Increased Risk of Attempted and Completed Suicide in Obsessive Compulsive Disorder: A Systematic Review of Follow-up Studies

Suicidality is common in the general population, and even more common in patients with mental health disorders (Hollander et al 1996/7, Batterham et al. 2017). Suicidality describes a spectrum from suicidal thoughts, ruminations, suicide ideations, suicide planning, suicide attempts and completed suicide. In individual subjects, suicidal thoughts, suicide ideations and intentions usually precede attempted and completed suicide. Subjects with previous thoughts of suicide and suicidal intentions may be more likely to make a suicide attempt. Subjects with previous suicide attempts may be more likely to perform new acts of suicide including further attempts and competed suicide.

However, even though this might be assumed in general, the proportion of subjects that go from one to the next stage may vary in different mental disorders. Cho et al. (2016) observed that the associations between different anxiety disorders and suicidal thoughts were different from associations of other individual disorders with suicidal attempt. Thus, subjects with OCD may have an increased prevalence of suicidal thoughts and ruminations but may not necessarily have an increased risk of suicide.

Obsessive compulsive disorders are common, often longterm disorders in the general population. The one month-prevalence was 1.1% in the British National Psychiatric Morbidity Survey of 2000 (Torres et al. 2006). OCD may appear independently or comorbid with other mental health disorders, especially depression and anxiety (Chaudhary et al. 2016, Chui et al. 1996, Torres et al. 2006).

Obsessive compulsive disorders are often stressful for the patients and their relatives. Suicidal thoughts, ideations and intentions can be a part of the obsessive ruminations experienced by patients with OCD. These can lead to severe and constant frustration and, consequently, may cause or add to the wish to end one’s life. OCD has been associated with suicidal ideations (De Berardis et al. 2015). Suicide ideations were common in college students with OCD; however, most of the suicidal ideations were associated with symptoms of depression (Huz et al. 2016). In contrast, Gupta et al. (2014) observed that suicide ideations were highly prevalent in patients with OCD; and although depression may be a contributory factor for suicidal ideations in patients with OCD, it was not the sole risk factor for suicidal ideations.

In a recent large multinational study, suicidal ideation within the last month was reported by 6.4% (n = 200) of patients with OCD, a lifetime history of previous suicide attempts was observed in 9.0% (n = 314) (Brakoulias et al. 2017). Cross-sectional studies reveal that OCD is not only associated with a high risk of depression but also of recent suicidal behaviour (Chaudhary et al. 2016). OCD is common among adolescents in India; it is associated with ADHD, sexual abuse, psychological distress, poorer academic performance and suicidal behaviour (Jaisoorya et al. 2015).

In the American Epidemiological Catchment Area Study, OCD – both in its uncomplicated and comorbid form – had significantly higher rates of childhood conduct symptoms, adult antisocial personality disorder problems, and of suicide attempts than did ‘no or any other psychiatric disorders’ (the latter include all psychiatric disorders except schizophrenia without individual disorders or numbers being given, Hollander et al. 1996/7).

In contrast to this, some studies have indicated that the risk of suicide may be reduced in comparison to other mental health disorders such as depression, bipolar disorder and schizophrenia (Coryell 1981). A systematic review and meta-analysis by Kanwar et al. (2013) provided evidence that the rates of suicides are higher in patients with any type of anxiety disorders compared to the subjects without anxiety disorders, except those with OCD who had the same suicide rates as those without OCD.

Due to these conflicting publications, the present study intends to evaluate the risk of attempted and completed suicide in patients with OCD. We excluded studies on suicidal ideation, suicidal thoughts and ruminations in OCD, as they may not provide consistent information on the actual suicidal behaviour in subjects with OCD. Previous suicide attempts may be more often reported by subjects with severe OCD than by controls. This may result in an awareness, or a hindsight bias, that is, the increased awareness of subjects with severe OCD of their previous suicide attempts. A selection bias might be caused by a higher admission or specialist care referral rate for OCD subjects with previous suicide attempts.

To avoid hindsight, awareness, recall bias and selection bias, we did not include cross-sectional studies where the risk factor, that is, OCD is assessed at the same time as the outcome, that is, suicidal attempts. Thus, the present systematic review will focus on the prospective follow-up studies assessing the risk of suicidal behaviour, that is, suicide attempts and completed suicide in subjects with OCD in comparison to that of the general population or to that of subjects with other mental disorders. We included suicide attempts in our analysis as completed suicide is luckily a very rare event in prospective studies.

We performed a systematic review on the incidence of attempted and completed suicide in patients with obsessive compulsive disorders. We performed a literature research in Pubmed and Medline using the search terms [obsessive-compulsive disorder or OCD] AND suicide in November 2017.

We used the following inclusion criteria: English language, papers providing data on the incidence of suicidal events, that is, attempted and completed suicide over a specified time period.

We excluded non-English papers due to our language limitations. Editorials, reviews, case reports, were excluded as they do not provide original data on the suicide risk in OCD. Papers on suicidal thoughts, ideations and ruminations on patients and controls were excluded as they do not provide an adequate assessment of suicide risks.

We excluded papers which focused on retrospective assessment of suicidal events by patients as such information is very much prone to a hindsight and awareness bias. Patients with a history of suicide might also be more prone to be admitted to hospitals, thus creating a selection bias.

The following information was extracted from the papers: authors, year of publication, country of publication, study focus, study type, subjects investigated, their gender, their age (mean), the assessments performed, the outcomes of the assessments, the study results, the study conclusions, the limitations and possible bias of the studies. The biases were assessed using the Cochrane criteria (Higgins et al. 2017).

Using the above mentioned search criteria (that is, OCD and suicide) in Pubmed and Medline, 404 papers on obsessive compulsive disorder and suicide were initially identified in November 2018. Two other papers were later identified through the reference list of the full papers received as below. Of the 404 paper, we screened all the available titles and abstracts for possible eligibility. We excluded the following articles:

We finally identified 73 papers, that is, 71 plus two papers from the reference lists of other papers, which might focus on the prevalence or incidence of suicidal acts in subjects with OCD. The PDF or full prints of these 73 eligible papers were received and assessed for their relevance to the study questions.

Finally, eight papers were identified that provided prospective estimates of the incidence of suicidal events, that is, provided data on attempted or completed suicide over a defined observation period. These reports were included in the final review.

Table 1 provides a summary of the six research studies finally included in the systematic analysis.

In our review, we observed that the incidence of suicide attempts in OCD subjects varies considerably between studies and samples.

The rates of attempted suicide were increased in OCD subjects identified in the general population versus control subjects, even though the first three studies did not reveal statistical significance:

The rates of attempted suicides were usually high in patients in treatment for OCD even though there were no control samples investigated, that is, neither OCD subjects from the general population nor subject without OCD:

Completed suicides were rare in all studies in the general population, thus showing no significant increases of completed suicides in OCD subjects versus controls in all, but one study. Assuming an average observation period of 22 years, Fernandez et al. (2017) observed significantly increased rates of completed suicides in OCD subjects versus control subjects, that is, 0.067%/year in OCD subjects versus 0.0072% in control subjects.

Suicides were relatively high in inpatient samples (even thought there were no control samples assessed):

Psychiatric comorbidity seems to be the most common risk factor for attempted and completed suicide: Alonso et al. (2010) revealed that suicidal behaviour in OCD subjects was associated with being unmarried, symmetry/ordering obsessions and a history of affective disorders. Angst et al. (2005) reported that comorbid OCD/OCS was associated with significantly higher levels of treatment seeking, impairment, distress and suicidality compared with pure OCD/OCS. According to Fernandez de la Cruz et al. (2017), a previous suicide attempt, comorbid substance abuse or personality disorder increased the risk of death by suicide in subjects with OCD.

Bolton et al. 2008 showed that any anxiety disorders at baseline was significantly associated with subsequent onset suicide attempts (adjusted odds ratio 2.20, however, predictors were not individually assessed for subjects with OCD. Studies by Rueck et al. 2017 and Sareen et al, (2005) did not allow to assess risk factors of suicidal behaviour in OCD due to the low samples sizes of subjects at risk.

The one large Swedish national register study showed clear evidence that the rate of attempted and completed suicide is significantly higher in subjects with OC as compared to control subjects from the general population (Fernandez de la Cruz et al. 2017). The numbers of subjects, the duration of observations and the comprehensiveness of patient and control sample identification are impressive and support the validity of the evidence.

Other prospective studies included in our review that provided data on the suicide risk in OCD were in support of this observation. They showed non-significant trends in the same direction (Angst et al. 2005, Bolton et al. 2008, Sareen et al. 2005). However, due to the low rates of attempted and completed suicide combined with the low number of subjects at risk, the power of these studies is far too low to reach any statistical significance. The studies by Alonso at al. (2010) and Rueck et al. (2017) can also be seen as supportive for an increased risk of suicide in OCD as they provide relatively high rates of suicide compared to the above mentioned studies, but their relevance is limited due to their lack of a control group.

The observed increase of attempted and completed suicide in prospective studies, especially the one from Fernandez et al. (2017), is comparable to the reports from previous cross-sectional and retrospective studies: Data from 3711 adult patients with primary OCD that came from several countries indicated that suicidal ideation within the last month was reported by 6.4% (n = 200) of patients with OCD and a lifetime history of suicide attempt was observed in 9.0% (n = 314) (Brakoulias et al. 2017). In agreement, patients with a previous suicide attempt had significantly higher rates of comorbid psychiatric disorders (Dell’Osso et al. 2017). This data were supported by the studies reviewed by Angelakis et al. (2015).

This recent review on OCD and suicidality was performed by Angelakis et al. (2015). This review included the literature up to March 2014, however, it focussed mostly on suicidal thoughts and ideations and could not include the most recent prospective studies. This review revealed a significant association between OCD and suicide through ideation and suicide attempts, but did not provide sufficient information about completed suicide. Most of the studies used by the authors were cross sectional, thus being prone to a selection and hindsight bias. Some prospective studies were used twice reporting slightly different figures, even though they referred to identical or overlapping samples. For clarity, we used the initial publication of both prospective studies (Angst et al. 2005, update reported by Fineberg t al. 2013; Sareen et al. 2005, update reported by ten Have et al. 2009). The updates were not included a second time as they provided highly redundant information that did not affect the general outcomes of the initial reports, but the double-presentations would give the false impression of a higher number of relevant studies and samples.

The observed association between co-morbidity and suicidality has also been observed in other disorders (Al-Asadi et al. 2014, 2015). Adolescents with OCD have increased numbers of suicidal ideations, the number of previous suicide attempts was found to be lower than that of other patients with mental disorders (Apter et al. (2003). However, such claims are based on cross sectional and retrospective studies that are prone to selection and awareness bias. Sadly, there are no follow up studies that have included different psychiatric disorders including OCD and would allow a relevant valid comparison.

Psychiatric comorbidity seems to be the most common risk factor for attempted and completed suicide. According to Fernandez de la Cruz et al. (2017), a previous suicide attempt, comorbid substance abuse or personality disorder increased the risk of death by suicide in subjects with OCD. Two smaller studies supported the relevance of additional psychiatric comorbidity but also identified other risk factors for suicide behaviour in OCD, that is, being unmarried, symmetry/ ordering obsessions and a history of affective disorders (Alonso et al. 2010). Angst et al. 2005 reported psychiatric comorbidity to be associated with significantly higher levels of treatment seeking, impairment, distress and suicidality compared with pure OCD/OCS.

The risk factors identified in these prospective studies are confirmed through reports from cross sectional studies, which usually have more subjects at risks and thus a higher statistical power, but sadly are more prone to selection and awareness bias:

In agreements with the above studies, the prevalence of suicidal thought was higher for those with more than one psychiatric disorders (Al-Asadi et al. 2015). According to Hung et al. (2010), depression and anxiety symptoms contribute significantly to suicidal ideation among patients with OCD. According to Kamath et al. (2007), depression and hopelessness were the major correlates of suicidal behaviour in OCD. Chia et al. (1996) also assumed that depression was a main contributing factor in two of the subjects who had committed suicide. Torres et al. (2011) assessed 582 OCD patients from Brazil; 36 percent of the OCD patients reported lifetime suicidal thoughts, 20% had made suicidal plans, 11% had already attempted suicide, and 10% presented current suicidal thoughts. In the logistic regression, only lifetime major depressive disorder and posttraumatic stress disorder (PTSD) remained independently associated with all the aspects of suicidal behaviours.

Angelakis et al. (2015) identified more predictors of suicidality in OCD: Comorbid Axis I disorders, increased severity of comorbid depressive and anxiety symptoms, increased severity of obsessions, feelings of hopelessness and past history of suicide attempts were associated with worsening levels of suicidality in OCD. In OCD patients without current depression, suicide ideations were associated with female gender, presence of comorbid psychiatric illness (other than depression), contamination obsessions and cleaning/washing compulsions (Gupta et al. 2014).

Similar observations were also found in the presence of one or more obsessive compulsive symptoms that were associated with an increased odds ratio of suicide risk of approximately 2.4 in a sample of 474 students; although this was no longer a significant risk factor when controlling for depressive symptoms (Huz et al. 2016). Of the OCS assessed, only obsessions about speaking or acting violently as well as depression remained independent risk factors for suicidality. In agreement, Ching et al. (2017) observed that violent obsessions were associated with suicidality in an OCD-analog sample of college students. It might be argued that these associations were driven by the associations between OCD and suicidal thoughts, and suicide attempts were retrospectively assessed and thus were prone to hindsight and awareness bias.

The conclusion that there is an increased risk of suicide in OCD and is mainly based on one prospective study (Fernandez de la Cruz et al. 2017), which prospectively assessed the attempted and completed suicide. This study is also the only prospective study that had enough statistical power to identify some predictors of suicide in OCD. Other studies (Angst et al. 2005, Alonso et al. 2010) are supportive of the outcome but they are usually based on smaller samples of OCD subjects or suffer a lack of a reasonable control group. As completed suicide luckily is a very rare event, none of the studies is sufficiently powered to calculate reasonably accurate risks, that is expected incidence data for suicide in OCD with low statistical confidence intervals. Previous studies that show an increase of suicidality in OCD focussed on suicidal thoughts and ideations, but less on suicide attempts or completed suicides.

To be able to generalize the results from this Swedish Register based study to other countries, one would need to access similarly large national patients’ and mortality registers. To be able to compare the results of future studies on the relationship of OCD and suicidality, it would be helpful if such studies would use similar diagnostic criteria for psychiatric diagnoses and suicidality, use comparable assessment tools for all types of suicidality, and apply predetermined comparable time-frames.