Risk factors associated with suicide among leukemia patients: A Surveillance, Epidemiology, and End Results analysis

Abstract Previous studies have found that the risk of suicide is higher in patients diagnosed with cancer than in the general population. We aimed to identify potential risk factors associated with suicide in leukemia patients by analyzing data obtained from the Surveillance, Epidemiology, and End Results (SEER) database. We screened the SEER database for leukemia patients added between 1975 and 2017, and calculated their suicide rate and standardized mortality rate (SMR) relative to the total United States population from 1981 to 2017 as a reference. Univariate and multivariate Cox regression analyses were used to determine the risk factors for suicide in leukemia patients. We collected 142,386 leukemia patients who had been added to the SEER database from 1975 to 2017, of whom 191 patients committed suicide over an observation period of 95,397 person‐years. The suicide rate of leukemia patients was 26.41 per 100,000 person‐years, and hence the SMR of the suicided leukemia patients was 2.16 (95% confidence interval [CI] = 1.85–2.47). The univariate and multivariate Cox regression analyses showed that a high risk of suicide was associated with male sex (vs. female: hazard ratio [HR] = 4.41, 95% CI = 2.93–6.63, p < 0.001), older age at diagnosis (60–69 years vs. ≤39 years: HR = 2.60, 95% CI = 1.60–4.23, p < 0.001; 70–79 years vs. ≤39 years: HR = 2.84, 95% CI = 1.72–4.68, p < 0.001; ≥80 years vs. ≤39 years: HR = 2.94, 95% CI = 1.65–5.21, p < 0.001), white race (vs. black: HR = 6.80, 95% CI = 1.69–27.40, p = 0.007), acute myeloid leukemia (vs. lymphocytic leukemia: HR = 1.59, 95% CI = 1.09–2.33, p = 0.016), unspecified and other specified leukemia (vs. lymphocytic leukemia: HR = 2.72, 95% CI = 1.55–4.75, p < 0.001), and living in a small city (vs. large city: HR = 2.10, 95% CI = 1.23–3.60, p = 0.007). Meanwhile, being a non‐Hispanic black (vs. Hispanic: HR = 0.06, 95% CI = 0.01–0.62, p = 0.019) was a protective factor for suicide. Male sex, older age at diagnosis, white race, and acute myeloid leukemia were risk factors for suicide in leukemia patients, while being a non‐Hispanic black was a protective factor. Medical workers should, therefore, provide targeted preventive measures to leukemia patients with a high risk of suicide.


| BACKGROUND
Suicide refers to an individual deliberately or voluntarily ending their own life under the action of complex psychological activities. [1][2][3] Suicide is a complex global social phenomenon. 4,5 The World Health Organization reports that 817,000 people died of suicide globally in 2016, accounting for 1.49% of the total number of deaths, and corresponding to one person committing suicide every 40 s. 6 The suicide mortality rate in the United States was 14.78 per 100,000 in 2018, which is relatively high compared with other countries. 7 The prevalence of cancer has been increasing worldwide in recent years, and it is now the third leading cause of death. 8 It was estimated that there were 18.1 million new cancer cases in 2018, with 9.6 million cancer deaths globally. 9 Cancer treatments previously focused on prolonging life, while neglected the huge losses caused by cancer patients and their families during a long course of treatment and rehabilitation, such as the financial burden and mental illness. 10,11 The increasing incidence of cancer is also increasing the incidence of mental illness in patients during diagnosis and treatment, such as depression, fear of cancer recurrence, and suicidal thoughts. 9,12,13 Several studies have found increases in suicidal ideation among cancer patients and that the suicide rate is higher in these patients. [2][3][4][14][15][16][17][18] The suicide rate in the United States is significantly higher in cancer patients than in the general population, with the suicide standardized mortality rate (SMR) reaching 4.44. 3,19 This situation indicates the importance of identifying the risk factors for suicide in cancer patients in order to be able to prevent their suicide behaviors. Some studies have found that male sex, white race, marital status, and other risk factors are strongly correlated with the suicide rate of patients with certain types of cancer, such as kidney cancer, breast cancer, lung cancer, and head and neck cancer. 13,[20][21][22] Leukemia is a type of malignant proliferative disease involving hematopoietic stem cells. 23 There were 437,033 new leukemia patients identified worldwide in 2018, accounting for 2.4% of all new cancer patients and ranking 15th among new cancer patients. 9 Leukemia is a clinically common malignant tumor in the United States. 24 Several studies have analyzed potential risk factors associated with suicide in common cancers in the United States using data obtained from the Surveillance, Epidemiology, and End Results (SEER) database. 13,19,21,25 However, no research into the relationship between leukemia and suicide has been reported in the literature. Therefore, the purpose of this study was to identify the potential risk factors related to leukemia and suicide by analyzing the SEER database.

| Data source
The SEER database of the National Cancer Institute covers 30% of the United States population and is considered to be representative of that population. 26,27 The SEER database provides registered researchers with free access to a large amount of research materials, including patient demographic, cancer incidence, and survival data. All of the leukemia patients selected for inclusion in this study had been added to the SEER database between 1975 and 2017. We obtained permission to access the database after signing and submitting the SEER Research Data Agreement form via email. The SEER*Stat software (version 8.3.6) was used to identify relevant patients for inclusion in this study.

| Study population and inclusion criteria
We collected all leukemia patients diagnosed during 1975-2017 as defined by International Classification of Disease for Oncology codes (third edition). Patients were identified using the primary site code (C42.1) and morphology codes (9800-9948) for leukemia. We considered that suicide had occurred in cases in which the cause of death was coded as "suicide and self-mutilation." The exclusion criteria included patients with incomplete variables and no diagnosis or microscopy findings. The information collected for all patients included the year of diagnosis, age at diagnosis, sex, race, type of leukemia, surgery status, survival period, and cause of death. The application of our selection criteria identified 142,386 leukemia patients who had been added to the SEER database from 1975 to 2017, of whom 191 had suicided. The steps used to select leukemia patients for inclusion in this study are shown in Figure 1

| Statistical analysis
We divided the patients into the following three groups to allow basic data comparisons: suicidal death, non-suicidal K E Y W O R D S leukemia, risk factors, SEER, suicide death, and survival. The chi-square test was used to compare suicide rates among patients in the different groups. We used the Web-Based Injury Statistics Query and Reporting System of the United States Centers for Disease Control and Prevention (CDC) 7 to calculate the SMR values for suicide in each group, with the general population of the United States from 1981 to 2018 used as the reference. The 95% confidence interval (CI) of the SMR was obtained using Byar's approximation. 28 Subsequent univariate and multivariate Cox regression analyses were used to generate raw and adjusted hazard ratios (HR) and their associated 95% CI for identifying potential risk factors for suicide. All statistical analyses were performed using R software (version 3.6.3). All tests were double-sided and had a significance criterion of p < 0.05.
Since the CDC does not have data on the suicide mortality rate of the general population from 1975 to 1980, the population suicide benchmark from 1981 to 1983 was used in this study to adjust the suicide rate of leukemia patients from 1975 to 1980. 7 It was found that between 1975

| Factors associated with suicide
The univariate analysis showed that the factors associated with a higher risk of suicide were male sex (vs.  Table 3. A subsequent Cox survival regression analysis revealed that the risk of suicide was higher in male leukemia patients than in female patients (Figure 3).

| DISCUSSION
Multiple studies have confirmed that the suicide mortality rate is higher in cancer patients than in the general population. 3,4, 17 Innos K reported that the suicide rate in Estonian cancer patients was higher than that in the general population, 29 while Crocetti E found the same result for Italian cancer patients. 16 The suicide rate is about twofold higher in cancer patients than the general population in the United States. 19 Using the SEER database, we found that the suicide rate of leukemia patients in the United States from 1975 to 2017 was approximately 26.41 per 100,000 person-years, compared to 12.24 per 100,000 person-years in the general population, giving an SMR of 2.16 (95% CI = 1.85-2.47), which is similar to the result that other cancer patients have a higher suicide rate than the general population. 19 The important risk factors for suicide in leukemia patients include male sex, older age at diagnosis, white race, and acute myeloid leukemia. We found that the proportion of male leukemia patients in the United States is similar to that of females (Table 1), whereas most suicided leukemia patients are males (85.9%), with a suicide rate of 39.96 per 100,000 person-years, compared to 8.63 per 100,000 person-years in female leukemia patients (p < 0.001) ( Table 2). In addition, male leukemia patients in the United States are more likely to commit suicide than are than female patients, with an HR of 4.41 (Table 3), which is similar to the suicide rates for male patients with other cancers. 19,30 Despite the similar prevalence rates of male and female leukemia patients, the much higher suicide rate in male leukemia patients may be related to the poor ability of males to withstand psychological pressures and their tendency for self-directed violence. 31,32 We found that most of the leukemia patients were older than 60 years (65.1%), and the suicide rate was higher in this age group (60-69 years vs. ≤39 years: HR = 2.60, 95% CI = 1.60-4.23, p < 0.001; 70-79 years vs. ≤39 years: HR = 2.84, 95% CI = 1.72-4.68, p < 0.001; ≥80 years vs. ≤39 years: HR = 2.94, 95% CI = 1.65-5.21, p < 0.001) ( Table 3). In general, the incidence rates of leukemia and suicide increased with the age of the patients, which is consistent with previous studies of elderly patients having other types of cancer. 22,33,34 The higher suicide rates of elderly leukemia patients might be related to their quality of life, outlook about life and death, and psychological condition such as the presence of depression. 11,35,36 Previous research has shown that the suicide rate among cancer patients in the United States is highest for whites and lowest for blacks. 37,38 Our results similarly found that among all racial groups in the United States, the risk of suicide was on average 6.80-fold higher for whites than for blacks (95% CI = 1.69-27.40, p = 0.007) (  non-Hispanic Asians were 15% and 31% lower, respectively (Table 3). These results indicate that being white is a risk factor for suicide, while being black is a protective factor, which might be related to differences in the level of knowledge and culture among racial groups, religious beliefs, and economic conditions. 12,36,39,40 We also found that acute myeloid leukemia and unspecified and other specified leukemia were high-risk factors for suicide in leukemia patients. In the SEER database, most of the leukemia patients in the United States have lymphocytic leukemia (50.7%) who also account for the majority of suicided patients (58.1%), and their SMR is also higher than in the general population (SMR = 1.76, 95% CI = 1.40-2.06) ( Tables 1 and 2). However, compared with lymphoid leukemia, patients with acute myeloid leukemia had a 2.27-fold higher risk of suicide (95% CI = 1.53-3.37, p < 0.001), while among those with unspecified and other specified leukemia it was 3.21-fold higher (95% CI = 1.82-5.66, p < 0.001) ( Table 3). The prognosis varies with the type of cancer, and this often causes changes in many characteristics of affected patients, such as physical status, psychological status, and quality of life, which in turn lead to an increased risk of suicide. 19,36 Acute myelogenous leukemia often occurs in young adults, and the disease progresses rapidly, 41 with some patients dying within a few months after the disease onset. 42 The unspecified and other specified leukemia have unclear disease types, poor treatment effects, and short survival times. 43 The low remission rate for these two types of leukemia patients exerts huge economic pressures and psychological burdens on them. Resulting factors such as discrimination, inferiority, and depression have led to their high suicide rate compared with other types of leukemia patients. 44 Finally, as shown in Figure 1, we found that the SMR for suicide among leukemia patients peaked in the 1980s and 2010s, and reached a nadir during 1975-1977 (SMR = 0.93, 95% CI = 0. 25-2.37). Because the CDC lacks data on suicide mortality for the general population before 1980, the suicide rate of leukemia patients from 1975 to 1980 was adjusted to the population suicide benchmark from 1981 to 1983, 7 which may have reduced the adjusted SMRs. Although the SMR of leukemia patients generally fluctuated between 1.5 and 3.0 after the 1980s, it peaked in the 2010s at 3.84 (95% CI = 2.41-5.82). This may be related to factors such as financial crises, high medical costs, declining quality of life, and psychological burden. 45,46 Previous research has confirmed that some interventions can help to prevent suicide, 5 and that the suicide behavior of cancer patients is affected by various factors such as the psychological status, economic status, and religious beliefs. 34,45 Psychological factors such as depression are especially important risk factors for suicide. 12 Our research results indicated that the suicide rate of leukemia patients is higher than that of the general population. In order to reduce suicide in leukemia patients, psychiatric evaluations should be applied to this population. Some tools are available for identifying the risk of depression, including the Baker Depression Scale. 31 Providing psychological treatment to patients with leukemia at risk of depression can help to reduce the risk of suicide, as can the application of active treatment, improving their quality of life, and strengthening communication between family members. 47 Our research was subject to some limitations. We used SEER*Stat software (version 8.3.6) to collect patient information, but some information is currently missing from the system, such as the marital status, medication status, chemotherapy status, social factors, and psychological factors. These factors may impact suicide, and so the lacking data might have biased the risk assessments of suicide factors in leukemia patients. In this study, we collected leukemia patients from 1975 to 2017 who may have affected their health or even committed suicide due to the psychiatric illness. The SEER database is unable to obtain information on the diagnosis of leukemia patients with psychiatric, which affects the results of the study. Moreover, the SEER database only collects data from a proportion of cancer patients in the United States. Therefore, the suicide risk of leukemia patients should also be assessed in other countries.

| CONCLUSIONS
We have investigated the risk of suicide in leukemia patients and the underlying independent risk factors. Male sex, older age at diagnosis, white race, and acute myeloid leukemia were found to be significant risk factors for suicide in leukemia patients, while being a non-Hispanic black was a protective factor. Medical workers can use our research results to screen leukemia patients with a higher risk of suicide and apply targeted preventive measures to them.