Trends in incidence and associated risk factors of suicide mortality in patients with non‐small cell lung cancer

Abstract Lung cancer patients have an increased risk for committing suicide. But no comprehensive study about the suicide issues among non‐small‐cell lung cancer (NSCLC) patients has been published. We aimed to estimate the trend of suicide rate and identify the high‐risk group of NSCLC patients. Patients diagnosed with primary NSCLC were identified from Surveillance, Epidemiology, and End Results (SEER) database (1973‐2013). Suicide mortality rate (SMR) were calculated. Multivariable logistic regression was employed to find out independent risk factors for suicide. Among 495 889 NSCLC patients, 694 (0.14%) of them died from suicide. The suicide mortality rates have significantly decreased (before 1993: 0.21%, 1994‐2003: 0.16%, after 2004: 0.09%, P < .001). Male (OR 6.22, 95% CI: 4.96‐7.98, P < .001), white (OR 3.89, 95% CI: 2.66‐5.97, P < .001), being unmarried (OR 1.43, 95% CI: 1.22‐1.67, P < .001), the elderly (60‐74 vs <60: OR 1.24, 95% CI: 1.03‐1.50, P = .024, >75 vs <60: OR 1.31, 95% CI: 1.05‐1.63, P = .018) were independently associated with higher risk of suicide mortality. Surgery (OR: 1.44, 95% CI: 1.19‐1.73, P < .001) was also relative with higher risk of suicide. Our study observed significant decrease in suicide mortality among NSCLC patients in US over past decades. Older age, male sex, unmarried status, and surgery were risk factors of committing suicide. Clinicians should be aware of these high‐risk groups.

lung cancer patients were Asians, men, older, widowed, small cell lung carcinoma, metastatic, and refusing treatment. 7 Lung cancer is second most common cancer, and 85% of them are non-small cell lung carcinoma (NSCLC). 8 Although several researchers have observed a high risk of suicide among lung cancer patients (being discussed as a single cancer entity), further examinations of patients with the most common subtype of lung cancer (NSCLC) are required, because of totally different distribution, treatment strategy and prognosis between subtypes. 2,7,9 However, to our knowledge, a comprehensive study about the suicide issues among NSCLC patients has not been specifically published. Given that potential suicide prevention, knowing the trend of suicide rate and the high-risk patient is of great importance. Therefore, we conducted this study using a large populationbased database to estimate the trend of suicide rate and identify the high-risk group of NSCLC patients. In addition, we also performed a sub-analysis of patients diagnosed from 2004 to 2013 to depict recent issues.

| MATERIAL AND METHODS
National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database is an authoritative source of information on cancer incidence and survival in the United States. SEER database encompasses about 28% of United State population and collect cases diagnosed between 1973 and 2013.
Patients whose cause of death variable coded as "Suicide and self-inflicted injury" were identified. We obtained the demographic and clinicopathological data from the SEER database, including age, sex, race, marital status, year at diagnosis, state, tumor site, grade, histologic type, stage, surgery, cause of death, survival time, vital status and radiation. Patients were divided into 3 groups according to age at diagnosis (younger than 60 years, 60-74 years, and older than 75 years). Race were sorted by white, black and others. We classified patients as married or unmarried. Year of diagnosis were separated into 3 groups (before 1993, 1994-2003, after 2004). We classified the tumor site as upper lung, middle lung, lower lung, and bronchus/others. Grade of tumor were categorized into I/II, III/IV, and unknown groups. Surgery and radiation were both classified as performed, not performed, and unknown. Disease stage for the analysis was coded based on the variable "SEER Historic Stage A." We give a value of 0.5 months to those who didn't survive for a full month after diagnosis, because SEER record their survival time in months.
Univariate analysis using chi-square test was used to compare patients committed suicide with those died from other causes. Multivariable logistic regression was employed to find out independent risk factors for suicide. All statistical analyses were performed using R version 3.4.2 software (Institute for Statistics and Mathematics, Vienna, Austria; www.r-project. org). Statistical significance was set at two-sided P < .05.

| Sub-analysis of patients diagnosed from 2004 to 2013
This subgroup of patients can better represent the demographic and clinicopathological character of recent patients. So we do the sub-analysis of patients diagnosed from 2004 to 2013.
Univariate analysis displayed that higher suicide mortality rate was associated with male patients (P < .001), white patients (P < .001), and patients didn't have radiation therapy (P = .115). Concerning the time after diagnosis, the highest suicide mortality rate was found to be the first year after diagnosis (P = .008).

| DISCUSSION
Our study observed significant improvement in suicide prevention among NSCLC patients in US over past decades. Urban et al 9 found that suicide has not changed significantly decreased in lung cancer over time. However, in contrast to rising suicide rate of US general population, the suicide mortality rate of NSCLC patients has decreased considerably over past decades, which is consistent with previous study about suicide trend among cancer patients. [12][13][14] This result may be associated with relatively better prognosis of NSCLC, because of early screening test for lung cancer and significant advances in treatment, such as chemotherapy and targeted therapy. [15][16][17] Demographic characteristics associated with an increased rate of suicide in the NSCLC patients, such as older age, male sex, race were similar to those in general population. 18 Earlier research showed that older people tend to commit complete suicide among general population. 19,20 Older patients with cancer are also high-risk group, which is consistent with our research. [21][22][23] Older patients usually encountered with greater disease burden, and social psychological pressure. Higher suicide rate of older NSCLC patients may be related to pressure and depression. 20 Another possible reason is that, they hold the rational will of ending their life at the right time. 24 Male sex is a risk factor of suicide in NSCLC patients, and it is accordant with trends in general population and those with other cancer. 2,25 Although depression seemed to be higher in female patients with NSCLC, male patients are more likely to succeed in ending their own life. 26,27 However, the incidence of female suicidal behavior in NSCLC patients may be underrepresented, because failed suicide attempts were not recorded in the SEER database. 22 Race has a significant impact on suicidal ideation. The risk of dying from suicide was more than double for the white NSCLC patients than for the black patients. 28 The reason is for higher suicide rate in white patients with NSCLC is still unknown, and hopelessness in those patients is likely to associate with suicidal behavior. 29 In addition, unmarried NSCLC patients are easier exposed to suicide attempts. Married patients have a greater socioeconomic status than unmarried. 30,31 Many cancer research studies have reported a poor prognosis in unmarried patients. [32][33][34] Interestingly, characteristics of NSCLC seems to be not relevant to suicide of patients, which is controversial with former analysis. 9 In patients diagnosed between 1973 and 2013, clinical characteristics such as primary site, histologic type, historic stage have no significant influence on suicide. The same as those diagnosed recently. A possible reason is that multiple primary tumor may interfere with the result. In our study, patients with multiple primary tumor were excluded to prevent interference of other tumors. Another possible reason is that the prognosis of patients with advanced NSCLC is poor, they may die from the disease itself rather than other causes. Patients diagnosed between 1973 and 2013 who undergo treatment like surgery are more likely to suicide. This is also different from previous research. 9 It can be explained by debility and loss of autonomy brought by the curative surgery toward NSCLC. 35 When considering patients diagnosed between 1973 and 2013, Suicide mortality rate (SMR) is not significantly associated with time after diagnosed. But SMR is observed to be higher within the first year of NSCLC diagnosis among patients diagnosed between 2004 and 2013. It has been reported that cancer patients were at high risk of suicide within the first year of diagnosis and associated demographic and clinical factors were analyzed. 36 The different result can be possibly explained by the improvement of life quality of long-term survivor of NSCLC patients, and the major reason for suicidal behavior of recent NSCLC patients is possibly shock of cancer diagnosis. 37,38 While the demographic risk factors of cancer-related suicide are focused on, psychological and social risk factors are frequently missed. Suicide is a complicated phenomenon that biological, psychological and social risk factors would interact and influence on it. Cancer diagnosis may lead to demoralization of NSCLC patients, such as hopelessness and helplessness, which can lead to suicidal ideation. 39 Poor consequences of cancer treatment may bring physical and mental pain to NSCLC patients. 40 It has been reported that cancer patients with low socioeconomic status and family support are more likely to suicide. 41 The National Comprehensive Cancer Network provides a distress management guideline that recommends screening all patients for distress. 42 Our findings may assist oncologists to effectively identify those NSCLC patients at higher risk of suicide, specifically for older, white, unmarried male patients with surgery. For those NSCLC patients with high risk of suicide, we should pay more attention, because appropriate psychosocial interventions have a positive impact on quality of life. 43 To reduce cancer-related suicide, patients' understanding of cancer diagnosis and treatment options should be ensured. 44 Besides, promoting family communication combine with encouraging selfdetermination and participation in treatment can mitigate social risk of suicide. 44,45 There are some limitations in our study. Suicide is a complicated phenomenon affected by factors such as economic level and education level. These factors are not included in SEER database. Additionally, SEER database only contains data of US patients, so our study is limited to US patients, and research over the world is still needed. Moreover, details of treatment to NSCLC were not taken in, and we only know whether patients have undergone surgery and radiation or not. Details of treatment such as the time after surgery and chemotherapy may be associated with suicide of cancer patient. 3 Furthermore, our study can't obtain suicide attempts data of NSCLC patients, and patients potential to suicide were likely to be underestimated.

| CONCLUSIONS
In summary, our study observed a significant decrease in suicide mortality among NSCLC patients in US over past decades. Older age, male sex, unmarried status, and surgery were risk factors of committing suicide. Grade, stage, histologic type and primary site of NSCLC appear not relate to suicide. It can help clinicians identified these NSCLC patients for better support and suicide prevention. Further studies are still needed.