Risk of re-attempts and suicide death after a suicide attempt: A survival analysis

Suicide is a global health issue and since 2008, it is the primary cause of unnatural death in Spain [1]. A history of previous suicide attempt is the strongest predictor for future suicidal ideation and behaviour (SIB), including suicide ideation, suicide attempts, and suicide [2–5]. For instance, in a 5 years follow-up of 302 individuals admitted to an inpatient psychiatric unit for medically serious suicide attempts, 37% of them made at least one further suicide attempt, and 6.7% eventually died by suicide [6]. Furthermore, most suicides occur in people with mental disorders [1], but most people with mental disorders, even severe, never attempt suicide [7]. In other words, this risk factor and many others have poor predictive power. Therefore, a better differentiation between suicide attempters who eventually suicide and suicide attempters who will not is critical to developing preventive plans.

In a systematic review of 14 cohorts (n = 21,385), Neeleman estimated that individuals with antecedents of self-harm were 25 times more likely to die by suicide than the general population [8]. Owens et al. [9] reviewed 80 observational and empirical studies and concluded that the risk of another SIB ranged between 16% (first year) and 23% (follow-up of 4 years or longer), whereas for suicide it ranged from 2% (first year) to 7% (follow-up of 9 years). Christiansen et al. [10] estimated the risk of another SIB in a five-year follow-up study at about 31%. These authors stressed that the risk of another SIB was higher during the first two-years after the index suicide attempt. Female gender and the presence of mental disorders are well-known risk factors for repeated SIB [10]. Other authors have stressed the role of personality disorders, particularly borderline personality disorder, in future SIB [11]. On the other hand, between 1 and 6% of individuals evaluated because of a suicide attempt eventually suicide in the year following. The risk of suicide is higher in older patients and those individuals with a higher number of lifetime suicide attempts [12–15], counter to clinical lore about frequent attempters not being at risk for suicide because they only engage in low risk SIB.

Even if evidence is scarce, recent studies have demonstrated that it is possible to reduce the risk of re-attempt or even suicide in individuals at risk [16, 17]. For instance, we previously reported that a one-year telephone intervention program was effective in reducing an 8% the proportion of patients who re-attempted suicide compared to the control population [18]. This is in keeping with some [19] but not all [20] previous literature on the effectiveness of telephone intervention programs.

Aims of the study: The main objective of the current study is to identify risk factors for re-attempt and suicide using survival analysis.

In keeping with the literature, we found that younger age, and presence of personality disorders and alcohol use disorder were risk factors for re-attempting suicide in our sample of suicide attempters [10, 24, 25]. Furthermore, alcohol use and older age were risk factors for suicide. As for the telephone management program, around 50% of suicide attempters and 90% of those who died by suicide, respectively, did not complete the telephone follow-up at month 12. Those who completed the telephone follow-up were less likely to die by suicide. However, it was discouraging that most suicides (n = 127, nearly 90% of all suicides) were not evaluated in the ED during the study period.

Overall, there were more index suicide attempts for women than men in our sample, which is consistent with literature [26], and might be explained by the higher risk of depression among women [27]. Twenty percent of suicide attempters re-attempted suicide during the follow-up period. This is also in keeping with previous literature [13, 28]. In one study, 25% of the initial cohort of suicide attempters (n = 150; 38% had previous suicide attempts) re-attempted suicide during the 10-years follow-up [29]. In a Danish register-based survival analysis of 2614 suicide attempters matched with 39,210 non-attempters, 31.33% of suicide attempters re-attempted suicide within the follow-up period –average follow-up period was nearly 4 years- [10]. The authors stressed that the probability of suicide attempters re-attempting suicide was stronger during the first 2 years after the index suicide attempt. In a 10-year follow-up study between 1993 and 2002, from the initial 3690 suicide attempters admitted to Christchurch Hospital, 28.1% were readmitted for a further non-fatal suicide attempt [26]. Again, risk of readmission and rates of mortality from suicide were higher in the first 2 years after the index suicide attempt, but occurred throughout the 10 years follow-up period. The findings from both studies perfectly match our results as 88% of suicide attempters in our sample re-attempted within the first-2 years of follow-up.

Regarding risk factors for re-attempting suicide, we found three risk factors: 1) younger age; 2) presence of personality disorders; and 3) presence of alcohol use disorder, which is also in keeping with the literature. In the Danish registry-based survival analysis mentioned above [10], both younger age and alcohol abuse were risk factors for re-attempting suicide. Our results also match those reported by Osvath and colleagues when comparing first-time suicide attempters (n = 549) with repeaters (n = 609): both alcohol abuse, and particularly, the presence of personality disorders were associated with an increased risk of re-attempting suicide [30]. In a study following a similar case-control design comparing 112 first-time attempters and 159 repeaters, alcohol misuse was again one of the strongest factors associated with repetition of suicide attempts [31]. In a 20-year follow-up of first-ever suicide attempters, alcohol intoxication at index suicide attempt predicted repetition of suicide attempt at 5 years [32]. In the 10-year follow-up carried out by Gibb and colleagues, the factors associated with repetition were female gender, younger age, and use of a low-lethality suicide method [26]. In a previous study of 446 suicide attempters, we also reported that younger female attempters with severe personality disorders were prone to repeat suicide attempts [11].

On the other hand, 1.2% (15/1241) of suicide attempters evaluated died by suicide within the follow-up period, which is in the lower range of reported studies. For instance, in a follow-up of 11,563 patients who presented to hospital after deliberate self-harming, 1.5% and 3% died by suicide after 5 and 15 years of follow-up, respectively [33]. Also, in the 10-year follow-up mentioned above, of the initial 3690 suicide attempters admitted to Christchurch Hospital, 4.6% died by suicide [26]. In another cohort of 150 suicide attempters followed-up during 10 years in Catalonia, 12% completed suicide, and the risk was highest during the first 2 years after the index suicide attempt [29].

We found that the risk factors for suicide among suicide attempters followed-up in our study were: 1) being older; and 2) the presence of an alcohol use disorder. These results are also in keeping with literature [34–37]. Indeed, one of the most consistent findings in Suicidology is that suicide rates are higher among adults aged 60 and older [38]. Accordingly, suicide prevention programs should specifically be designed for this population [38]. On the other hand, alcohol use disorders not only increased the risk for re-attempting suicide but for suicide. Our finding is also in keeping with previous literature [39–41]. For instance, in a sample of 1018 unselected deliberate self-poisoning patients followed-up 14 years, of the 22.7% who had suicided by the end of the study, 85 (38.5%) showed clear evidence of long-term alcohol misuse [41]. These authors stressed that more attention should be paid to alcohol use disorders in suicide attempters [41].

However, our data cannot be generalized, as the most relevant and discouraging aspect of our study was that most suicides (127 out of 142, 87.5%) took place in people who were not evaluated in the ED during the study period. In other words, the 15 suicides among our sample of suicide attempters are likely not representative of suicide completers in our catchment area. For instance, most individuals who died by suicide in our study were women with a history of previous suicide attempts. However, most suicides within our catchment area, but who were not evaluated in the ED during the study period, were men (70.1%). Literature is clear in this respect: most suicides are male in most countries. Our finding that most suicides in our catchment area were not included in our sample might be explained by the fact that 60% of suicides in our area died during the first attempt, and 92.3% of suicides occurred during the first or second attempt [42]. Furthermore, most individuals who suicide are not followed up in mental health services, but rather in primary health services [5], thus making it difficult to identify individuals at risk.

Finally, 90% of those who eventually died by suicide, were not followed-up at month 12. This finding, paired with the above mentioned data that most suicides are followed up in primary health services, strongly suggests that, for the prevention of suicide, it is critical to implement “multiple practice improvements over several years” [16].

Younger age and the presence of either a personality disorder or an alcohol use disorder are risk factors for re-attempt in suicide attempters. Alcohol use and older age were risk factors for suicide. Most suicides within the period of study were not included in our study. Thus, our study raises an important question: longitudinal, follow-up studies are methodologically sound studies that allow drawing etiological connections between risk factors and suicide. The problem is that, if previously published follow-up studies on suicide had the same problem that we faced in our study –that most suicides were not “detected”-, most literature published to date on follow-ups could be extrapolating data from populations affected by a selection bias, thus not reflecting the real predictive characteristics of suicide deaths. Furthermore, research is clear on this: most who die by suicide do not even seek mental health services, and attempters and completes are different, although partially overlapping, populations [37, 47]. Thus, until we are able to detect most individuals at risk, and bearing always in mind that the prediction of suicide is impossible, probably the most intelligent interventions to decrease the daunting suicide rate are reducing access to means and a population-based strategy [47] directed to the prevention of depression in the general population by using different measures at different levels of the health system as recommended by the EAAD [16, 17, 48].