Welcome to the Schizophrenia Resource Centre

Welcome, this website is intended for healthcare professionals in EMEA with an interest in the treatment of schizophrenia. By clicking the link below you are declaring and confirming that you are a healthcare professional

You are here

Long-term follow-up of all-cause and unnatural death in young people with first-episode psychosis

Schizophrenia Research, 1, 159, pages 70 - 75



To determine mortality-related estimates and causes of death in young people with first-episode psychosis (FEP), and to identify baseline predictors of mortality.


Mortality outcomes in 723 young people presenting to an early psychosis service were prospectively ascertained up to 20 years. Predictors of all-cause and unnatural death were investigated using survival techniques.


Forty-nine participants died by study end. Most deaths (n = 41) occurred within 10 years of service entry. All-cause mortality was 5.5% at 10 years, rising to 8.0% after 20 years. Unnatural death rates at 10 and 20 years were 5.0% and 5.9%, respectively. Three risk factors consistently predicted all-cause mortality and unnatural deaths.


A substantial proportion of excess mortality was due to non-suicide unnatural death, and, later, natural deaths. This suggests that mental health services should expand their current focus on suicide to incorporate strategies to prevent accidental death and promote healthier lifestyles.

Keywords: Mortality, Unnatural death, Psychotic disorders, Follow-up studies.

1. Introduction

Life expectancy of persons diagnosed with a psychotic illness is substantially lower than that of other members of the community ( Saha et al., 2007 ). Risk of suicide, particularly in patients with a diagnosis of schizophrenia has been the focus of much research ( Limosin et al., 2007 ). However, understanding and possibly ameliorating excess mortality require the study of a broad spectrum of causes of death in persons ascertained at the onset of their first episode of illness.

Most studies investigating incidence and predictors of mortality in patients with first-episode psychosis (FEP) have relatively short follow-up intervals ( Bertelsen et al., 2007 ) and substantial heterogeneity in the age of participants (Craig et al, 2006 and White et al, 2009). Mortality studies need to span sufficient time to observe their outcome. Risk of type of death changes with age, with older patients more likely to die of natural causes, albeit prematurely. Curtailing the period of observation potentially distorts outcomes. Other studies are limited by their retrospective designs (for example,De Hert et al, 2001 and Robinson et al, 2010a), which potentially introduces more sources of bias and confounding than prospective studies.

Recently, Dutta et al. (2011) determined the mortality status of a large cohort of FEP patients (n = 2132) from four UK clinical services an average 13.4 years after initial diagnosis. Male gender and overall symptom load were established as predictors of risk. This is an important contribution but was necessarily retrospective and covered patients diagnosed so long ago as to make uncertain the applicability to patients currently entering treatment.

In common with many studies of first-episode samples (Limosin et al, 2007, Harris et al, 2008, and Pompili et al, 2011), focus on suicide alone fails to accommodate the competing nature of different causes of mortality. Premature death may also result from heightened risk of accident, drug use and mortality related to lifestyle factors ( Ruschena et al., 1998 ). It may be difficult to distinguish suicide from other causes of mortality. Failure to accommodate competing causes may result in overestimation of the cumulative incidence of individual causes of mortality ( Satagopan et al., 2004 ).

1.1. Aims of the study

This study sought to prospectively examine long-term mortality outcomes of 723 young people with FEP aged 14–30 years at service entry by reporting the incidence and causes of death over a 20-year period, and to identify the prognostic significance of a range of features of the premorbid clinical profile and early course of illness.

2. Material and methods

2.1. Sample

The study cohort comprised 723 FEP patients in a prospective, long-term follow-up study full details of which have been published ( Henry et al., 2007 ). Briefly, at ascertainment patients were aged 14–30 years and had a DSM-IV diagnosis of a psychotic disorder (schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder, bipolar psychotic disorder, major depressive disorder with psychotic features, brief reactive psychosis/brief psychosis and psychosis not otherwise specified).

2.2. Predictors of mortality

Selection of predictors was informed by existing suicide and mortality literature, and consideration of variables available from the comprehensive research assessment battery documenting the demographic, diagnostic and clinical features of the illness. Predictors used in the initial, univariate analyses are presented in Table 1 .

Table 1 Baseline characteristics of participants: Means with standard deviations (SD) or percentages (%) with counts.

Baseline characteristics Evaluable cases Mean or percentage (%) SD or count (N)
Sociodemographic and clinical features
Age at service entry, years 723 21.9 3.6
Age at onset of psychotic symptoms, years 571 21.5 3.6
Gender, % male 723 69.4 (502)
DSM-IV diagnosis, % 723    
 Schizophrenia/schizophreniform   57.3 (414)
 Depressive psychosis   12.0 (87)
 Bipolar   13.0 (94)
 Schizoaffective   9.5 (69)
 Other   8.2 (59)
Duration untreated psychosis, days 579 186.6 430.3
 Median   48.0  
Duration of prodrome, days 566 339.2 534.8
 Median   134.5  
Duration of untreated illness, days 566 503.7 659.0
 Median   253.0  
Marital status, % 614    
 Never married   85.0 (522)
 Married/defacto   10.7 (66)
 Prev partnered   4.2 (26)
Educational level, % 570    
 Incomplete secondary   59.1 (337)
 Completed secondary/trade/tech   34.0 (194)
 Completed tertiary   6.8 (39)
Work status, % 571    
 Employed   64.8 (370)
 Student/home duties   21.7 (133)
 Unemployed   1.6 (68)
Maximum regular daily dose of neuroleptics (CPZ equivalent), % 561    
 ≤ 250   50.4 (283)
 > 250 to 500   28.5 (160)
 > 500 to 750   8.9 (50)
 > 750 to 1000   5.7 (32)
 > 1000   6.4 (36)
Previous self-harm 563 20.6 (116)
Percentage of episode depressed (more than 50%) 570 28.8 (164)
Hopelessness 557 25.3 (183)
Poor premorbid social adjustment or work history 549 25.0 (137)
Family history of suicide 560 4.8 (27)
Family history of psychiatric illness 566 56.5 (320)
Suicidal tendencies 552 29.0 (210)
Poor insight 571 70.9 (405)
Problem alcohol use 571 13.7 (78)
Problem drug use 570 54.2 (309)
Parental status 571 10.3 (59)
Living alone 571 8.6 (49)
Drug therapy prior to service entry 565 22.5 (127)
History of violent behaviour 561 16.0 (90)
Psychosocial stressor in previous year 550 40.4 (222)
Stress immediately before onset 541 17.4 (94)
Baseline BPRS total score 507 28.5 9.6
Baseline BPRS psychotic subscale 506 10.5 3.8
Baseline SANS total score 455 23.1 15.4
BPRS total score at stabilisation 544 13.7 8.3
BPRS psychotic subscale at stabilisation 544 3.4 3.5
SANS at stabilisation 507 18.2 15.0
BDI at stabilisation 491 7.2 7.0

2.3. Diagnosis

Psychotic diagnoses at baseline for the 723 subjects were systematically ascertained using the Royal Park Multidiagnostic Instrument for Psychosis (RPMIP;McGorry et al, 1990a and McGorry et al, 1990b) or the SCID-I/P ( First et al., 2002 ). Cases were classified into one of five diagnosis groups: schizophrenia and schizophreniform disorders (SCZ-S); schizoaffective disorder, bipolar disorder; depressive psychosis; and other psychotic disorders (see Section 2.1 ).

2.4. Demographic, illness duration and clinical variables

The majority of the 723 subjects (79%) were assessed with the RPMIP during the initial episode to assess illness duration components, and other clinical and demographic variables. The remaining 21% received less detailed assessments in accordance with study design protocols.

Information derived from the RPMIP included age at service entry and age at onset of prodromal features, and of psychosis, and duration of untreated psychosis, sex, marital status, education, work status, parental status, whether the subject was living alone, previous drug therapy, history of violent behaviour and family histories of psychiatric illness and suicide. The presence/absence of the following prior to presentation was also established: psychosocial stressor in previous year; stress immediately before onset, deliberate self-harm; and poor premorbid social adjustment/work history. The RPMIP also assessed symptomatology in the presenting episode, including: suicidal tendencies; poor insight; hopelessness; and percentage of episode depressed. Alcohol and illicit drug use was classified as non-problematic versus problematic use. Maximum regular daily dose of neuroleptic medication was recorded in CPZ equivalents.

2.5. Psychopathology

Subjects assessed with the RPMIP also received psychopathology assessments scheduled to be administered within the first few days following entry into treatment (index presentation; T1) and again at symptom remission/stabilisation (median 2.0 months after index presentation; T2). Psychopathology was assessed using the 18-item Brief Psychiatric Rating Scale (Overall and Gorham, 1962, Lukoff et al, 1986, and McGorry et al, 1988) and the Schedule for the Assessment of Negative Symptoms (SANS; Andreasen, 1982 ). A positive symptom subscale (BPRSPS; score range 0–24) was derived from the BPRS, comprising items measuring conceptual disorganization, hallucinatory behaviour, unusual thought content and suspiciousness. The 13-item Beck Depression Inventory (BDI; Beck and Beck, 1972 ) assessed level of depression at T2.

2.6. Follow-up of survival status

Survival status of all 723 subjects was established by linkage with the National Mortality Database (NMD) maintained at the Australian Institute of Health and Welfare (AIHW). COD was coded according International Classification of Diseases version 9 (ICD9) ( World Health Organization, 1977 ) for deaths prior to 1997, and ICD10 ( World Health Organization, 1993 ) for deaths in and after 1997. At the time of linkage (8th December, 2010), COD was unavailable in the NMD for six deaths but were obtained from the Coroner's Court of Victoria or the Victorian Registry of Births, Deaths and Marriages. We included only perfect matches with NMD database except in a small number of patients where apparently discrepant information could be reconciled.

2.7. Statistical analysis

Each death was classified either as a natural or unnatural event, and each unnatural death further classified as suicide, accident or cause undetermined, using ICD codes for underlying and other causes of death provided by the ABS.

The primary outcome was duration of survival measured from T1 to date of death or close-out date, whichever occurred first. Percentage dead at a given time and its standard error (SE) were estimated using the Kaplan–Meier method. Potential follow-up time was defined as the time from service entry to the designated closeout date according to the reverse Kaplan–Meier method ( Schemper and Smith, 1996 ). Unnatural death and suicide rates were obtained using a competing risks model ( Kalbfleisch and Prentice, 1980 ) for the following non-overlapping events: suicide, unnatural death other than suicide, and natural death.

Associations of potential predictors with mortality outcomes were assessed in a two-stage process; firstly univariate analyses using Mantel–Cox log-rank tests estimated unadjusted associations between each predictor and the particular mortality outcome; secondly, univariate predictors significant or near significant at P < 0.10 were entered in Cox proportional hazards models. Wald tests assessed the independent significance of each predictor adjusting for other variables. Variables retained in these models according to a stepwise backward elimination procedure. Two-sided tests were used throughout. The validity of the proportional hazards assumption underlying the Cox regression model in all univariate and multivariate analyses was established by inspection of scaled Schoenfeld residuals and tests for deviations from proportionality. No tests were significant or near significant. Analyses were conducted using the R package, v2.12.0 ( R Development Core Team, 2010 ).

3. Results

3.1. Baseline characteristics

More than two-thirds (69%) of the 723 subjects were male. Mean age at service entry was 21.9 years (SD = 3.6; range 14–30 years). The majority (57%) received a diagnosis of schizophrenia or schizopheniform disorder (SCZ-S), 13% bipolar psychotic disorder, 12% depressive psychosis, 10% schizoaffective and 8% received another psychotic diagnosis, comprising delusional disorder, psychotic disorder NOS or brief reactive psychosis (see Table 1 ).

3.2. Mortality rates

Forty-nine of the 723 subjects died by the close-out date. The median potential follow-up time was 14.9 years, and ranged from 9.6 to 21.3 years. There were 41 unnatural deaths (suicide 22, accident 15, cause undetermined 4). Age at time of death ranged from 19 to 43 years (median 26 years). Forty-one participants died within 10 years of service entry. The remaining 8 died between 12 and 18 years following service entry. All-cause mortality rate was 5.5% (SE 0.9%) at 10 years, rising to 8.0% (SE 1.3%) after 20 years ( Fig. 1 ). Unnatural death rates at 10 and 20 years were 5.0% and 5.9%, respectively. The suicide rate was 2.6% (SE 0.6) at 10 years, rising to 3.2% (SE 0.7%) at 20 years. Death rates in the schizophrenia-spectrum subgroup followed the same trend: at 10 and 20 years, suicide rates were 2.7% and 3.6%, unnatural death rates were 5.3% and 6.9% and all-cause mortality rates were 6.0% and 10.5%, respectively.

3.3. Causes of death

Among the 22 suicide cases, the following methods were documented: self-inflicted injury by hanging, strangulation or suffocation by plastic bag (n = 9); jumping or lying before moving object (n = 7); self-poisoning (n = 4) (carbon monoxide 2, clozapine 1, clozapine with doxylamine 1); and self-harm using firearm or sharp object (n = 2). Ten of the 15 deaths classified as accidental were due to poisoning from use of illicit drugs, principally. The remaining five accidental deaths were accounted for as follows: two deaths due to car accidents; one case by firearm; one by machinery, and for one additional case, the nature of the accident was not specified. Of the four cases of unnatural death of undetermined intent, two of the cases were a result of poisoning with psychotropic medication, one resulted from contacting with a sharp object and one was an unattended death for which the cause of mortality was unknown. Eight cases were classified as natural death with underlying causes stated as asthma, cardiomyopathy, deep vein thrombosis, myocarditis, pulmonary embolism, disease of pericardium, multi-organ infarction and congestive cardiac failure.

3.4. Associations of prior and presenting features with mortality risks

The first part of the two-stage analysis (univariate tests) included variables listed in Table 1 . Several predictors were associated with both all-cause mortality and unnatural death outcomes ( Table 2 ), at a probability level less than 0.10. These included male gender, higher maximum regular daily dose of neuroleptics ( Fig. 2 ), history of violent behaviour, suicidal tendencies, problem alcohol use, problem illicit drug use and poor premorbid social adjustment/work history.

Table 2 Univariate associations of each predictor with risk of (i) all-cause mortality and (ii) unnatural death, for all FEP subjects and for schizophrenia-spectrum (SCZ-S) diagnoses, at or below a priori specified probability value of 0.10.

    All FEP cases SCZ-S subgroup
All-cause mortality Unnatural death All-cause mortality Unnatural death
Predictor Level HR a 95% CI P HR a 95% CI P HR a 95% CI P HR a 95% CI P
Gender Male 3.27 1.79, 6.0 0.003 4.18 2.16, 8.1 0.002            
Maximum regular daily dose of neuroleptics (CPZ equivalents) b ≤ 250 1.0   0.027 1.0   0.005 1.0   0.096 1.0   0.014
> 250 to 500 1.72     1.81     1.65     1.69    
> 500 to 750 1.78     1.91     1.72     1.94    
> 750 to 1000 1.63     2.26     1.26     2.01    
> 1000 2.97     4.08     2.66     4.29    
History of violent behaviour Yes 2.07 0.91, 4.70 0.056 2.25 0.93, 5.44 0.038            
Suicidal tendencies Yes 2.96 1.59, 5.51 < 0.001 3.88 1.97, 7.63 < 0.001 3.34 1.54, 7.24 0.001 4.91 2.04, 11.8 < 0.001
Alcohol use Problem use 3.81 1.58, 9.17 < 0.001 3.48 1.35, 8.97 < 0.001 4.67 1.66, 13.2 < 0.001 4.41 1.38, 14.1 < 0.001
Illicit drug use Moderate to severe use 1.8 1.00, 3.24 0.072 1.89 1.00, 3.57 0.074            
Poor premorbid social adjustment or work history Yes 2.12 1.07, 4.22 0.021 2.27 1.08, 4.76 0.021            
Family history of psychiatric illness Yes                   0.46 0.21, 1.04 0.067
Living alone Yes 1.99 0.70, 5.69 0.10 2.03 0.65, 6.37 0.13 2.79 1.15, 6.79 0.024 3.1 0.70, 13.8 0.035

a Hazard ratio.

b Test of linear association.


Fig. 1 Cumulative incidence of causes of death (curves are additive).


Fig. 2 Overall mortality by maximum regular daily dose of neuroleptics during initial episode.

Within multivariate models, significant predictors of all-cause mortality ( Table 3 ) were higher maximum regular daily dose of neuroleptics (P = 0.001), suicidal tendencies (P = 0.003), problem alcohol use (P < 0.001) and poor premorbid social adjustment/work history (P = 0.026). In the adjusted analysis of risk of unnatural death, significant factors were maximum regular daily dose of neuroleptic (P < 0.001), suicidal tendencies (P < 0.001), problem alcohol use (P = 0.016) and being male (P = 0.031).

Table 3 Significant baseline factors identified in multi-predictor analyses.

  Factor Overall mortality Risk of unnatural death
HR a 95% CI P HR a 95% CI P
All FEP cases Max. regular CPZ dose 2.21 1.37, 3.57 0.001 2.52 1.50, 4.22 < 0.001
Suicidal tendencies 2.63 1.38, 5.02 0.003 4.15 2.00, 8.62 < 0.001
Problem alcohol use 3.16 1.63, 6.12 < 0.001 2.49 1.19, 5.24 0.016
Poor premorbid adj/work history 2.03 1.09, 3.78 0.026      
Male gender       3.17 1.11, 9.05 0.031
SCZ-S subgroup Max. regular CPZ dose 2.30 1.32, 4.02 0.003 3.88 1.88, 7.99 < 0.001
Suicidal tendencies 3.48 1.61, 7.51 0.001 8.06 2.76, 23.5 < 0.001
Problem alcohol use 4.29 1.97, 9.34 < 0.001 3.13 1.15, 8.55 0.026
Living alone 3.20 1.30, 7.92 0.012      

a Hazard ratio.

3.4.1. Schizophrenia-spectrum subgroup

Analyses were repeated including only patients with schizophrenia-spectrum diagnoses (Table 2 and Table 3). Univariate analyses identified the following predictors associated with both mortality outcomes at or below the probability level of 0.10: higher maximum regular daily dose of neuroleptics, suicidal tendencies, problem alcohol use and living alone. Additionally, the absence of family history of psychiatric illness predicted unnatural death. Multi-predictor analyses predicting all-cause mortality and risk of unnatural deaths identified maximum regular daily dose of neuroleptic, suicidal tendencies, and problem alcohol use as predictors in both cases. Living alone significantly predicted all-cause mortality (P = 0.012) when effects of other predictors were adjusted, however it was not a statistically significant predictor of unnatural death.

4. Discussion

Patients in this epidemiologically representative FEP sample were found to have a high risk of premature death (8.0% at 20-years; 10.5% in SCZ-S subgroup), principally from unnatural causes. Rates of unnatural deaths and suicides were the highest in the first 10 years (5.0% and 2.6% respectively; 5.3% and 2.7% in SCZ-S subgroup), followed by a steady decline over the next 10 years (5.9% and 3.2% respectively at 20-years; 6.9% and 3.6% in SCZ-S subgroup).

4.1. Comparison of mortality rates with general population estimates

In order to obtain an estimate of the suicide rates over the years from 2000 to 2009 in the general population, we accumulated the official annual suicide rates for the 15–44 year age groups in the population over this period ( ABS, 2011 ), resulting in an estimated suicide rate of 143.3 per 100,000 over 10 years. The age groups included in the estimates were those most closely reflecting the age range of the sample at recruitment and duration of followup. Unsurprisingly, the suicide rate in this cohort of young people diagnosed with FEP is substantially elevated compared with that in the comparably aged general population. The suicide rate in this study at 10 years is 2.6%, that is, 2600 per 100,000. Thus, the suicide rates are more than 18 times that in the general population. Although these comparative estimates are derived from a slightly different age range to the sample, the magnitude of difference is overwhelming. Unfortunately, official statements did not allow estimation of comparable rates of unnatural death in the general population.

4.2. Comparison of mortality rates with previous studies

Our mortality rates are lower than those reported elsewhere. Bromet et al. (2005) estimated suicide at 10% for schizophrenia cohorts prospectively followed up over 10 years, whilst Kua et al. (2003) reported similarly elevated rates in their schizophrenia cohort at 10 year (8.5%) and 20 year (9.7%) follow-up. In comparison, suicide rates in our SCZ-S subgroup were substantially lower, estimated at 2.7% at 10-years and 3.6% at 20-years. Unnatural death rates in a recently published 10-year FEP follow-up ( White et al., 2009 ) were also higher (9.2%) than that in our FEP 10-year (5.0%) and 20 year (5.9%) unnatural death rates. One explanation could be that early intervention (EI) programmes such as those in place at the catchment site may afford protection from suicide and other mortality outcomes. For example, a recent FEP study ( Chen et al., 2011 ) compared 700 EI subjects with 700 historical controls over 3-years, finding significantly lower rates of suicide (1.0% vs. 2.9%) and all-cause mortality (1.1% vs. 3.1%) in the EI group. Furthermore, in adjusted analyses reported by Harris et al. (2008) , suicide risk was 50% lower in the first three years following first contact with mental health services for those exposed to specialist early psychosis treatment compared to those not exposed, however this protective effect was lost over longer periods of follow-up. Nonetheless, caution should be exercised in making comparisons between studies, given different methodologies, age ranges and diagnostic mixes.

4.3. Suicide is not the only significant issue

Notably, a substantial proportion of excess mortality was due to non-suicide unnatural death (accident 15, cause undetermined 4) and, at longer follow-up, natural deaths (n = 8). Although suicide is clearly a significant issue, individuals with FEP are also at increased risk of death potentially resulting from accident-proneness, comorbid substance use, financial and social deprivation and concurrent physical conditions ( Ruschena et al., 1998 ).

The importance of presenting unnatural death rates alongside suicide rates is highlighted, as the actual suicide rate is likely to lie somewhere between the two. It is probable that suicides are under-reported for a number of reasons ( Suicide Prevention Australia, 2010 ) as it is often difficult to establish intentionality with certainty. Even given reasonable evidence; some coroners do not believe that young people have the capacity to understand the finality of their actions, hence findings of accidental/undetermined intent are often made. (Nearly half of unnatural deaths in our data were classified this way.) Further, the stigma of suicide may influence investigators to make an undetermined finding to protect the family. All these factors would lead to underestimation of suicide rates. Previous estimates of the lifetime suicide rate for schizophrenia (4.0%; Inskip et al., 1998 ) (4.9%; Palmer et al., 2005 ) lie between the suicide and unnatural death rates reported for our schizophrenia/schizophreniform subgroup (3.6% and 6.9% at 20 years). By this logic, in our FEP sample, the true suicide rate is likely to lie between 3.2% (reported suicide rate) and 5.9% (unnatural death rate) at 20 years following index presentation, whilst all-cause mortality is 8.0%.

4.4. Risk factors for excess mortality

Identifying risk factors for mortality at the beginning of treatment may enable clinicians to implement specialized treatments targeted at prevention of premature death. Clinical predictors examined in our prognostic analysis were characteristics of subjects either at service entry or symptom stabilisation. Three characteristics consistently and independently predicted mortality outcomes among all FEP subjects and the schizophrenia-spectrum subgroup: suicidal tendencies, problem alcohol use and maximum regular daily dose of neuroleptic. Our findings regarding suicidal tendencies are consistent with Robinson et al. (2010a) and De Hert et al. (2001) . In another study, Robinson et al. (2010b) compared single and multiple suicide attempts over a 7.4 year follow-up using a subset of our cohort and reported that a history of problem alcohol use increased the risk of multiple suicide attempts.

No study has previously reported maximum regular daily dose of neuroleptic at baseline as a risk factor for mortality. The underlying causal mechanism for this association is unclear. It may serve as an indicator of illness severity or treatment resistance. In this role, our results suggest two distinct prognostic groups: those who received only low doses of medication and who had a low risk of mortality and those who received very high doses who had an extremely elevated risk of death. Intermediate dosages did not reliably differentiate moderate levels of risk. This interpretation is supported by Joukamaa et al. (2006) , who found that the number of neuroleptics used at baseline showed a graded relation to mortality.

4.5. Strengths of the study

Strengths of this study are its large sample size, prospective collection of clinical data, complete-as-possible enumeration of deaths, and long follow-up duration. Most studies investigating incidence and predictors of mortality in FEP have been retrospective, had relatively short follow-up intervals ( Bertelsen et al., 2007 ) or substantial heterogeneity in participant age (Craig et al, 2006 and White et al, 2009). Our study is one of the few examining mortality in FEP which is unhindered by these methodological limitations.

4.6. Limitations

Limitations include the low base-rate of suicide events which precluded prognostic analyses for risk of suicide and non-coverage of death outside Australia. It is also possible that FEP patients not recruited to the study may have been at greater risk of unnatural death. However, those recruited are comparable on a range of socio-demographic and clinical variables ( Harrigan et al., 2003 ).

4.7. Clinical implications

Although suicide is a serious issue, individuals with FEP are also at increased risk for non-suicide unnatural death. Mental health services may need to expand their current focus beyond suicide in FEP to incorporate strategies aimed at preventing accidental deaths and promoting healthy lifestyles. Risk factors for mortality have been identified that may assist the targeting of preventive treatment to high-risk individuals.

Role of funding source

This research was supported by the Victorian Health Promotion Foundation Victoria, Australia (grant number 91-0084C); the Australian National Health and Medical Research Council Canberra, Australian Capital Territory, Australia (grant number 350241); and the Colonial Foundation, Victoria, Australia. Funding sources played no role in the execution and analysis of the study or preparation of the manuscript.


KY designed and led the statistical analyses, undertook the literature search and prepared the first draft of the manuscript. SH and AJM contributed to the design of the study and to further critical manuscript drafts. KY and MGH facilitated linkage with the National Mortality Database, whilst MGH, HPY, LPH, HJJ, HH and PDMc were involved in the design of the long-term follow-up study. All authors contributed to and have approved the final manuscript.

Conflict of interest

PDM has received research grants from Janssen Cilag and AstraZeneca, and honoraria from Janssen Cilag, Roche and Eli Lilly. No other declarations of interest.


The authors wish to thank all of the young people and family members who took part in the study. We are particularly grateful to the Australian Institute of Health and Welfare, the Coroners Court of Victoria and the Victorian Registry of Births, Deaths and Marriages for their assistance in providing mortality data for the study. PDM is supported by NHMRC Senior Principal Research Fellowship 1060996.


  • ABS, 2011 ABS. Causes of death, Australia, 2009. ABS Cat. No. 3303.0 (ABS, Canberra, 2011)
  • Andreasen, 1982 N.C. Andreasen. Negative symptoms in schizophrenia. Definition and reliability. Arch. Gen. Psychiatry. 1982;39(7):784-788
  • Beck and Beck, 1972 A.T. Beck, R.W. Beck. Screening depressed patients in family practice. A rapid technic. Postgrad. Med.. 1972;52(6):81-85
  • Bertelsen et al., 2007 M. Bertelsen, P. Jeppesen, L. Petersen, A. Thorup, J. Ohlenschlaeger, P. le Quach, T.O. Christensen, G. Krarup, P. Jorgensen, M. Nordentoft. Suicidal behaviour and mortality in first-episode psychosis: the OPUS trial. Br. J. Psychiatry. 2007;(Supplement 51):s140-s146
  • Bromet et al., 2005 E.J. Bromet, B. Naz, L.J. Fochtmann, G.A. Carlson, M. Tanenberg-Karant. Long-term diagnostic stability and outcome in recent first-episode cohort studies of schizophrenia. Schizophr. Bull.. 2005;31(3):639-649
  • Chen et al., 2011 E.Y.H. Chen, J.Y.M. Tang, C.L.M. Hui, C.P.Y. Chiu, M.M.L. Lam, C.W. Law, C.W.S. Yew, G.H.Y. Wong, D.W.S. Chung, S. Tso, K.P.M. Chan, K.C. Yip, S.F. Hung, W.G. Honer. Three-year outcome of phase-specific early intervention for first-episode psychosis: a cohort study in Hong Kong. Early Interv. Psychiatry. 2011;5:315-323
  • Craig et al., 2006 T.J. Craig, Q. Ye, E.J. Bromet. Mortality among first-admission patients with psychosis. Compr. Psychiatry. 2006;47:246-251
  • De Hert et al., 2001 M. De Hert, K. McKenzie, J. Peuskens. Risk factors for suicide in young people suffering from schizophrenia: a long-term follow-up study. Schizophr. Res.. 2001;47(2–3):127-134
  • Dutta et al., 2011 R. Dutta, R.M. Murray, J. Allardyce, P.B. Jones, J. Boydell. Early risk factors for suicide in an epidemiological first episode psychosis cohort. Schizophr. Res.. 2011;126(1–3):11-19
  • First et al., 2002 First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B., 2002. Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition. (SCID-I/P) Biometrics Research, New York State Psychiatric Institute, New York
  • Harrigan et al., 2003 S. Harrigan, P. McGorry, H. Krstev. Does treatment delay in first-episode psychosis really matter?. Psychol. Med.. 2003;33:97-110
  • Harris et al., 2008 M.G. Harris, P.M. Burgess, D.C. Chant, J.E. Pirkis, P.D. McGorry. Impact of a specialized early psychosis treatment programme on suicide. Retrospective cohort study. Early Interv. Psychiatry. 2008;2(1):11-21
  • Henry et al., 2007 L.P. Henry, M.G. Harris, G.P. Amminger, H.P. Yuen, S.M. Harrigan, M. Lambert, P. Conus, O. Schwartz, A. Prosser, S. Farrelly, R. Purcell, H. Herrman, H.J. Jackson, P.D. McGorry. Early Psychosis Prevention and Intervention Centre long-term follow-up study of first-episode psychosis: methodology and baseline characteristics. Early Interv. Psychiatry. 2007;1(1):49-60
  • Inskip et al., 1998 H.M. Inskip, E.C. Harris, B. Barraclough. Lifetime risk of suicide for affective disorder, alcoholism and schizophrenia. Br. J. Psychiatry. 1998;172:35-37
  • Joukamaa et al., 2006 M. Joukamaa, M. Heliovaara, P. Knekt, A. Aromaa, R. Raitasalo, V. Lehtinen. Schizophrenia, neuroleptic medication and mortality. Br. J. Psychiatry. 2006;188:122-127
  • Kalbfleisch and Prentice, 1980 J. Kalbfleisch, R. Prentice. The Statistical Analysis of Failure Time Data. (John Wiley & Sons, New York, NY, 1980)
  • Kua et al., 2003 J. Kua, K.E. Wong, E.H. Kua, W.F. Tsoi. A 20-year follow-up study on schizophrenia in Singapore. Acta Psychiatr. Scand.. 2003;108(2):118-125
  • Limosin et al., 2007 F. Limosin, J.-Y. Loze, A. Philippe, F. Casadebaig, F. Rouillon. Ten-year prospective follow-up study of the mortality by suicide in schizophrenic patients. Schizophr. Res.. 2007;94:23-28
  • Lukoff et al., 1986 D. Lukoff, K. Nuechterlein, J. Ventura. Manual for the expanded brief psychiatric rating scale. Schizophr. Bull.. 1986;12:594-602
  • McGorry et al., 1988 P.D. McGorry, R.J. Goodwin, G.W. Stuart. The development, use, and reliability of the brief psychiatric rating scale (nursing modification)—an assessment procedure for the nursing team in clinical and research settings. Compr. Psychiatry. 1988;29(6):575-587
  • McGorry et al., 1990a P.D. McGorry, D.L. Copolov, B.S. Singh. Royal park multidiagnostic instrument for psychosis: part I. Rationale and review. Schizophr. Bull.. 1990;16(3):501-515
  • McGorry et al., 1990b P.D. McGorry, B.S. Singh, D.L. Copolov, I. Kaplan, C.R. Dossetor, R.J. van Riel. Royal park multidiagnostic instrument for psychosis: part II. Development, reliability, and validity. Schizophr. Bull.. 1990;16(3):517-536
  • Overall and Gorham, 1962 J. Overall, D. Gorham. The Brief Psychiatric Rating Scale. Psychol. Rep.. 1962;10:799-812
  • Palmer et al., 2005 B.A. Palmer, V.S. Pankratz, J.M. Bostwick. The lifetime risk of suicide in schizophrenia: a reexamination. Arch. Gen. Psychiatry. 2005;62(3):247-253
  • Pompili et al., 2011 M. Pompili, G. Serafini, M. Innamorati, D. Lester, A. Shrivastava, P. Girardi, M. Nordentoft. Suicide risk in first episode psychosis: a selective review of the current literature. Schizophr. Res.. 2011;129(1):1-11
  • R Development Core Team, 2010 R Development Core Team. R: A Language and Environment for Statistical Computing. (R Foundation for Statistical Computing, Vienna, Austria, 2010)
  • Robinson et al., 2010a J. Robinson, M. Harris, S. Cotton, A. Hughes, P. Conus, M. Lambert, B.G. Schimmelmann, P. McGorry. Sudden death among young people with first-episode psychosis: an 8–10 year follow-up study. Psychiatry Res.. 2010;177(3):305-308
  • Robinson et al., 2010b J. Robinson, M.G. Harris, S.M. Harrigan, L.P. Henry, S. Farrelly, A. Prosser, O. Schwartz, H. Jackson, P.D. McGorry. Suicide attempt in first-episode psychosis: a 7.4 year follow-up study. Schizophr. Res.. 2010;116(1):1-8
  • Ruschena et al., 1998 D. Ruschena, P.E. Mullen, P. Burgess, S.M. Cordner, J. Barry-Walsh, O.H. Drummer, S. Palmer, C. Browne, C. Wallace. Sudden death in psychiatric patients. Br. J. Psychiatry. 1998;172:331-336
  • Saha et al., 2007 S. Saha, D. Chant, J. McGrath. A systematic review of mortality in schizophrenia. Arch. Gen. Psychiatry. 2007;64(10):1123-1131
  • Satagopan et al., 2004 J.M. Satagopan, L. Ben-Porat, M. Berwick, M. Robson, D. Kutler, A.D. Auerbach. A note on competing risks in survival data analysis. Br. J. Cancer. 2004;91:1229-1235
  • Schemper and Smith, 1996 M. Schemper, T.L. Smith. A note on quantifying follow-up in studies of failure time. Control. Clin. Trials. 1996;17(4):343-346
  • Suicide Prevention Australia, 2010 Suicide Prevention Australia. Position Statement Youth Suicide Prevention. (Suicide Prevention Australia, 2010)
  • White et al., 2009 C. White, J. Stirling, R. Hopkins, J. Morris, L. Montague, D. Tantam, S. Lewis. Predictors of 10-year outcome of first-episode psychosis. Psychol. Med.. 2009;39(9):1447-1456
  • World Health Organization, 1977 World Health Organization. International Classification of Diseases ICD-9. (World Health Organization, Geneva, 1977)
  • World Health Organization, 1993 World Health Organization. International Statistical Classification of Diseases and Related Health Problems ICD-10. (World Health Organization, Geneva, 1993)


a Orygen Youth Health Research Centre, Centre for Youth Mental Health, The University of Melbourne, Parkville, Victoria 3052, Australia

b Queensland Centre for Mental Health Research, School of Population Health, The University of Queensland, Australia

c School of Psychological Sciences, The University of Melbourne, Parkville, Australia

lowast Corresponding author at: Orygen Youth Health Research Centre, Centre for Youth Mental Health, The University of Melbourne, 35 Poplar Road, Parkville, 3052 Victoria, Australia. Tel.: + 61 3 9342 2800; fax + 61 3 9342 2941.

1 Present address: Australian Mathematical Sciences Institute, The University of Melbourne, Parkville, Victoria 3010, Australia.