Introduction

Attention has been drawn to the public health importance of suicide worldwide [1–3]. Suicidal behaviors commonly run on a continuum from developing an idea to kill self, to making a plan for the act, acting out the plan, with or without suicide as the final outcome. There is probably a complex interrelationship between the stages in this continuum, much of which remains poorly understood.

The few studies of suicide so far conducted in Nigeria suggest that the rate of suicide may be lower than has been reported in other parts of the world [4, 5]. Asuni [4] argued that the low suicide rate was due to the near non-existence of depression in Africa, implying that suicide is always due to depression and is in fact an index of depression. However, it is now generally agreed that the presumed rarity or absence of depression in Africa, propagated by early workers in the continent, is mistaken [6]. Besides, even though a number of psychological autopsy studies suggest that the majority of patients who committed suicide had a mental disorder at the time of suicide [7–19], these studies have not examined the risk factors for each of the behaviors that constitute the continuum of suicidality. For example, mental disorders have been shown to be variously associated with suicidal ideation, suicidal plans and suicidal attempts [20–22]. Suicidal attempts are more common than completed suicides [23, 24] and many people who have suicidal ideation do not proceed to plan and only a minority of those who plan to take their own lives eventually make an attempt. Knowledge of the factors that predict the transition from one stage to another is likely to open the possibility for prevention.

There is an important question as to the specificity of any single mental disorder as a risk for or predictor of suicidality given that co-occurrence of multiple disorders is a common observation [25, 26, 27, 28]. In this regard, there are reports suggesting that comorbidity may be important in considering the association between mental disorders and suicidal outcomes. For example, there is some evidence that the apparent link between panic disorder and suicide attempts becomes attenuated when co-occurring depression and substance use disorders are controlled for [29–32]. On the other hand, there is conflicting evidence about whether comorbidity is responsible for the apparent association between suicidal outcomes on the one hand and the co-occurrence of anxiety and bipolar disorders [33–39]. There is, therefore, continuing interest in the relationship between suicidal outcomes and comorbidity of mental disorders. Specifically, it is of interest to determine (1) whether the association between a specific mental disorder and suicidal outcome is direct or is dependent on the co-occurrence of other mental disorders; (2) whether increasing number of mental disorders constitutes an independent risk for suicidal outcomes, irrespective of which disorders are involved. In a previous paper, we reported the profile and risks of suicidal behaviors in the Nigerian Survey of Mental Health and Wellbeing [40]. In this report, we explore the relationship between suicidality and comorbid mental disorders using data from the same survey.

Methods

Details of the Nigerian National Survey of Mental Health and Well being have been published elsewhere [26, 41]. Here, we provide only a brief summary of the methodology.

Sample

The study employed a four-stage area probability sample of households to select non-institutionalized adults. The survey was conducted in five of the six geo-political regions of Nigeria: south–west (Lagos, Ogun, Osun, Oyo, Ondo, and Ekiti); south-east (Abia, Anambra, Enugu, Ebonyi, and Imo); south–south (Akwa Ibom, Cross-river, and Rivers); north–central (Kaduna, Kogi, and Kwara), and north–east (Adamawa, Borno, Gombe, and Yobe). Collectively, these states represent about 57% of the national population. An eligible member of a household had to be 18 years of age or over and able to speak one of the languages of the study. Only one such person was selected per household. When the primary respondent was either unavailable following repeated calls (five repeated calls were made) or refused to participate, no replacement was made within the household. On the basis of this selection procedure, face-to-face interviews were carried out on 6,752 respondents. The overall response rate was just over 79%. The survey assessments were conducted in Yoruba, Igbo, Hausa, and Efik languages.

Field work was conducted between February 2002 and May 2003. The survey was administered in two parts: part I consisted of a core of diagnoses and was administered to all respondents; part II consisted of sections for the assessment of risk factors, consequences and correlates of disorders as well as a few disorders not included in the core diagnoses. Part II was administered to respondents who had a history of past or recent part I disorders plus a probability sub-sample of other respondents. A total of 6,752 respondents completed part 1; 2,143 completed part II.

Respondents were informed about the study and provided consent, mostly verbal but sometimes signed, before interviews were conducted. Verbal consent was the norm because of widespread illiteracy and because some respondents seemed somewhat uneasy about the implications of appending their signature to a document. These survey procedures were approved by the University of Ibadan/University College Hospital, Ibadan joint Ethical Review Board.

Measures

Diagnostic assessments were conducted with the use of the World Health Organization’s (WHO) Composite International Diagnostic Interview, version 3 (CIDI 3.0) administered by trained lay interviewers [26]. The CIDI is a fully structured diagnostic interview that can generate diagnoses according to the criteria of both the Diagnostic and Statistical Manual of Mental Disorders, 4th edition and the International Classification of Diseases, 10th edition [42, 43]. The language versions of the CIDI 3.0 used in the present survey were derived using standard protocols of iterative back translation and harmonization conducted by panels of bilingual experts.

The suicidality section of the CIDI 3.0 first asks about three possible suicidal experiences: ideas, plans and attempts. When a lifetime experience of any of the outcomes is reported, subsequent questions determine the age of the subject at the first ever occurrence of the experience and the last episode of the outcome. The subject is further asked to state the number of times each event has ever occurred and if suicidal attempt resulted in some injury or needed medical attention. If the subject admits to having made suicide attempt, the interviewer asks whether the attempt was really serious and the subject survived only as a matter of luck or the subject either simply wanted to attract attention or used a suicide method knowing it was not fool-proof.

Subjects who met the DSM-IV criteria of specified mental disorders were identified and comorbidity was defined as the co-existence of more than one identified mental disorder in the same individual. The core mental disorders considered include mood disorders, i.e., major depressive disorder, mania, irritable depression, dysthymia and bipolar disorder; anxiety disorders, i.e., panic disorder, specific phobia, social phobia, agoraphobia, generalized anxiety disorder, posttraumatic stress disorder and separation anxiety disorder. The third major group of disorders consists of impulse control disorders, i.e., intermittent explosive disorder, oppositional defiant disorder, conduct and attention deficit hyperactivity disorder. Substance and drug use disorders were also assessed.

In view of the small numbers for several specific diagnostic entities, this analysis is based on groups of disorders: i.e., any mood, any anxiety, any impulse and drug/substance disorder.

Quality control

The core sections of the CIDI 3.0 had been available in Yoruba language before the commencement of the NSMHW. Further translations and adaptations to Igbo, Hausa and Efik languages were conducted, using standard WHO translation guidelines. Each completed questionnaire and worksheet returned by the field interviewers were checked for accuracy and consistency by research supervisors. The supervisors also conducted random field checks on 10% of the selected households to verify household listing, appropriate use of the sample selection procedure, and whether interviewers had conducted the interview in full. When interviewers had reported non-response, supervisors also checked on the reason.

Results

Prevalence of lifetime DSM-IV disorders among suicidality outcomes

Persons with lifetime suicide attempt were more likely than those without attempt to have experienced lifetime DSM-IV disorders (Table 1). For example, whereas the prevalence of any anxiety disorder among persons who ever made a suicide attempt was 22.0%, it was 6.3% among persons who had never made such attempt. The respective estimates for any mood disorder were 49.7 versus 3.0% and 20.6 versus 3.6% for alcohol abuse or dependence. Lifetime attempters were also more likely to have comorbid conditions. Compared with an estimate of 0.4% for three or more disorders among persons who had never attempted suicide, the estimate was 11.1% among lifetime attempters. Higher rates of mental disorders as well as comorbidities were also more common among ideators than non-ideators. The trend was similar, even though less consistent, for ideators with a lifetime plan and ideators without a plan. Among ideators with a lifetime plan, those who made attempts were generally more likely to have lifetime mental disorders and more likely to have comorbid conditions than those who had made no attempts. On the contrary, persons who made no attempts among lifetime ideators without a plan were more likely to have lifetime disorders than those who had not plans but nevertheless made attempts. That is, persons who had thought about suicide but had made no plans and had made no attempts were more likely to have lifetime disorders than impulsive attempters.

Table 1 Prevalence of lifetime DSM-IV disorders among persons with suicidal behaviors
Full size table

Comorbidities of lifetime suicide attempts with lifetime DSM-IV disorders and number of other disorders

Table 2 shows the effects of controlling for other disorder groups and number of disorders on the independent associations of each group of disorders with lifetime attempt. A striking observation is that while each disorder group is associated with lifetime attempt in bivariate analysis, the associations changed substantially with multivariate analysis. Thus, while mood disorder remains an independent risk factor for attempt even after controlling for the effects of other disorder groups as well as number of disorders, substance use disorder is no longer associated with a significant elevation of risk of attempt. After controlling for the effects of other disorders as well as number of disorders, anxiety disorder continued to constitute an independent but somewhat attenuated significant risk while impulse disorder lost its significant association. On the whole, comorbidity seems to have accounted for the association of impulse disorders with suicidal attempt and to have attenuated but not accounted for that of anxiety disorders.

Table 2 Comorbidities of lifetime suicide attempts with lifetime DSM-IV disorders
Full size table

Comorbidities of lifetime suicidality with lifetime DSM-IV disorders and number of other disorders

We explored the association of specific disorder groups and their comorbidities with lifetime suicidal outcomes. Table 3 presents the results of multivariate analyses with survival modeling in which all the disorder groups as well as number of disorders other than the one in focus were included in the model. There was an independent association of each of the disorder groups (anxiety, mood, impulse, and substance use) with lifetime ideation with the risk being highest among persons with lifetime impulse disorder. Number of disorders did not independently increase the likelihood of suicidal ideation. Lifetime mood disorder was the only disorder group with independent association with making a plan among lifetime ideators. No associations were found for disorder groups or number of lifetime disorders for making either a planned or impulsive attempt among ideators. On the other hand, the risk of lifetime attempt was elevated among persons with each of the disorder groups, with the pattern almost mirroring the associations with lifetime ideation except that, for attempts, only the associations with anxiety and mood disorders reached statistical significance. Among persons with impulsive attempt or those with planned attempt, independently, number of disorders remained unrelated to risk of attempt.

Table 3 Comorbidities of lifetime suicidality with lifetime DSM-IV mental disorders and number of other disorders
Full size table

Discussion

In this representative community dwelling sample of Nigerians, our major findings agree with the prevalence reports of many previous studies. The prevalence figures for mental disorders were much lower among subjects without the five suicidal behavior outcomes of interest investigated in this report compared with those with the outcomes. For example, while only 3% of persons who had never made a suicide attempt had experienced a lifetime mood disorder, about 50% of those with previous suicide attempt had experienced a mood disorder. Also, while only about 12% of persons with no previous suicide attempt had experienced any lifetime DSM-IV disorder, 57% of those with a previous suicide attempt had met the criteria for at least one DSM-IV disorder. Associations of mental disorders were clearly found for lifetime suicidal ideation as well as plan following ideation. Impulsive attempt (attempt among ideators without a plan) was not associated with lifetime mental disorder.

Results of bivariate analysis suggest that each disorder group significantly elevated the risk of lifetime suicide attempt, with impulse and mood disorders doing so more strongly than anxiety and substance use disorders. Contrary to these findings, multivariate analyses in which comorbid conditions as well number of co-occurring conditions were controlled for indicate that while the salience of anxiety and mood disorders was hardly affected, substance use disorder was no longer an independent risk factor for lifetime suicide attempt. The importance of impulse disorder as an independent risk factor was considerably attenuated but there was still a noticeable trend that just barely missed statistical significance. Indeed, while mood disorder remained by far the greatest independent risk factor for suicidal attempts, it would appear that, but for small numbers and the attendant loss of statistical power, impulse disorders appeared to be next in importance as a risk factor for suicidal attempt with the associated odd ratio considerably higher in value than that of anxiety disorder. Overall, in this sample, therefore, the association of substance use disorder with lifetime suicide attempt seems explainable solely on the basis of comorbid conditions.

Our multivariate analyses suggest that each of the disorder groups constituted an independent risk for lifetime suicidal ideation with three- and eightfold increase in the likelihood of this outcome for the different disorder groups. Once the effect of a particular group was accounted for, the number of other co-occurring disorders did not make an independent contribution to risk of ideation. Also, except for plan among ideators for which mood disorder was an independent risk factor, neither the occurrence of any of the other disorder groups nor the number of disorders was an independent risk for transiting from ideation to plan or from ideation to either planned attempt or impulsive attempt. Indeed, for lifetime attempt in general, number of disorders was not a significant factor in risk once the effect of specific disorder group was accounted for. Thus, while our findings support those of several others suggesting that most mental disorders are risks for suicide attempts [44], we observed that the critical step in this link between mental disorders and suicidal attempt is suicidal ideation. Once suicidal ideation was experienced, only mood disorders predicted those who would make a plan. No disorder group was related to whether planned or impulsive attempt would be made and the number of disorders did not matter to any of the transitions either. The analytic strategy we used, in which each of the disorders was examined independently in bivariate analysis, before adding all other disorders in a multivariate model, permits us to assert that, in this sample, mood disorder as a group is the main predictor of suicidal attempt.

As earlier pointed out, suicidal behavior is without a doubt complex and unlikely to have simple relationship with mental disorders. Our findings suggest that comorbidity is important in regard to the relationship of suicidal behavior to some but not all disorder groups we studied. The relevance of comorbidity to suicidality may derive from different perspectives. Psychiatric comorbidity commonly results from the limitations of current diagnostic manuals which have remained largely based on phenomenology rather than etiology therefore making it impossible to draw a precise line of demarcation between different mental disorders. One possible explanation for the association of mental disorders and suicide is that the comorbid mental disorders may have one common final chemical pathway to suicidality. For example, serotonergic transmission has been implicated in impulsivity, depression, substance disorders, suicide and a number of other disturbances [18, 45, 46]. Also, while the different comorbid mental disorders may actually be independently associated with suicidality, they may have a tendency to cluster with cumulative impairment, severity, and distress thus leading to increased suicide risks [47].

There are limitations in our report which must be borne in mind in interpreting and applying our results. First, we did not take the severity of the disorders used in this analysis into account. This was largely because of the small numbers of the individual disorders in the population. It is possible that the severity of a particular mental disorder or its comorbidity with other disorders may have different salience for suicidality outcomes. However, previous works have reported that objective severity of illness did not distinguish patients with a history of suicide attempts [48–50]. Suicide attempt is associated in various ways with the entire spectrum of suicidal behavior and is one of the strongest predictors for future suicidal acts. Since previous reports have not shown substantial effects of disorder severity on attempt, we suspect that this may also probably be the case for ideation and plan and thus our omitting disorder severity may not have substantially compromised our results. Second, in considering comorbidity, we did not include all the mental disorders which have been previously reported to be associated with suicide. Specifically, our investigation did not make formal diagnosis of DSM-IV Axis II (personality) disorders and schizophrenia which have been variously reported as known risks for suicide [51, 52]. However, schizophrenia, which has a population prevalence of 1%, is often not common in non-clinical samples and we may not have missed a great number that would significantly affect the results. Also, cluster B personality disorders have been reported to have particular importance for suicidality in the presence of bipolar disorders [53]. However, there were virtually no cases of bipolar disorder in our sample that would have permitted any specific exploration of personality disorders. Third, as in all cross-sectional surveys where suicidal behaviors are assessed on the basis of retrospective reports, recall bias may have led to underestimation of the suicidal spectrum in this report. Finally, we did not investigate the complex comorbidity of mental disorders and physical illnesses. Medical conditions have been reported as risk factors for suicidality and such conditions may occur with mental disorders [54–56], but our focus here was on DSM-IV Axis I comorbidity.

Despite these limitations, our findings are important because it is the first large community-based study of the association of mental disorders with suicidal spectrum behaviors in Sub-Saharan Africa. As recent studies have shown, these behaviors are common in the region and constitute important public health concern [57]. Evidence is beginning to emerge that perhaps the quantity rather than just the type of psychiatric disorder may be of unique importance in suicidality and our study adds to this emerging literature. While mood disorder remains an important risk for suicidal ideation and attempt, understanding comorbidity may be crucial in teasing out the link of several other disorders with suicidal behavior. The biological, genetic and psychosocial underpinning of comorbidity among mental disorders should be the focus of future studies addressing the association of these disorders with suicidal outcomes.