Correlates of violent suicide attempts in patients with bipolar disorder

Suicide is one of the leading causes of death in bipolar disorder (BD); violent suicide attempts are associated with the highest level of lethality. We aimed to evaluate factors related to the risk of violent suicide in a large naturalistic sample of patients with BD; in addition, we analyzed the rates of lifetime suicide attempts and the variables associated with suicidal behavior.

We recruited 847 patients with BD. Patients were grouped according to whether they had a lifetime history of suicide attempts and, among suicide attempters, subjects who had used a violent suicide method were compared with those who had attempted suicide with a nonviolent method. Comparisons were performed using χ2 tests for categorical variables and ANOVA for continuous variables. Logistic regression (LogReg) was used to identify explanatory variables associated with violent suicide attempts (dependent variable).

Two hundred and two patients (24%) had a lifetime history of suicide attempts. Subjects with at least one lifetime suicide attempt showed longer duration of illness (22.4±14.1 years vs 19.9±14.2 years: p 0.028), more lifetime hypomanic episodes (3.3±4.3 vs 2.3±3.1: p0.001), more lifetime depressive episodes (6.0±4.4 vs 4.7±4.1: p<0.001), higher rates of lifetime psychiatric comorbidities (50.0% vs 41.3%: p 0.029), higher rates of lifetime medical comorbidities (58.0% vs 48.9%: p 0.028) and higher rates of reduced HDL cholesterol (46.2% vs 36.7%: p 0.030). Among suicide attempters, fifty-two patients (30.6%) attempted suicide with a violent method. We found more men in the group of violent suicide attempters than in the group of nonviolent suicide attempters (65% vs 28%; p: <0.001). Moreover subjects with previous violent attempts showed higher mean values of weight (80.5±18.3 vs 69.4±14.7: p<0.001), body mass index (27.8±5.6 vs 25.2±4.7: p<0.003) and waist circumference (98.7±18.5 vs 92.4±14.3: p 0.032). The LogReg analysis confirmed the association of violent attempts with male gender (p: <0.001; Phi: 0.35) and higher waist circumference (p: <0.001; Cohen's d: 0.39).

In our research we analyzed lifetime suicide attempts, but the sample does not include completed suicides, meaning that we are unable to test whether the results are generalizable to suicide deaths. Moreover, some relevant variables, such as medical comorbidities/metabolic parameters at the time of suicide attempts and previous medication, were not collected. Another limitation concerns the heterogeneity of recruited patients in terms of clinical characteristics (e.g.: medical conditions, drug treatments), with potential confounding factors.

The present study confirms the association between male gender and violent suicide and suggests a correlation between obesity and the use of violent suicide methods. The relationship between obesity and suicidal behaviour is worthy of interest and deserves to be explored by further studies.