Introduction
In Ireland, ∼500 people (475–554) have died by suicide on an annual basis between 2001 and 2013, resulting in a suicide rate of 10.3–13.5 per 100 000 (http://www.nsrf.ie/statistics/suicide/; Central Statistics Office 2013), with ∼4.5 times as many men dying by suicide over this time period, compared with women (http://www.nsrf.ie/statistics/suicide/). Internationally, Ireland has the eighth lowest rate of death by suicide out of 26 countries examined, based on 2013 data extracted from the World Health Organisation’s Statistical Information System (http://www.who.int/whosis/en) for both males and females. However, within the 15–24-year age group, Ireland has the fourth highest rate of suicide with an average suicide rate of 15.4 per 100 000 (Health Service Executive 2012).
Ascertaining putative causes for death by suicide have revealed a large array of potential risk factors. Psychological autopsy studies, on completed suicides have found that 60–90% of individuals, have suffered from a psychiatric illness before their death (Cavanagh et al. Reference Cavanagh, Carson, Sharpe and Lawrie2003), with mood disorders particularly prevalent (Bertolote et al. Reference Bertolote, Fleischmann, De Leo and Wasserman2004; Fleischmann et al. Reference Fleischmann, Bertolote, Belfer and Beautrais2005), although this is not a universal finding (Burgess et al. Reference Burgess, Pirkis, Morton and Croke2000). The abuse or dependence of alcohol and/or psycho-active substances (Rossow & Amundsen Reference Rossow and Amundsen1995; Johnsson & Fridell Reference Johnsson and Fridell1997; Bertolote et al. Reference Bertolote, Fleischmann, De Leo and Wasserman2004; Schneider et al. Reference Schneider, Schnabel, Sargk, Maurer, Weber and Wetterling2005) and the presence of a personality disorder (particularly borderline personality disorder; Isomesta et al. Reference Isomesta, Henriksson, Heikkinen, Aro, Marttunen, Kuoppasalmi and Lonnqvist1996; Park et al. Reference Park, Kim and Hong2013) have also previously been highlighted as other putative diagnostic risk factors for suicide. A multitude of psychosocial factors have also been implicated in increasing the risk of suicide. Childhood traumas such as exposure to family violence (Reinhertz et al. Reference Reinhertz, Giaconia, Silverman, Friedman, Pakiz and Frost1995), the loss of a parent or separation of parental figures (Grossi & Violato Reference Grossi and Violato1992) and childhood sexual abuse (CSA; Dube et al. Reference Dube, Anda, Felitti, Chapman, Williamson and Giles2001), have been especially implicated in this increased risk of suicide. However, caution is required with interpretation of this data, as several confounding factors including an increased risk of recurrent depressive disorder (Liu et al. Reference Liu, Jager-Hyman, Wagner, Alloy and Gibb2012) and alcohol and psycho-active substance abuse (Maniglio Reference Maniglio2011) may also be present in individuals who have experienced childhood trauma, which independently have been associated with an increased risk of suicide. In addition, epidemiological studies have suggested that genetic factors are a significant risk factor for suicide, independent of the presence of mental illness (Qin et al. Reference Qin, Agerbo and Mortensen2002; Mann Reference Mann2003).
Other commonly reported findings associated with completed suicide, include previous contact with mental health services and previous suicide attempts. Of individuals who die by suicide, 39–63% have had some contact with the mental health services previously (Luoma et al. Reference Luoma, Martin and Pearson2002) with 24–32% of individuals having had contact in the year before their death (Evans Reference Evans1994; Appleby et al. Reference Appleby, Shaw, Amos, McDonnell, Harris, McCann, Kiernan, Davies, Bickley and Parsons1999).The period of time shortly after discharge from an acute in-patient mental health unit has also been identified on several occasions as a risk factor for suicide (Geddes et al. Reference Geddes, Juszczak, O'Brien and Kendrick1997; Hoyer et al. Reference Hoyer, Olesen and Mortensen2004) with the first 2 weeks following discharge identified as the period of greatest risk (Swinson et al. Reference Swinson, Ashim, Windfuhr, Kapur, Appleby and Shaw2007). Studies evaluating self-harm in individuals who subsequently die by suicide, have demonstrated that 52–63% of individuals have had one or more episodes of self-harm (Boardman et al. Reference Boardman, Grimbaldeston, Handley, Jones and Willmott1999; Houston et al. Reference Houston, Hawton and Shepperd2001).
For individuals who die by suicide, death by hanging has consistently been demonstrated as the most common method use (Burgess et al. Reference Burgess, Pirkis, Morton and Croke2000). Gender differences have been demonstrated, with hanging (Hunt et al. Reference Hunt, Swinson, Palmer, Turnbull, Cooper, While, Windfuhr, Shaw, Appleby and Kapur2010) and use of firearms (Callanan & Davis Reference Callanan and Davis2012) more prevalent in men and self-poisoning more prevalent in women (Shen et al. Reference Shen, Hackworth, McCabe, Lovett, Aumage, O'Neil and Bull2006; Health Service Executive 2010).
In this study, we comprehensively investigate the demographic, psychosocial-- and clinical characteristics of individuals who have died by probable suicide in the West of Ireland using in-depth postmortem reports and their entire psychiatric case records. We want to explore in particular, the individuals’ diagnoses, methods employed leading to their death, their previous rates of self-harm and suicidal ideation, their interaction with the mental health services, their toxicology status at the time of their death and any documented stressors in their life, particularly in their childhood.
Methods
We included all postmortem reports carried out in University College Hospital Galway (UCHG) between 1 January 2006 and 31 May 2012 (n=1991). These postmortem reports were undertaken on individuals who died in three Health Service Executive (HSE) West regions [West Galway (including Galway city), East Galway and Roscommon]. The population of this region in 2011 was 314 718 (www.cso.ie). Approximately 10% of postmortem examinations from the East Galway region since January 2011 have not been carried out in UCHG. A senior psychiatrist (M.D./B.H.) and an advanced nurse practitioner (A.v.L.) jointly determined based on postmortem records if individuals had died by ‘probable suicide’. Postmortem reports included in-depth reports from the consultant pathologist, witness and police (garda) summaries and toxicology screens on all individuals. If uncertainty was present in relation to determining if an individual had died by suicide, a third senior clinician was consulted (M.D./B.H.) and consensus was attained. If doubt was present in relation to the possible cause of death (suicide/accidental death), a decision of accidental death was determined. Important considerations in determining if an individual had died by probable suicide included the method of death, the presence of a suicide note or other final acts, the toxicology screen (drug levels in the toxic range) and witness and police reports.
One hundred and eighty-one individuals were determined to have died by probable suicide based on the above screening mechanisms, with 153 individuals having resided in the HSE West region. We cross-checked each of these 153 individuals against the computer records of attendees (both in-patient and out-patient) of the three associated mental health services (West Galway Mental Health Services, East Galway Mental Health Services and Roscommon Mental Health Services) and ascertained that 58 individuals (38%) had previously had contact with the mental health services in at least one of these regions. The clinical case notes (all mental health records) were available for 57 of these individuals. All clinical notes were examined by both J.K. and B.H.
Recorded data from the postmortem reports included the method of death, the presence of a suicide note or any other ‘final acts’ and blood toxicology for alcohol, psycho-active substances and medication levels. Demographic and clinical data were attained from a comprehensive clinical note review. Demographic data collected included information in relation to gender, age, marital status, education level attained, socio-economic status, employment status, family structure and nationality. As demographic data could potentially change over time, we detail from the clinical notes the most recent employment status, socio-economic status and marital status. Clinical data recorded included the number of hospital admissions to the mental health services, psychiatric diagnoses, the date (where applicable) of most recent discharge from hospital and last attendance with the mental health services, the status of psychiatric admissions (formal/informal) and treatments received. Information in relation to any previous episodes of self-harm and suicidal ideation was carefully recorded. This included ascertaining the number of previous episodes of deliberate self-harm (DSH), the methods of self-harm, the perceived lethality of these episodes and any documented reports of suicidal ideation.
Other clinical data recorded included any known psycho-social stressors, childhood trauma, medical illnesses including chronic pain conditions, history of forensic difficulties and family history of psychiatric illness or suicide. Where possible, psychiatric diagnoses were based according to the International Classification of Mental and Behavioural Disorders 10 (ICD-10) diagnostic criteria from clinical note review and a diagnosis of personality disorder was only given if this was clearly reported with accompanying supportive symptomatology evident in the clinical notes. The treating consultant psychiatrist was contacted for clarification where uncertainty was present in relation to any clinical or demographic data, including ICD-10 diagnoses.
Ethical approval was attained before the commencement of this study from the Clinical Research Ethics Committee for Galway University Hospitals, and the ethics committees at St Brigid’s Hospital, Ballinasloe, and Roscommon County Hospital, Roscommon.
Statistical analysis was performed using the Statistical Package for Social Sciences 20.0 for Windows (SPSS Inc., IBM, USA). We utilised the student t-test for parametric data and the χ 2-test or where there was a small sample size Fishers exact test for non-parametric data.
Results
The demographic and clinical characteristics of the 57 included individuals who died by probable suicide and who attended the mental health services in the HSE west regions of West Galway, East Galway and Roscommon are detailed in Table 1. In brief, the ratio of males to females was 4.2:1, the most common marital status was ‘single’ (47%), the most common employment status was ‘unemployed’ (58%). The lowest socio-economic class (SEC) SEC-V was most common (54%). The most common psychiatric diagnosis noted in the clinical records was recurrent depressive disorder or a major depressive episode (44%), followed by alcohol dependence syndrome or harmful use of alcohol (35%). Eight individuals (32%) with depression had comorbid alcohol dependence or harmful use of alcohol. Thirteen individuals (23%) suffered with chronic pain, with six of these individuals also describing a chronic medical illness. A history of childhood adversity was described in the case notes in 25 individuals (44%), with a history of CSA documented in 13 individuals (23%). Individuals with chronic pain (23%) had higher rates of childhood adversity compared with those without such a history (77% v. 34%, *df=1, p=0.010). Seventeen individuals (30%) had a past forensic history (either arrested by Gardai and/or prosecuted in court for a criminal offence).
Demographic and clinical characteristics of patients while attending the mental health services
a Excludes alcohol or drug dependence or harmful use.
b Considerable co-morbidity between these groups.
c Includes one individual with anorexia nervosa, one individual with substance-induced psychosis and one individual with a diagnosis of an organic induced psychosis.
Demographic and clinical data attainable at the time of the individuals’ death are detailed in Table 2. The mean age of individuals was 46.8 years (s.d.±17.4; range 18–85 years). Thirty-three (58%) individuals who died by probable suicide were still registered (actively attending) with one of the three mental health services at the time of their death. Twenty individuals (35% of the entire cohort and 61% of individuals actively attending the mental health services) had been in contact with the mental health services in the month before their death, with 15 of these individuals having had contact with the mental health services in the previous 2 weeks. Forty individuals (70%) had previously been admitted to a psychiatric unit. Of these, eight individuals (20%) had been discharged from an in-patient psychiatric unit within a month of their death.
Demographical and clinical factors at time of death
a Other methods of death included gunshot and road traffic accident.
The most commonly employed method of death was by hanging with 33 individuals (58%) dying by this method. The other methods employed in order of frequency were drowning, overdose, self-inflicted gunshot and crashing a vehicle at high speed (Table 3). At least one previous suicide attempt had been undertaken by 37 individuals (65%), with 11 individuals (19%) having three or more previous suicide attempts. Attempted overdose was the most common previous method of attempted suicide employed by individuals (n=26, 46%), followed by attempted hanging (n=14, 25%). Fourteen individuals had a previous documented attempted hanging, with 11 (79%) of these individuals subsequently dying by the same method. All individuals who had a documented history of chronic pain (n=13) had at least one previous suicide attempt.
Previous suicide attempts and deliberate self-harm in patients who died by suicide
DSH, deliberate self-harm.
a Three individuals died by gunshot and one by crashing a vehicle at high speed.
When we compared individuals who had three or more previous suicide attempts (n=11) to those with two or less previous suicide attempts, adverse childhood experiences [n=8 (73%) v. n=17 (37%), df=1, p=0.04] and a forensic history [n=7 (64%) v. n=10 (22%), df=1, p=0.01] were more prevalent. No other demographic differences were evident between individuals with three or more previous suicide attempts compared with individuals with less than three attempts.
A history of DSH (without intent to end life) was present in 14 individuals (25%). Forty-three individuals (75%) had either a history of a suicide attempt or DSH. A documented history of suicidal ideation or ‘suicide talk’ was present in 40 individuals (70%). Eleven individuals (19%) had no previous documented history of suicidal ideation, DSH or suicide attempts.
The toxicology reports accompanying the postmortems are outlined in Table 4 and were available for 56 individuals. Thirty-four of these individuals (61%) had no ethanol detected in their blood at the time of postmortem. Antidepressants (34%) and benzodiazepines (29%) were the most prevalent medications detected on toxicology screening. For individuals with a documented history of depression (n=28), only nine individuals had antidepressants (32%) detectable with toxicology screening. Only one of the five individuals with a diagnosis of schizophrenia had antipsychotic agents detectable on toxicology screening.
Toxicology data obtained from postmortem reports
Toxicology data available for 56 individuals.
a Several individuals had positive toxicology for several substances.
Discussion
In this study, we comprehensively evaluated the detailed lifelong clinical and psycho-social data available from the psychiatric case records and postmortem reports of 57 individuals who died by probable suicide who previously attended the mental health services. While several previous studies have examined data in the year before the date of an individuals’ death (King et al. Reference King, Baldwin, Sinclair, Baker, Campbell and Thompson2001; Deisenhammer et al. Reference Deisenhammer, Huber, Kemmler, Weiss and Hinterhuber2007), our data based on a comprehensive examination of the entirety of an individuals’ psychiatric contact enabled us to identify several clinical characteristics throughout their lifetime of contact, which were potential risk factors for their later death by probable suicide.
Twenty-three per cent of individuals who died by probable suicide had contact with the mental health services in the year before their death (35 of the 153 individuals in the HSE-West region), which is consistent with some previous findings (Appleby et al. Reference Appleby, Shaw, Amos, McDonnell, Harris, McCann, Kiernan, Davies, Bickley and Parsons1999; Ho Reference Ho2003; Hunt et al. Reference Hunt, Windfuhr, Swinson, Shaw, Appleby and Kapur2011), but lower than that found in a number of other European studies (Barraclough et al. Reference Barraclough, Bunch, Nelson and Sainsbury1974; Deisenhammer et al. Reference Deisenhammer, Huber, Kemmler, Weiss and Hinterhuber2007). Overall, 38% of individuals (58 of the 153 individuals in the HSE-West region) had some contact with the mental health services in their lifetime, which represents an identical figure to that ascertained by Luoma et al. (Reference Luoma, Martin and Pearson2002) in their review of 40 studies; however, they ascertained in their review a wide range of lifetime contact (39–63%) with more elderly individuals who died by suicide having less lifetime contact with the mental health services. We did not ascertain any association between the rates of lifetime contact with the mental health services and age in this study. One potential reason for the relatively low rate of lifetime contact with mental health services in this cohort is that many individuals with mental illness and particularly those suffering from affective or neurotic disorders in Ireland are treated in primary care and are consequently not referred to the mental health services. Thus, although individuals may not have had contact with a mental health service, many will have had contact with primary care services for mental health difficulties.
Our finding of a 4.2:1 male to female ratio of death by probable suicide is largely in line with previous research findings (Shen et al. Reference Shen, Hackworth, McCabe, Lovett, Aumage, O'Neil and Bull2006; Hepp et al. Reference Hepp, Ring, Frei, Rössler, Schnyder and Ajdacic-Gross2010), except for studies in China where some studies have demonstrated a greater preponderance of females dying by suicide (Ping QinMortensen Reference Ping QinMortensen2001; Rebholz et al. Reference Rebholz, Dongfeng, Wenjie, Jing, Xigui, Jian-feng, Ji-chun, Chung-Shiuan, Kelly, Xiufang, Bazzano and Jiang2011). Other socio-demographic features that were over-represented in individuals who died by probable suicide in this study, such as been single, unemployed and living alone are also consistent with previous research findings (Boardman et al. Reference Boardman, Grimbaldeston, Handley, Jones and Willmott1999; Burgess et al. Reference Burgess, Pirkis, Morton and Croke2000; Qin et al. Reference Qin, Agerbo and Mortensen2003; Lee et al. Reference Lee, Lin, Liu and Lin2008; Law et al. Reference Law, Wong and Yip2010).
A previous diagnosis of depression was present in almost half of the individuals that died by probable suicide according to their case records, which is in line with several European studies (Arsenault-Lapierre et al. Reference Arsenault-Lapierre, Kim and Turecki2004; Bertolote et al. Reference Bertolote, Fleischmann, De Leo and Wasserman2004; Hunt et al. Reference Hunt, Kapur, Robinson, Shaw, Flynn, Bailey, Meehan, Bickley, Burns, Appleby and Parsons2006; Hirokawa et al. Reference Hirokawa, Matsumoto, Katsumata, Kitani, Akazawa, Takahashi, Kawakami, Watanabe, Hirayama, Kameyama and Takeshima2012) but is in contrast to a number of Asian studies (Thong et al. Reference Thong, Su, Chan and Chia2008; Law et al. Reference Law, Wong and Yip2010) and a study in Australia (Burgess et al. Reference Burgess, Pirkis, Morton and Croke2000) that found psychotic disorders, such as schizophrenia the most prevalent diagnosis in individuals who died by suicide. Only five individuals (9%) in this study had a schizophrenia spectrum disorder. Alcohol or substance abuse disorders were the second most frequent diagnosis in this study, a finding, which is also consistent with a number of previous findings (Berglund & Ojehagen Reference Berglund and Ojehagen1998; Pirkola et al. Reference Pirkola, Isometsa, Heikkinen and Lonnqvist2000). Thirty-two per cent of individuals with a diagnosis of depression in this study had a comorbid diagnosis of harmful use or dependence on alcohol. Previous studies have demonstrated that the co-occurrence of alcohol dependence or harmful use and major depressive episode is associated with an increased risk of suicide (Bartels et al. Reference Bartels, Coakley, Oxman, Constantino, Oslin, Chen, Zubritsky, Cheal, Durai, Gallo, Llorente and Sanchez2002; Dumais et al. Reference Dumais, Lesage, Lalovic, Seguin, Tousignant, Chawky and Turecki2005; Sher et al. Reference Sher, Oquendo, Galfalvy, Grunebaum, Burke, Zalsman and Mann2005).
The commonest methods of death were hanging followed by drowning, which is consistent with several previous studies (Fushimi et al. Reference Fushimi, Sugawara and Shimizu2005; Chen et al. Reference Chen, Lee, Chang and Liao2009; Holmgren & Jones Reference Holmgren and Jones2010; Huisman et al. Reference Huisman, van Houwelingen and Kerkhof2010; Lin et al. Reference Lin, Chang and Lu2010) including an Irish study (Casey et al. Reference Casey, Gemmell, Hiroeh and Fulwood2012). However, jumping from a height, noted in some previous studies as the most common method of death among individuals with mental illness (Gunnell & Nowers Reference Gunnell and Nowers1997; Beautrais Reference Beautrais2007), was not a method employed in this study, and perhaps this reflects the geographical variation of method and corresponding perceived availability of such methods. A history of past suicide attempts has been long established as one of the strongest risk factors for eventual death by suicide (Gunnell & Frankel Reference Gunnell and Frankel1994; Harris & Barraclough Reference Harris and Barraclough1997; Neeleman Reference Neeleman2001). Similarly, 65% of our cohort had a past suicide attempt (and 75% had a history of either a suicide attempt or DSH) with overdose the most common method employed, accounting for almost half of all previous attempts, followed by hanging. Of the 14 individuals with a previous attempt of hanging; 11 subsequently died by this method; whereas individuals who subsequently died by other methods, rarely had a previous attempt of hanging. This suggests that a previous attempt of hanging may be an independent risk factor for death by hanging in the future, whereas attempted overdose was less associated with a subsequent similar method of death. This finding is consistent with a large Swedish cohort study, which evaluated if the method of attempted suicide predicted successful suicide during a follow-up period of 21–31 years (Runeson et al. Reference Runeson, Tidemalm, Dahlin, Lichenstein and Langstrom2010), with over 90% of individuals who used hanging in the ‘index attempt’ died by the same method. However, our numbers are relatively modest in relation to deaths by drug overdose, to draw any firm conclusions in relation to this hypothesis.
Individuals with three or more previous suicide attempts, were more likely to have a forensic history and a history of family disharmony; however, our numbers are too modest (n=11) and so we are unable to infer any direct associations. In particular, although we did not ascertain an increased rate of individuals with a personality disorder in those who had three or more previous suicide attempts or who had a forensic history, no formal psychometric measures of personality disorder or impulsivity were routinely undertaken in this cohort.
A history of DSH, including self-harm with no intent to end one’s life, is a well-recognised risk factor for future death by suicide (Boardman et al. Reference Boardman, Grimbaldeston, Handley, Jones and Willmott1999; Jenkins et al. Reference Jenkins, Hale, Papanastassiou, Crawford and Tyrer2002), with one-quarter of the cohort in this study having a history of DSH (with no intent to end their life). Previous studies have demonstrated that individuals with both a past history of a suicide attempt and a past history of DSH demonstrate more severe psychopathology (Guertin et al. Reference Guertin, Lloyd-Richardson, Spirito, Donaldson and Boergers2001; Stanley et al. Reference Stanley, Gameroff, Michalsen and Mann2001; Muehlenkamp & Gutierrez Reference Muehlenkamp and Gutierrez2007) and greater suicidal ideation (Stanley et al. Reference Stanley, Gameroff, Michalsen and Mann2001; Muehlenkamp & Gutierrez Reference Muehlenkamp and Gutierrez2007) than those with a history of a suicide attempt alone.
Eight of the forty individuals (20%) who had a previous in-patient admission to a psychiatric unit, died by probable suicide within 4 weeks of discharge. This finding is consistent (albeit slightly lower) with other studies (Roy Reference Roy1982; Yim et al. Reference Yim, Yip, Li, Dunn, Yeung and Miao2004; Sinclair et al. Reference Sinclair, Harriss, Baldwin and King2005; Thong et al. Reference Thong, Su, Chan and Chia2008). Over 40% of the in-patient cohort died by suicide within 1 year of discharge, which is again consistent but slightly lower than previous reports (Roy Reference Roy1982; Hunt et al. Reference Hunt, Kapur, Robinson, Shaw, Flynn, Bailey, Meehan, Bickley, Burns, Appleby and Parsons2006). The period after psychiatric in-patient discharge has consistently been demonstrated to be a period of increased risk of suicide (Geddes et al. Reference Geddes, Juszczak, O'Brien and Kendrick1997; Holley et al. Reference Holley, Fick and Love1998; King et al. Reference King, Baldwin, Sinclair, Baker, Campbell and Thompson2001; Hoyer et al. Reference Hoyer, Olesen and Mortensen2004; Skeem et al. Reference Skeem, Silver, Aippelbaum and Tiemann2006; Karvonen et al. Reference Karvonen, Hakko, Koponen, Meyer-Rochow and Rasanen2009), a finding corroborated by this study.
The presence of childhood adversity and in particular CSA (Dube et al. Reference Dube, Anda, Felitti, Chapman, Williamson and Giles2001) has previously been associated with increased rates of suicide. Here, 23% of individuals had a documented history of CSA. Given that individuals who attend mental health services are often not asked about CSA or are either reluctant or refuse to inform their treating clinician about previous childhood trauma or CSA (Read et al. 2006), it is very likely that 23% is an under-estimate of the actual rate of CSA in this group and tentatively supports previous reports of an association between CSA and completed suicide (Dube et al. Reference Dube, Anda, Felitti, Chapman, Williamson and Giles2001). Consistent with previous findings (Jones et al. Reference Jones, Power and Macfarlane2009; Gonzalez et al. Reference Gonzalez, Boyle, Kyu, Georgiades, Duncan and MacMillan2012), higher rates of childhood adversity were noted in individuals with chronic pain in adulthood.
Three quarters of this cohort had either no alcohol detectable or low levels of alcohol (<80 mmol/l) detectable by toxicology. Our figures of alcohol use prior to an individuals’ death by probable suicide (39%) are similar to some previous studies which have noted that approximately one-third of individuals who died by suicide had alcohol consumed prior to their death (Cherpitel et al. Reference Cherpitel, Borges and Wilcox2004; Boenisch et al. Reference Boenisch, Bramesfeld, Mergl, Havers, Althaus, Lehfeld, Niklewski and Hegerl2010; Holmgren & Jones Reference Holmgren and Jones2010). Only 11% of individuals in this study had high levels of alcohol (>200 mmol/l) detectable by toxicology. Thus, although alcohol can have a disinhibiting effect and increase the risk of someone attempting suicide (Borges & Rosovsky Reference Borges and Rosovsky1996), in this study the association between probable suicide and alcohol intoxication was not demonstrated.
According to the toxicology reports, one third of individuals with a diagnosis of depression had antidepressants detectable on postmortem. However, it is possible that some individuals had their treatment stopped by a health professional outside the mental health services rather than discontinuing their medication themselves. In relation to the cohort with schizophrenia, where it would be less likely that their medication would be discontinued by a health professional outside the mental health services (none had their medication discontinued by their treating mental health professional), only one individual (out of 5) had antipsychotic medication detectable at toxicology. It is thus likely that a large proportion of individuals with previous mental illness were non-concordant with their treatment prior to their death. There have been multiple studies demonstrating high rates of treatment non-concordance in individuals with mental health difficulties (Melartin et al. Reference Melartin, Rytsala, Leskela, Lestela-Mielonen, Sokero and Isometsa2005; Cantrell et al. Reference Cantrell, Eaddy, Shah, Regan and Sokol2006; Akincigil et al. Reference Akincigil, Bowblis, Levin, Walkup, Jan and Crystal2007; Sawada et al. Reference Sawada, Uchida, Suzuki, Watanabe, Kikuchi, Handa and Kashima2009), with particularly high rates of treatment non-concordance in individuals suffering from schizophrenia (Hunt et al. Reference Hunt, Kapur, Robinson, Shaw, Flynn, Bailey, Meehan, Bickley, Burns, Appleby and Parsons2006), with the Clinical Antipsychotic Trials of Intervention Effectiveness study outlining that only one-quarter of individuals after 18 months of treatment were concordant with their medication (Manschreck & Boshes Reference Manschreck and Boshes2007). Treatment non-concordance has previously been demonstrated to be associated with both increased rates of suicidal ideation (Qurashi et al. Reference Qurashi, Kapur and Appleby2006) and suicide (Hawton et al. Reference Hawton, Sutton, Haw, Sinclair and Deeks2005; Ward et al. Reference Ward, Ishak, Proskorovsky and Caro2006). Our study tentatively supports these findings.
Limitations
There are a number of limitations with this study. First, our numbers are relatively modest, and thus caution is required with interpreting some of our findings. For example, only seven individuals died by drug overdose and thus determining if there is an association between death by overdose and previous methods of attempted suicide is not feasible. Second, almost 40% of the sample had no contact with the mental health services in the year before their death, and thus it is possible that they had mental health difficulties that either were unaddressed or were addressed by their general practitioner alone (without referral to the local mental health services). Third, no formal psychometric instruments were routinely administered to study participants. However, all diagnoses were based on ICD-10 diagnostic criteria, as documented in the clinical case notes, and were corroborated by the presence of documented symptoms of that diagnosis. The notes were comprehensively evaluated over several hours by two individuals including one experienced clinician (B.H.) and where uncertainty was present (no obvious diagnosis, multiple diagnoses present), the diagnosis or diagnoses was confirmed from their treating psychiatrist. A treating consultant psychiatrist who knew the individual who died by probable suicide was available for all individuals in this study. Fourth, although we distinguish between suicide attempts and DSH without an attempt to end one’s life, caution is required with such a distinction on the basis of clinical notes alone. Finally, although we report that there were high rates of treatment non-concordance, it is possible that health professionals outside the mental health services discontinued individuals’ medications, particularly if they had neurotic or affective disorders.
Conclusion
In this study, we comprehensively evaluated the detailed lifelong clinical and demographic data available from the psychiatric case records and postmortem reports of 57 individuals who died by probable suicide who previously attended the mental health services. Individuals who died by probable suicide, most commonly died by hanging and drowning; with previous attempts of hanging particularly prevalent in the group who subsequently died by hanging. The high rates of suicide attempts and/or DSH in this cohort emphasise the importance of the National clinical programme for the management of self-harm among individuals presenting to hospital emergency departments. High rates of childhood adversity were documented in the case notes of individuals who died by probable suicide. At the time of death, less than one-third of individuals according to toxicology reports were taking the medication that was last prescribed to them by the mental health services suggesting a high rate of treatment non-concordance in individuals who died by probable suicide.
Acknowledgements
The authors would like to thank staff at the department of pathology at University Hospital Galway and administrative staff in the psychiatry departments at University Hospital Galway, St. Brigids Hospital Ballinasloe and Roscommon County Hospital.
