|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2007.14.3990 on August 11 2008 © 2008 American Society of Clinical Oncology. Cancer and the Risk of Suicide in Older Americans
From the Harvard School of Public Health; and Brigham and Women's Hospital, Boston, MA; Center for State Health Policy, Rutgers University, New Brunswick, NJ Corresponding author: Matthew Miller, MD, ScD, Harvard School of Public Health, 677 Huntington Ave, Kresge Bldg, Rm 305, Boston, MA 02115; e-mail: mmiller{at}hsph.harvard.edu
Purpose: To determine whether the risk of suicide is greater among patients with cancer than among patients with other medical illnesses. Patients and Methods: A case-control study of the suicide risk associated with medical illness among older Americans that used healthcare utilization data linked to prescription and mortality files. The patient population was comprised of 1,408 New Jersey residents age 65 years or older who were enrolled in Medicare and in a pharmaceutical insurance program. Patient cases (n = 128) died as a result of suicide during the study period of 1994 to 2002. Control patients (n = 1,280) were frequency-matched to patient cases on age and sex. Data were analyzed by using the odds ratio (OR) of suicide adjusted for age, sex, ethnicity, medical and psychiatric comorbidity, and use of prescription medications. Results: In adjusted analyses, the only medical condition that remained associated with suicide was cancer (OR, 2.3; 95% CI, 1.1 to 4.8). Suicide also remained associated with a diagnosis of affective disorder (OR, 2.3; 95% CI, 1.3 to 4.2), anxiety/personality disorder (OR, 2.2; 95% CI, 1.3 to 3.6), treatment with antidepressants (OR, 2.0; 95% CI, 1.2 to 3.2), and treatment with opioid analgesics (OR, 1.6; 95% CI, 1.0 to 2.5). Conclusion: The risk of suicide in older adults is higher among patients with cancer than among patients with other medical illnesses, even after psychiatric illness and the risk of dying within a year were accounted for.
In the United States, rates of suicide and medical illness are higher among older adults than among other age groups.1 Controlled studies of the independent contribution of medical illness to the risk of elder suicide have, however, yielded conflicting results.2-22 Because of sample size limitations, several studies with rich contextual information about participants’ psychiatric conditions have aggregated diverse medical conditions into a single generic category.7,8,11,12 Several larger studies have used disease-specific registry data that have detailed information about a particular medical condition—most often cancer—but not about potential confounders in the source population, including psychiatric comorbidity.4,9,10,14-18,22 The two largest studies to examine the suicide-medical illness connection used Canadian administrative health care datasets linked to mortality files.19,20 Both studies included information about psychiatric comorbidity for patient cases and for controls. The first19 compared patient cases who died as a result of suicide with controls who died in motor vehicle collisions—most of whom were younger adults—and found that cancer and psychiatric illness were associated with suicide. The second study,20 the largest and only truly population-based study on this topic, used condition-specific prescription records (ie, not diagnostic codes) to identify medical and psychiatric conditions. Suicide was associated with treatment for several common medical and psychiatric conditions, including congestive heart failure, chronic obstructive pulmonary disease, anxiety disorders, and depression. The relation between suicide and some leading causes of medical morbidity, notably most cancers, was not evaluated. The current investigation is, to our knowledge, the first population-based study to examine the relative risk of suicide associated with medical conditions among older Americans while controlling for measures of medical and psychiatric comorbidity.
Study Population Eligible participants included all persons who were simultaneous beneficiaries of Medicare and the New Jersey Pharmaceutical Assistance Program for the Aged and Disabled (PAAD) in New Jersey. The PAAD program provides prescription drug insurance for low-income elderly who have annual household incomes between $10,000 and approximately $17,000. Medicare data include information on all inpatient and outpatient diagnoses, procedures, and admissions. Pharmacy data from PAAD include all prescriptions, dosages, and days’ supplies. Pharmacy data were linked to Medicare data by using unique patient identifiers that were destroyed after linkage. This population has full coverage for prescription drug use, physicians’ services, and hospital care. The study protocol received institutional review board approval. To be eligible for selection onto our study, participants had to be at least 65 years of age and had to be enrolled and active users of Medicare and PAAD for at least 12 consecutive months during 1994 to 2002. Beneficiaries demonstrated active use by filling at least one prescription and by having at least one health care encounter in each of the 6-month periods before the index date, which ensured a uniform, 12-month eligibility period before the index date.
Identification of Patient Cases and Controls Controls were beneficiaries who had not become suicides. For every patient case, 10 living controls were selected and were frequency-matched (to patient cases) with respect to age at index date (within 1 year), sex, and ethnicity (white v African American v other).
Ascertainment of Illness and Other Covariates
Additional covariates, defined using data from the 12-month period before the index date, included age, sex, ethnicity, number of outpatient visits in the year before the index date, and the patient's Charlson comorbidity scale.23 The Charlson score was calculated for each patient by using 19 categories of medical comorbidity that are based on 16 comorbid conditions, which are defined with ICD-9-CM diagnosis and procedure codes. Each category has an associated weight, which is based on the adjusted risk of 1-year mortality. The overall comorbidity score reflects the cumulative increased likelihood of 1-year mortality; the higher the score, the more severe the burden of comorbidity. Prescription medication data filled within 180 days before the index date was available for opioids, non-narcotic analgesics (nonsteroidal anti-inflammatory drugs), benzodiazepines, antidepressants, and other psychoactive medications (neuroleptics and sedative hypnotics).
Statistical Analyses
One hundred twenty-eight suicides in patients age 65 years or older were identified. Most suicide decedents carried a diagnosis for one or more medical illnesses, as did a comparable proportion of controls (68% of suicides v 63% of controls; Table 2). In unadjusted analyses, suicides were significantly more likely than controls to have congestive heart failure (19% v 10%), chronic obstructive pulmonary disease (33% v 23%), malignancy (15% v 5%), painful conditions of the neck and back (17% v 10%), and stroke (30% v 19%). Persons with acute myocardial infarction, angina, diabetes, or osteoarthritis were not more likely to die as a result of suicide than persons without these conditions. The mean Charlson comorbidity score among patient cases was higher than among controls (2.8 v 1.8), and higher comorbidity scores were monotonically associated with a greater risk of suicide (Table 2).
Suicides were significantly more likely than controls to have mental illness (Table 2). Thirty suicides (23%) carried a diagnosis of an affective disorder compared with 5% of controls; 36% had a diagnosis of an anxiety and/or personality disorder compared with 11% of controls. Compared with controls, suicide victims were two to four times as likely to have filled a prescription for opioid analgesics, benzodiazepines, sedative-hypnotics, antipsychotics, or antidepressants (Table 2). In multivariate analyses, the only medical illness that remained associated with suicide was malignancy (OR, 2.3; 95% CI, 1.1 to 4.8; Table 2). Suicide also remained associated with a diagnosis of affective disorder (OR, 2.3; 95% CI, 1.3 to 4.2), anxiety/personality disorder (OR, 2.2; 95% CI, 1.3 to 3.6), treatment with antidepressants (OR, 2.0; 95% CI, 1.2 to 3.2), and treatment with opioid analgesics (OR, 1.6; 95% CI, 1.0 to 2.5). The Charlson comorbidity score did not remain significantly associated with suicide in multivariate analyses that included (collinear) medical conditions that contribute to the Charlson score itself (Table 2). However, the comorbidity score did remain significantly associated with suicide in models that included psychiatric comorbidity (OR, 1.1; 95% CI, 1.0 to 1.2) but not if patients with cancer were excluded from analyses (OR, 1.0; 95% CI, 0.9 to 1.1; not shown). More than 40% of suicide victims used a firearm to kill themselves, and almost two of three suicides among patients with cancer involved firearms (Table 3). Hanging/suffocation was the second-most frequent method, both for cancer patients and overall; self-poisoning was used in 16% of suicides overall and in 11% of all suicides among patients with cancer.
Prostate cancer was the predominant cancer diagnosis among both patient cases and controls (not shown). Of the 19 cases of suicide among patients with cancer, eight (42%) had prostate cancer; of the 64 patients with cancer who did not die by suicide, 27 (43%) had prostate cancer. Cancer of the lung or bronchus accounted for 16% of patient cases and 6% of controls. Colorectal cancer accounted for 11% of patient cases and 14% of controls. Hematologic malignancy was present in one patient case (5%) and in five controls (8%). The most salient difference between patient cases and controls was the disproportionate presence of metastatic disease in those who subsequently died by suicide (62% of nonhematologic malignancies among patient cases v 33% among controls; not shown). The predominance of metastatic disease among suicide patient cases was apparent within discrete cancer diagnoses (eg, among patients with prostate cancer, 63% of suicide patient cases had metastatic disease v 33% of controls), though our ability to observe this pattern for less common malignancies was limited by the small number of suicide patient cases with these less frequent diagnoses.
The first case series to examine the relation between medical illness and suicide among older adults concluded that medical illness contributed to the suicides of three quarters of study patient cases.25 In our study as well, most suicides (68%) had at least one of the study illnesses at the time of their death, but so too did most controls (63%). Our finding that suicide was significantly associated with several prominent medical conditions in bivariate analyses, but that only cancer remained significantly associated with suicide in multivariate analyses, suggests that the experience of being a cancer patient may confer a suicide risk beyond that associated with other medical and psychiatric conditions, at least among our population of older Americans. Our finding that suicide remained significantly associated with cancer after we controlled for the use of opioid analgesics suggests that pain may have been inadequately controlled and/or that pain per se does not fully explain the special status of cancer among suicides. Unfortunately, we do not have direct information about pain severity or about how the experience of cancer might otherwise have differed from the experience of other medical illness (eg, the effect of cachexia or the social meaning of malignancy in our society). Because only 19 suicides in our study had cancer, we lacked statistical power to examine how the risk of suicide may have differed according to the distribution of cancer diagnoses or the regimen of cytotoxic therapy received. Among our patients with cancer, one finding nevertheless stood out: metastatic disease was more common among patients with cancer who died by suicide than among patients with cancer who were controls. The predominance of metastatic disease among suicide patient cases with cancer held, not only in a comparison of the 19 suicides with cancer to the 64 patients with cancer who were controls, but also within the dominant cancer diagnosis among patient cases and controls (prostate cancer); this suggests that characteristics of advanced disease, its treatment, or the psychological and/or social response to progressive disease may have contributed to the risk of suicide (eg, intractable pain, inadequacy of poorer prognosis, use of higher doses of analgesics, social isolation). Almost two thirds of our cancer patients who died as a result of suicide used a firearm. By contrast, only two (11%) of the 19 suicide victims with cancer poisoned themselves; moreover, self-poisoning with opioid analgesics was not the method of suicide among these patients. This finding, consistent with other work,26-29 highlights the importance of considering the availability of household firearms as a suicide risk factor among this population. Our findings must be viewed in light of additional limitations. Study participants were all older adults and beneficiaries of Medicare and a drug prescription program for low-income elderly. Generalization to more affluent older adults or to younger adults may not be appropriate. In addition, our prevalence estimates for mental illness among our cohort are likely to underestimate the true prevalence, because unrecognized depression is common in the elderly.30-32 For example, psychological autopsy studies have estimated that 44% to 87% of elder suicides involve major depression,31,33 whereas only 23% of suicides in our cohort carried the diagnosis of depression or other schizoaffective disorders. If physicians are less likely to diagnose depression (when depression is actually present) in some medical diseases (eg, cancer) than in others, multivariate analyses that control for diagnosed depression will overestimate the independent association between suicide and these conditions. Our cross-sectional study does not address the unresolved and contentious issue of whether antidepressants increase, decrease, or have an indifferent effect on rates of suicide. Our finding that the use of antidepressants remained associated with suicide even after we controlled for diagnosed mental illness should not, therefore, be viewed as evidence for or against a causal link between antidepressants and suicide. The antidepressant-suicide association in our analyses could arise because antidepressants actually cause an increased risk of suicide, because of preferential prescribing of antidepressants to people who are at increased risk of suicide, or because antidepressant use is related to some other characteristic that increases the risk of suicide independent of the drug's direct effect (eg, antidepressant use as a proxy for intractable pain when used as an analgesic adjuvant). Despite these limitations, our study adds potentially valuable information to what is currently known about the relation between suicide and medical illness. Most patients who died as a result of suicide in our study, as in other studies,19,20 had visited a physician in the month before their death, and 25% were seen within a week of their suicide. These visits might have presented an opportunity to intervene. Our finding that patients with cancer appear to be at particularly high risk for suicide, beyond the risk accounted for by their probability of dying within a year (as proxied by the Charlson comorbidity score), the presence of diagnosed mental illness, and their use of psychoactive medication, suggests that additional research is needed to identify and to address potential tractable factors, such as body regions involved, presence of metastasis, pain control, social support, treatment regimens, and the stigma still associated with a diagnosis of cancer, that may put some patients with malignancy at an elevated risk of suicide.
The authors indicated no potential conflicts of interest.
Conception and design: Matthew Miller, Daniel H. Solomon Provision of study materials or patients: Katherine Hempstead Collection and assembly of data: Helen Mogun, Katherine Hempstead Data analysis and interpretation: Matthew Miller, Helen Mogun, Deborah Azrael, Daniel H. Solomon Manuscript writing: Matthew Miller, Deborah Azrael, Daniel H. Solomon Final approval of manuscript: Matthew Miller, Deborah Azrael, Daniel H. Solomon
published online ahead of print at www.jco.org on August 11, 2008 Supported by the American Foundation for Suicide Prevention and by the Centers for Disease Control Grants No. R49/CCR115279-09 for Injury Control Research Centers. Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Hoyert DL, Kung HC, Smith BL: Deaths: Preliminary data for 2003. Natl Vital Stat Rep 53:1-48, 2005[Medline] 2. Conwell Y, Duberstein PR, Caine ED: Risk factors for suicide in later life. Biol Psychiatry 52:193-204, 2002[CrossRef][Medline] 3. Carney SS, Rich CL, Burke PA, et al: Suicide over 60: The San Diego study. J Am Geriatr Soc 42:174-180, 1994[Medline] 4. Barraclough BM: Suicide in the elderly: Recent developments in psychogeriatrics. Br J Psychiatry 119:87-97, 1971 (suppl 6) 5. Ross RK, Bernstein L, Trent L, et al: A prospective study of risk factors for traumatic deaths in a retirement community. Prev Med 19:323-334, 1990[CrossRef][Medline] 6. Turvey CL, Conwell Y, Jones MP, et al: Risk factors for late-life suicide: A prospective, community-based study. Am J Geriatr Psychiatry 10:398-406, 2002 7. Conwell Y, Lyness JM, Duberstein P, et al: Completed suicide among older patients in primary care practices: A controlled study. J Am Geriatr Soc 48:23-29, 2000[Medline] 8. Duberstein PR, Conwell Y, Conner KR, et al: Suicide at 50 years of age and older: Perceived physical illness, family discord, and financial strain. Psychol Med 34:137-146, 2004[CrossRef][Medline] 9. Allebeck P, Bolund C: Suicides and suicide attempts in cancer patients. Psychol Med 21:979-984, 1991[Medline] 10. Preville M, Hebert R, Boyer R, et al: Physical health and mental disorder in elderly suicide: A case-control study. Aging Ment Health 9:576-584, 2005[CrossRef][Medline] 11. Beautrais AL: A case control study of suicide and attempted suicide in older adults. Suicide Life Threat Behav 32:1-9, 2002[CrossRef][Medline] 12. Rubenowitz E, Waern M, Wilhelmson K, et al: Life events and psychosocial factors in elderly suicides: A case-control study. Psychol Med 31:1193-1202, 2001[CrossRef][Medline] 13. Waern M, Rubenowitz E, Wilhelmson K: Predictors of suicide in the old elderly. Gerontology 49:328-334, 2003[CrossRef][Medline] 14. Campbell PC: Suicide among cancer patients. Conn Health Bull 80:207-212, 1966 15. Fox BH, Stanek EJ, Boyd SC, et al: Suicide rates among cancer patients in Connecticut. J Chronic Dis 35:89-100, 1982[CrossRef][Medline] 16. Marshall JR, Burnett W, Brasure J: On precipitating factors: Cancer as a cause of suicide. Suicide Life Threat Behav 13:15-27, 1983[Medline] 17. Louhivuori KA, Hakama M: Risk of suicide among cancer patients. Am J Epidemiol 109:59-65, 1979 18. Allebeck P, Bolund C, Ringback G: Increased suicide rate in cancer patients: A cohort study based on the Swedish Cancer-Environment Register. J Clin Epidemiol 42:611-616, 1989[CrossRef][Medline] 19. Quan H, Arboleda-Florez J, Fick GH, et al: Association between physical illness and suicide among the elderly. Soc Psychiatry Psychiatr Epidemiol 37:190-197, 2002[CrossRef][Medline] 20. Juurlink DN, Herrmann N, Szalai JP, et al: Medical illness and the risk of suicide in the elderly. Arch Intern Med 164:1179-1184, 2004 21. Grabbe L, Demi A, Camann MA, et al: The health status of elderly persons in the last year of life: A comparison of deaths by suicide, injury, and natural causes. Am J Public Health 87:434-437, 1997 22. Bjorkenstam C, Edberg A, Ayoubi S, et al: Are cancer patients at higher suicide risk than the general population? Scand J Public Health 33:208-214, 2005 23. Schneeweiss S, Seeger JD, Maclure M, et al: Performance of comorbidity scores to control for confounding in epidemiologic studies using claims data. Am J Epidemiol 154:854-864, 2001 24. SAS Institute: SAS/STAT software: Changes and enhancements, release 8.2. Cary, NC, SAS Institute Inc, 2001 25. Dorpat TL, Anderson WF, Ripley HS: The relationship of physical illness to suicide, in Resnik HLP (ed): Suicidal Behaviors: Diagnosis and Management. Boston, MA, Little, Brown, 1968, pp 209-219 26. Miller M, Hemenway D, Azrael D: Firearms and suicide in the northeast. J Trauma 57:626-632, 2004[Medline] 27. Miller M, Lippmann S, Azrael D, et al: Household firearm ownership and rates of suicide across the 50 United States. J Trauma 62:1029-1034, 2007[Medline] 28. Bailey JE, Kellermann AL, Somes GW, et al: Risk factors for violent death of women in the home. Arch Intern Med 157:777-782, 1997[CrossRef][Medline] 29. Kellermann AL, Rivara FP, Somes G, et al: Suicide in the home in relation to gun ownership. New Engl J Med 327:467-472, 1992[Abstract] 30. Conwell Y: Suicide in the elderly: Bias, infirmity, and suicide. Crisis 16:147-148, 1995[Medline] 31. Conwell Y, Brent D: Suicide and aging: I—Patterns of psychiatric diagnosis. Int Psychogeriatr 7:149-164, 1995[CrossRef][Medline] 32. Caine ED, Conwell Y: Suicide in the elderly. Int Clin Psychopharmacol 16:S25-S30, 2001 (suppl 2) 33. Conwell Y: Management of suicidal behavior in the elderly. Psychiatr Clin North Am 20:667-683, 1997[CrossRef][Medline] Submitted September 10, 2007; accepted December 12, 2007. Related Articles
Related Editorial
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2008 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
= .05 level. All models also controlled for matching factors (age, sex, and ethnicity) to avoid introduction of confounding by the matching processes itself. 
65 Years
