Annals of Oncology Advance Access originally published online on October 19, 2006
Annals of Oncology 2007 18(2):381-387; doi:10.1093/annonc/mdl385
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2006 European Society for Medical Oncology
oncology practice |
Suicide and cancer: a gender-comparative study
Division of Radiation Oncology, The Ottawa Hospital Regional Cancer Centre, Ottawa, Ontario; The Ottawa Hospital Research Institute, Ottawa, Ontario, Canada
* Correspondence to: Dr W. S. Kendal, Division of Radiation Oncology, The Ottawa Hospital Regional Cancer Center, 503 Smyth Road, Ottawa, Ontario, K1H 1C4 Canada. Tel: +1-613-737-7700; Fax: +1-613-247-3511; E-mail: wkendal{at}ottawahospital.on.ca
 |
Abstract |
|---|
 Top Abstract introductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Background: Persons with cancer commit suicide more frequently than those without, and males generally commit suicide more frequently than females. A population-based analysis of cancer patients was carried out here, comparing suicide risk between the genders, to elucidate the features specific to each gender.
Patients and methods: A total of 1.3 million cancer cases from the Surveillance, Epidemiology, and End Results program were analyzed. Cox proportional hazards models were fitted to personal, tumor-related, and social variates.
Results: A total of 265 female and 1307 male suicides were enumerated, reflecting 0.04% and 0.19% from each gender, and providing an overall hazard ratio for male suicide of 6.2 [95% confidence interval (CI) 5.4–7.1]. Females with colorectal (P = 0.01) and cervical (P < 0.0001) cancers showed decreased suicide rates. Males with head and neck cancers (P < 0.0001) and myeloma (P = 0.02) had increased rates, whereas rates were decreased in males with lung cancer (P = 0.01), liver (P = 0.01), brain tumors (P = 0.04), and leukemia (P = 0.007). The hazard ratio associated for male suicide with distant metastasis was 2.84 (95% CI 2.49–3.24); for married status, 0.46 (95% CI 0.39–0.54); and for African-American ancestry, 0.24 (95% CI 0.17–0.34)—comparable ratios were seen here for female suicides. In head and neck cancers, with both genders analyzed together, the suicide hazard was increased if surgery was contraindicated (3.0, 95% CI 1.3–6.8), but not if refused.
Conclusions: The high-risk patient was male, with head and neck cancer or myeloma, advanced disease, little social or cultural support, and limited treatment options. Oncologists and allied health professionals should be aware of the potential for suicide in cancer patients and their associated risk factors.
Key words: cancer, gender, suicide
|
introduction |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Each year in America about 1.4 million new cancers will be diagnosed [1], 570 000 persons will die of cancer [1], and another 1700 persons with cancer will commit suicide [2]. The causes underlying suicide are many [3]: in the general population, mental illness and substance abuse figure prominently [4, 5]. Male gender is a risk factor [6, 7] that can apparently affect other risk factors [3]. It remains unclear exactly how having cancer might influence suicide risk, and how gender differences might modulate these effects.
A number of studies have corroborated an increased rate of suicide among cancer patients. In Finland, suicides by female and male cancer patients were 1.9 and 1.3 times that of the general population [8]. In Connecticut [9] and Estonia [10], male cancer patients committed suicide at a higher rate, particularly just after diagnosis of their cancers; the rate in females was comparable to the general population. Swedish [11, 12], Swiss [13, 14], Italian [15], Japanese [16], Norwegian [17], and Danish [18] studies have all noted higher rates for both genders, maximal within the first few months of diagnosis.
Notwithstanding existential and quality of life (QoL) issues which might prompt suicide [19–21], psychological autopsy of cancer suicides commonly reveals affective disorders, a sense of loss, prior experience with cancer or debilitating disease, and inflexible personality traits [22]. Suicides in terminally ill patients seem to differ from those during cancer remission, the latter group exhibits higher frequencies of personal/familial mental disorder, as well as prior suicide attempts [23, 24].
In the general American population, males commit suicide about four times more frequently than females [25], although females will admit to attempts three times more frequently [26]. Comparatively little is known about how gender differences interplay with cancer suicide. A gender-comparative study of suicide in cancer patients could improve our understanding of the underlying factors, and possibly influence preventative and social policies. The Surveillance, Epidemiology, and End Results (SEER) registry [2] provides population-based data regarding the pathology, disease extent, treatment, social factors, and causes of death among individuals with cancer in the United States. This database was employed here to characterize gender-specific risk factors for suicide in cancer patients, with the aim to ascertain those most in need of supportive and preventative help.
|
patients and methods |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
study cohorts and data
The study population was derived from the April 2004 release of the November 2003 submission to the SEER 9 Public-Use population-based registry [2]. Data were retrieved from individuals with invasive cancer from all SEER anatomic sites that were diagnosed from 1973 to 2001. The study population was initially restricted to individuals with a single primary cancer, to better correlate suicide risk with each particular site. A further analysis of head and neck cancers included multiple primary cancers.
Information retrieved for each individual from the SEER database included each individual's sex, age at diagnosis, cause of death, race and ethnicity, marital status at diagnosis, and survival time. As well, the anatomic site, grade, and SEER summary stage for each tumor were recorded. The SEER database did not provide the marital status of the study individuals at the time of last follow-up, nor information regarding substance abuse, associated psychiatric disorders, sexual orientation, employment status, family history, financial history, or QoL.
statistical analysis
Statistical analysis was carried out with SPSS. Two retrospective cohorts were defined, distinguished by gender. An assessment was carried out for all anatomic sites together, and separately for cancers of the breast, prostate, lung and bronchus, colon and rectum, urinary bladder, corpus uteri, pancreas, kidneys and renal pelvis, stomach, ovary, thyroid, brain, cervix uteri, liver, testis, and esophagus. As well, individuals with non-Hodgkin's lymphoma, Hodgkin's disease, leukemia, myeloma, and melanoma were examined as separate categories.
A number of Cox proportional hazard analyses were carried out. These provided 95% confidence intervals for the hazard ratios, 
, as well as the Wald 
2 test [27]. Age at diagnosis was analyzed as a continuous variable; the remaining covariates were analyzed categorically.
Because completed suicide was a relatively rare event, relative to other causes of death, an assessment of the competing hazards was included. The cumulative incidence of completed suicide was calculated as a function of time from diagnosis, and corrected for competing hazards by the method of Satagopan et al. [28].
|
results |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Data from 1 316 762 individuals were retrieved from the SEER database (Table 1). For both genders considered together, the median age at diagnosis was 66 years with a range of 0–115 years. Female cancer patients tended to be younger at diagnosis, had a lower age at diagnosis, and had longer follow-up periods than males. The frequency of completed suicides for females was 0.02%, about one-fifth that of males; the hazard ratio for male suicide relative to females was 6.2. Females also experienced a lower death rate from other causes, when compared with the study population (Table 1). The male cohort revealed disproportionately more white and African-Americans than the female cohort; more males were married at the time of diagnosis, whereas more females were widowed or divorced. For females, the cancers more frequently revealed regional nodal involvement at diagnosis; for males, they were more frequently localized or metastatic. Figure 1A provides the comparative frequencies of completed suicide by females and males, plotted against age. Suicide occurred most frequently during the adult decades. These age and gender-specific frequencies appeared to be in proportion to those seen within the general American population [29].
|
|
Table 2 details the frequencies of suicidal death specific to each cancer site. Data particular to each site were compared with the suicide frequency from the remaining sites. When the pooled female and male data were analyzed, cancers of the lung and bronchus, urinary bladder, head and neck, esophagus, and myeloma revealed relatively high suicide rates, whereas the rates were lower for cancers of the breast and liver. This pattern, however, did not persist on gender-specific analysis. The suicide rates for most anatomic sites from females were essentially similar, except for colorectal and cervical cancers which revealed lower rates relative to the remaining sites analyzed from the female cohort. The suicide rates for male head and neck cancer, liver cancer, and myeloma remained elevated relative to the remaining male cohort, recapitulating the results from the pooled female and male data. The rates for cancers of the urinary bladder, esophagus, and lung and bronchus in males, however, were not elevated when compared with the remaining male cohort. Moreover, the suicide rates for male leukemia and prostate and brain cancers were actually lower than those seen with the other male sites.
|
The cumulative incidence of suicidal death within each cohort was then calculated and plotted versus time from diagnosis (Figure 1B). Corrections were applied here for both competing hazards and for the censoring of survivors. Male suicides occurred with higher incidence than for females throughout, and the rate was highest immediately after cancer diagnosis. In contrast, the incidence of female suicide remained more or less constant over time, as reflected by a nearly linear relationship between the cumulative incidence and time. The combined incidence of suicide within the SEER cancer population, and estimated by actuarial methods, was elevated at 24 cases per 100 000 when compared with the observed 10.6 yearly suicides per 100 000 from the general American population [30].
Proportional hazard models were constructed to determine the factors that associated with suicide within each gender (Table 3). The strongest association with female suicide was with distant metastasis at diagnosis ( = 2.3). Married status at diagnosis carried with it a reduced hazard ratio ( = 0.67). Race and ethnicity provided the next strongest association: the suicide hazard was reduced for African-American females compared with white females ( = 0.17).
|
For males, the strongest association with suicide was with distant metastasis (
= 2.8, Table 3). Here, the intermediate stages also afforded elevated hazard ratios. As with females, divorced status carried an increased suicidal hazard. The suicide hazard for married males was significantly decreased ( = 0.46). African-American males experienced a reduced suicidal hazard, similar to that seen with females ( = 0.24). As well, the hazard ratio for suicide increased by = 1.03/year of age in males.
Since head and neck cancers provided the highest incidence of suicide of any site, an additional proportional hazards model was constructed for this site alone (Table 4). SEER did not provide data with respect to substance abuse in this population, and thus, this covariate could not be examined. Of the available data, male gender remained the dominant risk factor ( = 4.7). If the head and neck cancer was the first primary site, the hazard for suicidal death was significantly decreased ( = 0.5) compared with cases where the head and neck tumor was not the first primary cancer. White individuals with head and neck cancers exhibited a greater suicide hazard. As well, those with distant spread in their cancer showed an increased suicidal hazard, in keeping with the earlier analyses. The suicide risk with these head and neck cancers was also increased with age and was decreased with married status.
|
A number of additional covariates were included with the head and neck cancer analysis. The hazard for suicidal death increased with histological grade, in contrast to the analyses from Table 3, where it did not have any significant role in the regression models (data not provided). If cancer-directed surgery had been contraindicated, the suicide hazard was also increased. No similar significant relationships could be demonstrated between suicide and either the use, non-use, or refusal of therapeutic irradiation (data not provided). When the different head and neck sites were analyzed separately and compared with the index group drawn from laryngeal cancer, only those with oropharyngeal cancer revealed a statistically significant association with suicide (
= 3.0). Indeed, when the 32 598 laryngeal cancer cases were analyzed separately (data not provided), there was no major association with the use of either radiotherapy or cancer-directed surgery, and neither did the sequence of surgery followed by radiotherapy reveal any statistically significant association with suicidal death. For these laryngeal cancers, such surgery would likely have involved major procedures such as laryngectomy, and so, the likely functional losses did not seem associated with an increased suicide rate. |
discussion |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
It was not possible to differentiate, on the basis of the SEER data, between suicides associated with affective illness or substance abuse and those motivated by the desire for relief from terminal illness or the avoidance of being a burden to others. The SEER database did provide information regarding potential risk factors for depression in cancer, such as advanced stage [31], pancreatic cancer [32, 33], refusal of treatment [34], unmarried patients with head and neck cancer [35], and (through the documentation of marital status) a lack of family support [36]. One aim of this study had been to examine how surrogate factors such as these might relate to suicide risk.
Advanced stage was one of the stronger risk factors for suicide in this analysis. On the other hand, this study showed that pancreatic cancer carried with it no more suicide risk than that experienced by the general population of cancer patients. Similarly, refusal of either surgical treatment or radiotherapy did not figure as significant risk factors in the present analysis. There was, however, a significant association of suicidal death when surgery was deemed contraindicated, a feature likely indicative of advanced tumor features or additional comorbidities. The data presented here did confirm that married status conferred a lower risk of suicide. Overall, it would seem plausible to conclude that depression played a likely role in many of the suicides from the study population.
Poor prognostic features for cancer cure also associated strongly with suicide risk. Metastatic disease at diagnosis, cancer-directed surgery that could not be carried out (for head and neck cancers), high-grade tumors, or treatment contraindications, all were associated with increased risk. On the other hand, for aggressively treated patients there was no increased risk, even if the treatment might presumably have lead to major loss of function, as might occur with head and neck cancers. The head and neck cancer analysis revealed that oropharyngeal cancer was particularly associated with suicide risk. These tumors are also associated with major QoL deficiencies and depression [37, 38]. Even among persons with oropharyngeal cancer, an aggressive approach can provide improvement in their emotional functioning and QoL [37, 39].
Race and ethnicity had a significant influence on suicide risk. The suicide hazard experienced by African-Americans was decreased, consistent with the relative reduction of suicides by African-Americans in the general population [40]. Religious beliefs, family support, and a cultural rejection of suicide have been thought responsible for their decreased suicide hazard [41].
In this study, there had been a 4.8 times excess in the overall number of male suicides over females (Table 1), consistent with the 4.5 times ratio seen among the general American population [25]. Other studies have noted the rate of suicide to be highest soon after the diagnosis of cancer [9, 10], this observation was confirmed here for males, but not for females. This difference might relate to the smaller numbers of female suicides analyzed or possibly to the corrections employed here for competing hazards and censored observations.
The suicide risk was decreased in females with colorectal and cervical cancer. On the other hand, breast cancer in females revealed a risk similar to the overall female cancer population, a result at variance with an earlier report of an increased risk of suicide among Danish women with breast cancer [18]. In males, suicide risk was decreased for cancers of the prostate, brain, liver, and leukemia. Their risk was increased for head and neck cancer and myeloma, but a similar pattern was not seen with females. These results for males with head and neck cancers would seem consistent with an increased level of psychological distress particular to this site [35]; however, pancreatic cancer patients purportedly also suffer from increased distress, yet no increased risk was observed here. Similarly, an increased suicide rate has been claimed for prostate cancer [42], but not supported in the present analysis.
Many of the extremes seen within the male cohort were not as apparent to the female cohort. Particularly with sites at greatest risk for male suicide, such as head and neck cancer and myeloma, issues with respect to QoL, coping, symptom control, and psychological distress would seem foremost [37, 43]. The lack of such findings within the female cohort could be attributable to a lower statistical power from the almost five-fold fewer recorded female suicides. At the same time, however, female cancer patients likely experienced similar problems, but they may have been less inclined to react through self-directed violence, much as seen within the general female population [6].
The composite picture thus obtained for suicide risk was that of a white male, with a new diagnosis of possibly head and neck cancer or myeloma, widowed, with widely disseminated and perhaps high-grade disease, or maybe a history of other cancers. In contrast, for decreased risk, one might envision a female of African-American heritage, with perhaps colorectal or cervical cancer, localized disease, and living with her spouse. This high-risk picture might be associated with substance abuse, QoL, or psychological issues framed in the context of poor family and cultural supports, as well as associated with a poor prognostic outlook from the cancer. If we are to help people cope with their cancer, medical and allied health care workers should be perceptive to the risk for suicide in all cancer patients—particularly those individuals with high risk factors such as poor social support, substance abuse, and a history of mental health problems.
|
Acknowledgements |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
The author would like to thank Dr Libney Eapen for his helpful discussions during the conduct of this study.
Received for publication June 7, 2006. Revision received August 24, 2006. Accepted for publication September 11, 2006.
|
References |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionAcknowledgementsReferences |
|---|
1. Jemal A, Murray T, Ward E, et al. (2005) Cancer statistics, 2005. CA Cancer J Clin 55:10–30.
2. National Cancer Institute. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—SEER 9 Regs Public-Use, Nov 2003 Sub (1973–2001), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004, based on the November 2003 submission.
3. Qin P, Agerbo E, Mortensen PB. (2003) Suicide risk in relation to socioeconomic, demographic, psychiatric, and familial factors: a national register-based study of all suicides in Denmark, 1981 –1997. Am J Psychiatry 160:765–772.
4. Cavanagh JT, Carson AJ, Sharpe M, Lawrie SM. (2003) Psychological autopsy studies of suicide: a systematic review. Psychol Med 33:395–405.[CrossRef][ISI][Medline]
5. Harris EC and Barraclough B. (1997) Suicide as an outcome for mental disorders. A meta-analysis. Br J Psychiatry 170:205–228.
6. Hawton K. (2000) Sex and suicide: gender differences in suicidal behaviour. Br J Psychiatry 177:484–485.
7. Qin P, Mortensen PB, Agerbo E, et al. (2000) Gender differences in risk factors for suicide in Denmark. Br J Psychiatry 177:546–550.
8. Louhivuori KA and Hakama M. (1979) Risk of suicide among cancer patients. Am J Epidemiol 109:59–65.
9. Fox BH, Stanek EJ III, Boyd SC, Flannery JT. (1982) Suicide rates among cancer patients in Connecticut. J Chronic Dis 35:89–100.[CrossRef][ISI][Medline]
10. Innos K, Rahu K, Rahu M, Baburin A. (2003) Suicides among cancer patients in Estonia: a population-based study. Eur J Cancer 39:2223–2228.[CrossRef][ISI][Medline]
11. Allebeck P, Bolund C, Ringback G. (1989) Increased suicide rate in cancer patients. A cohort study based on the Swedish Cancer-Environment Register. J Clin Epidemiol 42:611–616.[CrossRef][ISI][Medline]
12. Bjorkenstam C, Edberg A, Ayoubi S, Rosen M. (2005) Are cancer patients at higher suicide risk than the general population? Scand J Public Health 33:208–214.[ISI][Medline]
13. Levi F, Bulliard JL, La Vecchia C. (1991) Suicide risk among incident cases of cancer in the Swiss Canton of Vaud. Oncology 48:44–47.[CrossRef][ISI][Medline]
14. Chatton-Reith J, el May H, Raymond L. (1990) The risk of suicide in cancer patients derived from a cancer registry. Rev Epidemiol Sante Publique 38:125–131.[ISI][Medline]
15. Crocetti E, Arniani S, Acciai S, et al. (1998) High suicide mortality soon after diagnosis among cancer patients in central Italy. Br J Cancer 77:1194–1196.[ISI][Medline]
16. Tanaka H, Tsukuma H, Masaoka T, et al. (1999) Suicide risk among cancer patients: experience at one medical center in Japan, 1978 –1994. Jpn J Cancer Res 90:812–817.[CrossRef][ISI]
17. Hem E, Loge JH, Haldorsen T, Ekeberg O. (2004) Suicide risk in cancer patients from 1960 to 1999. J Clin Oncol 22:4209–4216.
18. Yousaf U, Christensen ML, Engholm G, Storm HH. (2005) Suicides among Danish cancer patients 1971 –1999. Br J Cancer 92:995–1000.[CrossRef][ISI][Medline]
19. Passik SD and Breitbart WS. (1996) Depression in patients with pancreatic carcinoma. Diagnostic and treatment issues. Cancer 78:615–626.[ISI][Medline]
20. Filiberti A and Ripamonti C. (2002) Suicide and suicidal thoughts in cancer patients. Tumori 88:193–199.[ISI][Medline]
21. Wilson KG, Curran D, McPherson CJ. (2005) A burden to others: a common source of distress for the terminally ill. Cogn Behav Ther 34:115–123.[Medline]
22. Conwell Y, Caine ED, Olsen K. (1990) Suicide and cancer in late life. Hosp Community Psychiatry 41:1334–1339.
23. Hietanen P and Lonnqvist J. (1991) Cancer and suicide. Ann Oncol 2:19–23.
24. Bellini M and Capannini D. (1994) Increased suicide risk in cancer patients. Minerva Psichiatr 35:175–186.[Medline]
25. Minino AM, Arias E, Kochanek KD, et al. (2002) Deaths: final data for 2000. Natl Vital Stat Rep 50:1–119.[Medline]
26. Weissman MM, Bland RC, Canino GJ, et al. (1999) Prevalence of suicide ideation and suicide attempts in nine countries. Psychol Med 29:9–17.[CrossRef][ISI][Medline]
27. Woodward M. (1999) Epidemiology: study design and data analysis(Chapman & Hall/CRC, New York).
28. Satagopan JM, Ben-Porat L, Berwick M, et al. (2004) A note on competing risks in survival data analysis. Br J Cancer 91:1229–1235.[CrossRef][ISI][Medline]
29. Stockard J and O'Brien RM. (2002) Cohort variations and changes in age-specific suicide rates over time: explaining variations in youth. Social Forces 81:605–642.
30. National Institute of Mental Health. (2003) In harm's way: suicide in America(National Institutes of Health, US Department of Health and Human Services, Bethesda, MD).
31. Ciaramella A and Poli P. (2001) Assessment of depression among cancer patients: the role of pain, cancer type and treatment. Psychooncology 10:156–165.[CrossRef][Medline]
32. Green AI and Austin CP. (1993) Psychopathology of pancreatic cancer. A psychobiologic probe. Psychosomatics 34:208–221.
33. Holland J, Korzun A, Tross S, et al. (1986) Comparative psychological disturbance in patients with pancreatic and gastric cancer. Am J Psychiatry 143:982–986.
34. Goldberg RJ. (1983) Systematic understanding of cancer patients who refuse treatment. Psychother Psychosom 39:180–189.[ISI][Medline]
35. Kugaya A, Akechi T, Okuyama T, et al. (2000) Prevalence, predictive factors, and screening for psychologic distress in patients with newly diagnosed head and neck cancer. Cancer 88:2817–2823.[CrossRef][ISI][Medline]
36. Nordin K and Glimelius B. (1999) Predicting delayed anxiety and depression in patients with gastrointestinal cancer. Br J Cancer 79:525–529.[CrossRef][ISI][Medline]
37. de Graeff A, de Leeuw JRJ, Ros WJG, et al. (1999) A prospective study on quality of life of patients with cancer of the oral cavity or oropharynx treated with surgery with or without radiotherapy. Oral Oncology 35:27–32.[CrossRef][ISI][Medline]
38. Pourel N, Peiffert D, Lartigau E, et al. (2002) Quality of life in long-term survivors of oropharynx carcinoma. Int J Radiat Oncol Biol Phys 54:742–751.[CrossRef][ISI][Medline]
39. Shepherd KL and Fisher SE. (2004) Prospective evaluation of quality of life in patients with oral and oropharyngeal cancer: from diagnosis to three months post-treatment. Oral Oncol 40:751–757.[CrossRef][ISI][Medline]
40. Garlow SJ, Purselle D, Heninger M. (2005) Ethnic differences in patterns of suicide across the life cycle. Am J Psychiatry 162:319–323.
41. Neeleman J, Wessely S, Lewis G. (1998) Suicide acceptability in African- and white Americans: the role of religion. J Nerv Ment Dis 186:12–16.[CrossRef][ISI][Medline]
42. Llorente MD, Burke M, Gregory GR, et al. (2005) Prostate cancer: a significant risk factor for late-life suicide. Am J Geriatr Psychiatry 13:195–201.
43. Chaturvedi SK, Shenoy A, Prasad KM, et al. (1996) Concerns, coping and quality of life in head and neck cancer patients. Support Care Cancer 4:186–190.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. J. Kerr Editor-in- New wave Ann. Onc., January 1, 2008; 19(1): 3 - 4. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||