- Email: [email protected]
Unnatural Causes of Death and Suicide Among Former Adolescent Psychiatric Patients
Journal of Adolescent Health 52 (2013) 207–211
www.jahonline.org Original article
Unnatural Causes of Death and Suicide Among Former Adolescent Psychiatric Patients Subin Park, M.D., Ph.D.a, Chang Yoon Kim, M.D., Ph.D.b, and Jin Pyo Hong, M.D., Ph.D.b,* a b
Division of Child and Adolescent Psychiatry, Department of Psychiatry, Seoul National University College of Medicine, Seoul, Republic of Korea Department of Psychiatry, Asan Medical Center, Ulsan University College of Medicine, Seoul, South Korea
Article history: Received December 20, 2011; Accepted May 24, 2012 Keywords: Adolescence; Suicide; Mortality; Mental disorders; Korea
A B S T R A C T
Purpose: Compared with the general population, adolescent psychiatric patients are subject to premature death from all causes, but suicide-specific mortality rates in this population have not been carefully investigated. Therefore, we examined the high mortality due to unnatural causes, particularly suicide, using standardized mortality ratios (SMRs) relative to sex, diagnosis, and type of psychiatric service. Methods: A total of 3,029 patients aged 10 –19 years presented to the outpatient clinic of a general hospital in Seoul, Korea, or were admitted to that hospital for psychiatric disorders from January 1995 to December 2006. Unnatural causes mortality risk and suicide mortality risk in these patients were compared with those in sex- and age-matched subjects from the general Korean population. Results: The SMR for unnatural causes was 4.6, and for suicide it was 7.8. Female subjects, the young, and inpatients had the highest risks for unnatural causes of death or suicide. Among the different diagnostic groups, patients with psychotic disorders, affective disorders, and personality disorders had significantly increased SMRs for unnatural causes, and those with psychotic disorders, affective disorders, and disruptive behavioral disorders had significantly increased SMRs for suicide. Conclusions: The risks of unnatural death and suicide are high in adolescent psychiatric inpatients in Korea, but not as high in adolescent outpatients. Effective preventative measures are required to reduce suicide mortality in adolescent psychiatric patients, particularly female patients admitted for general psychiatric care. 䉷 2013 Society for Adolescent Health and Medicine. All rights reserved.
Korea has the highest suicide rate of all the countries in the Organization for Economic Cooperation and Development (OECD). The mean age-standardized rate of suicide in Korea is 11.3 per 100,000 [1], and suicide is the most frequent cause of death in adolescents and young adults [2]. In addition, previous studies of suicide in adult psychiatric patients have shown that the younger the patients, the higher the standardized mortality
* Address correspondence to: Jin Pyo Hong, M.D., Ph.D., Department of Psychiatry, Asan Medical Center, 388-1 Pungnap-2dong, Songpa-gu, Seoul 138 –736, South Korea. E-mail address: [email protected] (J.P. Hong).
IMPLICATIONS AND CONTRIBUTION
The present study indicates that the risks of unnatural death and suicide are particularly high in female adolescent patients who have been admitted for general psychiatric care. Effective preventative measures are required to reduce suicide mortality among this highrisk group.
ratios (SMRs) [3–7]. Therefore, we were particularly interested in suicide-specific mortality risk in adolescent psychiatric patients. Most studies on mortality in child and adolescent psychiatric patients report all causes SMRs only [8 –12], and cause-specific mortality rates are not well-investigated in this population owing to small sample sizes or difficulty in obtaining causes of death. We are only aware of one study of adolescent psychiatric patients that differentiates unnatural causes mortality from natural causes mortality [10]. In that study, which included only psychiatric inpatients and was conducted in Oslo, Norway, SMR for unnatural causes (including drug overdose, suicide, homicide, and accident) was significantly elevated in both sexes. The SMRs were 8.5 (95% confidence intervals [CI]: 6.7–10.6) for male patients and 15.8 (95% CI: 10.8 –22.4) for female patients; suicide-
1054-139X/$ - see front matter 䉷 2013 Society for Adolescent Health and Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jadohealth.2012.05.018
208
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
specific SMRs were 6.8 (95% CI: 3.4 –12.1) for male patients and 19.0 (95% CI: 10.6 –31.3) for female patients; and drug overdose SMRs were 54.1 (95% CI: 32.9 – 84.3) for male patients and 83.3 (31.0 –178.0) for female patients. To establish effective strategies to prevent the premature deaths of adolescent psychiatric patients, detailed information on the cause-specific mortality in various diagnostic groups and different treatment settings is required. Therefore, we have investigated unnatural causes mortality and suicide mortality further by calculating SMRs relative to diagnosis, age, and type of hospital service (inpatient/outpatient), as well as sex-specific SMRs. Methods Subjects The subjects consisted of all patients who presented to a psychiatric outpatient clinic of a general hospital located in Seoul, Korea, or who were admitted to that hospital for a psychiatric disorder, during a 12-year period (from January 1995 to December 2006), and who were aged 10 –19 years at the time of the hospital visit. The data were censored either on the date of death or on December 31, 2009, because this is the date when the latest available data on the National Statistical Office (NSO) mortality were obtained. Person-years (the number of years that each individual was under observation) were calculated for each patient. Information about whether the patients were alive on December 31, 2009, was provided by linkage to the database of the NSO. For those who were dead, causes of death were also established through linkage to the database of the NSO. All deaths in Korea are reported to the NSO by a document of death notice, which contains the cause of death. However, death by suicide is sometimes misreported as death by other cause, such as car accident, by a bereaved family because of a social stigma of suicide. On the other hand, all deaths caused by suicide, homicide, or accidents are investigated by the National Police Agency (NPA). The police interview the individuals who knew the dead person well, such as family members, partner, friends, and/or neighbors, and make an investigation report on suicide, homicide, or accidents. When comparing the NSO and NPA data using national identification numbers, 96.6% of suicide cases in the NPA data were also counted as suicide cases in the NSO data [13]. For the purposes of the present study, we identified subjects who had died because of unnatural causes (ICD-10 External causes of morbidity and mortality V01-Y98). The NSO data and hospital records were matched using the unique national identification number assigned to all Korean citizens. Statistical analysis The unnatural causes mortality and suicide-specific mortality of the study subjects were compared with those of the corresponding general population of Korea. SMRs and 95% CIs were calculated using a person-years method with a free software program Person-years and Mortality COMputation Program 1.41 (Dirk Taeger, Dortmund, Germany) [14]. The SMR compared the observed numbers of deaths with the expected numbers of deaths. The person-years method calculated the total number of years that a person was at risk for each calendar year. The expected numbers of unnatural deaths and suicide-specific deaths
were calculated from the age-, sex-, and cause-specific mortality rates of the general population in Korea, for each calendar year. These specific mortality rates were obtained from the NSO. First, the sex-specific unnatural causes and suicide-specific SMRs were calculated. Next, the patients were divided into those who had received general psychiatric care (inpatients) and those who had not (outpatients). Unnatural causes and suicide-specific SMRs were then calculated for each type of care. Three psychiatric residents reviewed and confirmed the ICD-10 diagnoses that had been recorded in the electronic medical records by a boardcertified psychiatrist. The medical record requirements of the participating hospital included one primary diagnosis and several auxiliary diagnoses based on the ICD-10 coding. Diagnosis of those with multiple psychiatric disorders was made according to the primary ICD-10 diagnosis provided in the electronic medical records. Although a primary psychiatric diagnosis existed in most medical records, 6.3% of those records included an unclear diagnosis (i.e., mental disorder, not otherwise specified, or no ICD-10 F code diagnosis). For these cases, a board-certified psychiatrist made the diagnosis based on the clinical information provided in the medical records. For the purposes of the present study, these diagnoses were collapsed into eight diagnostic groups: psychotic disorder (F20 –F29), affective disorders (F30 – F33), neurotic disorders (F40 –F48), disruptive behavioral disorders (F90 –F92), mental retardation (F70 –F79), personality disorders (F60 –F69), eating disorders (F50), and other psychiatric disorders (diagnoses not included in the aforementioned diagnoses). Although previous studies indicated that substance abuse conferred a risk for death by suicide in adolescents [10,15], substance use disorders were not separated from the other diagnoses because of insufficient statistical power (only 13 patients [4.3%] had a primary diagnosis of substance use disorders). Unnatural causes and suicide-specific SMRs were then calculated for these diagnoses. The observation period for each patient was divided into three age spans: from first presentation to hospital to 19 years of age; from 20 to 24 years of age; and 25 years and older. Unnatural causes and suicide-specific SMRs were then calculated for these ages. To identify independent predictors for death by unnatural causes and suicide in adolescent psychiatric patients, sex, type of hospital service, diagnosis, and age were concurrently entered into the Cox proportional hazards model. Table 1 Characteristics of the 3,029 adolescent psychiatric patients Characteristics
N (%)
Sex, male Age at first hospital visit, mean ⫾ SD (years) Type of care Inpatient treatment Outpatient treatment only Psychiatric diagnosis Psychotic disorders Affective disorders Neurotic disorders Disruptive behavioral disorders Mental retardation Personality disorders Eating disorders Other diagnosesa
1,847 (61.0) 15.3 ⫾ 2.8
a
709 (23.4) 2,320 (76.6) 355 (11.7) 615 (20.2) 781 (25.8) 558 (18.4) 101 (3.3) 109 (3.6) 62 (2.0) 448 (14.8)
“Other diagnoses” includes organic mental disorders, substance use disorders, developmental disorders, tic disorders, and other childhood behavioral and emotional disorders.
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
Table 2 Unnatural causes and suicide-specific SMRs in male and female adolescent psychiatric patients Sex (person-years)
All (25,468) Male (15,390) Female (10,078)
Table 4 Unnatural causes and suicide-specific SMRs in relation to psychiatric diagnosis Diagnosis (personyears)
Death by unnatural causes Suicide O
E
SMR
95% CI
O
E
SMR
95% CI
36 19 17
7.7 6.1 2.1
4.6 3.1 8.3
3.3–6.4 1.9–4.9 4.8–13.2
25 11 14
3.2 2.1 1.2
7.8 5.4 11.5
5.0–11.4 2.7–9.7 6.3–19.3
Death by unnatural causes
Suicide
O
O
E
SMR 95% CI
Psychotic disorder 14 1.3 11.0 (3,538) Affective disorder 8 1.4 5.7 (4,608) Neurotic disorder 5 2.2 2.3 (6,917) Disruptive behavioral 3 1.0 2.9 disorder (4,198) Mental retardation 2 .3 6.7 (1,020) Personality disorder 3 .4 7.3 (1,133) Eating disorder (520) 1 .2 6.0
CI ⫽ confidence interval; E ⫽ expected death; O ⫽ observed death; SMR ⫽ standardized mortality ratio.
All statistical analyses were performed with SPSS (version 12.0; SPSS Inc., Chicago, IL), and an ␣-level of ⬍.05 was considered significant. Results During the 12-year study period, 3,029 adolescent psychiatric patients (709 inpatients and 2,320 outpatients) were enrolled and 36 of those patients died of unnatural causes (25 by suicide, 7 by accidents, and 4 undetermined). The characteristics of the study subjects are shown in Table 1. Unnatural causes SMRs were 4.6 (95% CI: 3.3– 6.4) in all patients, 3.1 (95% CI: 1.9 – 4.9) in male patients, and 8.3 (95% CI: 4.8 –13.2) in female patients. Suicide-specific SMRs were particularly high ⫺7.8 (95% CI: 5.0 –11.4) in all patients, 5.4 (95% CI: 2.7–9.7) in male patients, and 11.5 (95% CI: 6.3–19.3) in female patients (Table 2). In patients who had been admitted for general psychiatric care, the mortality risk from unnatural causes was 7.8 times expectation and from suicide it was 14.2 times expectation. In patients who received outpatient care only, the mortality risk from unnatural causes was 3.2 times expectation and from suicide it was 4.9 times expectation (Table 3). Among the different diagnostic groups, patients with psychotic disorders (11.0, 95% CI: 6.0 –18.4), affective disorders (5.7, 95% CI: 2.4 –11.2), and personality disorders (7.3, 95% CI: 1.5– 21.5) had significantly increased SMRs for unnatural causes. Patients with psychotic disorders (20.0, 95% CI: 10.3–37.1), affective disorders (8.2, 95% CI: 2.7–19.2), and disruptive behavioral disorders (7.3, 95% CI: 1.5–21.4) had significantly increased SMRs for suicide (Table 4). The mortality risk was highest in the 11–20 age-group, with an unnatural causes mortality risk 34 times expectation and a suicide mortality risk 71 times expectation (Table 5). Multivariate analysis using the Cox proportional hazards model indicated that younger age (hazard ratio [HR] ⫽ .68, per 1-year increase, 95% CI: .62–.74, p ⬍ .001), previous psychiatric admission (HR ⫽ 3.66, 95% CI: 1.75–7.67, p ⫽ .001), and diagnosis
209
6.0–18.4
E
SMR 95% CI
11 .5 20.7
10.3–37.1
2.4–11.2
5 .6
8.2
2.7–19.2
.7–5.3
3 .9
3.3
.7–9.5
.6–8.4
3 .4
7.3
1.5–21.4
.7–24.1
1 .1
8.0
.1–44.5
1.5–21.5
1 .2
5.9
.8–3.3
.8–33.1
1 .1 14.3
.2–79.5
of psychotic disorder (HR ⫽ 7.23, 95% CI: 2.42–21.58, p ⬍ .001) and personality disorder (HR ⫽ 5.54, 95% CI: 1.27–24.07, p ⫽ .022) were significantly and independently associated with increased risk of death from unnatural causes in the adolescent psychiatric patients. In addition, younger age (HR ⫽ .65, per 1-year increase, 95% CI: .59 –.73, p ⬍ .001), previous psychiatric admission (HR ⫽ 3.98, 95% CI: 1.62–9.76, p ⫽ .003), and diagnosis of psychotic disorder (HR ⫽ 10.12, 95% CI: 2.64 –38.86, p ⫽ .001) were significantly and independently associated with increased risk of death by suicide. Discussion In the adolescent psychiatric patients, the SMRs for unnatural causes and suicide were 4.6 and 7.8, respectively. Female patients, the young, and inpatients had the highest risks for unnatural causes of death or suicide. Among the different diagnostic groups, patients with psychotic disorders, affective disorders, and personality disorders had significantly increased SMRs for unnatural causes, and those with psychotic disorders, affective disorders, and disruptive behavioral disorders had significantly increased SMRs for suicide. Our finding of a high unnatural causes mortality, and of a particularly high suicide mortality, in adolescent psychiatric patients is in accord with the results of a previous study conducted in Oslo, Norway [10]. Direct comparison of our results with those of the previous study is difficult because of different follow-up periods (5–14 years vs. 15–33 years after index hospital visit), compositions of subjects (both inpatients and outpatients vs. inpatients only), and classifications of causes of death (the cited author differentiated drug overdose from suicide, but we did
Table 3 Unnatural causes and suicide-specific SMRs in relation to type of care received and sex Type of care (person-years)
Inpatient (7,020) Male (3,722) Female (3,298) Outpatient (18,448) Male (11,667) Female (6,780)
Death by unnatural causes
Suicide
O
E
SMR
95% CI
O
E
SMR
95% CI
19 11 8 17 8 9
2.4 1.7 .7 5.3 4.4 1.3
7.8 6.3 11.0 3.2 1.8 6.8
4.7–12.3 3.1–11.3 4.8–21.8 1.9–5.1 .8–3.6 3.1–12.8
14 7 7 11 4 7
1.0 .6 .4 2.2 1.5 .8
14.2 12.2 16.5 4.9 2.7 8.8
7.7–23.7 4.9–25.2 6.6–33.9 2.5–8.8 .7–7.0 3.5–18.2
210
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
Table 5 Unnatural causes and suicide-specific SMRs in relation to age-group Age (person-years)
10–19 (2,829) 20–24 (5,826) 25–34 (16,813)
Death by unnatural causes
Suicide
O
E
SMR
95% CI
O
E
SMR
95% CI
12 19 5
.4 1.4 6.0
34.0 13.6 .8
17.6–59.5 8.2–21.2 .3–1.9
9 12 4
.13 .60 2.5
71.0 20.0 1.6
32.4–134.9 10.3–34.9 .4–4.1
not). With that in mind, we compared the unnatural causes SMR of our inpatients with that in the previous study, instead of comparing the suicide-specific SMRs for the total subjects in the two studies. The unnatural causes SMRs of our inpatients were lower than those of the discharged adolescent patients in the study by Kjelsberg (6.3 vs. 9.3 for boys and 12.2 vs. 15.8 for girls). This indicates that the difference in unnatural causes mortality between adolescent psychiatric inpatients and communitydwelling adolescents is less prominent in Seoul than in Oslo, possibly owing to the high suicide mortality in the general Korean population. One possible explanation for this is that in Korea, many high-risk adolescents die by suicide rather than seeking hospital care. Consistent with findings from previous studies [16 –18], suicide-specific SMRs were substantially higher for girls than for boys, particularly among inpatients. This finding suggests that girls who have been admitted to hospital for psychiatric problems are more at risk than boys, whereas more boys than girls die by suicide in the general population. Furthermore, the current findings suggest that suicide attempts among girls with psychiatric disorders merit clinical attention to safeguard against suicide risk. Thus, although men routinely have higher rates of death by suicide in the general population [19], among Korean youth admitted for general psychiatric care, female patients may be at a heightened suicide risk. The unnatural causes SMR and suicide-specific SMR of the outpatients were much lower than those of the inpatients (3.2 vs. 7.8 and 4.9 vs. 14.2, respectively). In fact, the unnatural causes SMR and suicide-specific SMR of male outpatients did not differ significantly from those of male adolescents in the general population (p ⬎ .05). Our multivariate Cox regression analysis indicated that a history of general psychiatric care was independently associated with death due to unnatural causes and death by suicide. This result is consistent with the previous finding that inpatients are 3.5 times more likely than outpatients to die by suicide [8]. All the individual psychiatric diagnoses were associated with increased unnatural death and suicide-specific SMRs, although some of the associations were not statistically significant (p ⬎ .05) because of low numbers. Consistent with findings from previous work examining adult psychiatric patients [3–7,20,21], we found that psychotic disorder, affective disorder, and personality disorder had a substantial impact on the risks of unnatural death and suicide in adolescent psychiatric patients. The increased suicide mortality in adolescents with disruptive behavioral disorders is consistent with the result of a follow-up study of adolescents with attention-deficit hyperactivity disorder, which reported a suicide rate of .63%–.78% and an estimated relative risk ratio of 2.91 (95% CI: 1.47–5.70) [9]. The suicide-specific SMR decreased from 71 in 10 –19 year olds to 20 in 20 –24 year olds, and to 1.6 in 25–34 year olds (Table 5). This decline may be because of the low suicide mortality rate
of adolescents in the general population [2]. Another explanation is that the risk of death from suicide is highest in the early stages of a mental illness [22] and during the first year after discharge [3,5]. This study has three major strengths. First, most other studies calculated all causes SMRs only, whereas we calculated unnatural causes SMRs and suicide-specific SMRs. Second, we performed suicide-specific mortality analyses in relation to age, type of care, and psychiatric diagnosis, as well as sex. Third, most other studies were concerned only with adolescent psychiatric inpatients, whereas we investigated both inpatients and outpatients and compared their suicidal mortality risks indirectly via SMRs. The present study had several limitations. First, the relatively low number of deaths by suicide limited the statistical power to detect significant differences in suicide mortality between our sample and the general population. Although suicide was more common in our psychiatric population than in the general population, large sample sizes are needed to study rare outcomes. Second, NSO data may underestimate the number of deaths due to suicide because some suicides may have been classified as “undetermined” deaths [23,24]. Among our patients, there were four deaths that were classified as “undetermined.” Third, although three psychiatric residents reviewed and confirmed the diagnoses that had been recorded in the electronic medical records, the reliability of such diagnoses was not systematically assessed. Fourth, the effect of comorbid psychiatric conditions on mortality could not be examined in this study because only a primary diagnosis was considered. Finally, as this study was conducted in a single general hospital, our findings may not be representative of all Korean psychiatric inpatients. Thus, further research should explore the nature of adolescent suicide risk while accounting for psychiatric comorbidity in many hospitals nationwide. Despite these limitations, the present study indicates that in Korea, the risks of unnatural death and suicide are high in adolescent psychiatric patients, particularly female patients who have been admitted for general psychiatric care. These high-risk adolescents should be asked directly to indicate whether they have suicidal ideation, and they may need to participate in specific suicide prevention programs in addition to receiving conventional psychiatric treatment. References [1] OECD. OECD health data 2011: Statistics and Indicators for 34 countries. Paris: Organization for Economic Cooperation and Development, 2011. [2] Korea National Statistical Office. National statistics report 2010. Seoul: Korea National Statistical Office, 2010. [3] Goldacre M, Seagroatt V, Hawton K. Suicide after discharge from psychiatric inpatient care. Lancet 1993;342:283– 6. [4] Qin P, Nordentoft M. Suicide risk in relation to psychiatric hospitalization: Evidence based on longitudinal registers. Arch Gen Psychiatry 2005;62: 427–32.
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
[5] Geddes JR, Juszczak E, O’Brien F, Kendrick S. Suicide in the 12 months after discharge from psychiatric inpatient care, Scotland 1968 –92. J Epidemiol Community Health 1997;51:430 – 4. [6] Ho TP. The suicide risk of discharged psychiatric patients. J Clin Psychiatry 2003;64:702–7. [7] Yim PH, Yip PS, Li RH, et al. Suicide after discharge from psychiatric inpatient care: A case-control study in Hong Kong. Aust N Z J Psychiatry 2004; 38:65–72. [8] Engqvist U, Rydelius PA. Death and suicide among former child and adolescent psychiatric patients. BMC Psychiatry 2006;6:51. [9] James A, Clacey J, Seagroatt V, Goldacre M. Adolescent inpatient psychiatric admission rates and subsequent 1-year mortality in England: 1998 –2004. J Child Psychol Psychiatry 2010;51:1395– 404. [10] Kjelsberg E. Adolescent psychiatric in-patients. A high-risk group for premature death. Br J Psychiatry 2000;176:121–5. [11] Rydelius PA. Deaths among child and adolescent psychiatric patients. Acta Psychiatr Scand 1984;70:119 –26. [12] Thomsen PH. A 22- to 25-year follow-up study of former child psychiatric patients: A register-based investigation of the course of psychiatric disorder and mortality in 546 Danish child psychiatric patients. Acta Psychiatr Scand 1996;94:397– 403. [13] Im JS, Choi SH, Hong D, et al. Proximal risk factors and suicide methods among suicide completers from national suicide mortality data 2004 – 06 in Korea. Compr Psychiatry 2011;52:231–7. [14] Taeger D, Sun Y, Keil U, Straif K. A stand-alone windows applications for computing exact person-years, standardized mortality ratios and confidence intervals in epidemiological studies. Epidemiology 2000;11:607– 8.
211
[15] Schilling EA, Aseltine RH, Glanovsky JL, et al. Adolescent alcohol use, suicidal ideation, and suicidal attempts. J Adolesc Health 2009;44: 335– 41. [16] Cooper J, Kapur N, Webb R, et al. Suicide after deliberate self-harm: A 4-year cohort study. Am J Psychiatry 2005;162:297–303. [17] Laursen TM, Munk-Olsen T, Nordentoft M, Mortensen PB. Increased mortality among patients admitted with major psychiatric disorders: A registerbased study comparing mortality in unipolar depressive disorder, bipolar affective disorder, schizoaffective disorder, and schizophrenia. J Clin Psychiatry 2007;68:899 –907. [18] Meloni D, Miccinesi G, Bencini A, et al. Mortality among discharged psychiatric patients in Florence, Italy. Psychiatr Serv 2006;57:1474 – 81. [19] Mos´cicki EK. Gender differences in completed and attempted suicides. Ann Epidemiol 1994;4:152– 8. [20] Troister T, Links PS, Cutcliffe J. Review of predictors of suicide within 1 year of discharge from a psychiatric hospital. Curr Psychiatry Rep 2008; 10:60 –5. [21] King CA, Segal H, Kaminski K, et al. A prospective study of adolescent suicidal behavior following hospitalization. Suicide Life Threat Behav 1995; 25:327–38. [22] Pompili M, Amador XF, Girardi P, et al. Suicide risk in schizophrenia: Learning from the past to change the future. Ann Gen Psychiatry 2007;6:10. [23] Evenson RC, Cho DW, Holland R. Identifying psychiatric suicides for research purposes. J Clin Psychol 1988;44:1029 –32. [24] Marusic A, Roskar S, Zorko M. Undetermined deaths: Are they suicides? Croat Med J 2003;44:550 –2.
www.jahonline.org Original article
Unnatural Causes of Death and Suicide Among Former Adolescent Psychiatric Patients Subin Park, M.D., Ph.D.a, Chang Yoon Kim, M.D., Ph.D.b, and Jin Pyo Hong, M.D., Ph.D.b,* a b
Division of Child and Adolescent Psychiatry, Department of Psychiatry, Seoul National University College of Medicine, Seoul, Republic of Korea Department of Psychiatry, Asan Medical Center, Ulsan University College of Medicine, Seoul, South Korea
Article history: Received December 20, 2011; Accepted May 24, 2012 Keywords: Adolescence; Suicide; Mortality; Mental disorders; Korea
A B S T R A C T
Purpose: Compared with the general population, adolescent psychiatric patients are subject to premature death from all causes, but suicide-specific mortality rates in this population have not been carefully investigated. Therefore, we examined the high mortality due to unnatural causes, particularly suicide, using standardized mortality ratios (SMRs) relative to sex, diagnosis, and type of psychiatric service. Methods: A total of 3,029 patients aged 10 –19 years presented to the outpatient clinic of a general hospital in Seoul, Korea, or were admitted to that hospital for psychiatric disorders from January 1995 to December 2006. Unnatural causes mortality risk and suicide mortality risk in these patients were compared with those in sex- and age-matched subjects from the general Korean population. Results: The SMR for unnatural causes was 4.6, and for suicide it was 7.8. Female subjects, the young, and inpatients had the highest risks for unnatural causes of death or suicide. Among the different diagnostic groups, patients with psychotic disorders, affective disorders, and personality disorders had significantly increased SMRs for unnatural causes, and those with psychotic disorders, affective disorders, and disruptive behavioral disorders had significantly increased SMRs for suicide. Conclusions: The risks of unnatural death and suicide are high in adolescent psychiatric inpatients in Korea, but not as high in adolescent outpatients. Effective preventative measures are required to reduce suicide mortality in adolescent psychiatric patients, particularly female patients admitted for general psychiatric care. 䉷 2013 Society for Adolescent Health and Medicine. All rights reserved.
Korea has the highest suicide rate of all the countries in the Organization for Economic Cooperation and Development (OECD). The mean age-standardized rate of suicide in Korea is 11.3 per 100,000 [1], and suicide is the most frequent cause of death in adolescents and young adults [2]. In addition, previous studies of suicide in adult psychiatric patients have shown that the younger the patients, the higher the standardized mortality
* Address correspondence to: Jin Pyo Hong, M.D., Ph.D., Department of Psychiatry, Asan Medical Center, 388-1 Pungnap-2dong, Songpa-gu, Seoul 138 –736, South Korea. E-mail address: [email protected] (J.P. Hong).
IMPLICATIONS AND CONTRIBUTION
The present study indicates that the risks of unnatural death and suicide are particularly high in female adolescent patients who have been admitted for general psychiatric care. Effective preventative measures are required to reduce suicide mortality among this highrisk group.
ratios (SMRs) [3–7]. Therefore, we were particularly interested in suicide-specific mortality risk in adolescent psychiatric patients. Most studies on mortality in child and adolescent psychiatric patients report all causes SMRs only [8 –12], and cause-specific mortality rates are not well-investigated in this population owing to small sample sizes or difficulty in obtaining causes of death. We are only aware of one study of adolescent psychiatric patients that differentiates unnatural causes mortality from natural causes mortality [10]. In that study, which included only psychiatric inpatients and was conducted in Oslo, Norway, SMR for unnatural causes (including drug overdose, suicide, homicide, and accident) was significantly elevated in both sexes. The SMRs were 8.5 (95% confidence intervals [CI]: 6.7–10.6) for male patients and 15.8 (95% CI: 10.8 –22.4) for female patients; suicide-
1054-139X/$ - see front matter 䉷 2013 Society for Adolescent Health and Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jadohealth.2012.05.018
208
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
specific SMRs were 6.8 (95% CI: 3.4 –12.1) for male patients and 19.0 (95% CI: 10.6 –31.3) for female patients; and drug overdose SMRs were 54.1 (95% CI: 32.9 – 84.3) for male patients and 83.3 (31.0 –178.0) for female patients. To establish effective strategies to prevent the premature deaths of adolescent psychiatric patients, detailed information on the cause-specific mortality in various diagnostic groups and different treatment settings is required. Therefore, we have investigated unnatural causes mortality and suicide mortality further by calculating SMRs relative to diagnosis, age, and type of hospital service (inpatient/outpatient), as well as sex-specific SMRs. Methods Subjects The subjects consisted of all patients who presented to a psychiatric outpatient clinic of a general hospital located in Seoul, Korea, or who were admitted to that hospital for a psychiatric disorder, during a 12-year period (from January 1995 to December 2006), and who were aged 10 –19 years at the time of the hospital visit. The data were censored either on the date of death or on December 31, 2009, because this is the date when the latest available data on the National Statistical Office (NSO) mortality were obtained. Person-years (the number of years that each individual was under observation) were calculated for each patient. Information about whether the patients were alive on December 31, 2009, was provided by linkage to the database of the NSO. For those who were dead, causes of death were also established through linkage to the database of the NSO. All deaths in Korea are reported to the NSO by a document of death notice, which contains the cause of death. However, death by suicide is sometimes misreported as death by other cause, such as car accident, by a bereaved family because of a social stigma of suicide. On the other hand, all deaths caused by suicide, homicide, or accidents are investigated by the National Police Agency (NPA). The police interview the individuals who knew the dead person well, such as family members, partner, friends, and/or neighbors, and make an investigation report on suicide, homicide, or accidents. When comparing the NSO and NPA data using national identification numbers, 96.6% of suicide cases in the NPA data were also counted as suicide cases in the NSO data [13]. For the purposes of the present study, we identified subjects who had died because of unnatural causes (ICD-10 External causes of morbidity and mortality V01-Y98). The NSO data and hospital records were matched using the unique national identification number assigned to all Korean citizens. Statistical analysis The unnatural causes mortality and suicide-specific mortality of the study subjects were compared with those of the corresponding general population of Korea. SMRs and 95% CIs were calculated using a person-years method with a free software program Person-years and Mortality COMputation Program 1.41 (Dirk Taeger, Dortmund, Germany) [14]. The SMR compared the observed numbers of deaths with the expected numbers of deaths. The person-years method calculated the total number of years that a person was at risk for each calendar year. The expected numbers of unnatural deaths and suicide-specific deaths
were calculated from the age-, sex-, and cause-specific mortality rates of the general population in Korea, for each calendar year. These specific mortality rates were obtained from the NSO. First, the sex-specific unnatural causes and suicide-specific SMRs were calculated. Next, the patients were divided into those who had received general psychiatric care (inpatients) and those who had not (outpatients). Unnatural causes and suicide-specific SMRs were then calculated for each type of care. Three psychiatric residents reviewed and confirmed the ICD-10 diagnoses that had been recorded in the electronic medical records by a boardcertified psychiatrist. The medical record requirements of the participating hospital included one primary diagnosis and several auxiliary diagnoses based on the ICD-10 coding. Diagnosis of those with multiple psychiatric disorders was made according to the primary ICD-10 diagnosis provided in the electronic medical records. Although a primary psychiatric diagnosis existed in most medical records, 6.3% of those records included an unclear diagnosis (i.e., mental disorder, not otherwise specified, or no ICD-10 F code diagnosis). For these cases, a board-certified psychiatrist made the diagnosis based on the clinical information provided in the medical records. For the purposes of the present study, these diagnoses were collapsed into eight diagnostic groups: psychotic disorder (F20 –F29), affective disorders (F30 – F33), neurotic disorders (F40 –F48), disruptive behavioral disorders (F90 –F92), mental retardation (F70 –F79), personality disorders (F60 –F69), eating disorders (F50), and other psychiatric disorders (diagnoses not included in the aforementioned diagnoses). Although previous studies indicated that substance abuse conferred a risk for death by suicide in adolescents [10,15], substance use disorders were not separated from the other diagnoses because of insufficient statistical power (only 13 patients [4.3%] had a primary diagnosis of substance use disorders). Unnatural causes and suicide-specific SMRs were then calculated for these diagnoses. The observation period for each patient was divided into three age spans: from first presentation to hospital to 19 years of age; from 20 to 24 years of age; and 25 years and older. Unnatural causes and suicide-specific SMRs were then calculated for these ages. To identify independent predictors for death by unnatural causes and suicide in adolescent psychiatric patients, sex, type of hospital service, diagnosis, and age were concurrently entered into the Cox proportional hazards model. Table 1 Characteristics of the 3,029 adolescent psychiatric patients Characteristics
N (%)
Sex, male Age at first hospital visit, mean ⫾ SD (years) Type of care Inpatient treatment Outpatient treatment only Psychiatric diagnosis Psychotic disorders Affective disorders Neurotic disorders Disruptive behavioral disorders Mental retardation Personality disorders Eating disorders Other diagnosesa
1,847 (61.0) 15.3 ⫾ 2.8
a
709 (23.4) 2,320 (76.6) 355 (11.7) 615 (20.2) 781 (25.8) 558 (18.4) 101 (3.3) 109 (3.6) 62 (2.0) 448 (14.8)
“Other diagnoses” includes organic mental disorders, substance use disorders, developmental disorders, tic disorders, and other childhood behavioral and emotional disorders.
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
Table 2 Unnatural causes and suicide-specific SMRs in male and female adolescent psychiatric patients Sex (person-years)
All (25,468) Male (15,390) Female (10,078)
Table 4 Unnatural causes and suicide-specific SMRs in relation to psychiatric diagnosis Diagnosis (personyears)
Death by unnatural causes Suicide O
E
SMR
95% CI
O
E
SMR
95% CI
36 19 17
7.7 6.1 2.1
4.6 3.1 8.3
3.3–6.4 1.9–4.9 4.8–13.2
25 11 14
3.2 2.1 1.2
7.8 5.4 11.5
5.0–11.4 2.7–9.7 6.3–19.3
Death by unnatural causes
Suicide
O
O
E
SMR 95% CI
Psychotic disorder 14 1.3 11.0 (3,538) Affective disorder 8 1.4 5.7 (4,608) Neurotic disorder 5 2.2 2.3 (6,917) Disruptive behavioral 3 1.0 2.9 disorder (4,198) Mental retardation 2 .3 6.7 (1,020) Personality disorder 3 .4 7.3 (1,133) Eating disorder (520) 1 .2 6.0
CI ⫽ confidence interval; E ⫽ expected death; O ⫽ observed death; SMR ⫽ standardized mortality ratio.
All statistical analyses were performed with SPSS (version 12.0; SPSS Inc., Chicago, IL), and an ␣-level of ⬍.05 was considered significant. Results During the 12-year study period, 3,029 adolescent psychiatric patients (709 inpatients and 2,320 outpatients) were enrolled and 36 of those patients died of unnatural causes (25 by suicide, 7 by accidents, and 4 undetermined). The characteristics of the study subjects are shown in Table 1. Unnatural causes SMRs were 4.6 (95% CI: 3.3– 6.4) in all patients, 3.1 (95% CI: 1.9 – 4.9) in male patients, and 8.3 (95% CI: 4.8 –13.2) in female patients. Suicide-specific SMRs were particularly high ⫺7.8 (95% CI: 5.0 –11.4) in all patients, 5.4 (95% CI: 2.7–9.7) in male patients, and 11.5 (95% CI: 6.3–19.3) in female patients (Table 2). In patients who had been admitted for general psychiatric care, the mortality risk from unnatural causes was 7.8 times expectation and from suicide it was 14.2 times expectation. In patients who received outpatient care only, the mortality risk from unnatural causes was 3.2 times expectation and from suicide it was 4.9 times expectation (Table 3). Among the different diagnostic groups, patients with psychotic disorders (11.0, 95% CI: 6.0 –18.4), affective disorders (5.7, 95% CI: 2.4 –11.2), and personality disorders (7.3, 95% CI: 1.5– 21.5) had significantly increased SMRs for unnatural causes. Patients with psychotic disorders (20.0, 95% CI: 10.3–37.1), affective disorders (8.2, 95% CI: 2.7–19.2), and disruptive behavioral disorders (7.3, 95% CI: 1.5–21.4) had significantly increased SMRs for suicide (Table 4). The mortality risk was highest in the 11–20 age-group, with an unnatural causes mortality risk 34 times expectation and a suicide mortality risk 71 times expectation (Table 5). Multivariate analysis using the Cox proportional hazards model indicated that younger age (hazard ratio [HR] ⫽ .68, per 1-year increase, 95% CI: .62–.74, p ⬍ .001), previous psychiatric admission (HR ⫽ 3.66, 95% CI: 1.75–7.67, p ⫽ .001), and diagnosis
209
6.0–18.4
E
SMR 95% CI
11 .5 20.7
10.3–37.1
2.4–11.2
5 .6
8.2
2.7–19.2
.7–5.3
3 .9
3.3
.7–9.5
.6–8.4
3 .4
7.3
1.5–21.4
.7–24.1
1 .1
8.0
.1–44.5
1.5–21.5
1 .2
5.9
.8–3.3
.8–33.1
1 .1 14.3
.2–79.5
of psychotic disorder (HR ⫽ 7.23, 95% CI: 2.42–21.58, p ⬍ .001) and personality disorder (HR ⫽ 5.54, 95% CI: 1.27–24.07, p ⫽ .022) were significantly and independently associated with increased risk of death from unnatural causes in the adolescent psychiatric patients. In addition, younger age (HR ⫽ .65, per 1-year increase, 95% CI: .59 –.73, p ⬍ .001), previous psychiatric admission (HR ⫽ 3.98, 95% CI: 1.62–9.76, p ⫽ .003), and diagnosis of psychotic disorder (HR ⫽ 10.12, 95% CI: 2.64 –38.86, p ⫽ .001) were significantly and independently associated with increased risk of death by suicide. Discussion In the adolescent psychiatric patients, the SMRs for unnatural causes and suicide were 4.6 and 7.8, respectively. Female patients, the young, and inpatients had the highest risks for unnatural causes of death or suicide. Among the different diagnostic groups, patients with psychotic disorders, affective disorders, and personality disorders had significantly increased SMRs for unnatural causes, and those with psychotic disorders, affective disorders, and disruptive behavioral disorders had significantly increased SMRs for suicide. Our finding of a high unnatural causes mortality, and of a particularly high suicide mortality, in adolescent psychiatric patients is in accord with the results of a previous study conducted in Oslo, Norway [10]. Direct comparison of our results with those of the previous study is difficult because of different follow-up periods (5–14 years vs. 15–33 years after index hospital visit), compositions of subjects (both inpatients and outpatients vs. inpatients only), and classifications of causes of death (the cited author differentiated drug overdose from suicide, but we did
Table 3 Unnatural causes and suicide-specific SMRs in relation to type of care received and sex Type of care (person-years)
Inpatient (7,020) Male (3,722) Female (3,298) Outpatient (18,448) Male (11,667) Female (6,780)
Death by unnatural causes
Suicide
O
E
SMR
95% CI
O
E
SMR
95% CI
19 11 8 17 8 9
2.4 1.7 .7 5.3 4.4 1.3
7.8 6.3 11.0 3.2 1.8 6.8
4.7–12.3 3.1–11.3 4.8–21.8 1.9–5.1 .8–3.6 3.1–12.8
14 7 7 11 4 7
1.0 .6 .4 2.2 1.5 .8
14.2 12.2 16.5 4.9 2.7 8.8
7.7–23.7 4.9–25.2 6.6–33.9 2.5–8.8 .7–7.0 3.5–18.2
210
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
Table 5 Unnatural causes and suicide-specific SMRs in relation to age-group Age (person-years)
10–19 (2,829) 20–24 (5,826) 25–34 (16,813)
Death by unnatural causes
Suicide
O
E
SMR
95% CI
O
E
SMR
95% CI
12 19 5
.4 1.4 6.0
34.0 13.6 .8
17.6–59.5 8.2–21.2 .3–1.9
9 12 4
.13 .60 2.5
71.0 20.0 1.6
32.4–134.9 10.3–34.9 .4–4.1
not). With that in mind, we compared the unnatural causes SMR of our inpatients with that in the previous study, instead of comparing the suicide-specific SMRs for the total subjects in the two studies. The unnatural causes SMRs of our inpatients were lower than those of the discharged adolescent patients in the study by Kjelsberg (6.3 vs. 9.3 for boys and 12.2 vs. 15.8 for girls). This indicates that the difference in unnatural causes mortality between adolescent psychiatric inpatients and communitydwelling adolescents is less prominent in Seoul than in Oslo, possibly owing to the high suicide mortality in the general Korean population. One possible explanation for this is that in Korea, many high-risk adolescents die by suicide rather than seeking hospital care. Consistent with findings from previous studies [16 –18], suicide-specific SMRs were substantially higher for girls than for boys, particularly among inpatients. This finding suggests that girls who have been admitted to hospital for psychiatric problems are more at risk than boys, whereas more boys than girls die by suicide in the general population. Furthermore, the current findings suggest that suicide attempts among girls with psychiatric disorders merit clinical attention to safeguard against suicide risk. Thus, although men routinely have higher rates of death by suicide in the general population [19], among Korean youth admitted for general psychiatric care, female patients may be at a heightened suicide risk. The unnatural causes SMR and suicide-specific SMR of the outpatients were much lower than those of the inpatients (3.2 vs. 7.8 and 4.9 vs. 14.2, respectively). In fact, the unnatural causes SMR and suicide-specific SMR of male outpatients did not differ significantly from those of male adolescents in the general population (p ⬎ .05). Our multivariate Cox regression analysis indicated that a history of general psychiatric care was independently associated with death due to unnatural causes and death by suicide. This result is consistent with the previous finding that inpatients are 3.5 times more likely than outpatients to die by suicide [8]. All the individual psychiatric diagnoses were associated with increased unnatural death and suicide-specific SMRs, although some of the associations were not statistically significant (p ⬎ .05) because of low numbers. Consistent with findings from previous work examining adult psychiatric patients [3–7,20,21], we found that psychotic disorder, affective disorder, and personality disorder had a substantial impact on the risks of unnatural death and suicide in adolescent psychiatric patients. The increased suicide mortality in adolescents with disruptive behavioral disorders is consistent with the result of a follow-up study of adolescents with attention-deficit hyperactivity disorder, which reported a suicide rate of .63%–.78% and an estimated relative risk ratio of 2.91 (95% CI: 1.47–5.70) [9]. The suicide-specific SMR decreased from 71 in 10 –19 year olds to 20 in 20 –24 year olds, and to 1.6 in 25–34 year olds (Table 5). This decline may be because of the low suicide mortality rate
of adolescents in the general population [2]. Another explanation is that the risk of death from suicide is highest in the early stages of a mental illness [22] and during the first year after discharge [3,5]. This study has three major strengths. First, most other studies calculated all causes SMRs only, whereas we calculated unnatural causes SMRs and suicide-specific SMRs. Second, we performed suicide-specific mortality analyses in relation to age, type of care, and psychiatric diagnosis, as well as sex. Third, most other studies were concerned only with adolescent psychiatric inpatients, whereas we investigated both inpatients and outpatients and compared their suicidal mortality risks indirectly via SMRs. The present study had several limitations. First, the relatively low number of deaths by suicide limited the statistical power to detect significant differences in suicide mortality between our sample and the general population. Although suicide was more common in our psychiatric population than in the general population, large sample sizes are needed to study rare outcomes. Second, NSO data may underestimate the number of deaths due to suicide because some suicides may have been classified as “undetermined” deaths [23,24]. Among our patients, there were four deaths that were classified as “undetermined.” Third, although three psychiatric residents reviewed and confirmed the diagnoses that had been recorded in the electronic medical records, the reliability of such diagnoses was not systematically assessed. Fourth, the effect of comorbid psychiatric conditions on mortality could not be examined in this study because only a primary diagnosis was considered. Finally, as this study was conducted in a single general hospital, our findings may not be representative of all Korean psychiatric inpatients. Thus, further research should explore the nature of adolescent suicide risk while accounting for psychiatric comorbidity in many hospitals nationwide. Despite these limitations, the present study indicates that in Korea, the risks of unnatural death and suicide are high in adolescent psychiatric patients, particularly female patients who have been admitted for general psychiatric care. These high-risk adolescents should be asked directly to indicate whether they have suicidal ideation, and they may need to participate in specific suicide prevention programs in addition to receiving conventional psychiatric treatment. References [1] OECD. OECD health data 2011: Statistics and Indicators for 34 countries. Paris: Organization for Economic Cooperation and Development, 2011. [2] Korea National Statistical Office. National statistics report 2010. Seoul: Korea National Statistical Office, 2010. [3] Goldacre M, Seagroatt V, Hawton K. Suicide after discharge from psychiatric inpatient care. Lancet 1993;342:283– 6. [4] Qin P, Nordentoft M. Suicide risk in relation to psychiatric hospitalization: Evidence based on longitudinal registers. Arch Gen Psychiatry 2005;62: 427–32.
S. Park et al. / Journal of Adolescent Health 52 (2013) 207–211
[5] Geddes JR, Juszczak E, O’Brien F, Kendrick S. Suicide in the 12 months after discharge from psychiatric inpatient care, Scotland 1968 –92. J Epidemiol Community Health 1997;51:430 – 4. [6] Ho TP. The suicide risk of discharged psychiatric patients. J Clin Psychiatry 2003;64:702–7. [7] Yim PH, Yip PS, Li RH, et al. Suicide after discharge from psychiatric inpatient care: A case-control study in Hong Kong. Aust N Z J Psychiatry 2004; 38:65–72. [8] Engqvist U, Rydelius PA. Death and suicide among former child and adolescent psychiatric patients. BMC Psychiatry 2006;6:51. [9] James A, Clacey J, Seagroatt V, Goldacre M. Adolescent inpatient psychiatric admission rates and subsequent 1-year mortality in England: 1998 –2004. J Child Psychol Psychiatry 2010;51:1395– 404. [10] Kjelsberg E. Adolescent psychiatric in-patients. A high-risk group for premature death. Br J Psychiatry 2000;176:121–5. [11] Rydelius PA. Deaths among child and adolescent psychiatric patients. Acta Psychiatr Scand 1984;70:119 –26. [12] Thomsen PH. A 22- to 25-year follow-up study of former child psychiatric patients: A register-based investigation of the course of psychiatric disorder and mortality in 546 Danish child psychiatric patients. Acta Psychiatr Scand 1996;94:397– 403. [13] Im JS, Choi SH, Hong D, et al. Proximal risk factors and suicide methods among suicide completers from national suicide mortality data 2004 – 06 in Korea. Compr Psychiatry 2011;52:231–7. [14] Taeger D, Sun Y, Keil U, Straif K. A stand-alone windows applications for computing exact person-years, standardized mortality ratios and confidence intervals in epidemiological studies. Epidemiology 2000;11:607– 8.
211
[15] Schilling EA, Aseltine RH, Glanovsky JL, et al. Adolescent alcohol use, suicidal ideation, and suicidal attempts. J Adolesc Health 2009;44: 335– 41. [16] Cooper J, Kapur N, Webb R, et al. Suicide after deliberate self-harm: A 4-year cohort study. Am J Psychiatry 2005;162:297–303. [17] Laursen TM, Munk-Olsen T, Nordentoft M, Mortensen PB. Increased mortality among patients admitted with major psychiatric disorders: A registerbased study comparing mortality in unipolar depressive disorder, bipolar affective disorder, schizoaffective disorder, and schizophrenia. J Clin Psychiatry 2007;68:899 –907. [18] Meloni D, Miccinesi G, Bencini A, et al. Mortality among discharged psychiatric patients in Florence, Italy. Psychiatr Serv 2006;57:1474 – 81. [19] Mos´cicki EK. Gender differences in completed and attempted suicides. Ann Epidemiol 1994;4:152– 8. [20] Troister T, Links PS, Cutcliffe J. Review of predictors of suicide within 1 year of discharge from a psychiatric hospital. Curr Psychiatry Rep 2008; 10:60 –5. [21] King CA, Segal H, Kaminski K, et al. A prospective study of adolescent suicidal behavior following hospitalization. Suicide Life Threat Behav 1995; 25:327–38. [22] Pompili M, Amador XF, Girardi P, et al. Suicide risk in schizophrenia: Learning from the past to change the future. Ann Gen Psychiatry 2007;6:10. [23] Evenson RC, Cho DW, Holland R. Identifying psychiatric suicides for research purposes. J Clin Psychol 1988;44:1029 –32. [24] Marusic A, Roskar S, Zorko M. Undetermined deaths: Are they suicides? Croat Med J 2003;44:550 –2.