treatment. It is important to consider which of these suicide prevention strategies may be best suited to supplement these treatment plans.

This study had four important strengths. To our knowledge, this is the first study of the association between cancer and suicide risk specific to Veterans engaged in VHA care. Second, study analyses adjusted for important mental health and substance use disorder confounders. This was not possible in many prior US studies of cancer and suicide, which relied on the National Cancer Institute's SEER database.^{3, 4, 7} Third, use of comprehensive National Death Index death certificate search results from the VA/DoD MDR, the gold standard for death data, enhanced suicide mortality identification. Fourth, by adjusting for multiple comparisons we limited the likelihood of false positives (type I errors), which had not been accounted for in previous studies.

We note four important study limitations. First, statistical power was limited for cancer diagnoses that are uncommon within the VHA patient population, which is predominantly older males. Second, study data were limited to measures available in the VHA electronic health record, and there may be misclassification of the exposure and incomplete assessment of potential confounders, including mental health morbidity. If, as observed here, cancer is associated with increased suicide risk, then incomplete capture of cancer diagnoses may have attenuated the observed effect of cancer on suicide. Third, due to limited computing power, adjusted analyses were completed using a data driven approach; only cancer subtypes with bivariate associations with suicide risk were assessed using the full model, which adjusted for mental health comorbidities. Finally, it was beyond the scope of this study to investigate the impact of cancer treatment type, cancer progression, and VHA services utilization on suicide risk but these may be important mediators of suicide risk that should be investigated in the future.

5 CONCLUSION

Study findings document a positive association between a new cancer diagnosis and suicide risk among Veterans engaged in VHA care. Cancer subtypes with increased risk for suicide included esophageal, lung, and head and neck cancer as well as cancers diagnosed at stages 3 or 4 or of an unknown stage. Veterans were at highest risk for suicide in the first 3 months following diagnosis and among suicide decedents during this period firearms were more likely to be involved. VHA oncologists should be aware of and attentive to the mental health needs of their patients. Lethal means safety counseling, routine mental health screening, normalizing discussion about suicide, improving coordination and awareness between patient teams, and engaging cancer patients in mental health treatment may facilitate suicide prevention efforts within VHA oncology.

AUTHOR CONTRIBUTIONS

Kallisse Dent curated data, conducted formal analysis, and was responsible for writing the original draft of the manuscript. Benjamin Szymanski also provided analytic input for formal analysis. John McCarthy, Ira Katz, and Michael Kelley provided supervision. All authors contributed to conceptualization, methodology, and review, and editing of the manuscript.

FUNDING INFORMATION

This work was conducted as part of ongoing operations in the Veterans Affairs Office of Mental Health and Suicide Prevention. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the VA.

CONFLICT OF INTEREST

The authors have no conflict of interest to disclose.

ETHICS STATEMENT

All analyses were conducted as a part of VA's Office of Mental Health and Suicide Prevention operations and quality improvement activities, and institutional review board was not required.