Medical Hypotheses 82 (2014) 16–19
Contents lists available at ScienceDirect
Medical Hypotheses journal homepage: www.elsevier.com/locate/mehy
Are suicide rates higher in the cancer population? An investigation using forensic autopsy data Geoffroy Lorin de la Grandmaison a,⇑, Laurence Watier b, Stéphanie Cavard a, Philippe Charlier a a b
Department of Forensic Medicine and Pathology, Raymond Poincaré Hospital, AP-HP, Versailles Saint-Quentin University, 104 Boulevard Raymond Poincaré, 92380 Garches, France Department of Public Health (INSERM U657), Versailles Saint-Quentin University (EA4499), 104 Boulevard Raymond Poincaré, 92380 Garches, France
a r t i c l e
i n f o
Article history: Received 24 June 2013 Accepted 28 October 2013
a b s t r a c t Previous population-based studies have identified increased suicide rates among cancer patients. Available post mortem data on the contribution of cancer to completed suicide are limited, however. This study examines forensic autopsy data in order to assess whether cancer is significantly more frequent in a suicide population compared with a control population. In total, 232 cases were included in both the suicide and the control groups. Based on a complete standard histological examination, cancer was significantly more often found in the suicide group than in the control one (8.6% vs. 3.9%, p = 0.03). The multivariate analysis also showed that the presence of cancer increased the risk of suicide. Moreover, cancer was not known to the deceased in 70% of cases, while the most frequent mental disease found in cancer-related suicide cases was depression (75%). In the 20 cancer-related suicide cases analysed herein, it was difficult to ascertain whether malignancy was the only motive for committing suicide, as cancer could be considered to be either a major causative factor for suicide or an incidental finding. Therefore, further research is warranted with the help of the psychological autopsy method to investigate the link between suicide and cancer further. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Several epidemiological studies in the United States, Europe and Asia have identified increased suicide rates among cancer patients [1–13]. Indeed, cancers in certain anatomic sites are closely associated with increased suicide rates. According to published studies [10,14,15], lung, stomach, oesophagus, pancreas, prostate and head and neck cancers are associated with the highest suicide rates. Moreover, women with gynaecologic malignancies have a greater incidence of suicide than women with other cancer types [16], while suicide rates are also higher among patients with advanced disease at diagnosis [3,5,12]. Further, men have a higher suicide risk than women [17]. It has also been shown that patients who have been recently diagnosed have an increased risk of suicide compared with cancer-free persons [17], suggesting that cancer diagnosis constitutes a major stress event [18]. The above-reported population-based studies show some limitations, however. Firstly, their findings are based on the information provided by death certificates, for which manner and cause of death may be subject to misclassification bias. Suicide among cancer patients may thus be underreported, especially concerning accidental overdoses. Secondly, epidemiological studies compare suicide victims with cancer with the general population, for which ⇑ Corresponding author. Tel.: +33 1 47 10 76 81; fax: +33 1 47 10 76 83. E-mail address: [email protected] (G.L. de la Grandmaison). 0306-9877/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.mehy.2013.10.025
cancer diagnosis may be unknown by the physician who completed the death certificate. The hypothesis Very few post mortem data are available in the literature regarding the link between suicide and malignancy [19,20]. Most data on the sudden deaths of cancer patients are related to undiagnosed tumours or medical malpractice. The presence of malignant disease was revealed by forensic autopsy in 1.7–5.9% of cancer cases, according to published autopsy series [19,21,22]. Of the 7020 forensic autopsy series analysed by Gezelius and Eriksson [21], 171 cases (2.4%) showing malignancy were found. The manner of death for 22 of these cases (13%) was suicide, of which tumour was considered to be the major causative factor for the suicide in 14. Similarly, in the forensic autopsy series presented by Bogdanovic et al. [22], neoplasms were the suicide motive in 1.9% of the 1931 suicide cases, but no statistical comparison with control cases was performed. To test the hypothesis that cancer is associated with increased suicide rates as suggested by clinical studies, we used autopsy data in order to compare the rate of cancer in a suicide population with that in a control population. To avoid biased analysis, we chose a specific control group, as the quality of the inclusion criteria is a major factor for improving the significance of the statistical results on the link between cancer and suicide.
17
G.L. de la Grandmaison et al. / Medical Hypotheses 82 (2014) 16–19
Evaluation of the hypothesis
Table 1 Characteristics of suicide cases and control cases.
A retrospective study was carried out on adult cases that underwent forensic autopsy from January 2002 to December 2012 in our Department of Forensic Medicine and Pathology at the Raymond Poincaré Hospital in Garches. During this period, 4585 forensic autopsies were carried out. All the forensic autopsies requested by the prosecutor following police investigations led in the western suburbs of Paris are carried out in our department. The western suburbs of Paris comprise three and a half administrative territorial divisions (i.e., Yvelines, Eure et Loire, Val d’Oise and part of Hauts de Seine), with an estimated population of 3,309,000. Cases were selected according to the following two criteria: - Manner of death was clearly determined after the police investigation and autopsy results. - A complete standard histological examination was performed according to our autopsy protocol (including microscopic sections of the heart, lungs, liver, kidneys, pancreas, spleen, thyroid, adrenal glands, prostate and neuropathological study after brain formalin fixation). Breasts and testes were sampled at the time of autopsy only if a tumour was seen macroscopically. For each case, the following parameters were reported: age, sex, manner of death, cause of death, presence of malignant tumour, histological type of tumour and tumour extent (presence of metastasis). Two groups were defined and compared in this study: a suicide group and a control group, which comprised those who had died from an accident. For suicide cases with malignant disease found by autopsy, knowledge of the diagnosis of cancer by the deceased during his/her life and mental disease were recorded when available. Bodies showing an advanced state of putrefaction were excluded from the study. Finally, medical health issues other than cancer were also recorded for the control group. Frequency distribution (%) and mean ± standard deviation (SD) were computed according to qualitative and continuous variables. We constructed two variables to consider the presence of cancer: a binary variable (no cancer as the reference, cancer with and without metastasis) and one with three classes (no cancer as the reference, cancer without metastasis and cancer with metastasis). First, the studied variables of the cases and controls were compared. Chi-square or Fisher exact tests were used to test differences in proportions and the Student t-test to test differences in means. Second, multiple logistic regression was used. Candidate variables for multivariate modelling were selected from among the variables with a p-value 60.20 according to univariate analysis. When appropriate, interactions between variables were also tested. A two-tailed p-value
Contents lists available at ScienceDirect
Medical Hypotheses journal homepage: www.elsevier.com/locate/mehy
Are suicide rates higher in the cancer population? An investigation using forensic autopsy data Geoffroy Lorin de la Grandmaison a,⇑, Laurence Watier b, Stéphanie Cavard a, Philippe Charlier a a b
Department of Forensic Medicine and Pathology, Raymond Poincaré Hospital, AP-HP, Versailles Saint-Quentin University, 104 Boulevard Raymond Poincaré, 92380 Garches, France Department of Public Health (INSERM U657), Versailles Saint-Quentin University (EA4499), 104 Boulevard Raymond Poincaré, 92380 Garches, France
a r t i c l e
i n f o
Article history: Received 24 June 2013 Accepted 28 October 2013
a b s t r a c t Previous population-based studies have identified increased suicide rates among cancer patients. Available post mortem data on the contribution of cancer to completed suicide are limited, however. This study examines forensic autopsy data in order to assess whether cancer is significantly more frequent in a suicide population compared with a control population. In total, 232 cases were included in both the suicide and the control groups. Based on a complete standard histological examination, cancer was significantly more often found in the suicide group than in the control one (8.6% vs. 3.9%, p = 0.03). The multivariate analysis also showed that the presence of cancer increased the risk of suicide. Moreover, cancer was not known to the deceased in 70% of cases, while the most frequent mental disease found in cancer-related suicide cases was depression (75%). In the 20 cancer-related suicide cases analysed herein, it was difficult to ascertain whether malignancy was the only motive for committing suicide, as cancer could be considered to be either a major causative factor for suicide or an incidental finding. Therefore, further research is warranted with the help of the psychological autopsy method to investigate the link between suicide and cancer further. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Several epidemiological studies in the United States, Europe and Asia have identified increased suicide rates among cancer patients [1–13]. Indeed, cancers in certain anatomic sites are closely associated with increased suicide rates. According to published studies [10,14,15], lung, stomach, oesophagus, pancreas, prostate and head and neck cancers are associated with the highest suicide rates. Moreover, women with gynaecologic malignancies have a greater incidence of suicide than women with other cancer types [16], while suicide rates are also higher among patients with advanced disease at diagnosis [3,5,12]. Further, men have a higher suicide risk than women [17]. It has also been shown that patients who have been recently diagnosed have an increased risk of suicide compared with cancer-free persons [17], suggesting that cancer diagnosis constitutes a major stress event [18]. The above-reported population-based studies show some limitations, however. Firstly, their findings are based on the information provided by death certificates, for which manner and cause of death may be subject to misclassification bias. Suicide among cancer patients may thus be underreported, especially concerning accidental overdoses. Secondly, epidemiological studies compare suicide victims with cancer with the general population, for which ⇑ Corresponding author. Tel.: +33 1 47 10 76 81; fax: +33 1 47 10 76 83. E-mail address: [email protected] (G.L. de la Grandmaison). 0306-9877/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.mehy.2013.10.025
cancer diagnosis may be unknown by the physician who completed the death certificate. The hypothesis Very few post mortem data are available in the literature regarding the link between suicide and malignancy [19,20]. Most data on the sudden deaths of cancer patients are related to undiagnosed tumours or medical malpractice. The presence of malignant disease was revealed by forensic autopsy in 1.7–5.9% of cancer cases, according to published autopsy series [19,21,22]. Of the 7020 forensic autopsy series analysed by Gezelius and Eriksson [21], 171 cases (2.4%) showing malignancy were found. The manner of death for 22 of these cases (13%) was suicide, of which tumour was considered to be the major causative factor for the suicide in 14. Similarly, in the forensic autopsy series presented by Bogdanovic et al. [22], neoplasms were the suicide motive in 1.9% of the 1931 suicide cases, but no statistical comparison with control cases was performed. To test the hypothesis that cancer is associated with increased suicide rates as suggested by clinical studies, we used autopsy data in order to compare the rate of cancer in a suicide population with that in a control population. To avoid biased analysis, we chose a specific control group, as the quality of the inclusion criteria is a major factor for improving the significance of the statistical results on the link between cancer and suicide.
17
G.L. de la Grandmaison et al. / Medical Hypotheses 82 (2014) 16–19
Evaluation of the hypothesis
Table 1 Characteristics of suicide cases and control cases.
A retrospective study was carried out on adult cases that underwent forensic autopsy from January 2002 to December 2012 in our Department of Forensic Medicine and Pathology at the Raymond Poincaré Hospital in Garches. During this period, 4585 forensic autopsies were carried out. All the forensic autopsies requested by the prosecutor following police investigations led in the western suburbs of Paris are carried out in our department. The western suburbs of Paris comprise three and a half administrative territorial divisions (i.e., Yvelines, Eure et Loire, Val d’Oise and part of Hauts de Seine), with an estimated population of 3,309,000. Cases were selected according to the following two criteria: - Manner of death was clearly determined after the police investigation and autopsy results. - A complete standard histological examination was performed according to our autopsy protocol (including microscopic sections of the heart, lungs, liver, kidneys, pancreas, spleen, thyroid, adrenal glands, prostate and neuropathological study after brain formalin fixation). Breasts and testes were sampled at the time of autopsy only if a tumour was seen macroscopically. For each case, the following parameters were reported: age, sex, manner of death, cause of death, presence of malignant tumour, histological type of tumour and tumour extent (presence of metastasis). Two groups were defined and compared in this study: a suicide group and a control group, which comprised those who had died from an accident. For suicide cases with malignant disease found by autopsy, knowledge of the diagnosis of cancer by the deceased during his/her life and mental disease were recorded when available. Bodies showing an advanced state of putrefaction were excluded from the study. Finally, medical health issues other than cancer were also recorded for the control group. Frequency distribution (%) and mean ± standard deviation (SD) were computed according to qualitative and continuous variables. We constructed two variables to consider the presence of cancer: a binary variable (no cancer as the reference, cancer with and without metastasis) and one with three classes (no cancer as the reference, cancer without metastasis and cancer with metastasis). First, the studied variables of the cases and controls were compared. Chi-square or Fisher exact tests were used to test differences in proportions and the Student t-test to test differences in means. Second, multiple logistic regression was used. Candidate variables for multivariate modelling were selected from among the variables with a p-value 60.20 according to univariate analysis. When appropriate, interactions between variables were also tested. A two-tailed p-value
