General Practitioners' Accounts of Patients Who Have Self-Harmed: A Qualitative, Observational Study.

BACKGROUND: The relationship between self-harm and suicide is contested. Self-harm is simultaneously understood to be largely nonsuicidal but to increase risk of future suicide. Little is known about how self-harm is conceptualized by general practitioners (GPs) and particularly how they assess the suicide risk of patients who have self-harmed. AIMS: The study aimed to explore how GPs respond to patients who had self-harmed. In this paper we analyze GPs' accounts of the relationship between self-harm, suicide, and suicide risk assessment.
METHOD: Thirty semi-structured interviews were held with GPs working in different areas of Scotland. Verbatim transcripts were analyzed thematically.
RESULTS: GPs provided diverse accounts of the relationship between self-harm and suicide. Some maintained that self-harm and suicide were distinct and that risk assessment was a matter of asking the right questions. Others suggested a complex inter-relationship between self-harm and suicide; for these GPs, assessment was seen as more subjective. In part, these differences appeared to reflect the socioeconomic contexts in which the GPs worked.
CONCLUSION: There are different conceptualizations of the relationship between self-harm, suicide, and the assessment of suicide risk among GPs. These need to be taken into account when planning training and service development.

Nonfatal self-harm and suicide are generally understood to be related, but distinct, behaviors. While many people who have self-harmed deny any intent to die (Adler & Adler, 2011), there is considerable evidence that self-harm is a major risk factor for subsequent completed suicide (Hawton, Zahl, & Weatherall, 2003). This presents a challenge for frontline health-care professionals who see patients with a wide range of self-harming behavior and must assess risk of subsequent suicide in each case.

In addition to increased risk of suicide, individuals who have self-harmed appear likely to be at greater risk of a range of other clinical and social challenges, including substance misuse and mental health problems (Hasking, Momeni, Swannell, & Chia, 2008). The findings of a recent longitudinal study of a general population sample of young adults suggests that the association between self-harm and such adverse outcomes is stronger where self-harm has been identified as suicidal in nature (Mars et al., 2014).

The relationship between self-harm and suicidality is highly contested among researchers. While some argue that it is possible to differentiate between self-harming acts that are suicidal and those that are not (Plener & Fegert, 2012), others point to the difficulty of making meaningful distinctions (Kapur, Cooper, O'Connor, & Hawton, 2013). The inclusion of nonsuicidal self-injury (NSSI) as a proposed diagnosis in the latest version of the American Psychiatric Association's Diagnostic and Statistical Manual has triggered a heated debate (American Psychiatric Association, 2013; De Leo, 2011; Gilman, 2013; Kapur et al., 2013). Published commentary on this issue highlights enduring differences between European and US perspectives (Arensman & Keeley, 2012; Claes & Vandereycken, 2007). In the UK the most widely used definition of self-harm is "self-injury or self-poisoning irrespective of the apparent purpose of the act" (National Institute for Clinical Excellence, 2011). However, there is evidence that, among lay groups in the UK, self-harm is often understood to refer to self-cutting that is accompanied by no or only minimal suicidality (Scourfield, Roen, & McDermott, 2011).

Some studies have found differences in stated suicidal ideation between young people who have taken overdoses and those who have engaged in self-cutting (Rodham, Hawton, & Evans, 2004). However, the relationship between self-harm and suicide is not straightforwardly related to the method used (Fortune, 2006). Whitlock and Knox (2007) found that rates of suicidal ideation were higher among those who had engaged in self-injurious behavior than among those who had taken overdoses in a community sample of college students. They argued that this finding underlined the importance of ongoing suicide risk assessment for young people who self-harm using any method. Further, Bergen et al. (2012), conducting research on hospital-treated self-harm, found that self-cutting was more closely related to completed suicide than self-poisoning was.

Evidence from psychological autopsy investigations suggests that a history of self-harm is one of the strongest risk factors for suicide, present in about 40% of cases (Cavanagh, Carson, Sharpe, & Lawrie, 2003). However, there is considerable variation in the prevalence of previous self-harm across studies (the range in the Cavanagh et al. review is 16–68%), reflecting heterogeneity in the samples being investigated (e.g., female nurses, Hawton et al., 2002; individuals not engaged with mental health services, Owens, Booth, Briscoe, Lawrence, & Lloyd, 2003) and limitations of the methodology (Pouliot & De Leo, 2006). The complex and sometimes contradictory nature of research evidence regarding the relationship between self-harm and suicide means that debates are unlikely to be resolved soon. This raises questions, however, as to how such complexities should be managed in clinical practice, particularly in primary care, where the range of self-harm that is treated may be more diverse and less clearly life-threatening than that seen in secondary care.

In the UK, rates of hospital-treated self-harm and suicide vary according to socioeconomic context and sociodemographic characteristics. People living in areas of socioeconomic deprivation have a higher likelihood of both dying by suicide and being treated in hospital for self-harm (Mok et al., 2012; Platt, 2011; Redley, 2003). Little is known about self-harm that is not treated in hospital, with most community-based research focusing on adolescent or college populations. Some studies indicate that there is little to no variation in reported self-harm among young people living in different socioeconomic contexts (Ross & Heath, 2002). Others have found that those living in areas of deprivation (Jablonska, Lindberg, Lindblad, & Hjern, 2009) and, in some areas of the US, those from African American groups (Gratz, 2012) are more likely to report self-harm. Studies of self-harm treatment in primary care are limited; consequently, the frequency and features of self-harm in such settings are relatively unknown.

Although there is a dearth of research in primary care, this setting would appear to offer clear opportunities for contributing to suicide prevention (Appleby, Amos, Doyle, Tomenson, & Woodman, 1996; Cole-King & Lepping, 2010; Pearson et al., 2009; Saini et al., 2010). About half of patients who go on to die by suicide visit their general practitioner (GP) in the month leading up to their death (Luoma, Martin, & Pearson, 2002; Pearson et al., 2009). Further, following hospital treatment for self-harm, patients in the UK are usually referred back to their GP for follow-up (Mitchell, Kingdon, & Cross, 2005). Outcomes relating to a primary care intervention for patients who have engaged in suicidal self-harm have been explored (Bennewith et al., 2002), while other studies have examined GP responses to suicidal self-harm using qualitative (Kendall & Wiles, 2010) and quantitative (Rothes, Henriques, Leal, & Lemos, 2014) approaches.

To date, there has been no research on GPs' responses to self-harm as defined in UK clinical guidelines, that is, including cases of self-harm that are not treated in hospital and are not deemed suicidal. This study is the first – to our knowledge – to explore GPs' accounts of self-harm in general, avoiding a narrow focus on suicidal self-harm. The aims of the study were: to explore how GPs talked about responding to and managing patients who had self-harmed; to identify potential gaps in GPs training; and to assess the feasibility of developing a multifaceted training intervention to support GPs in responding to self-harm in primary care. We focus here on GPs' accounts of the relationship between self-harm and suicide and approaches to carrying out suicide risk assessments on patients who had self-harmed. (A separate paper will address accounts of providing care for patients who had self-harmed; the present paper should not be taken as evidence that GPs talked only about managing suicide risk among these patients.)

Method

A narrative-informed, qualitative approach (Riessman, 2008) was adopted, in order to explore in depth how GPs talked about patients who had self-harmed, including how they addressed suicide risk. Through this we sought to examine GPs' understandings of self-harm, and reflect upon how the meanings attached to self-harm, including the relationship with suicide, might affect clinical practice.

Participants were GPs recruited from two health boards in Scotland. We obtained a sample of interviewees working in practices from diverse geographic and socioeconomic areas. Recruitment was in two stages: an initial mailing via the Scottish Primary Care Research Network, followed by a targeted approach, using personal networks to recruit GPs working in practices located in areas of socioeconomic deprivation. We did not selectively recruit participants based on particular experience of self-harm or psychiatry either in training or practice. An overview of the characteristics of the final sample of 30 GPs is shown in Table 1. The socioeconomic characteristics of the practice were calculated using the Scottish Index of Multiple Deprivation. Those classed as deprived were located in areas in deciles 1–3; middle-income practices were in deciles 4–6; affluent practices in deciles 7–10. Rural/urban practices were classified using the Scottish Government sixfold urban/rural classification.

Table 1

Overview of the characteristics of the final sample of 30 GPs

Characteristics	Number of participants
Practitioner gender
Male	16
Female	14
Geography of practice area
Urban	21
Rural	9
Socioeconomic status of area
Deprived	12
Middle-income	3
Affluent	13
Mixed	2
Total sample	30

All participants gave informed, written consent. Participants were reimbursed for practice time spent on the research study, and were provided with a package of educational materials for use toward continuing professional development at the end of the study period.

GPs participated in a semistructured interview with one of the authors (King). They were offered either telephone or face-to-face interviews, with all but one opting for a telephone interview. No particular reason was provided for preferring a face-to-face interview, and the interview did not differ substantially from those conducted via telephone. During the interview, and leading from our narrative approach, participants were invited to discuss two or more recent cases (suitably anonymized) where they had treated a patient who had self-harmed. This approach allowed us to generate rich narratives from GPs regarding the types of patients they understood to have self-harmed, along with their accounts of treating such patients. Subsequently, the following topics were explored: understandings of self-harm; assessment of suicide risk in the context of self-harm; and training and education needs and experiences. The topic guide was developed directly from the research aims. Interviews were planned to last 30 min and ranged from 20 to 40 min.

Interviews were recorded, transcribed verbatim, and entered into the NVivo 10 qualitative data analysis package (NVivo, version 10) to facilitate data management and content coding. Analysis was thematic, informed by narrative approaches that sought to avoid fracturing participants' responses and retained a focus on each GP participant as a case. Chandler carried out deductive coding, based on the interview schedule, followed by inductive, open coding to identify common themes in the data (Hennink, Hutter, & Bailey, 2011; Spencer, Ritchie, & O'Connor, 2005). Table 2 presents an overview of the deductive codes, along with the inductive subcodes within the code on self-harm and suicide, which are the focus of this paper. Proposed themes were shared, discussed, and agreed on within the research team. In relation to the coding presented in this paper, theoretical data saturation was achieved. The present paper is based on analysis of a deductive code containing all talk about the relationship between self-harm and suicide, and the assessment of suicide risk in the context of self-harm.

Table 2

Overview of deductive codes (bold) with inductive codes within self-harm and suicide

Reflecting on self-harm
Self-harm and suicide
Relationship between self-harm and suicide
– Distinct relationship – self-harm indicates low risk
– Complex relationship – self-harm may indicate high risk, relationship difficult to untangle
Assessing suicide risk in the context of self-harm
Straightforward
Challenging
Just ask them
Identify risk/protective factors
Best practice
Training needs and experience

Results

The Relationship Between Self-Harm and Suicide

When asked to reflect on the relationship between self-harm and suicide, GPs' accounts tended to embody one of two understandings: (a) that there was a very weak relationship between the practices; and (b) that there was a close and complex relationship between the practices. Some GPs' accounts introduced elements of each of these understandings.

Self-Harm and Suicide as Distinct

Some GPs portrayed self-harm and attempted suicide as distinct in several ways, addressing differences with intent, methods used, and help-seeking behavior. GPs sometimes identified a theoretical link between self-harm and risk of completing suicide; however, this formal knowledge was contrasted with practice experience of treating patients who had self-harmed as a way of "releasing" problematic emotions:

Their [people who have self-harmed] risk of actual suicide is more than the general population, as far as I can remember, going back to teaching days […] most people don't want to kill themselves. […] this is just, again, an anecdotal – cases we've looked after, that most people don't want to kill themselves. That it's a sense of frustration and danger in themselves, and it's a form of releasing anger. (GP5, F, mixed socioeconomic area)

Thus, unlike attempted suicide, which entailed an intense wish to die, self-harm was believed to be carried out for other, different, reasons, in particular tension release:

It seems like there's two different sides to the coin: those that it's sort of [a] response to stress and that's how they deal with their anxiety and they get some, you know, instant relief from their anxieties and stresses with that, and then you've got the other ones where it's maybe a more serious sort of cry for help and it's not something that they've done on a regular basis. (GP7, F, rural, affluent area)

GP7 suggests that there are differences between self-harm and suicide, both in terms of intent (anxiety relief vs. a serious cry for help) and frequency (nonsuicidal self-harm would be likely to recur more regularly than a suicide attempt). Framing self-harm and suicide in this manner led to a perception that certain methods of self-harm were especially likely to be associated with low suicidality, in particular self-cutting: "The people cutting their forearms and things, they're definitely not trying to kill themselves I don't think" (GP15, F, rural, deprived area).

The phrase cry for help was often used in GPs' accounts, although the meaning ascribed to this appeared to vary. Thus, in the account of GP7, the cry for help indicated a serious act (attempted suicide); other GPs associated the cry for help with nonfatal self-harm, which posed a lower risk of eventual suicide:

In my experience it seems like the majority of self-harmers didn't seem to have that high a risk of completing a suicide. In my experience most of them are fairly low risk […] A lot of them were cry for helps. (GP10, M, rural, affluent area)

GPs used the term cry for help to describe both the perceived intention of an act of self-harm (communication of distress) and also the help-seeking behavior of the patient. Some of these accounts suggested that those patients who were seriously suicidal would be less likely to seek (or cry for) help. By contrast, patients whose actions were characterized as self-harm were framed as "seeking help" and therefore "not really trying to kill themselves" (GP6, M, urban, middle-income area).

It's a very gray area […] people who are really suicidal, you often don't find out, because they just go and do it […] the population I see is enormously skewed towards people who have a lower degree of suicidality in it, if you like, are seeking help from me […] they're using these attempts at self-harm as a way of expressing how bad they feel. (GP20, M, urban, affluent area)

It's a classic cliché that self-harm is a cry for help […] whereas true suicide […] folk who kill themselves the chances are they are going to do it, and the folk who are really serious about doing it will do it, and you won't know about it. (GP13, M, semi-urban, affluent area)

While GPs differed in their use of the term cry for help, particularly whether this was infused with positive or negative connotations, in most cases it served to differentiate self-harm from suicide.

Self-Harm and Suicide as Related

Unlike the accounts above, which constructed self-harm and suicide as distinct practices, other GPs emphasized the difficulty of distinguishing meaningfully between self-harm and suicide. One way in which this was accomplished was through accounts that framed suicide as an ongoing concern when treating patients who had self-harmed:

I think it's always a fear that's in the background for us. (GP4, F, semi-urban, deprived area)

My feeling would be that most people who are self-harming have at some point had more suicidal thoughts. (GP19, M, mixed socioeconomic area)

When GPs talked about self-harm and suicide as related, reference was often made to patients' difficult lives. GPs mentioned the adverse structural and interpersonal conditions in which many of their patients lived, emphasizing high levels of poverty and financial uncertainty, drug or alcohol dependence, lack of stable accommodation, and poor or abusive relationships. In the context of such challenges, GPs suggested it was particularly hard to separate self-harm from suicidality.

I think it's very difficult, actually, in my patients, because I think there's just a gross ambivalence about being alive. (GP28, M, urban, deprived area)

I think many of them have a wish not to be there. You know, they have passive suicidal ideation; they just wish they didn't exist anymore. (GP29, F, urban, deprived area)

GPs providing these accounts challenged interview questions that asked them to consider self-harm and suicidality as distinct.

Researcher: How often in your experience is self-harm accompanied by some degree of suicidality? […]

GP: I'm sorry not to answer your question very helpfully, but that's the trouble. There are degrees of suicidality and often teasing out whether somebody who's referring to suicidal thoughts of one kind or another is actually meaning to self-harm with no actual intention to kill themselves, or they are truly meaning to kill themselves. That's not particularly easy. (GP18, M, semi-urban, deprived practice)

Such accounts questioned whether concepts of suicidality or suicidal ideation were useful when treating patients who had self-harmed, because the issue of intent was often unclear (including to the patients themselves) and the separation between self-harm and suicide was indistinct. The majority of GPs providing these accounts were working in practices located in socioeconomically deprived areas, or had significant experience working with marginalized patient groups. There were exceptions, however. For instance, GP22 (F, urban, affluent area) suggested that one of her patients was self-harming: "Probably more a cry for help but I think she is so vulnerable that she could make mistakes, a mistake easily enough to kill herself […] we always live with uncertainty."

Establishing the presence or absence of suicidal intent among patients with difficult lives was described as problematic. GPs noted that such patients might live with suicidal thoughts over long periods and/or be at high risk of accidental self-inflicted death. In combination, these factors undermined any attempt to distinguish clearly between suicidal and nonsuicidal self-harm.

The Challenges of Suicide Risk Assessment Among Patients Who Had Self-Harmed

All GPs were asked how they assessed suicide risk in patients who had self-harmed. In contrast to their responses to questions about the relationship between self-harm and suicide, GPs' accounts in relation to this issue were more similar. The majority emphasized the difficulty of assessing suicide risk among patients who self-harmed, although different explanations for this difficulty were given.

Challenges: Time Constraints and Establishing Intent

Time constraints were frequently identified as presenting a barrier in assessing suicide risk:

In a ten-minute consultation, under enormous working pressure, yes, [assessing suicide risk is] very difficult actually. (GP26, M, urban, deprived area)

Indeed, time constraints were described more generally as posing a challenge when treating patients who had self-harmed and who were therefore framed as being complex or difficult cases. GPs' accounts suggested the adoption of different approaches to managing time constraints, which may have been shaped by local contexts and resources.

The problem of assessing intent among patients who self-harmed was raised, with some GPs highlighting the limitations of asking patients direct questions:

So, it's easy for the ones who are willing to speak about it, but it's very difficult for the ones who are really wanting to do it […]. In one [patient] there was contact with a complaint of depression, but they had basically said that they weren't suicidal but unfortunately they were. (GP12, M, urban, middle-income area)

As with GP12, some of these accounts drew on understandings of suicide as a practice that was generally difficult to identify and prevent, since people who "really want to do it" may not disclose their plans.

GPs working with marginalized, disadvantaged patient groups were particularly like to suggest that assessing suicide risk was an inherently imprecise endeavor, since people's lives were volatile and dangerous.

You can never be confident I guess with a mental health assessment, about when someone feels like they are genuinely at acute risk of suicide or when they're at risk of self-harm and possible death through misadventure. (GP10, F, urban, deprived area)

Again, this type of account emphasized the limitations of asking patients about suicidal thoughts, since absence of such thoughts may not necessarily preclude future self-inflicted death in the context of inherently risky living.

Challenges: Carrying Out Suicide Risk Assessments

While GPs often noted the difficulty and limitations of assessing suicide risk, they nevertheless provided accounts of how they carried out assessments. These narratives emphasized the importance of asking patients about suicidal thoughts and plans, but also addressed wider risk and protective factors, such as social isolation and drug and alcohol use, as well as relying on what was often described as gut feeling (a mixture of intuition and experiential learning).

Yeah, I know, it's not easy. When you think about it, it's … I think I just sort of go with my gut feeling. I think you sort of get a feeling about a person when you meet them as to whether it's a cry for help, is it just a stress response, it is something more serious. (GP7, F, rural, affluent area)

To be honest, I tend to go more on … well, if I know a patient, then I would go more on my gut feeling […]. I don't think always because people have suicidal ideas or even suicide intent… I'm not always sure that we need to intervene, and I think a lot of what I try and do is to reflect back to the patient in terms of them taking responsibility […]. So in terms of assessment, I don't use a risk assessment tool or anything, and I kind of weigh what they're actually saying, in terms of what they're planning and what's their history, so I guess I do take that into consideration, and their social situation as well. (GP27, M, urban, deprived area)

While GP7 and GP27 both referred to using gut feeling to guide suicide risk assessments, there were differences in their accounts. GP7 indicated a preference for referring patients who self-harmed to specialists, as she felt that carrying out suicide risk assessments was not well-supported in primary care. By contrast, GP27 provides a more assured account that suggests a greater level of comfort in responding to patients who self-harm and who may experience continuing suicidality. Further, the account of GP7 indicated a view that self-harm and suicide were distinct, while GP27 emphasized the difficulty of making such distinctions.

GPs' accounts of assessing suicide risk among patients who self-harmed were diverse. Some, such as GP7, indicated that the difficulty lay in a lack of specialist knowledge to ascertain whether self-harm was serious (suicidal) or a cry for help (nonsuicidal); such accounts were based on an understanding of self-harm and suicide as distinct. Others, such as GP12, highlighted that patients may not be able, or feel able, to disclose suicidality even when present. Again, these accounts tended to assume that suicide and self-harm were distinct practices. By contrast, others suggested suicide risk assessment was difficult because of the close and complex relationship between self-harm and suicide. GP27 noted that intention was not necessarily the most important factor in understanding completed suicide among disadvantaged patient groups, where risk of death in general was perceived as heightened, and disclosure of suicidality pervasive.

Straightforward Accounts of Risk Assessment

A minority of GPs provided confident, assured accounts of carrying out suicide risk assessments.

How easy it is to assess risk? I don't think it's difficult to assess risk. I've been a GP for over 20 years, and I've done a bit of psychiatry as well, so I don't think it's a too difficult thing to do. (GP16, M, urban, affluent area)

GP16 emphasized his comfort and capability in treating patients who had self-harmed, and in assessing suicide risk. GPs providing such accounts highlighted the importance of asking direct questions about suicidality to patients who had self-harmed:

I think a lot of the time it [assessing suicide risk] is relatively straightforward if you just ask them the right questions and always distract them away from the self-harm bit and talk about normal things […] you have to be direct to them about killing themselves. (GP2, M, urban, affluent area)

GP2 highlighted the importance of getting a sense of patients' wider life circumstances, using these, along with direct questions about suicidal intent, to build up a picture of suicide risk. These accounts did not necessarily downplay the complexity of assessing suicide risk, but nonetheless indicated a greater level of comfort, and confidence, in doing so. The context in which these accounts were provided is significant here. GPs taking part in the study were opening themselves up to potential or perceived critique, and not all participants may have been comfortable discussing uncertainty.

Descriptions of suicide risk assessment that focused on asking about intent may have been limited by being grounded in an understanding of self-harm and suicide as distinct practices. If a patient referred to self-harm as a form of coping with emotions or tension release, and denied any wish to die, suicide risk was interpreted as low. However, these descriptions of straightforward suicide risk assessment sit uneasily with the accounts provided by other GPs, which problematized the role of intent when assessing suicide risk.

Discussion

Our research suggests that GPs have diverse understandings of the relationship between self-harm and suicide, paralleling the plurality of views on this topic in other disciplines (Arensman & Keeley, 2012; Gilman, 2013; Kapur et al., 2013). These findings indicate the importance of attending to GPs' working definitions of suicide and self-harm, and point to the potential limitations of previous work that has focused narrowly on suicidal self-harm (Bennewith et al., 2002). GPs may have very different opinions on what constitutes suicidal self-harm, or indeed whether it is practical to make distinctions between suicidal and nonsuicidal self-harm. Understandings are likely to be shaped in part by different practice contexts and patient characteristics.

Defining Self-Harm and Suicide

As well as demonstrating that defining self-harm continues to be a challenge (Chandler, Myers, & Platt, 2011), GPs' accounts further unsettle attempts to define suicidality. Is it is a facet of personality (trait) that is found to greater or lesser degree in each individual; a transient state that fluctuates according to external circumstances and context; or a post hoc description of someone who goes on to die by suicide? Our findings resonate with work on the sociological construction of suicide, in problematizing the process whereby deaths come to be understood as suicides (Atkinson, 1978; Timmermans, 2005). However, rather than debating whether a death was a true suicide, GPs in our sample were engaged in deliberating about the extent to which self-harming patients' practice was truly suicidal.

These discussions reflect wider debates about the categorization of self-harm: as deliberate self-harm, nonsuicidal self-injury, a psychiatric diagnosis, a symptom of distress, or a sign of a difficult patient. Crucially, our analysis indicates variation in understanding of the relationship between self-harm and suicide, and the consequent impact on practice in the primary care setting.

Practice Context and Suicide Risk Assessments Among Patients Who Self-Harm

GPs' accounts of treating patients who self-harm, and especially of addressing suicide risk assessments with high-risk groups of patients, highlight a potential challenge for current approaches to responding to self-harm in primary care. The question of intent is, for instance, central to some proposed treatment guidelines for patients in general practice who self-harm. Thus, Cole-King and colleagues suggest that establishing whether self-harm is oriented toward suicide or the relief of emotional pain should be the "first priority" (Cole-King, Green, Wadman, Peake-Jones, & Gask, 2011, p. 283). This approach reflects the accounts of many of the GPs in our sample, who similarly indicated a focus on distinguishing between nonsuicidal self-harm and self-harm with suicidal intention. However, other GPs highlighted significant problems with ascertaining intent, particularly when treating high-risk populations who have a generally higher risk of premature death and where the presence or absence of suicidal intent may be unclear.

It may be significant that GPs working in more deprived, disadvantaged areas appeared more likely to describe suicidal self-harm and nonsuicidal self-harm as intertwined, fluid, and unstable categories, thus making suicide risk assessments especially difficult. By contrast, GPs working in areas that were more rural or affluent tended to discuss suicidal self-harm and nonsuicidal self-harm as distinct, separate practices, characterized by very different methods and intent. It is likely that these differences are rooted in the socioeconomic patterning of rates of both self-harm and suicide (Gunnell, Peters, Kammerling, & Brooks, 1995; Mok et al., 2012), thus highlighting the importance of context in shaping GPs' experience with, and interpretation of, self-harming patients.

Limitations

This was a study of 30 GPs' accounts of treating patients who had self-harmed in two regions of Scotland. It thus carries risks of insufficient sampling and of over-generalization. We addressed these by: (a) purposively sampling from very diverse practices within these regions and ensuring participants varied in age, gender, and experience; as with all such studies, participants may have had a particular interest in psychiatry or suicide; however, interviewees reported a range of experiences and levels of interest in these topics; (b) conducting in-depth analysis of the GPs' accounts; and (c) obtaining data saturation on several key themes. The finding that GPs differ substantially in the way in which they conceptualize associations of self-harm and suicide occurred independently of context, so is likely to be generalizable. Our cautious proposal that the differences in accounts may relate to socioeconomic setting may be more sensitive to context and certainly warrants further investigation in order to confirm or refute this suggestion.

Our research used a fairly blunt and imprecise measure of socioeconomic context (matching the postcode of the practice with the Scottish Index of Multiple Deprivation). Future research should adopt a more sensitive measure that takes more account of the socioeconomic characteristics of the patient population, rather than the location of the practice itself.

Conclusions

GPs in our sample understood self-harm in different ways, reflecting definitional inconsistency and uncertainty in the academic literature. GPs varied in their account of the relationship between self-harm and suicide and in how they described suicide risk assessment. Some patterns emerged in our findings. In particular, GPs who provided accounts of self-harm and suicide as related in complex ways also tended to frame suicide risk assessment as a challenging, continuing process. GPs providing such accounts were more likely to describe working in practices that served populations with high levels of social isolation and economic deprivation. On the basis of these findings, we suggest that there is a clear need for enhanced and accessible support, training, and education for GPs regarding the assessment and management of self-harm and suicidality. Such support, which could be provided as part of continuing professional development, should be responsive to GPs' practice experience, as this appears to shape attitudes toward, and views about, the nature of self-harm, how it relates to suicide, and the role of general practice in contributing to suicide prevention.