Research report

Characteristics of deaths by suicide in Northern Ireland from 2005to 2011 and use of health services prior to death

Siobhan O'Neill n, Colette V. Corry, Sam Murphy, Sharon Brady, Brendan P. BuntingBamford Centre for Mental Health and Well-Being, University of Ulster, Northland Road, Londonderry BT48 7JL, Northern Ireland, UK

a r t i c l e i n f o

Article history:Received 21 April 2014Received in revised form16 July 2014Accepted 18 July 2014Available online 29 July 2014

Keywords:SuicideService UseConflictMental Health

a b s t r a c t

Background: Service presentation may offer an opportunity for intervention prior to suicide. The studyaimed to examine the characteristics, disorders and service use profiles of those who had died by suicidein Northern Ireland (NI) from 2005 to 2011.Methods: An analysis of a database of deaths by suicide and undetermined intent based on data in the NICoronial files from 2005 to 2011 (N¼1667).Results: Males are three times as likely to die by suicide as females and suicide rates similar among thoseaged 20–60 years. Females have increased service use prior to suicide; males tend to disengage withservices prior to death. Females are more likely to have recorded prior attempts, service use, diagnosisand referral. The most common health service used was primary care.Limitations: Despite the inclusion of undetermined deaths (probable suicides) a proportion of deaths bysuicide remain unrecorded as such. Data on marital status and mental and physical disorders were basedon information recorded by police officers from relatives, other informants and medical records. Thereliability of this data may therefore be questioned.Conclusions: Primary care has an important role in suicide prevention. Gendered patterns in service useprior to death should be considered in suicide prevention programmes. It is important to strengthenclinicians' knowledge of the manifestations of suicidal ideation in males and ways of encouraging serviceuse in males. The NI population who were exposed to the height of the violence of the conflict appear tobe at increased risk of suicide as they age.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

Suicide and suicidal behaviour are recognised as a wide-reaching social and important public health issues with annualrates of 11.8 per 100,000 in the United Kingdom (UK). It is also amajor economic concern with combined costs of d1.4 million persuicide in the UK (Knapp et al., 2011). Northern Ireland (NI) is thesole country in the UK to have demonstrated an overall increase inrecorded suicides in the last decade (Snowcroft, 2013; Tomlinson,2012; NISRA, 2014). However, disparity in coronial reportingsuggest unreliability, particularly with regard to those ‘narrative'verdicts which are increasingly used in England, Scotland andWales (Gunnell et al., 2011; Carroll et al., 2012). Such verdicts tendnot to be used in NI where there is now a single coroner's servicethus increasing the consistency of the recording procedures.

Self-harm, suicidal ideation and mental disorders are impor-tant precipitating factors for death by suicide. Psychologicalautopsy studies indicate that over 90% of those who die by

suicide have a psychiatric disorder (Foster et al., 1997). NI has ahistory of conflict, and there is evidence that those who havebeen exposed to the conflict have a higher risk of mentaldisorders (Bunting et al., 2013; Ferry et al., 2013). The mentalhealth needs of the NI population are higher than those of otherparts of the UK. It is estimated that 24% of women and 17% menin NI have a current mental health disorder, a figure 20% higherthan England and Wales (Appleby et al., 2013). Results from theWorld Mental Health Survey initiative demonstrated that NIconsistently ranked in the top three countries with respect torates of mental disorders, and the NI prevalence of Post-Traumatic Stress Disorder was the highest of all the countriessurveyed (Bunting et al., 2013). There is also evidence that peoplein NI who have experienced conflict related traumatic events aremore likely to have suicidal ideation and plans than those withother types of traumas, even when the effects of mental disordersare controlled for (O'Neill et al., 2014). In addition, schoolchildren in NI who have reported having experienced the conflicthave higher rates of self-harm (O'Connor et al., 2014).

In addition to mental disorders, people with physical disordershave an increased risk of self-harm and suicide (Singhal et al.,2014). Health service contact offers an opportunity for the delivery

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jad

Journal of Affective Disorders

http://dx.doi.org/10.1016/j.jad.2014.07.0280165-0327/& 2014 Elsevier B.V. All rights reserved.

n Corresponding author. Tel.: þ44 2871375354.E-mail address: sm.oneill@ulster.ac.uk (S. O'Neill).

Journal of Affective Disorders 168 (2014) 466–471

of suicide prevention interventions; however, the inconsistenciesin the recording of suicides have resulted in few studies of thepatterns of health service use prior to death by suicide. Studies ofUK primary care contact prior to suicide demonstrate that certainpopulations, younger people and females, are more likely topresent to their General Practitioner prior to death (Power et al.,1997; Stark et al., 2012). Studies of all health service use has shownthat a greater proportion of individuals who die by suicide havecontact with primary care providers than with mental healthspecialists (Luoma et al., 2002). Secondary care service users alsocontinue to contact access primary care services in the periodprior to death by suicide (Pearson, et al., 2009). The objective ofthe current study was to examine the characteristics and serviceuse history of those who died by suicide in NI, as well as mental/physical diagnoses at time of death.

2. Method

Approval was obtained from the University of Ulster ethicalcommittee to undertake the research. Cases were recorded by yearof death and deaths by suicide and undetermined intent weregenerated by staff from the NI Coroner's Service (CSNI) whichsubsequently directed file selection. The requirements for a coronialverdict of suicide were stricter than those required for classification asa probable suicide and inclusion in the database. Undetermineddeaths, which were probable suicides, were classified by both thesenior Coroner and also, following an analysis of the file, the ResearchAssociate. Undetermined deaths were classified as suicide where themeans of death was that of a common means of suicide and it waslikely that the individual took direct action that led to their death.

Case validation was undertaken with the assistance of NIStatistics and Research Agency (NISRA) personnel to ensure thatthe cases in the database were those included in the official NISRAstatistics on deaths by suicide. For each case data was extractedfrom physical files stored in CSNI archives and electronicallyrecorded in a database. Data on established risk factors includingprior suicidal behaviour, diagnosed mental and physical healthconditions, pharmacological profiles, demographics, substance useand prior adverse events were extracted. Health disorders andservice use was assessed using the deceased person's medicalnotes (where available), police reports and next of kin statements.Socioeconomic indicators were identified through the samesources as well as information included in pathology reportsregarding occupation and geographical position (coordinates).Information on adverse events prior to death was coded by boththe Research Associate and another investigator independently.There was a high level of concordance between the two and anydiscrepancies were resolved prior to statistical analysis.

Operational definitions of variables are as follows: age and maritalstatus refers to status at time of death; previous suicidal behaviour

includes hospital and non-hospital treated events. In terms of serviceuse, primary care refers to care under the General Practitioner;secondary care refers to outpatient mental health treatments; tertiarycare refers to psychiatric inpatient care. Mental disorders refer to bothmental and substance disorders.

Multinomial logistic regression was used to examine associationsbetween socio-demographic indicators and the contact with servicesprior to death. The reference categories were males, contact withtertiary level services, last contact with services over one year andsingle marital status. Age was based on the mean age. ‘Last point ofservice’ considered the last time contact with services was made bythe deceased, while ‘level of contact’ classified whether that treat-ment was primary, secondary or tertiary. A code of “none” in lastpoint of service use includes those cases were no information onservice use was available. Three mutually exclusive diagnostic out-comes were defined: (a) the presence of mental health disorder,(b) the presence of a physical health disorder, and (c) presence ofboth physical and mental health disorder. Analyses were implemen-ted using the IBM SPSS package (16).

3. Results

Information was gathered for those cases which occurredbetween 2005 and 2011 (N¼1667). Gender ratios for completedsuicides were 3:1, 77% male and 23% female (Table 1). Of these,gender proportions were similar in those under 19 years (9.3% and7.9% respectively), while males demonstrated somewhat higherrates aged between 20 and 39 years. Female suicides were highestin those aged between 40 and 69 years. These differences did notreach statistical significance.

A higher proportion of females were in contact with services inthe week prior to death relative to males (25.7% and 16.5%respectively). There was a small, though statistically significantincrease in the use of services by females in the two months priorto death (Table 2). A higher proportion of males last availed ofservices beyond this point, with a sevenfold higher rate amongthose who had not been in contact with health services for at leastone year before death (3.6% and 0.5% respectively).

Those aged over 40 years were more likely to engage withservices in the week before death (19.9%); however associationswith age group failed to reach statistical significance (χ2¼0.08,p40.05). The highest overall service presentations were recordedin those over 70 years (27%). A similar pattern emerged for this agegroup with regard to help seeking in the month prior to death,while those aged between 20 and 29 years engaged in servicesmore frequently in the two months prior to death (8.2%). The60–69 years age group were most frequent service users in theperiod up to four months preceding suicide (7%), while individualsaged below 19 years were more likely to access services betweensix months and one year prior to death (6.6%). (Table 3)

Table 1Gender and age.

Age group % (n) 10–19 yrs 20–29 yrs 30–39 yrs 40–49 yrs 50–59 yrs 60–69 yrs 70þ yrs Total

Female 25.5% (35) 17% (62) 18% (60) 25.5% (96) 29% (70) 28% (36) 22% (17) 22.6% (376)Male 74.5% (102) 83% (306) 82% (274) 74.5% (281) 71% (175) 72% (93) 73% (60) 77.4% (1291)

Table 2Gender and last health service interaction.

Gender % (n) Not known 1 wk 1–2 wks 2–4 wks 1–2 M 2–4 M 4–6 M 6–12 M 41 yr

Female 47.1% (180) 25.7% (98) 5.2% (20) 7.6% (29) 6.3% (24) 3.1% (12) 2.6% (10) 1.8% (7) 0.5% (2)Male 48.3% (623) 16.5% (213) 5% (65) 7.4% (95) 6% (78) 4.9% (63) 3.8% (49) 4.6% (59) 3.6% (46)

S. O'Neill et al. / Journal of Affective Disorders 168 (2014) 466–471 467

Statistically significant gender differences were identified withregard to number of previous suicide attempts (Table 4). Males weremore likely to have made only one previous attempt relative tofemales (17.3% and 15.4%); with females having increased proportionsof non-fatal attempts prior to completed suicide generally. Thesefindings prevailed among individuals who made 2–4 prior attemptsand those who made five attempts or more.

Under a third (31%) of the deceased had no recorded condition attime of death (n¼510). Combined physical and mental health condi-tions were recorded for 22% of the sample (n¼365), a mental disorderonly was recorded for 36% (n¼599) while 12% (n¼199) had only arecorded physical health condition (Table 5). Overall, 69.5% of all thosewho died by suicide had a diagnosed health condition.

Gender specific rates of mental, physical and combined condi-tions are presented in Table 5. Women were significantly morelikely than men to have a recorded health condition (75% vs. 67.8%,N¼1253) (χ2¼27.80, p¼o0.001). Women were more likely thanmen to have only a mental health condition; however thedifference was not statistically significant, (39.1% vs.34.8%, respec-tively). Almost twice as many men had only a physical condition(n¼173) (χ2¼11.47, p o0.001). In contrast, women were signifi-cantly more likely to have a combined mental and physical healthdisorders at point of suicide (n¼109) (χ2¼14.62, po0.001).

Figures for last point of service use and service use level prior todeath are presented in Table 6. Women were most likely to present forservices in the period up to 14 days before end of life, relative to men(18.2% vs. 23.9% respectively), (n¼326) (χ2¼6.12, po0.05). A higherproportion of men had not utilised any service in the year prior to death(3.1% vs. 1.1%), (n¼47) (χ2¼5.41, po0.05). Similarly, a statisticallysignificant difference was found among for gender and use of secondaryand tertiary services. Men were more likely not to progress beyondprimary service level (52.6% of men compared with 41.8% of women),(n¼839) (χ2¼13.67, po0.01). In contrast, females were more likely toutilise secondary level services, (n¼406) (χ2¼5.24, po0.01). Thisgender difference in services used was also evident for tertiary care,with higher rates of use among females (n¼98) (χ2¼5.01, po0.05).

Table 7 provides results of a multinomial logistic regressionillustrating socio-demographic associations with health condition attime of suicide. When the effects of the demographic, socio-economicand service type variables were considered simultaneously, severalassociations with suicide emerged. Men's odds of suicide were morethan three times those of women. Age was a significant indicator forall health conditions; a one year increase from themean of 40.31 yearsproduced an increased risk of 0.03 units for mental health disorder,0.07 units for physical disorder and 0.07 units for combined mentaland physical disorders (po0.001). While there was no significantdifference between males and females for physical disorders, womenwere statistically more likely to have amental disorder (po0.001) andalmost twice as likely to have combined mental and physical disordersat time of death (po0.001). Individuals who were married or co-habiting presented increased risk of mental disorder, while those whowere divorced or widowed were more likely to have combinedmentaland physical disorders. Those with mental disorders, or both mental

and physical disorders, were more likely to be in receipt of services atprimary care level prior to suicide (po0.05, po0.01). Last point ofservice contact up to six months before death was significantly morelikely among those with combined mental and physical disorders(p¼0.04), and in the two weeks prior to death among those with onlya mental health disorder (po0.05).

4. Discussion

The current research indicates that whilst suicide preventionefforts typically target the young, the average age of the indivi-duals in this population was 40 years and the rates of suicide werehighest in those aged 20–50 years. The cohort of people who weremost at risk of suicide several decades ago continue to remain atrisk as they grow older. In NI, this is the population who witnessed

Table 3Age group and last service interaction.

Age group % (n) Not known Up to 1 wk 1–2 wks 2–4 wks 1–2 mths 2–4 mths 4–6 mths 6 mths–1 yr 1 yr þ

Up to 19 yrs 52.6% (72) 16.1% (22) 3.6% (5) 5.8% (8) 3.6% (5) 4.4% (6) 4.4% (6) 6.6% (9) 2.9% (4)20–29 yrs 48.9% (180) 12.2% (45) 5.2% (19) 7.9% (29) 8.2% (30) 6.2% (23) 3.8% (14) 4.6% (17) 3% (11)30–39 yrs 49.7% (166) 17.7% (59) 3% (13) 5.7% (19) 6.3% (21) 4.5% (15) 4.8% (16) 3.9% (13) 3.6% (12)40–49 yrs 48.5% (183) 19.9% (75) 5.6% (21) 8.2% (31) 5% (19) 3.5% (13) 2.1% (8) 4.2% (16) 2.9% (11)50–59 yrs 43.7% (107) 23.7% (38) 5.7% (14) 8.2% (20) 6.5% (16) 2.4% (6) 4.1% (10) 2.9% (7) 2.9% (7)60–69 yrs 41.9% (54) 24.8% (32) 6.2% (8) 6.2% (8) 6.2% (8) 7% (9) 3.1% (4) 2.3% (3) 2.3% (3)70þ yrs 45.5% (35) 26.6% (20) 6.5% (5) 11.7% (9) 3.9% (3) 3.9% (3) 1.3% (1) 1.3% (1) 0Total 47.8% (797) 18.7% (311) 5.15% (85) 7.4% (124) 6.1% (102) 4.5% (75) 3.5% (59) 4% (66) 2.9% (48)

Table 4Prior suicidal behavior.

Gender % (n) Single prior attempt 2–4 prior attempts 5þ prior attempts

Female 15.4% (59) 20.2% (77) 21.5% (82)Male 17.3% (223) 11.9% (154) 11.5% (148)

Table 5Prevalence of mental, physical and combined conditions prior to death.

Disorder % (n) Total Men Women χ2

Mental 35.8% (599) 34.8% (452) 39.1% (147) 2.29Physical 11.9% (199) 13.3% (173) 6.9% (26) 11.47nn

Combined 21.8% (365) 19.7% (256) 29.9% (109) 14.62nn

None 30.5% (510) 32.1% (416) 25% (94) 6.88n

n po0.05.nn po0.001.

Table 6Last point of service use and service level prior to death.

Service profile n (%) Total Men Women χ2

Last service use1–2 weeks 19.5% (326) 18.2% (236) 23.9% (90) 6.12n

2 wks-2 months 9.9% (165) 9.7% (126) 10.4% (39) 0.142–6 months 6.8% (113) 7.2% (93) 5.3% (20) 1.586 months–1 year 2.8% (47) 3.3% (43) 1.1% (4) 5.41n

1 year4 3.2% (53) 3.5% (46) 1.9% (7) 2.69None/not known 57.9% (969) 58.1% (753) 57.4% (216) 0.05

Service levelprimary 50.1% (839) 52.6% (682) 41.8% (157) 13.67nn

Secondary 24.3% (406) 23% (298) 28.7% (108) 5.24nn

Tertiary 5.9% (98) 5.2% (67) 8.2% (31) 5.01n

n po0.05.nn po0.01.

S. O'Neill et al. / Journal of Affective Disorders 168 (2014) 466–471468

the years of the conflict when violence was at its peak. Severaltheorists have elucidated an association between exposure toviolence and the enaction of suicidal ideation (Klonsky and May,2014; Joiner, 2005; O'Connor, 2011). The possible link betweensuicide and the conflict is supported with the evidence of the roleof the conflict in elevating the rates of mental disorders (Buntinget al., 2013; Ferry et al., 2013) and the links between conflictrelated trauma and suicidal behaviour (O'Neill et al., 2014). This isin keeping with Tomlinson's (2012) contention that the increasedrates of suicide in NI since the peace agreements are a conse-quence of the decline in social cohesion and social connectednesswhich was characteristic of the conflict. The breakdown in socialconnectedness, along with exposure to violence and high levels ofmental disorders may have promoted the increase in suicide rates.

The highest proportion of suicides was among males andyounger males presented a higher risk relative to females, whodemonstrated higher incidence rates in later years. As such, thisstudy confirms the gender differences in suicide rates and theinternational trends of high rates of suicide among middle-agedmen (Snowcroft, 2013). The increased proportion of suicidesamong older women should not be neglected. Women were morelikely to have a history of suicide attempts, and had more recordedprevious attempts than males. This again supports prior research,however it may also reflect women’s' increased tendency to reportsuicidal behaviour. In keeping with the literature on gender andmental health service use (Kovess-Masfety et al., 2014), femaleswere more likely to engage in help seeking behaviour, to receivereferrals to secondary care, and to have a recorded mental disorderthan males. At 58%, the proportion of those with a recordedmental disorder is relatively low, particularly for males, when weconsider that psychological autopsy studies indicate rates ofmental disorders of over 90% among those who die by suicide(Foster et al., 1997; Cavanagh et al., 2003; Mann et al., 2005).Additionally, males were less likely to receive services beyondprimary care. There are a number of possible explanations for this.There is a wealth of evidence to indicate that men are reluctant to

disclose mental health concerns with their General Practitionerand other health care providers (Snowcroft, 2013). Men were morelikely than women to have a physical health diagnosis only;however their physical conditions may in fact reflect the physicalsymptoms of an undiagnosed affective disorder. Contemporarytheories of suicide view suicidal behaviour as goal directed, as ameans of addressing pain and distress (Klonsky and May, 2014;Joiner, 2005; O'Connor, 2011) and somatic symptoms are a keyfeature of depression. The patterns revealed in this study suggestthat some men attended to the physical or somatic, symptoms ofmental disorders such as depression or anxiety, rather thandisclosing emotional distress or low mood. In addition, men mayhave been unwilling to recognise suicidal ideation as indicative ofa mental health problem meriting disclosure. Further research onmen's experience of mental disorder and the role of physicalsymptoms in depression and suicidal ideation is necessary toobtain an understanding of the barriers to help seeking inthis group.

It may be argued that service contact within the two weeksprior to suicide represents an opportunity for intervention ifsuicidal ideation is assessed by clinicians and ideation reported.However only one in five of those who died by suicide in this study(18.2% and 23.9% for males and females respectively) presented tohealth services during this period of time. In fact these resultsshow that males had a tendency to disengage with services priorto death. Help-seeking should be promoted among those who arevulnerable to suicide and support the need to raise awareness ofthe benefits of disclosing mental health difficulties and suicidalideation. Efforts to reduce the stigma of mental disorder andrecognition of suicidal ideation as indicative of disorder, particu-larly among males, are therefore to be welcomed. The high rates ofsuicide across the age groups and the mean age of death (43 years)reflect current trends and create concern regarding the mentalhealth of this population cohort. The increased likelihood ofservice contact in those aged over 40 years and the high levelsof presentations in those over 70 years may be seen to reflect

Table 7Multinomial logistic regression analyses of risk factors associated with mental and physical disorders.

Outcome 1: mental health disorder Outcome 2: physical health disorder Outcome 3: combined mental and physical health disorders

Risk factor OR 95% CI OR 95% CI OR 95% CI

Femalea 2.2nn 1.4, 3.5 1.1 0.6, 2.0 2.8nn 1.7, 4.4Ageb 1.03nn 1.0, 1.04 1.07nn 1.05, 1.09 1.07nn 1.06, 1.09

Care levelc

None 0.01nn 0.002, 0.04 0.01nn 0.001,0.09 0.02nn 0.01, 0.09Primary 0.1nn 0.01, 0.3 0.4 0.08, 2.4 0.1nn 0.02, 0.4Secondary 0.5 0.1, 2.3 1.3 0.2, 8.2 0.7 0.1, 3.2

Last point of serviced

None 1.3 0.6, 2.7 1.9 0.7, 5.4 1.9 0.7, 5.11–2 weeks 2.6n 1.1, 6.0 2.4 1.7, 7.5 4.4nn 1.5, 12.62 weeks–2 months 1.9 0.8, 4.6 2.9 0.9, 9.4 4.4nn 1.5, 13.02–6 months 2.0 0.8, 5.0 2.5 0.7, 8.0 3.3n 1.0, 10.56 months–1 year 0.4 0.1, 1.2 1.6 0.4, 6.1 1.6 0.5, 5.8

Marital statuse

Not known 0.02 0.004–0.1 0.1nn 0.01, 0.5 0.01nn 0.001, 0.1Married/cohabiting 0.6n 0.4, 0.9 0.8 0.5, 1.3 0.7 0.5, 1.1Separated 0.9 0.5, 1.6 0.5 0.3, 3.6 0.8 0.4, 1.6Divorced/widowed 1.6 0.7, 3.8 1.3 0.5, 3.6 2.4n 1.0, 5.5

OR¼Odds ratio; CI¼Confidence interval.a Reference group¼male.b Age centred from mean¼40.3 years.c Reference group¼Tertiary level care.d Reference group¼more than 1 year.e Reference group¼Single.n po0.05.nn po0.01.

S. O'Neill et al. / Journal of Affective Disorders 168 (2014) 466–471 469

increased opportunities for intervention. However when weexamine the reasons for last contact, we again find that malesare less likely to report mental health concerns and are more likelyto seek treatment for physical health conditions.

The study shows that primary care remains the most commonhealth service used in the period prior to death by suicide. Thisfinding concurs with other studies of service used prior to deathby suicide (Luoma et al., 2002; Pearson et al., 2009). Those withonly mental disorders were more than twice as likely to havemade primary care contact in the two weeks preceding death,while those with combined disorders were over four times morelikely to have utilised services during this time. Being in contactwith health services for physical illnesses in the last three monthswas associated with an increased likelihood of receiving help formental-health problems. Primary care therefore continues to playan important role in suicide prevention and these findings addweight to the argument for enhanced screening for suicidalideation in this setting. It remains important to strengthenclinicians' knowledge of specific manifestations of suicidal idea-tion which may or may not be identified by service users as relatedto mental health. The capacity of primary care services to provideservices to people who are suicidal also requires examination,particularly since many deaths by suicide occur outside of GeneralPractitioner surgery opening hours.

5. Limitations

This was the first time that coronial flies were used to assessthe characteristics and service use history of those who died bysuicide in NI. The findings offer a unique insight into this popula-tion and it is important that they inform suicide preventionpolicies and service delivery. In common with the rest of the UK,the data on suicide in NI are subject to issues of data reliability,most prominent perhaps the delay between death and registra-tion. Also problematic is the issue of determining cause of death;this ambiguity was addressed in 1968 with the inclusion of an‘undetermined’ category whereby suicide is implied. In this study,both deaths categorised as suicide and undetermined deaths wereexamined in order to obtain a more accurate indicator of thecharacteristics of those who die by suicide. However there invari-ably remains a proportion of deaths by suicide which are notrecorded in either category (Tomlinson, 2012). The data on maritalstatus and mental and physical disorders were based on informa-tion recorded by police officers from relatives and other infor-mants. Information on medical records was inconsistent and insome cases medical records were not available. As such, this studycarries the risk of an under estimation of health disorders, relatedrisk factors and history of service use. This key limitation high-lights the need for improvement in the consistency of datacollected following a death by suicide and where possible, clearprotocols for the collection of information following a death byprobable suicide. In NI, as with other parts of the UK, out-of-hoursuicide prevention initiatives are also provided by organisationssuch as “Lifeline” (a 24-h crisis helpline and counselling service forpeople who are suicidal) and other suicide helplines, such as theSamaritans. Contact with these organisations was not routinelyincluded in the records of the deceased and as such we have noconsistent information on levels of contact with these servicesprior to death.

Finally, the time period included in the study has seen a trendof service reform and a suicide prevention strategy (DHSPSSNI,2006) and this may have impacted upon the service use profiles ofthose who died by suicide in more recent years.

6. Conclusion

The high levels of mental disorders in NI (Bunting et al., 2013),along with rising suicide rates in recent years (Snowcroft, 2013)and associations between conflict related trauma and suicidalbehaviour (O'Neill et al., 2014), indicate a need for heightenedconsideration of the ways in which people with mental disordersare identified and treated in NI. Importantly, the mental disordervariable refers to having been identified as having a mentaldisorder excluding people with an undiagnosed disorder. The ratesof undiagnosed disorders in this population are therefore a causeof concern. The low levels of recorded mental disorders anddisengagement from services among men in this study may be aconsequence of a reluctance to disclose mental health difficultiesor may reflect an emphasis on men's somatic symptoms. However,it also serves as a reminder that suicide is a behaviour, whichoccurs as a consequence of a combination of factors, in addition tomental disorders. These include adverse events, as well as otherindividual and environmental variables (O'Connor, 2011). It istherefore important that the effects of adverse life events, socialfactors and individual differences on suicide rates and suicidalbehaviour are continued to be examined to gain a more completeunderstanding of this behaviour. There remain uncertainties aboutthe role of mental disorders in the development of suicidalbehaviours; however access to relevant health services representsa crucial stepping stone towards timely identification and provi-sion of adequate treatment for persons at risk for suicide. Thisstudy presents, for the first time, an analysis of service use profilesof those who die by suicide in NI. The results identify some uniquecharacteristics of help-seeking behaviours among gender and agegroups with regard to service use which may be used to informinterventions and direct resource allocation in relation to suicideprevention.

Conflict of interestNo conflict declared.

Role of funding sourceThis study was funded by the Northern Ireland Public Health Agency, Research

and Development Division (COM/4027/08). The funders had no role in the researchor in the preparation of this manuscript for publication.

AcknowledgementThe authors acknowledge the Northern Ireland Coroner's Service and senior

Coroner Mr John Lecky for their support in providing the data for this study.

References

Appleby, L., Kapur, N., Shaw, J., Hunt, I., While, D., Windfuhr, K., Williams, A., 2013.National Confidential Enquiry into Suicide and Homicide by People with MentalIllness. University of Manchester, Manchester.

Bunting, B., Ferry, F., Murphy, S.D., O’Neill, S., Bolton, D., 2013. Trauma associatedwith civil conflict and posttraumatic stress disorder: evidence from theNorthern Ireland study of health and stress. J. Trauma. Stress 26, 134–141.

Carroll, R., Hawton, K., Kapur, N., Bennewith, O., Gunnell, D., 2012. Impact of thegrowing use of narrative verdicts by coroners on geographic variations insuicide: analysis of coroners' inquest data. J. Public Health 34, 447–453.

Cavanagh, J.T., Carson, A.J., Sharpe, M., Lawrie, S.M., 2003. Psychological autopsystudies of suicide: a systematic review. Psychol. Med. 33, 395–405.

Department of Health, Social Services and Public Safety, Northern Ireland. 2006.Protect life: a shared vision. DHSPSSNI, Belfast.

Ferry, F., Bunting, B., Murphy, S., O'Neill, S., Stein, D., Koenen, K., 2013. Traumaticevents and their relative PTSD burden in Northern Ireland: a consideration ofthe impact of the “Troubles”. Soc. Psychiatry Psychiatr. Epidemiol. 43, 435–446.

Foster, T., Gillespie, K., McClelland, R., 1997. Mental disorders and suicide inNorthern Ireland. Br. J. Psychiatry 170, 447–452.

Gunnell, D., Hawton, K., Kapur, N., 2011. Coroners' verdicts and suicide statistics inEngland and Wales. Increasing use of narrative verdicts raises serious concerns.BMJ, 343.

S. O'Neill et al. / Journal of Affective Disorders 168 (2014) 466–471470

Joiner, T.E., 2005. Why People Die by Suicide. Harvard University Press, Cambridge,MA.

Klonsky, E.D., May, A.M., 2014. Differentiating suicide attempters from suicide ideators:a critical frontier for suicidology research. Suicide Life-Threat. Behav. 1, 1–5.

Knapp, M., McDaid, M., Parsonage, M., 2011. Mental Health Promotion and MentalIllness Prevention: The Economic Case. Department of Health, London.

Kovess-Masfety, V., Boyd, A., van de Velde, S., de Graaf, R., Villagut, G., Haro, J.M.,Florescu, S., O'Neill, S., Weinberg, L., Alonso, J., 2014. the EU-WMH Investigators,2014. Are there gender differences in service use for mental disorders acrosscountries in the European union? Results from the EU –World Mental HealthSurvey. J. Epidemiol. Community Health (Published online 10.03.14), http://dx.doi.org/10.1136/jech-2013-202962.

Luoma, B., Martin, E., Pearsson, J., 2002. Contact with mental health and primarycare providers before suicide: a review of the evidence. Am. J. Psychiatry 159,909–916.

Mann, J.J., Apter, A., Bertolote, J., Beautrais, A., Currier, D., Haas, A., Hegerl, U.,Lonngvist, J., Malone, K., Marusic, A., Mehlum, L., Patton, G., Phillips, M.,Rutz, W., Rihmer, Z., Schmidtke, A., Shaffer, D., Silverman, M., Takahashi, Y., Varnik,A., Wasserman, D., Yip, P., Hendin, H., 2005. Suicide prevention strategies: asystematic review. JAMA 294, 2064–2074.

Northern Ireland Statistics and Research Agency Suicide Deaths, 2014. ⟨http://www.nisra.gov.uk/demography/default.asp31.htm⟩ (accessed 21.04.14).

O'Connor, R.C., Rasmussen, S., Hawton, K., 2014. Adolescent self harm: a schoolbased study in Northern Ireland. J. Affect. Disord. 159, 46–52.

O’Connor, R.C., 2011. The integrated motivational-volitional model of suicidalbehavior. Crisis 32, 295–298.

O'Neill, S., Ferry, F., Murphy, S.D., Corry, C.V., Bolton, B., Devine, B., Ennis, E.,Bunting, B.P., 2014. Patterns of suicidal ideation and behavior in NorthernIreland and associations with conflict related trauma. PLoS One, http://dx.doi.org/10.1371/journal.pone.0091532.

Pearson, A., Saini, P., Da Cruz, D., Miles, C., While, D., Swinson, N., Williams, A.,Shaw, J., Appleby, L., Kapur, N., 2009. Primary care contact prior to suicide inindividuals with mental illness. Br. J. Gen. Pract. 59, 825–832.

Power, K., Davies, C., Swanson, V., Gordon, D., Carter, H., 1997. Case-control study ofGP attendance rates by suicide cases with or without a psychiatric history. Br. J.Gen. Pract. 47, 211–215.

Singhal, A., Ross, J., Seminog, O., Hawton, K., Goldacre, M.J., 2014. Risk of self-harmand suicide in people with specific psychiatric and physical disorders: compar-isons between disorders using English national record linkage. J. R. Soc. Med.,http://dx.doi.org/10.1177/0141076814522033.

Snowcroft E., 2013. Samaritans Suicide Statistics Report 2013. Data for 2009–2011.Samaritans. ⟨http://www.samaritans.org/sites/default/files/kcfinder/files/research/Samaritans%20Suicide%20Statistics%20Report%202013.pdf⟩ (accessed 21.04.14).

Stark, C.R., Vaughan, S., Huc, S., O’Neill, N., 2012. Service contacts prior to death inpeople dying by suicide in the Scottish Highlands. Rural Remote Health 12,1876.

Tomlinson, M., 2012. War, peace and suicide: the case of Northern Ireland. Int.Sociol. 27, 464.

S. O'Neill et al. / Journal of Affective Disorders 168 (2014) 466–471 471