A patient's increased risk of suicide shortly after discharge from a psychiatric hospital is well documented (1,2). In a recent review that included 40 studies of the rates of contact of persons who committed suicide (3), it was reported that one-third of the victims had contacted mental health services within a month before the suicide. Almost half of the victims had been in touch with primary health care one month before the suicide and 77 percent within a year preceding the suicide.
A majority of the suicide victims had suffered from psychiatric disorders, mostly depression (4). An increased prevalence of suicidal ideation, suicide attempts, and completed suicides was also reported among persons with various types of physical illnesses (5). In fact, suicidality among somatic patients has been suggested to arise from comorbid psychiatric illness, especially depression (6).
The actual duration of survival between somatic hospitalization and suicide has hardly been investigated. We studied this issue with reference to a population-based collection of suicides with linkage to national hospital discharge registers. First, suicide victims were classified into those whose last hospitalization before the suicide was due to either a psychiatric disorder or a physical illness. The victims who had last been hospitalized for a physical illness were further subdivided into those with and those without a lifetime history of psychiatric disorders. Thus three study groups were formed and compared in terms of survival from the time of the last hospitalization. In addition, survival periods were compared among the victims with and without a history of psychiatric disorders after the last hospitalization for the main categories of somatic diagnoses.
The study population consisted of all 1,585 suicide victims during the years 1988 to 2000 in the province of Oulu in northern Finland. The data were based on the death certificates from forensic medical-legal investigations and hospital admissions from the national Finnish hospital discharge registers. The study protocol was approved by the ethics committee of Oulu University.
Three study groups were distinguished on the basis of the nature of the last hospitalization. Suicide victims were assigned to exclusive groups, according to discharge diagnoses at the last hospitalization before the suicide, whether that involved physical illness with (group 2) or without (group 1) a history of hospital-treated psychiatric disorder or psychiatric disorders alone (group 3).
All discharge diagnoses of infectious diseases were combined into a single group, because in ICD they are usually coded by site (7). Other physical diseases were grouped in accordance with the main ICD diagnostic categories. Psychiatric disorders included the following major categories: depression, schizophrenia, other psychotic disorders, substance-related disorders, and psychiatric diagnoses other than those mentioned.
Kaplan-Meier survival analyses and log rank tests were used to compare temporal associations between the last hospitalization and the suicide in the three study groups. Group differences in continuous variables were assessed with the Mann-Whitney U test and the Kruskal-Wallis test. The statistical software used was SPSS for Windows, version 11.
The results of the Kaplan-Meier analysis showed that survival estimates between the last hospitalization and the suicide differed significantly among the three study groups for both men (log rank=256.12, df=2, p<.001) and women (log rank=60.59, df=2, p<.001). Survival estimates also differed significantly in pairwise comparisons: group 1 compared with group 3 (men, log rank=213.90, df=1, p<.001; women, log rank=211.64, df=1, p<.001) and group 1 compared with group 2 (men, log rank=154.54, df=1, p<.001; women, log rank=21.39, df=1, p<.001).
Among men, median survival time after the last psychiatric hospitalization (group 3) was the shortest (73 days); it was 146 days and 1,351 days among victims whose last hospitalization was due to physical illnesses with (group 2) and without (group 1) a history of psychiatric disorders, respectively. Among women, the corresponding times were 18 days (group 3), 110 days (group 2), and 621 days (group 1) (Kruskal-Wallis test: men, χ2=252.8, df=2, p<.001; women, χ2=65.6, df=2, p<.001).
t1 presents prevalence and survival times by the main somatic ICD categories and history of psychiatric disorder. Somatic categories in which significantly shorter survival times were observed for suicide victims who had a history of psychiatric disorders than for those who did not have such a history included infections; diseases of the nervous, circulatory, respiratory, digestive, genitourinary, and musculoskeletal systems; injuries and poisonings; and symptoms, signs, and ill-defined conditions. Among victims without a history of psychiatric hospitalization, suicides occurred most rapidly (26 days) when the last hospitalization involved malignant neoplasms.
When the last hospitalization involved a psychiatric disorder, suicide occurred within two weeks among victims who were hospitalized for depression (124 victims, median of 15 days). Survival times in days for victims who were hospitalized because of schizophrenia, substance use, other psychoses, and other psychiatric disorders were 33 days (93 victims), 172 days (111 victims), 66 days (24 victims), and 113 days (95 victims), respectively (Kruskal-Wallis test, p<.001).
The main finding of this study was that previous psychiatric hospitalization was clearly associated with the time interval between hospitalization and suicide, even if the last hospitalization was for a somatic complaint. Thus our findings, supporting earlier ones, suggest that suicidal ideation among somatic patients may arise from comorbid psychiatric illness, especially depression (6).
Different types of physical illness have been suggested as long-term predictors of suicide (8). Our study revealed that the effect of previous psychiatric hospitalization on the timing of a suicide after somatic hospitalization was clearly apparent as a shorter survival time for the patients in a majority of the somatic diagnostic categories. Furthermore, the survival time after hospitalization due to malignant neoplasms was surprisingly short—about three days—for victims without a psychiatric history, which agrees with previous findings of increased risks of suicide among patients with cancer (9) and may indicate that the knowledge of harboring a malignant disease can cause a person to become more susceptible to suicidal ideation, regardless of whether the person has a history of a psychiatric disorder.
For a notable group of suicide victims in our study, the last hospitalization had been for injuries and poisonings. It is highly probable that some of the poisonings and injuries were masked suicide attempts rather than accidents (10). Generally, and applicable to almost every category of somatic disorders, victims with psychiatric histories commit suicide more rapidly than mentally healthy individuals.
A limitation of our study was that both physical and psychiatric diagnoses were based on information from hospitals, and thus only suicide victims with disorders severe enough to warrant inpatient treatment were present in the study groups. Second, because of the small number of cases, interpretation of the results in some somatic disease categories was tentative. Because all the study subjects were suicide victims, we could not estimate at an epidemiologic level the actual suicide risks among the victims with different types of comorbid psychiatric and physical disorders.
A strength of our study was that all suicides committed in the province of Oulu during the study period were evaluated. Thus the study was limited to neither certain age groups nor selected suicide data. Also, the Finnish hospital discharge register has been shown to be a reliable source of information about hospitalizations (7).
We reported that a history of hospital-treated psychiatric disorders was linked with shorter survival times between a last hospital discharge and a suicide, even if the last hospitalization had been due to a somatic complaint. Given that physical morbidity may also be a substantial risk factor in suicides, attention must be paid to this factor in treating somatic patients, especially if these patients are known to have a history of psychiatric hospitalization and to have made previous suicide attempts.
Dr. Lantto, Mr. Viilo, Dr. Hakko, Dr. Räsänen, and Dr. Timonen are affiliated with the department of psychiatry of the University of Oulu in Oulu, Finland. Mr. Viilo and Dr. Hakko are also with the department of psychiatry at Oulu University Hospital. Dr. Särkioja is with the department of forensic medicine at the University of Oulu. Dr. Meyer-Rouchow is with the department of biology at the School of Engineering and Science at the International University Bremen in Bremen, Germany, and with the department of physiology at the University of Oulu. Dr. Timonen is also with Oulu Health Center. Send correspondence to Dr. Timonen at the Department of Public Health Science and General Practice, University of Oulu, Finland (e-mail, firstname.lastname@example.org).
Median survival timesa among suicide victims who had been hospitalized for physical illness before the suicide, by history ofmental disorder
a Period between last hospital discharge and date of suicide