In a Finnish sample, 45 (51%) of 89 schizophrenics who committed suicide had their last contact with a health care provider within four days of death, 70% within two weeks, and 82% within a month.¹² Because the suicidality module of MINI 6 used in this study assesses both ‘within a month’ and ‘a life time risk’ of suicidal behaviour, an average prevalence of 82% was used as a guide to calculate the sample size.

Thus, using the formular by Araoye:

n = [z²p (1-p)/d²]¹³

Thus (n) = 3.84 x 0.82 (1–0.82) / 0.0025 which equates to 230.

The ‘psychotic module’ of the MINI 6 was administered to confirm the diagnosis of schizophrenia. The participants who met the inclusion criteria were then given the Sociodemographic and Clinical Questionnaire (SDQ) which included questions on age, gender, marital status, and educational level, amongst others. Each booklet was serialised, and the patients were given corresponding tally numbers. The English version 6.0.0 of MINI was also used by the researchers to determine the presence or absence of suicidality. The MINI diagnostic instrument is divided into different modules identified by letters corresponding to diagnostic categories. Each module screens for various psychiatric conditions such as major depression, dysthymia, suicidality, hypomania, and agoraphobia, to mention but a few. The ‘suicidality module’ of the MINI 6 was administered by the researchers to assess the respondents’ current and lifetime suicidal behaviour. Some examples of the questions in MINI are as follows: In the past month, did you suffer an accident? Did you plan or intend to hurt yourself in that accident, either actively or passively?

The respondents were expected to answer either ‘yes’ or ‘no’. The instrument has different allotted points for each ‘yes’ answered. The C1–C8 questions of MINI diagnostic instrument assess for the current risk of suicide whilst the last question – QC9 assesses for the lifetime risk. In our study, we used both the current suicide risk (QC1–QC8 of MINI) to calculate the total scores for the frequency and severity of suicidality amongst the participants. In contrast, the lifetime risk scores (QC9 of MINI) were used to determine the frequency of suicidal attempt alone. All questions were rated. A score of 1–8 points is low; 9–16 points is moderate, and >17 points is high on the suicidality scale module of MINI International Neuropsychiatric Interview (M.I.N.I 6).¹⁵

The lead researcher liaised with the medical record officer on the clinic day, selected the case files of all patients with recorded diagnosis of schizophrenia that were scheduled for appointment on each particular clinic day. The first person was selected by simple random sampling after which others were selected by systematic random sampling method. The maximum number that could be interviewed was 10 out of an average total attendance of 30 patients with the diagnosis of schizophrenia for each clinic day. Hence, the sampling interval rate of 3 was used for the study.

In a situation where a subject did not meet the selection criteria, the next subject meeting the selection criteria was recruited. This continued until the total number of calculated sample size was completed. The interview session took an average of 30 min.

The lead researcher who was a psychiatrist trainee was trained by a consultant psychiatrist (O.A.) who had also used it for research purposes. There was no need for a research assistant.

The data collection took eight months to be completed.

Ethical approval was obtained from the Health Research Ethics Committee of the LUTH. Written consent was obtained from each participant following a detailed explanation about the aims and objectives of the study. In those found to be positive for suicidality, the implications were explained to them, and they were appropriately referred to their attending Psychiatrists for review. Those requiring additional assistance, like the need for continuous cognitive behavioural therapy were also referred to the clinical psychologists.

Data were analysed using Statistical Package for Social Sciences (SPSS), version 18. Comparison between categorised sociodemographic/clinical variables and suicidality was made using the chi-square test. The description of the pattern of suicidality amongst participants was done by categorising them using the most severe form of suicidal behaviour in each respondent. In a situation where a participant had more than one suicidal event, for example, if a participant had a suicidal intention and suicidal attempt, the most severe form of suicidality in that continuum being suicidal attempt was chosen for that particular participant.

The non-parametric equivalent of the inferential statistics tests was used as appropriate: Mann-Whitney U test instead of independent t-test and Spearman’s correlation was used instead of Pearson correlation.

In order to determine the variables that were independently associated with suicidality in the patients with schizophrenia, all the variables that were associated with suicidality in the previous analysis were re-entered into binary logistic regression analysis using suicidality as the dependent variable.

The participants were aged between 20 and 71 years, with a mean age of 38.54 (±11.30) years (Table 1). Slightly more than half of the participants (50.6%) belonged to the 26–40 years age group. More than half of the participants were females (53.8%), single (51.9%) and employed (59.4%).

Less than half (43.8%) of the participants had their first episode of the illness at age 26–40 years (Table 2). Nearly seven out of 10 participants (67.5%) have had at least three episodes of schizophrenia. Likewise, about seven out of 10 participants (71.3%) have had three hospital admissions on account of the disorder.

In general, 69 (43.1%) participants had suicidal behaviour out of which 17 (24.7%) had severe suicidality, 7 (10.1%) had moderate suicidality whilst 45 (65.2%) had mild suicidality.

Of the 69 participants that met the criteria for suicidality, majority (75.4%) was assessed to have had suicidal attempt (Table 2).

The mean age of the participants who had suicidality was 36.75 (±10.54) years in comparison with the participants without suicidality who had a mean age of 39.89 (±11.7) years (Table 3). There was no significant statistical difference between the ages of participants who had suicidality and those who had none (t = –1.750, p = 0.082). The proportion of participants with suicidality was almost equally distributed within the gender with male and female being 43.2% and 43.0% respectively.

On the contrary, whilst Table 2 showed the percentage distribution of the income earned by the patients, Table 3 showed a significant statistical association between income earned per month as well as income spent per month on the illness and suicidality. This means, the less a participant earned in a month, the more he or she was likely to exhibit suicidal behaviour. The same applies to the amount participants spent monthly on their health. The more they spend on their health monthly, the more they are likely to exhibit suicidal behaviour (U = 2374.000, p = 0.008) and (U = 2374.000, p = 0.008) respectively.

Overall, there was no significant statistical relationship between suicidality and all the measured clinical attributes of the participants.

In order to determine all the variables that were independently associated with suicidality in the subjects, all variables that were associated with suicidality in previous analyses (i.e. income per month and amount spent on care monthly) were entered into binary logistic regression analysis using suicidality as the dependent variables (Table 4). However, none of these two factors was found to predict suicidality amongst the participants independently.

About 6 in 10 study participants had zero scores on current suicidality module of MINI, which indicated an absence of suicidality. In contrast, 43.1% had at least a score and above on the current suicidality risk module of the MINI questionnaire.

The result of this study may not be easily comparable with earlier studies as none of the studies available for review examined the prevalence of suicidality as a whole in patients with schizophrenia. Most of the previous studies focused on either suicidal ideation or completed suicide amongst patients with schizophrenia.

Even though religion and cultural belief in Nigeria are against suicide,²⁰ the high frequency of suicidality in this study shows that patients with schizophrenia are still not immune to having suicidal behaviour. The prevalence of suicidality in our study falls within the estimated 18% – 55% current risk of attempted suicide in patients with schizophrenia as found by Kasckow and colleagues in their research on the suicidal behaviour in older patients with schizophrenia.^4,35 The high frequency of suicidality found in this study might be attributed to the stigma associated with schizophrenia in this part of the world. Depression as an integral part of schizophrenia could be a very prominent factor that might have accounted for the high prevalence of suicidality amongst our study population. Feelings of depression are considered an integral part of suicidal behaviour.^36,37,38. More than half of the patients with schizophrenia will experience at least one major depressive episode.^39,40

Other psychopathology associated with schizophrenia include post schizophrenic depression, obsessive thoughts and hallucination. The presence of command auditory hallucinations (CAH) for suicide alone cannot predict suicidal behaviour, but individuals who are at risk for suicide are more likely to make an attempt when they have CAH.⁴¹

Poor insight, side effects of medication and poor quality of life amongst people with schizophrenia could be contributory. However, Radomsky et al. found in their study on suicidal behaviour before the first hospitalisation, for example, a lower prevalence of 30.2% lifetime history of suicide attempts amongst all patients with schizophrenia and other psychotic disorders.⁴² The variation of the results of the current study with that of Radomsky et al., which had a lower prevalence, could be because of the difference in the methodology used.⁴² Whilst our study looked at suicidality in its entirety (including ideation, intention, plan, attempt), the study conducted by Radomsky et al. focused on suicidal attempt alone. It is also not impossible that the clinic setting where this current study took place afforded the subjects ample opportunities to express themselves more freely and talk about their suicidal experiences, hence the reason for a higher frequency.⁴²

Other social factors which include pre-morbid social functioning, social trauma, and social support may affect the risk of suicidality in patients with schizophrenia. Many investigators suggest that persons with good pre-morbid social functioning who have schizophrenia become demoralised by the decrease in their ability to function and therefore become suicidal. This demoralisation syndrome has been demonstrated in patients who have insight about their illness.^43,44

Reductions in support – whether temporary, as in therapist or family vacations, or more long term, such as moving out of a family residence – are periods of increased risk.⁶

Gender wise, the proportion of subjects with suicidality in this study were almost equally distributed, with male and female being 43.2% and 43.0% respectively. Earlier studies have reported a preponderance of males at risk of suicidal behaviour.^6,45 The finding of more females who expressed suicidality in our study could have been because of the over-representation of females in our study. Other studies that reported lower risk of suicidal behaviour could have done so because it is assumed that several protective factors from suicide may be at work in women. These include the tendency to seek help when they are depressed, a less intense wish to die when attempting suicide and lower rates of substance use disorders.⁴⁶

Age was not significantly found to be associated with suicidal behaviour in the current study. Kessler et al., in their study on the prevalence and risk factors for lifetime suicide attempts in the National Co-morbidity Survey in 1999, reported that the young are more likely to harbour suicidal ideation and that ages less than 25 years are more strongly related to ideation.⁴⁷ The reason for the disparity could be because of the difference in the instrument used in the methodology.

There was no statistical significance between marital status and suicidality (p = 0.809), although some findings have suggested that being married reduces the risk of suicidal behaviour. This may be as a result of enhanced social support received from a primary partner.⁴⁸ The difference in this current study may then suggest that the assessment of the quantity of social contact may be more important than marital status alone for evaluating the risk of suicide in patients with schizophrenia. It has been shown that treatments designed to enhance social networks might have value in reducing the risk for suicidal behaviour amongst patients with psychosis.⁴²

Employment status was not found to be significantly associated with suicidality in this current study even though more than half of the group with suicidality were unemployed. Previous studies have, however, in contrast, demonstrated that unemployment is a significant risk factor for suicidal behaviour.^49,50

Education is presumed to bring about more economic empowerment and good quality of life and as a result reduces other risks of suicide such as poverty, homelessness and unemployment.^47,51 But this was not the case in our study as educational status was not found to be associated with suicidality (p = 0.112).

There was a significant relationship between the income received per month and suicidality as patients who earned more income were less likely to express suicidal behaviour than those earning less. This economic implication of schizophrenia as a chronic illness has been researched extensively.⁵²

In the same vein, this study revealed that the more patients spent on their illness monthly, the more they were likely to have suicidality.

There was no significant association between all the measured clinical variables and suicidal behaviour in this study group.

Generalisability of our study to the population will be limited because of bias inherent in tertiary-care based study with potential for more severe cases. This hospital-based study of outpatient clinic attendees may not adequately represent the general population of patients with schizophrenia. However, it provides data for comparison when future studies are conducted on similar sub-population in other mental health-related support service centres. There is a relative absence of local literature on the suicidality in patients with schizophrenia which would have enhanced useful comparison of methodologies and results. Cultural norms, such as moral objections to suicide, could have affected appropriate response to enquiries about suicidality. The researchers did not measure the severity of the psychopathology amongst the patients, so we do not know whether severity of the illness significantly associated with suicidal behaviours. The study also did not look at the prevalence of depressive symptoms in the study population. This would have been appropriate considering the established relationship between depression and suicide.

The strength of the study lies in the fact that international standardised instruments were deployed in all other measured aspects of the study except the sociodemographic and clinical questionnaire. Also, the study was able to subcategorise the suicidality module of MINI, which enabled us to determine the pattern of suicidality amongst the study group.