Suicide in the Medically Ill

By | March 6, 2016

Physical disease is present in a high proportion of people who commit suicide. A recent review cited several large studies that reported that medical illness was present in 30%-40% of the patients who committed suicide. However, most of these suicides do not occur during medical hospitalization. About 2% of Finnish suicides occurred in medical or surgical inpatients. In Montreal, Quebec, about 3% of the suicides were in general hospital inpatients, of which one-third (1%) were medical-surgical patients. During a 10-year period in a 3, 000-bed Chinese medical hospital, there were 75 self-destructive acts, only 15 of which proved fatal.

Sanders () reviewed six studies of inpatients at a general hospital who committed suicide. Most had one or more chronic or terminal illnesses or sequelae that were painful, debilitating, or both, including dyspnea, ostomies, or disfiguring surgery. Harris and Barraclough compiled a list of 63 medical disorders noted in the medical literature as potentially having elevated suicide risk. In their meta-analysis, they concluded that the only disorders that actually elevated suicide risk were HIV and AIDS, Huntington’s disease, cancer (particularly head and neck), multiple sclerosis, peptic ulcer disease, end-stage renal disease, spinal cord injuries, and systemic lupus erythematosus. More recent studies confirm or add to a seemingly arbitrary list of medical conditions associated with risk for suicide. In a Canadian study, cancer, prostate disease, and chronic pulmonary disease were associated with suicide ORsof 1.70-1.86 among adults older than 55 years with versus without the diseases. In the previously cited Chinese study of patients who committed suicide in a general hospital, 40% had cancer, 13% had neurological disease, 13% had cardiovascular disease, and 7% had liver failure. In the Montreal study, associated diagnoses included cardiovascular disease, abdominal pain, cerebrovascular disease, Parkinson’s disease, and rheumatoid arthritis. Of 12 patients who jumped from an Australian hospital between 1980 and 1991, 4 had delirium, 4 had terminal cancer, 2 had advanced lung disease, and 1 had irreversible cardiac failure. However, these studies were small and did not capture suicides in the medically ill attempted or completed outside the hospital, so they cannot be used to construct a list of “most suicidal” medical disorders.

A recent study drawing on the U.S. National Comor-bidity Survey identified a dozen general medical diagnostic categories with statistically significantly elevated ORs for suicide attempts, most ranging from 1.1 to 3.2, except for AIDS (133.9) and hernia (10.4). Clinically, however, use of a diagnosis alone in estimating suicide risk is not helpful. Even though the OR in each of the 12 categories achieved statistical significance, substituting a rate only slightly higher than the very low base rate offers little to guide clinical decision making, particularly if this is the only indicator being used to predict suicide.

What does appear useful is that suicides in the medically ill — as in the general population — appear to be related to frequently unrecognized comorbid psychiatric illnesses, including depression, substance-related disorders, delirium, dementia, and personality disorder. In their study of the role of physical disease in 416 Swedish suicides, Stensman and Sundqvist-Stensman () concluded that somatic disease was one important factor in the complexity of the suicidal act, but psychiatric conditions such as depression and alcohol abuse were more significant. Rather than focus on particular medical diagnoses, it will be more fruitful for the medical psychiatrist to determine whether a suicide-prone psychiatric condition is present in a medically ill patient, whether the patient is at a particularly emotionally difficult time in his or her illness course, and whether secondary effects of the medical illness — pain, physical disfigurement, cognitive dysfunction, and disinhibition — are present that add to the risk.

It must be emphasized that no matter how horrific the medical condition, significant suicide risk is not the rule. According to Brown et al. (1986), most terminally ill patients do not develop severe depression, and suicidality is closely associated with the presence of a depressive disorder. In the study of terminally ill cancer patients by Breit-bart et al., only 17% had a high desire for hastened death, for which depression and hopelessness were the strongest predictors. An important empirical finding in a Canadian study was that the will to live in the terminally ill fluctuates, mostly predicted by depression, anxiety, shortness of breath, and sense of well-being.

Three diagnoses — cancer, end-stage renal disease, and AIDS — are discussed here to illustrate these points further. These comparatively common conditions underscore principles that can be extrapolated to the breadth of diagnoses and situations encountered in medical settings.


Three large studies have found an increased suicide rate among patients with cancer. Luohivuori and Hakama () studied 63 suicides among 28, 857 Finnish cancer patients and found relative risks (compared with the general population) of 1.3 for women and 1.9 for men, with the highest excess mortality associated with gastrointestinal tumors. Fox et al. () studied 192 suicides between 1940 and 1973 among 144, 530 patients in the Connecticut Tumor Registry and calculated no increased suicide risk for women but a 2.3 relative risk for men. In the largest study of the relation between cancer and suicide, Alle-beck and colleagues () gathered statistics on 963 suicides between 1962 and 1979 among 424, 127 Swedes with a diagnosis of cancer and found an overall 1.9 relative risk for men and a 1.6 relative risk for women. Gastrointestinal tumors (excluding colorectal cancers) in men (relative risk=3.1) and lung tumors in either sex (relative risk=3.1 for men and 3.5 for women) were associated with the highest rate of death due to suicide.

Cancer patients who die by suicide are psychiatrically similar to noncancer patients, particularly when the cancer is in remission. In a case-control study of 60 suicides in individuals with cancer and 60 age- and sex-matched comparison suicides in individuals without a cancer history, Henriksson et al. found that most of the patients with cancer who committed suicide — as well as the control subjects without cancer — had a diagnosable psychiatric disorder. Terminally ill cancer patients had lower rates of depression and alcohol dependence than did patients in remission (72% vs. 96%), but nearly three-quarters still met criteria for a depressive disorder. As a group, cancer patients had fewer psychotic disorders than did control subjects. Allebeck et al. () observed that the longer the time from diagnosis of cancer, the lower the relative risk for suicide in a Swedish cohort. In the first year after diagnosis, the relative risk was 16.0 for men and 15.4 for women. From 1 to 2 years, the ratio decreased to 6.5 for men and 7.0 for women. By 3-6 years, the ratio was 2.1 for men and 3.2 for women. By 10 years after diagnosis, the rate, at 0.4, was actually less than one-half that in the general population. A study of Japanese cancer patients found the highest risk of suicide soon after patients had been discharged from the hospital, with an elevated relative risk the first 5 years after diagnosis compared with the general population and disappearing thereafter.

The fear of pain, disfigurement, and loss of function that cancer evokes in the patients’ imagination can precipitate suicide, especially early in the patients’ courses. In a large cohort of Italians with cancer, suicide accounted for only 0.2% of the deaths, but the relative risk during the first 6 months after diagnosis was 27.7. The high relative risk of suicide just after diagnosis comes at a time of overwhelming fear and cognitive overload. In individual patients, important contributing factors can include overly pessimistic prognosis, exaggerated impressions of anticipated suffering, a physician unintentionally undermining hope, fear of loss of control, or nihilism about treatment. Patients may fear or experience inadequate pain control, lost dignity, compromised privacy, or guilt at having habits that caused the disease. Surgical treatments may be disfiguring, chemotherapy debilitating, and side effects defemi-nizing or emasculating. As cancer patients live longer with their disease, most become less frightened and less susceptible to suicide.

End-Stage Renal Disease

More formidable than the suicide risk among cancer patients was the purported increase in relative risk of suicide among patients with end-stage renal disease. Abrams and colleagues () reported very high rates of suicide and suicidal behavior among 3, 478 renal dialysis patients studied at 127 dialysis centers. In their sample, 20 deaths were the result of suicide; 17 suicide attempts were unsuccessful; 22 patients withdrew from the program, knowing that doing so would hasten their deaths; and 117 deaths were attributed to noncompliance with treatment. The authors’ calculated suicide rate of 400 times that in the general population has been widely quoted but is misleading. In arriving at a 5% figure for suicidal behavior in dialysis patients, they used an extremely broad definition of suicide that encompassed death caused by a wide range of causes, from willful acts of self-destruction to noncompliance.

Most of the cases that Abrams and colleagues called suicide would never come to the attention of psychiatry today. Although their report has been widely cited, no other subsequent study (there have been nearly 20) has defined suicide so broadly. In extreme cases, noncompliance is better understood as a function of personality-disordered behavior; in less dramatic examples, it can be an understandable human response to a burdensome treatment. Deciding to forgo dialysis is not equivalent to suicide (“Renal Disease”). A recent United States study concluded that “most patients who decide to stop dialysis do not seem to be influenced by major depression or ordinary suicidal ideation”. Treatment withdrawal, negotiated among the patient, significant others, and the treatment team, has become routine as quality of life during dialysis fades.

In 1, 766 Minnesota dialysis patients followed up for 17 years, for example, only 3 killed themselves by frank suicide, representing only 2% of the 155 cases in which dialysis was discontinued. The suicide rate in this sample of dialysis patients was only about 15 times that in the general population, which is a considerable rate but much lower than Abrams and colleagues’ figure. Haenel et al. () also found less dramatic suicide rates among European patients undergoing chronic dialysis between 1965 and 1978. In Switzerland, dialysis patients killed themselves at about 10 times the rate in the general population. When patients who refused therapy and died as a result were included in the suicide group, the rate was 25 times higher. They also found no statistically significant difference between suicide rates among patients with functioning cadaveric renal transplants and patients undergoing maintenance dialysis, suggesting that transplantation may not in and of itself be associated with decreased suicide risk. Overall, among dialysis patients pooled from all countries belonging to the European Dialysis and Transplant Association, the suicide rate was 108 per 100, 000 per year. Whether compared with the general population suicide rate of 4-5 per 100, 000 in Mediterranean countries or 20-25 per 100, 000 in central European or Scandinavian countries, the figure of 108 per 100, 000 represents a higher suicide rate, although not orders of magnitude greater than that in the general population.


AIDS patients also have a higher relative risk of suicide, even though the risk appears to have decreased since the disease emerged. The existing data are primarily based on men who had sex with men in the United States in the 1980s. Extrapolation to the present is problematic because of many changes, including the demographics and geographic distribution of AIDS, advances in treatment, availability of mental health services, public education, and reduction in stigma and social hysteria. The perspective of suicide has changed as AIDS has evolved from a terminal illness to a chronic one. Another caveat in interpreting studies of suicidality in persons with HIV is that results will be confounded because the study populations (e.g., men who have sex with men, injection drug abusers, and poor minority heterosexual women) all differ in their sociodemographics and psychiatric epidemiology.

Marzuk et al. () found a suicide rate in persons with AIDS 36 times that in an age-matched sample of men without AIDS and 66 times that in the general population in New York City in 1985. Marzuk and colleagues reexamined this question based on all suicides in New York City in 1991-1993 and concluded that positive HIV serostatus was associated, at most, with a modest elevation in suicide risk. In California, in 1986, the rate was 21 times higher than that in the general population. In the largest study to date, Cote et al. charted a continuous decrease in suicide rates over 3 years among AIDS patients in 45 states and the District of Columbia. From 1987 to 1989, a total of 165 suicides among AIDS patients were reported to the National Center for Health Statistics. Of these, 164 were committed by men. The relative suicide risk calculated for AIDS patients was 10.5 in 1987, 7.4 in 1988, and 6.0 in 1989. The authors attributed the decrease to advances in medical care, diminishing social stigma, and improved psychiatric services, while noting probable underreporting of deaths due to both AIDS and suicide (Cote et al. 1991). In a review of 100 publications with information about suicide and HIV, Palmer et al. (in press) concluded that there has been a distinctive downward trend in HTV-related suicidality in the United States since the beginning of the epidemic.

Frierson and Lippman () suggested that suicide risk also may be increased among HIV-positive but asymptomatic people who fear the eventual illness, HTV-negative people who are worried about contracting the disease, and people who enter suicide pacts with dying loved ones. Rundell and colleagues compared 15 HTV-infected active-duty members of the air force who attempted suicide with 15 who did not and identified several risk factors equivalent to risk factors for suicide in general, including social isolation, perceived lack of social support, adjustment disorder, personality disorder, alcohol abuse, interpersonal or occupational problems, and history of depression.

Recent studies reflect both the changing demographics of HIV and AIDS and the stable classic risk factors for suicidality. Roy found that almost half of a cohort of HIV-positive substance-dependent patients had attempted suicide. Those who had attempted suicide were younger; were more likely to be female; and were more likely to have more childhood trauma, more depression, more family history of suicidal behavior, and higher neu-roticism. A survey in HlV-infected Americans living in rural areas found that 38% had thoughts of suicide during the past week, associated with greater depression and more stigma-related stress and less coping self-efficacy.

Finally, two recent studies, one in Italy and one in Brazil, found that although psychiatric morbidity and suicidal ideation or attempts are common in HIV-positive intravenous drug abusers, they are equally common in those who are HIV-negative. A Swiss study of men having sex with men found a high rate of suicide attempts in both HIV-negative and HIV-positive individuals, with moderately more suicidal ideation in those who were HIV-positive. Dannenberg et al. compared 4, 147 HIV-positive United States military service applicants and 12, 437 HIV-negative applicants disqualified from military service because of other medical conditions (matched on age, race, sex, and screening date and location) with the matched general population; the relative risk for suicide was similar for each group: 2.08 in the HTV-positive and 1.67 in the HIV-negative applicants. These studies reinforce the point that psychopa-thology is implicated more potently than any specific medical diagnosis in suicidality.

Prevention and Treatment

The first priority in preventing suicide in the medically ill is the early detection and treatment of the comorbid psychiatric disorders covered throughout this book. Patient and family education about the medical disease course and its treatment can help prevent excessive fear and pessimism. Direct questions and frank discussion about suicidal thoughts, ideally part of every primary physician’s care for any patient with a serious disease, can reduce suicidal pressures. One important role for psychiatrists is to restrain other physicians from automatically prescribing antidepressants for every medically ill patient who expresses a wish to die. Overdiagnosis of depression can lead to inappropriate pharmacotherapy, pathologization of normal feelings, or neglect of relevant personality traits potentially amenable to psychotherapeutic intervention. Soliciting patients’ wishes and preferences regarding pain management and end-of-life care early on may reduce the fear of having no control of their dying that lures some patients toward suicide.

Palliative care for the terminally ill is essential in offering relief to those for whom life has become (or is feared) unbearable “Palliative Care”. Psychiatrists can help elicit fears, guilt, impulses, and history that patients may be reluctant to share with their primary physicians. In addition to treating psychiatric symptoms, psychiatrists can monitor for illicit drug use, medication side effects, and emergent neuropsychiatric complications of the underlying medical illness. Psychotherapy can facilitate the exploration and expression of grief and restore a sense of meaning in life “Psychotherapy, ” “Palliative Care”. Psychotherapy also may be psychoeducational, reinforcing patients’ and family members’ accurate knowledge about the disease. Attention to patients’ spiritual needs is very important as well; spiritual well-being offers some protection against end-of-life despair. Finally, for both patients and family, support groups and other community resources may be critical in making the difference between feeling life is worth living and giving up.


Selections from the book: “Textbook of Psychosomatic Medicine”, 2005.