Incidence and Risk Profiles for Suicide Mortality in Patients With Schizophrenia Receiving Homecare Case Management in Taiwan

Abstract

Case management (CM)-based community therapy for patients with schizophrenia had little effect on reducing suicide mortality. We investigate the long-term suicide mortality outcome and associated risk factors in patients with schizophrenia receiving homecare (CM) in Taiwan. We enrolled a nationwide cohort of patients with schizophrenia who newly received homecare CM intervention (n = 13 317) between January 1, 2001, and December 31, 2015; their data were derived from Taiwan’s National Health Insurance Research Database. We calculated the incidence rate of suicide methods. We examined the demographic and medical utilization profile for suicide and then performed a nested case–control study and multivariate regression to identify independent risk factors for suicide mortality. Among the 13 317 patients who received homecare CM intervention, 1766 died during the study period, of whom 213 died by suicide, which is the leading cause of unnatural death. Jumping from a high place, self-poisoning, and hanging were the top 3 suicide methods. Increased medical utilization was noted for both psychiatric and non-psychiatric services within 3 months of suicide mortality. Comorbidities of depressive disorder, nonspecific heart diseases, pneumonia, and gastrointestinal ulcers were identified as independent risk factors for suicide mortality. Suicide was the leading cause of unnatural mortality in patients with schizophrenia receiving homecare CM intervention in Taiwan. We noted the preferred suicide methods, high medical utilization, and comorbidities before suicide. Thus, we suggest that the CM team should assess lethal methods for suicide and ensure that patients adhere to psychiatry treatment for improving the current care model for this specified population.

Introduction

Schizophrenia is a chronic and severe psychiatric disorder usually resulting in dysfunction in personal, social, and occupational areas.¹ Patients with schizophrenia have a high mortality rate, and their lifespan is decreased by more than 10 years relative to that of the general population.^2,3 Nearly half of patients with schizophrenia attempt suicide at some point in their life.⁴ Although most patients with schizophrenia die of natural causes, approximately 10% of them die by suicide, which is the leading cause of unnatural death and the largest contributor to the decrease in the life expectancy of individuals with schizophrenia.^5,6 Although the estimated lifetime risk of suicide in patients with schizophrenia has declined in recent years, its standardized mortality ratio (SMR) remains high worldwide.^7–9

Some patients with schizophrenia who experience financial problems and difficulty in accessing and adhering to medical treatment may require outreach care. Moreover, these factors increase the risk of mortality in patients requiring outreach care.¹⁰ Community outreach care can reduce the barriers to care for patients with schizophrenia and enable them to receive the required treatment. The homecare case management (CM) model is a component of community outreach–based care programs in Taiwan that aim to provide long-term treatment for individuals with severe mental illnesses, such as schizophrenia.¹¹

Our previous study demonstrated that the homecare CM model in Taiwan effectively reduced the frequencies of psychiatric hospitalization and involuntary admission but had little effect on medical utilization for comorbid physical illnesses.¹¹ Another study from our research reported high all-cause mortality in this population even after they received homecare CM.¹² Although intensive CM (ICM) is suggested to reduce suicidal behavior, the pooling results of randomized controlled trials indicate that ICM exerts little or no effect on reducing suicide mortality.^13,14 To date, no randomized trials or controlled observational studies have determined suicide risk in patients with schizophrenia receiving homecare CM. Enhancing knowledge on the effectiveness of this approach can facilitate the expansion of and improve access to community outreach–based care programs, thus the professional staff could reduce barriers encountered while caring for people with severe mental illness. The first aim of this study was to investigate the incidence rate of suicide and SMR for different methods of suicide in a large nationwide cohort of patients with schizophrenia receiving homecare CM in Taiwan. We examined whether these patients had increased medical utilization or medication use before their suicide mortality because receiving medical care can provide an opportunity for suicide intervention. The second aim of the study was to identify independent risk factors for testing causal hypotheses to understand the association of physical and psychiatric comorbidities with the risk of suicide mortality. The findings of this study can provide new knowledge on suicide profiles and facilitate clinical care for suicide risk identification and prevention in this population with a high risk of suicide.

Methods

Data Sources and Suicide Mortality Identification

Taiwan’s National Health Insurance Research Database (NHIRD) contains the medical claims data of the entire population of Taiwan. The database has been determined to be accurate, and the database has been used in numerous epidemiological and clinical studies published in peer-reviewed journals.^15,16 We identified patients who were diagnosed as having schizophrenia (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 295.xx and International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] codes F20.x and F25.x) and who had newly received homecare CM intervention between January 1, 2001, and December 31, 2015 (n = 13 317). We linked the data of the retrieved schizophrenia cohort with the national mortality database (January 1, 2001, to December 31, 2016) to determine the mortality status of each patient. In total, 1766 patients died during the study period. The cause of mortality of the patients was determined by the National Cause of Death Registry based on ICD-9-CM codes before December 31, 2014, and ICD-10-CM codes after January 1, 2015. We identified suicide based on suicide-related codes E950–E959 in ICD-9-CM and the codes X60–X84 and Y87.0 in ICD-10-CM. Among the 1766 patients, 213 died by suicide. figure 1 presents the study flow diagram. This study was approved by the Research Ethics Committee of Taipei City Hospital (approval number: TCHIRB-10905022-E). The requirement of informed consent was waived because of the retrospective nature of this study and the use of deidentified data in the NHIRD.

Fig. 1.

Study flow diagram.

Intervention

The homecare CM model used in Taiwan is a hospital-based community intervention that provides psychiatric treatment and basic medical care to patients with severe mental illness.¹¹ The homecare CM team mainly includes trained psychiatrists and psychiatric nurses. Patients can be referred to other paramedical staff, such as social workers, psychologists, and occupational therapists, if required. In-person sessions are conducted 1 to 4 times per month at patients’ homes to evaluate their clinical symptoms, possible adverse reactions to medications, and social functioning status. Additionally, medications and psychotherapy are provided to enhance medication compliance, engage noncooperative clients, and provide psychoeducation and counseling to patients and their families. In addition, telephone assistance and 24-hour emergency services are available. The homecare CM model followed in Taiwan is similar to the ICM model; however, the caseload of homecare CM model is higher and fewer home visits are conducted in the homecare CM program than in the ICM.

Case and Control Definition

We conducted a nested case–control study. The date of death by suicide was defined as the index date. Through risk set sampling, we identified controls who did not die by suicide from the schizophrenia homecare CM cohort and randomly matched 4 of these controls with each case by sex and age at the index date (±5 years). Furthermore, we included only controls who had at least one medical record after the index date to ensure that they remained covered by the National Health Insurance program. Finally, this study included 213 cases and 852 controls.

Variables

We collected information on demographics, diagnoses, prescriptions, and medical expenditures between January 1, 2001, and December 31, 2016. In addition, we recorded data on age at the time of the first homecare CM intervention and the urbanization level of the hospital location. We adopted the urbanization stratification¹⁷ specifically used in Taiwan and categorized the urbanization level into levels 1 (highly urbanized area), 2 (moderately urbanized area), 3 (township area), and 4 (rural area). We examined exposure to psychiatric drugs, including antipsychotics, mood stabilizers, antidepressants, benzodiazepines, concomitant medications, including cardiovascular, respiratory, antidiabetic, antithrombotic, and anti-parkinsonism drugs and corticosteroids for systemic use. Benzodiazepines with elimination half-lives of <12, 12–24, and >24 hours were categorized as short-, intermediate-, and long-acting drugs, respectively. Antipsychotics were divided into first-generation antipsychotics (FGAs), second-generation antipsychotics, and oral or long-acting injections. Medical utilization was determined by examining the number of admissions and outpatient visits. Moreover, we stratified medical utilization into psychiatric and non-psychiatric services. Psychiatric and physical comorbidities were assessed in accordance with ICD-9-CM codes. To exclude physical or psychiatric disorders directly resulting from suicide mortality, physical or psychiatric diagnoses were excluded if they were given on the index date.

Statistical Analysis

In part I, we estimated the causes of death and the incidence of suicide in the study cohort. Differences in the incidence of mortality cause and the methods of suicide between sexes were investigated through Gehan’s generalized Wilcoxon test and a life-table survival analysis.¹⁸ We calculated the SMR as the ratio of observed deaths in the study cohort to expected deaths in the general population of Taiwan. In particular, we calculated the expected number of deaths according to 10-year age bands by multiplying the average sex-specific mortality rate of the general population of Taiwan in the corresponding year and the contributed person-years of the cohort patients during the at-risk period by each age and sex category, and all values in each age and sex category were then summed to obtain the total number of expected deaths. We performed univariate Cox proportional hazards regression to evaluate social demographic variables and their correlations with the risk of suicide mortality. In part II, on the basis of the findings of the nested case–control study, we first performed univariate conditional logistic regression to determine the patterns of medical utilization and concomitant medication and antipsychotic use within 3 months before suicide mortality between the cases and controls. Second, to investigate the effects of psychiatric and physical comorbidities on suicide risk, we performed multivariable regression after adjustment for age, sex, psychiatric and physical comorbidities determined to be significant in the univariate logistic regression model, severity of schizophrenia (evaluated on the basis of the number of psychiatry admissions within 1 year before the index date), and Charlson comorbidity index values. All statistical analyses were conducted using SAS (version 9.4; SAS Institute Inc., Cary, NC, USA). A P-value of <.05 was considered significant.

Results

Total Mortality and Demographic Characteristics of Cases

Among the 13 317 patients with schizophrenia receiving homecare CM, 1766 died during the follow-up period (total mortality incidence = 1669.3 per 100 000 person-years). The incidence of natural death was higher than that of unnatural death (1332.8 and 336.5 per 100 000 person-years, respectively). Cardiovascular disease was the leading cause of natural death. Suicide was the leading cause of unnatural death, with an incidence of 201.3 per 100 000 person-years (SMR = 9.14 and 14.09 among the men and women, respectively; supplementary table1). A total of 130 men and 83 women died by suicide. The results of a univariate Cox proportional hazards regression analysis revealed that in those receiving homecare CM, the women had a lower risk of suicide than the men. The majority of the patients who died by suicide (85.4%) were aged <55 years, and 64.3% of the suicides occurred in urbanized areas. The urbanization level and unemployment status did not affect suicide risk (supplementary table 2).

Incidence of Suicide Methods

Jumping from a high place, self-poisoning, and hanging were the top 3 methods of suicide (table 1). The incidence of suicide by charcoal burning was higher in men than in women (incidence of 26.5 and 6.1 per 100 000 person-years, respectively). No differences were noted in the incidence of other methods of suicide between men and women. Compared with the general population, the schizophrenia cohort had the highest SMR for jumping from a high place (SMR = 30.8; 95% CI = 23.6–39.4), followed by drowning (SMR = 19.4; 95% CI = 11.9–30.0) and poisoning (SMR = 17.5; 95% CI = 13.1–22.8).

Table 1.

Incidence of Methods of Suicide in Patients with Schizophrenia Receiving Homecare Case Management Intervention

	Men (N = 7138)		Women (N = 6179)		Total (N = 13 317)			Men (N = 7138)	Women (N = 6179)	Total (N = 13 317)
	(PY = 56 510.75) a		(PY = 49 281.16) a		(PY = 105 791.91) a		P
Manner of Suicide Death	No. of Deaths	Incidence b	No. of Deaths	Incidence b	No. of Deaths	Incidence b		SMR c (95% CI)	SMR c (95% CI)	SMR (95% CI)
All-cause mortality	992	1755.4	774	1570.6	1766	1669.3	0.021*	4.8 (4.5–5.1)	6.3 (5.9–6.8)	5.4 (5.1–5.6)
Unnatural death	212	375.1	144	292.2	356	336.5	0.021*	7.6 (6.6–8.7)	15.6 (13.2–18.4)	9.6 (8.6–10.6)
Total suicides	130	230.0	83	168.4	213	201.3	0.027*	9.1 (7.6–10.9)	14.1 (11.2–17.5)	10.6 (9.2–12.1)
Jumping from high	36	63.7	27	54.8	63	59.6	0.555	32.6 (22.8–45.1)	28.7 (18.9–41.7)	30.8 (23.6–39.4)
Poisoning	33	58.4	21	42.6	54	51.0	0.258	16.2 (11.1–22.7)	20.1 (12.4–30.7)	17.5 (13.1–22.8)
Hanging	26	46.0	16	32.5	42	39.7	0.274	5.8 (3.8–8.5)	10.0 (5.7–16.2)	6.9 (5.0–9.3)
Drowning	8	14.2	12	24.4	20	18.9	0.234	15.8 (6.8–31.1)	23.0 (11.9–40.2)	19.4 (11.9–30.0)
Charcoal burning	15	26.5	3	6.1	18	17.0	0.020*	2.9 (1.6–4.7)	1.9 (0.4–5.6)	2.6 (1.6–4.2)
Cutting	4	7.1	1	2.0	5	4.7	0.265	14.5 (3.9–37.2)	12.3 (0.2–68.4)	14.0 (4.5–32.7)
Firearms	2	3.5	2	4.1	4	3.8	0.891	10.2 (1.1–36.8)	50.4 (5.7–181.9)	16.9 (4.6–43.4)
Others	6	10.6	1	2.0	7	6.6	0.126	17.3 (6.3–37.6)	9.6 (0.1–53.4)	15.5 (6.2–32.0)

^aPYs: person-years.

^bIncidence: number per 100 000 person-years.

^cSMR: standardized mortality ratio.

*P < .05.

Medical Utilization and Medication Use Before Suicide

table 2 presents the pattern of medical utilization within 3 months before suicide. The suicide cases had a high-risk ratio (RR) for the number of hospital admissions (RR = 2.48, 95% CI = 1.89–3.25) including non-psychiatric (RR = 7.49, 95% CI = 4.31–13.02) and psychiatric (RR = 1.67, 95% CI = 1.21–2.30) services. In addition, the suicide cases had more total outpatient visits (RR = 1.05, 95% CI = 1.03–1.07) within 3 months before suicide, both at non-psychiatric (RR = 1.05, 95% CI = 1.02–1.07) and psychiatric (RR = 1.12, 95% CI = 1.06–1.19) clinics. In terms of specialties, a higher proportion of the cases had visited surgery department (RR = 1.72, 95% CI = 1.05–2.83), emergency department (RR = 5.17, 95% CI = 3.35–7.98), and psychiatric department (RR = 1.90, 95% CI = 1.21–2.99) than did the controls.

Table 2.

Medical Utilization Within 3 Months Before Suicide (Not Including the Visit of the Index Date)

	Cases (N = 213)	Controls (N = 852)
	N (%)	N (%)	Crude Risk Ratio a	95% CI	P-value
ALL	204 (95.8)	758 (89.0)	2.83	1.40–5.71	.004**
Family medicine	52 (24.4)	182 (21.4)	1.20	0.84–1.72	.327
Internal medicine	60 (28.2)	191 (22.4)	1.36	0.97–1.92	.076
Surgery	24 (11.3)	58 (6.8)	1.72	1.05–2.83	.032*
Pediatrics	7 (3.3)	21 (2.5)	1.36	0.56–3.29	.498
Obstetrics and gynecology	15 (7.0)	38 (4.5)	1.66	0.88–3.14	.117
Psychiatry	188 (88.3)	681 (80.0)	1.90	1.21–2.99	.005**
Emergency department	52 (24.4)	49 (5.8)	5.17	3.35–7.98	<.001***
Other specialties	101 (47.4)	317 (37.2)	1.52	1.12–2.06	.007**
Within 3 months before the index date	Mean (SD)	Mean (SD)
Number of hospital admissions	0.44 (0.70)	0.13 (0.45)	2.48	1.89–3.25	<.001***
Non-psychiatric	0.23 (0.51)	0.02 (0.18)	7.49	4.31–13.02	<.001***
Psychiatric	0.21 (0.48)	0.11 (0.38)	1.67	1.21–2.30	.001**
Number of outpatient visits	9.12 (8.91)	6.24 (6.51)	1.05	1.03–1.07	<.001***
Non-psychiatric	5.02 (7.70)	2.98 (5.50)	1.05	1.02–1.07	<.001***
Psychiatric	4.10 (2.97)	3.25 (2.59)	1.12	1.06–1.19	<.001***

^aUnivariable conditional logistic regression.

* P < .05; ** P < .01; *** P < .001.

Medications used within 3 months before suicide mortality suggested a complicated medication profile (supplementary table 3). The use of beta-blockers, systemic use of corticosteroids, anti-parkinsonism drugs, antipsychotics, antidepressants, and benzodiazepines was higher in the patients who died by suicide than in the control group. Among the antipsychotics used before suicide, the use of oral and LAI FGAs was higher in patients who died by suicide. In the unadjusted model, differences in the distribution of medications used before suicide only indicated the factors related to 3-month suicide mortality and thus could not indicate the causal or protective effect of medications on suicide.

Psychiatric and Physical Comorbidities Before Suicide

table 3 presents the distribution of psychiatric and physical comorbidities within 3 months before suicide. In the nonadjusted model, the patients with schizophrenia who received homecare CM intervention and had psychiatry comorbidities, such as substance use disorder, depressive disorder, and anxiety, exhibited a higher risk of 3- month suicide mortality. In terms of physical comorbidities, unspecific heart disease, pneumonia, and peptic ulcer disease were associated with a higher risk of suicide. In the multivariable conditional logistic regression model adjusted for potential confounding factors, the patients with comorbid depressive disorder (adjusted RR = 3.19, 95% CI = 1.89–5.38), unspecific heart disease (adjusted RR = 6.35, 95% CI = 3.44–11.74), pneumonia (adjusted RR = 7.19, 95% CI = 2.34–22.12), and ulcer disease (adjusted RR = 3.48, 95% CI = 1.65–7.34) had an independent higher risk of suicide mortality.

Table 3.

Psychiatric and Physical Comorbidities Within 3 Months Before Suicide

	Case (N = 213)	Controls (N = 852)	Unadjusted Risk Ratio	95% CI	P-value	Adjusted Risk Ratio a	95% CI	P-value
Psychiatric comorbidity, N (%)
Substance use disorders	18 (8.5)	25 (2.9)	2.99	1.61–5.55	<.001
Depressive disorder	34 (16.0)	44 (5.2)	3.51	2.16–5.68	<.001	3.19	1.89–5.38	<.001*
Anxiety states	16 (7.5)	27 (3.2)	2.47	1.31–4.68	0.005
Intellectual Disabilities	3 (1.4)	16 (1.9)	0.75	0.22–2.57	0.648
Sleep disorder	31 (14.6)	87 (10.2)	1.48	0.96–2.29	0.077
Physical illnesses, N (%)
Cardiovascular diseases
Hypertension	19 (8.9)	75 (8.8)	1.02	0.59–1.74	0.956
Ischemic heart disease	3 (1.4)	11 (1.3)	1.09	0.30–3.91	0.894
Nonspecific heart disease	34 (16.0)	20 (2.4)	7.33	4.13–12.99	<.001	6.35	3.44–11.74	<.001*
Congestive heart failure	1 (0.5)	6 (0.7)	0.67	0.08–5.54	0.708
Cerebrovascular Diseases	3 (1.4)	8 (0.9)	1.50	0.40–5.66	0.549
Respiratory diseases
Pneumonia	15 (7.0)	8 (0.9)	10.69	3.85–29.63	<.001	7.19	2.34–22.12	<.001*
Chronic Obstructive Pulmonary Disease	6 (2.8)	15 (1.8)	1.62	0.62–4.25	0.324
Asthma	1 (0.5)	13 (1.5)	0.30	0.04–2.32	0.248
Upper respiratory tract infection	11 (5.2)	42 (4.9)	1.05	0.53–2.07	0.889
Gastrointestinal diseases
Chronic hepatic disease	3 (1.4)	30 (3.5)	0.38	0.12–1.28	0.119
Ulcer disease	18 (8.5)	20 (2.4)	4.09	2.07–8.09	<.001	3.48	1.65–7.34	<.001*
Endocrine diseases
Diabetes mellitus	13 (6.1)	78 (9.2)	0.64	0.34–1.18	0.150
Cancers	0 (0.0)	5 (0.6)	0.00	0-	0.985

^aFactors adjusted in multivariable regression model: Age, sex, significant psychiatric and physical comorbidities from univariate logistic regression, severity of schizophrenia (proxy as the psychiatry admission number within one year before index date), and Charlson comorbidity score. * P < .001.

Discussion

To the best of our knowledge, this is the first study to investigate the incidence of suicide mortality and methods of suicide in patients with schizophrenia receiving homecare CM intervention as well as medical utilization and comorbidities before suicide mortality in this population. Suicide was identified as the leading cause of unnatural death in the patients with schizophrenia receiving homecare CM in Taiwan. Jumping from a high place, self-poisoning, and hanging were the top 3 methods of suicide. We noted increased medical utilization and medication use 3 months before suicide in these patients. In addition, depressive disorder, nonspecific heart diseases, pneumonia, and ulcers in the gastrointestinal system were independent risk factors for suicide mortality. Our results indicate an unmet need for identifying high-risk patients and maintaining their treatment to prevent suicide in the current community model.

Incidence and Suicide Methods

Suicide risk might differ between patients with schizophrenia receiving homecare CM and those with schizophrenia in general. Most studies have suggested that the non-completion or interruption of psychiatric treatment mainly contributes to the risk of suicide in schizophrenia patients.^19,20 A nationwide cohort study⁵ in Taiwan reported that the incidence rate of suicide was 221.1 per 100 000 person-years for patients with schizophrenia, which is equivalent to the finding of this study (201.3 per 100 000 person-years). However, patients with schizophrenia who required outreach intervention usually had some unfavorable prognostic factors, such as poor insight, low education level, and low family support.²¹ These factors were also correlated with nonadherence to treatment. By contrast, adherence to effective treatment was a protective factor for suicide in patients with schizophrenia.²² In the current study, we could not determine whether homecare CM reduced suicide mortality. Additional community trial designs are warranted to answer these questions.

We highlighted the methods of suicide in patients receiving homecare CM in Taiwan, and some differences were noted between patients with schizophrenia and the general population. In Taiwan, charcoal burning was the second leading method of suicide in the general population; however, charcoal burning was not 1 of the top 3 suicide methods (ie, jumping from height, poisoning, and hanging) in the schizophrenia cohort receiving homecare CM in this study. In addition, the type of the suicide method adopted differed between sexes in patients with schizophrenia. However, sex differences were not significant in those receiving homecare CM, with the exception of the incidence of charcoal burning for suicide being higher in men.⁵ General patients with schizophrenia and those receiving homecare CM preferred jumping from a high place as a method of suicide; this finding is consistent with those of previous studies.^23,24 However, the general population in Taiwan is less likely to prefer jumping from a high place,⁵ maybe because this suicide method may affect the surrounding area and cause harm to others. These findings indicate that the patients receiving homecare CM intervention opted for relatively impulsive and accessible suicide methods instead of methods that require prior preparation, such as charcoal burning. In addition, the cognitive function deficits in this specific group may have interfered with their planning for suicide methods.²⁵

Medical Utilization and Comorbidities

Examinations of medical utilization and medication use can provide information on the health status of patients with schizophrenia receiving homecare CM before their suicide. For example, the use of antipsychotics and antidepressants was high in the patients who died by suicide. These findings indicate that the risk of suicide in patients with schizophrenia might be associated with their depressive or psychotic symptoms.²⁶ The utilization of psychiatric services was significantly increased in the patients who died by suicide. In addition, we also noted a high number of visits to non-psychiatric departments. We noted substantially increased visits to surgery and emergency departments before suicide; this result is consistent with that of a previous study indicating that emergency department visits or trauma management due to attempted suicide was strongly associated with subsequent suicide risk.²⁷ These findings highlight potential active suicide identification in patients receiving homecare CM intervention with a risk of suicide in emergency departments. Moreover, the increased psychiatry and nonpsychiatric medical utilization and medication use in this high-risk population before suicide may imply that patients with schizophrenia who died by suicide had severe mental and physical conditions noted by healthcare providers. However, the current study revealed that even the initiation of medication use or clinical service may not prevent suicide in the current homecare CM model. A large Nordic cohort study highlighted the importance of patient engagement and indicated that psychiatric treatment is crucial to prevent deaths by suicide after previous suicide attempts.²⁸ According to our results, higher medication utilization for both psychiatry and non-psychiatric services before suicide did not prevent all suicides. More intensive and active aftercare programs and social support are still required in the current homecare CM model to ensure timely psychiatric treatment for disadvantaged patients with schizophrenia.

The risk of suicide was high in patients with comorbid depressive disorders, nonspecific heart diseases, pneumonia, and gastrointestinal ulcers. Depression is a major risk factor for suicide in patients with schizophrenia.²⁹ In addition, comorbid depression is not uncommon in patients with schizophrenia and negatively affects patients’ quality of life.³⁰ The present study demonstrated that the patients receiving homecare CM intervention who died by suicide had increased risks of gastrointestinal ulcers, nonspecific heart diseases, and pneumonia within 3 months before suicide. A strong association of gastrointestinal ulcers with stress and anxiety can partially explain the high suicide risk in patients receiving homecare CM intervention.^31–33 Unspecific heart diseases usually include medically unexplainable chest discomfort, which can be a nonspecific functional somatic symptom or an ongoing heart disease without a definite diagnosis. Heart disease is associated with high levels of underlying psychosocial problems, such as depression, anxiety, stressful life events, and suicide.^34,35 Therefore, the association between suicide and stress-related physical disorders, such as heart disease and ulcers in the gastrointestinal system, should be considered when evaluating suicide risk in patients with schizophrenia receiving homecare CM. Few studies have indicated an association between pneumonia and suicide.³⁶ Some studies have suggested that substance misuses, such as smoking and alcohol consumption, is a risk factor for both suicide and increased admissions for pneumonia.^37,38 Increased suicide rates have been reported following admission with various infections. The association between infections and suicide can result from the link between inflammatory cytokines and the pathophysiological mechanisms of suicidal behaviors.³⁹

Studies should examine the interaction between increased use of psychoactive agents and physical comorbidities at the same time within 3 months in people who have died by suicide. For example, the increased use of antipsychotics (eg, FGAs) and antidepressants was associated with various side effects, such as akathisia, palpitation, rigidity and chocking pneumonia, and gastrointestinal upset. Such physical discomfort can lead to nonspecific heart disease, ulcers, and an increased risk of pneumonia and can thus lead to frequent visits to medical centers. We could not draw a definitive conclusion regarding the causality in the present study; however, both the physical and psychological side effects of medications can contribute to suicide risk.

Strengths and Limitations

This study has several strengths. This study investigated differences in medical utilization patterns and psychiatric and physical comorbidities between patients with schizophrenia receiving homecare CM who died by suicide and living controls. The inclusion of a large population-based cohort of patients receiving homecare CM intervention ensured the enrollment of an adequate number of cases and matched controls. In addition, complete details on medical utilization, used medications, and physical and psychiatric disease diagnoses are available in the NHIRD; this results in lower record bias compared with that encountered in self-report or psychological autopsy designs. Nevertheless, this study has several limitations that should be considered. First, we used claims data from a health insurance database, which provides information on only demographic variables and lacks details of some crucial social factors related to suicidal behaviors, such as education level, marital status, family support, socioeconomic conditions, and welfare status. Second, we used a national mortality database to track each suicide event; those who died from suicide but whose death was misclassified as mortality from other causes might have been neglected. For example, the most prominent misclassification could be accidental death, which was used to prevent the potential stigma associated with suicide. Third, this study did not compare its findings with those of patients who did not receive homecare CM in the current study. Finally, we did not know the potential warning signs that have been detected or were not in the present healthcare system. Although patients had higher medication utilization before suicide, this did not mean that integrative care was given to the patients. Future studies are warranted to evaluate the elements of a homecare CM intervention that can aid in preventing suicide in patients with schizophrenia.

Conclusions

This study examined the risk of suicide mortality in patients with schizophrenia receiving homecare CM intervention in Taiwan. Comorbidities of depressive disorder, nonspecific heart diseases, pneumonia, and gastrointestinal ulcers within 3 months before suicide were associated with an increased risk of suicide mortality. Increased medical resource utilization before suicide can provide an opportunity for early suicide identification and prevention. In addition, we should be aware of the possible side effects of antipsychotics and consider whether they increase the risk of suicide. Jumping from a high place, self-poisoning, and hanging were the top 3 suicide methods. CM teams should assess lethal methods during suicide screening. Moreover, the active care model in the current homecare CM intervention should be strengthened to reduce the suicide risk in this population.

Funding

This research was supported by grants from the Ministry of Science and Technology, Taiwan (MOST 108-2314-B-532-005 and 110-2314-B-532-003-MY3) and Taipei City Hospital (10901-62-009 and 10901-62-055). The funding sources had no involvement in the study design, data collection, analysis, interpretation of data, writing of the report, or the decision to submit the paper for publication.

Author Contributions

Drs W-Y Chen and Kuo conceived and designed the study. Dr Kuo acquired data. Mr. Su performed the statistical analysis. Drs Yang and CC Chen provided administrative and material support. Drs W-Y Chen and Kuo drafted the manuscript. Dr Pan made critical revisions to the manuscript for critical intellectual content, and Drs CC Chen and Kuo supervised the study. All authors approved the final version of the submitted manuscript.