|Home | About | Journals | Submit | Contact Us | Français|
Information on the relationship between characteristics of mental healthcare providers, including hospitals and psychiatrists, and postdischarge suicide is scanty. This study aims to identify the risk factors for suicide among schizophrenia patients in the 3-month postdischarge period. The study cohort comprised all patients with a principal diagnosis of schizophrenia discharged from psychiatric inpatient care from 2002 to 2004 who committed suicide within 90 days of discharge. The control cohort consisted of all surviving schizophrenia patients discharged from psychiatric inpatient care in the same period and were matched to cases for age, gender, and date of discharge. There were 87 and 348 cases in the study and control cohorts, respectively. For suicide cases, death most frequently occurred on the first day after leaving the hospital (16.1%). The adjusted hazard ratios for committing suicide during the 90-day postdischarge period were 2.639 times greater for patients without previous psychiatric admission than for those hospitalized more than 3 times in the year preceding the index hospitalization. The adjusted suicide hazard for schizophrenia patients treated by male psychiatrists was significantly higher than for patients treated by female psychiatrists, by a multiple of 5.117 (P = .032). The adjusted suicide hazard among patients treated by psychiatrists over age 44 years was 2.378 times (P = .043) that for patients treated by psychiatrists aged younger than 35 years. Risk factors related to psychiatric hospitalization, including number of psychiatric admissions in the previous year and length of stay, together with gender and age of the psychiatrist providing inpatient care, are identified.
Excess mortality has long been observed among psychiatric patients, particularly those with schizophrenia.1 Besides death from natural causes, schizophrenia patients are at an increased risk of unnatural death, such as suicide. The lifetime risk of suicide among patients with schizophrenia has been estimated at 10%.2 A recent large-scale meta-analysis still reveals that closer to 5% of this disadvantaged population will commit suicide during their lifetimes.3 Growing awareness of suicide risk has drawn attention to detection and management of suicidal tendencies among patients with schizophrenia.
Previous studies indicate a greater suicide risk for patients following discharge from psychiatric inpatient care.4,5 Based on data from Danish population registers, Qin and Nordentoft6 reported a sharp peak of suicide risk in the first week after discharge from psychiatric hospitalization. Among schizophrenia patients, half of those who commit suicide end their own lives in the first 3 months after discharge.7 Thus, preventive intervention should concentrate on the postdischarge period during which the clustering of suicides tends to occur.
Sociodemographic variables, including age, gender, unemployment, and living alone, have been proposed as associated with psychiatric patients’ suicides after leaving the hospital.8–11 A history of deliberate self-harm has been consistently reported to increase the risk of postdischarge suicide.10–13 Discharged patients are more likely to commit suicide if their most recent hospitalization was their first admission.12 In addition, compulsory admission may be another important risk factor for postdischarge suicide.11,13
Nevertheless, few studies have focused on risk factors specific to postdischarge suicide among schizophrenia patients. The majority of previous studies consisted of patient samples with varied psychiatric diagnoses. Because a substantial number of suicide victims suffer from affective disorders, some of the proposed risk factors could be inappropriate, applied to schizophrenia. Studies focusing on suicide risk among schizophrenics often classify all suicides into one group no matter when the deaths occurred.14 This type of study design may fail to identify suicide risk in relation to psychiatric inpatient care.
In terms of psychiatric hospitalization, patients receiving a shorter than median length of inpatient care have been found to be at an increased risk of committing suicide.6 Furthermore, Meehan et al15 reported in a large case series of suicides that 28% of postdischarge deaths occurred in the unplanned discharge group. Although the quality of care may be associated with specific practice patterns, there is scant information on the relationship between characteristics of mental healthcare providers, including hospitals and psychiatrists, and postdischarge suicide.
In this nationwide, matched case-controlled study, we aim to identify the risk factors for suicide among schizophrenia patients in the 3-month postdischarge period. We particularly examined risk factors related to psychiatric inpatient care, including characteristics of mental healthcare providers. Further insights about postdischarge suicide may be beneficial to develop effective suicide prevention strategies for schizophrenia patients in this critical transition period.
The dataset for this study was compiled from 2 large-scale population-based databases in Taiwan, the 2001–2004 National Health Insurance Research Database (NHIRD) and the Cause of Death File provided by the Taiwanese Department of Health (DOH).
The former is published by the National Health Research Institutes in Taiwan and includes all claims data from the National Health Insurance (NHI) program, which was implemented in 1995 as a means of providing comprehensive coverage of all inpatient and outpatient care for over 21 million enrolled individuals (around 96% of the Taiwanese population). The NHI program contracts with all healthcare institutions providing psychiatric inpatient care in Taiwan. As a single-payer payment system that offers unrestricted access to any mental healthcare provider of the patient's choice, the NHIRD offers a unique opportunity to identify risk factors related to psychiatric inpatient care prior to suicide death, particularly characteristics of mental healthcare providers.
The NHIRD consists of comprehensive hospitalization data, such as medical claims for inpatient expenditures, admissions, details of inpatient orders, and a registry of contracted medical facilities, medical personnel, board-certified specialists, and beneficiaries. In addition, one principal diagnosis is listed for each hospitalization, along with up to 4 secondary diagnoses; these are based upon the “International Classification of Diseases, Ninth Revision, Clinical Modification” (ICD-9-CM) codes.
It is mandatory for all deaths throughout Taiwan to be registered with the DOH, with this register providing information on the gender, age, along with suicide method, date, and place.
The NHIRD was linked to the Cause of Death File with the assistance of the DOH in Taiwan. Because these were deidentified secondary data, released for public access for research purposes, the study was exempt from full review by the Internal Review Board.
Our study design features a study cohort and a control cohort. The study cohort comprised all patients discharged from psychiatric inpatient care with a principal diagnosis of schizophrenia (any ICD-9-CM 295 code other than 295.7—schizoaffective disorder) from January 1, 2002, to December 31, 2004, who committed suicide (ICD-9-CM codes E950–E959) within 90 days of discharge. Only voluntary admissions were selected because not all hospitals provide compulsory inpatient care. Because compulsory admissions comprised less than 8% of total psychiatric admissions during the study period, excluding them was not likely to compromise the results. In total, there were 87 cases identified in the study cohort.
The control cohort comprised all surviving patients discharged from psychiatric inpatient care for the treatment of schizophrenia in the same period. We further refined our criteria for the control cohort by randomly selecting 348 cases (4 for every case in the study cohort) matched with the study cohort in terms of age, gender, and date of discharge.
The dependent variable was dichotomous: suicide death (or not) within 90 days of index discharge. The key independent variables consisted of patient, hospital, and psychiatrist characteristics at index hospitalization. Patient characteristics included both sociodemographic and clinical features. The former were age (grouped after examining the distribution of the selected sample: <30, 30–39, 40–49, >49), gender, and monthly income. Patients’ clinical features included length of stay (LOS) for the index hospitalization, unplanned discharge, and the number of psychiatric admissions within the year prior to the index hospitalization. Because there were no defined standard cutoff points for LOS, we ranked all hospitalizations for schizophrenia in the study period and selected cutoff points that most closely sorted them into 4 groups of similar size, in accordance with methods used by prior studies.16 Thus, the LOS was grouped into the following categories: <14, 14–26, 27–40, and >40. Because the administrative data used in this study does not include severity of illness, we used the number of psychiatric admissions within the year prior to index hospitalization as a proxy for severity of illness during the past year, as categorized into 3 groups: 0, 1–3, and >3.
Hospital characteristics included the type of ownership (public, not for profit, or for profit), teaching status (yes or no), and hospital level. Hospital level refers to the categorization of each hospital into medical centers (minimum 500 beds), regional hospitals (minimum 250 beds), or district hospitals (minimum 20 beds) and can be considered as a proxy for hospital size and clinical service capabilities.
The information on psychiatrist characteristics including age (as a surrogate for practice experience) and gender is included in the NHIRD. It is possible that a patient was associated with more than one psychiatrist during hospitalization. However, the database only shows the attending psychiatrist for any particular patient. “Attending psychiatrist” refers to the psychiatrist who admitted the patient and who is primarily responsible for oversight and care direction/coordination throughout the patient's stay. Psychiatrists’ age was categorized into 3 groups after examining the distribution of the study sample (≤34, 35–44, or ≥45).
The SAS statistical package (SAS System for Windows, Version 8.2) was used to perform the analyses. Pearson chi-square tests were used to examine differences between the 2 cohorts in terms of characteristics of patient, hospital, psychiatrist at index hospitalization, and the risk of committing suicide in the 3-month postdischarge period. Cox proportional hazard regression was also carried out as a means of computing the adjusted 3-month survival rate, following adjustment for the possible confounding factors. We kept only risk factors that were significantly related to suicide from the chi-square analyses in the regression model. The hazard ratios (HRs) are presented along with the 95% confidence intervals (CIs), a 2-sided P value of <.05 being considered statistically significant for this study.
Of the total sample of 435 patients, the mean age was 39.3 years, with an SD of 9.9 years. For suicide cases, deaths (16.1%) occurred most frequently on the first day after leaving the hospital, 32.2% occurred in the first 2 weeks, and almost half (46.0%) in the first month after discharge (table 1). Table 1 also shows that 28.7% of suicide victims killed themselves by self-poisoning and 58.6% committed suicides in their hometowns.
The details of the distribution of characteristics of patient, psychiatrist, and hospital at index hospitalization for these 2 cohorts are provided in table 2. There was no psychiatrist who cared for more than one sampled case in the study. The Pearson chi-square tests show that there were significant differences between these 2 cohorts in terms of LOS (P = .015), the number of psychiatric admissions within the year prior to index hospitalization (P = .005), and unplanned discharge (P = .016), as well as psychiatrist gender (P = .043) and age (P = .049). However, no significant differences were observed between these 2 cohorts in terms of hospital characteristics.
The results of regression analysis are presented in table 3. Rather surprisingly, the hazard of committing suicide for patients hospitalized between 27 and 40 days was only 0.352 times (95% CI = 0.172–0.721, P = .004) that of patients hospitalized less than 14 days. In addition, the adjusted HRs of committing suicide during the 3-month postdischarge period were 2.639 (reciprocal of 0.379) times greater for patients without previous psychiatric admission than for those hospitalized more than 3 times within the year prior to index hospitalization.
In terms of mental healthcare provider characteristics, the risk of schizophrenia patients committing suicide was significantly higher for those treated by male psychiatrists than that for those treated by female psychiatrists by a multiple of 5.117 (95% CI = 1.154–22.691, P = .032). The hazard of suicide among patients treated by psychiatrists over age 44 years was 2.378 times (95% CI = 1.028–5.499, P = .043) that for patients treated by psychiatrists aged younger than 35 years.
A 3-year national sample of 87 suicide victims suffering from schizophrenia who died in the 3-month postdischarge period was obtained utilizing a large population-based dataset. Because the number of schizophrenia patients voluntarily admitted to psychiatric inpatient care was approximately 25000 per year, the average suicide rate for such patients is almost 7 times the annual rate in the general population during the study period.
The age and gender distribution among suicide cases in this study are comparable to discharged suicide populations in previous studies.10–13,15 As regards suicide methods, our findings that self-poisoning, jumping, and hanging are the 3 leading means are similar to reports from the West.15 However, the rapid rise of suicide by burning charcoal among the general population in Taiwan and other Asian countries is not evident in our study sample.17 Because the spread of burning charcoal as a method of suicide has been attributed to the influence of mass media, this novel method of suicide might not emerge so dramatically among patients with schizophrenia, partly due to their social inattentiveness.
Similar to findings by Meehan et al,15 one-third of postdischarge suicides occurred in the first 2 weeks after leaving the hospital. The increased suicide risk among schizophrenia patients during this critical transition period is once again confirmed across cultures. Given that 41.4% of suicide victims died in places other than their hometowns, the continuity of mental health care before their deaths seems questionable.
To the best of our knowledge, this is the first matched case-controlled study attempting to explore the impact of mental healthcare providers, both hospital and psychiatrist, on postdischarge suicide among schizophrenia patients. There have been many calls to strengthen risk assessment skills of clinical staff with regard to the high suicide mortality risk during the postdischarge period. In fact, suicide risk is often underestimated during the victim's final contact with mental healthcare service.18 Suboptimal care may be associated with certain subgroups of mental healthcare providers in particular.
Results indicated that the risk for postdischarge suicide among schizophrenia patients is associated with the gender and age of the attending psychiatrist. In mental health care, the gender of the therapist was noted to have little effect on the outcome of psychotherapy.19 However, scant information on this issue is available for psychiatric inpatient care. It may be difficult to draw firm conclusions about the issue of psychiatrist's gender for there were only 2 suicide cases cared for by female psychiatrists. With regard to the age of healthcare professionals, it is generally believed that practice makes perfect. That is, physicians with more experience may have more abundant knowledge and better clinical practice skills enabling them to provide high-quality care. Surprisingly, however, a recent review concludes that physicians who have practiced longer are at risk of delivering lower quality care.20 Besides, senior psychiatrists may treat more patients on account of their reputation. It is plausible that senior psychiatrist's experience and reputation get them sicker patients. Moreover, an association between psychiatrists with higher caseload volume and higher 30-day readmission rates has been reported recently.21 This inverse volume-outcome relationship might provide another explanation for the increased risk of postdischarge suicide among schizophrenia patients cared by psychiatrists over 44 years of age. The distribution of psychiatrists’ ages in the study sample was not normal and was skewed to the right. This reflects the increasing number of young psychiatrists in Taiwan.
In partial agreement with findings by Appleby et al,12 in our study, schizophrenia patients without psychiatric admission in the year prior to index hospitalization were more likely to commit suicide. Although 40.2% of suicide victims had their last inpatient stay less than 2 weeks prior to death, both shorter and longer LOS would have increased the suicide risk during the postdischarge period based on the results of our regression analysis. A short LOS, noted as carrying risk for postdischarge suicide,6,22 may imply inadequate treatment. While a longer inpatient stay does not always guarantee good treatment outcomes, it could facilitate the development of a symbiotic relationship between patients and healthcare providers that makes discharge a painful experience.23
A particular strength of this study is the use of 2 nationwide population-based datasets, allowing us to trace schizophrenia patients’ utilization of psychiatric inpatient care prior to completed suicide. Furthermore, information obtained from administrative datasets can avoid recall bias. Nevertheless, we should make note of 3 inherent limitations to this study. Prior studies have reported that certain patient-specific clinical features, such as a history of self-harm, suicidal ideation prior to admission, mood symptoms, and recent life events, are important predictors of postdischarge suicide. Unfortunately, none of this information is available from the NHIRD.
Secondly, misclassification of cause of death and consequent underreporting of suicides within the registry system could potentially confound results. Furthermore, the diagnosis of schizophrenia reported by psychiatrists or hospitals may be less accurate than diagnoses made under a standardized schedule; however, the development of such a population-based dataset containing this sort of information would be extremely costly and difficult to accomplish.
Despite these limitations, this study replicates findings of studies in the West that indicate significant clustering of suicides soon after discharge from psychiatric care. Risk factors related to psychiatric hospitalization, including number of psychiatric admissions in the previous year, LOS, together with the gender and age of psychiatrists providing inpatient care, have been identified, after adjusting for patient and hospital characteristics. To confirm generalizability, the findings need to be replicated in other regions or countries. The relationship between postdischarge suicide and psychiatrists’ gender and age should therefore be regarded as exploratory rather than definitive. Further studies should be initiated to identify subgroups of mental health professionals who may need quality improvement interventions. With regard to suicide prevention, quality assurance, and performance evaluation of psychiatric care, focusing on suicide risk assessment and management should be emphasized during this critical transition period.
National Science Council, Taiwan (NSC 96-2314-B-038-021).