Assessment and associated factors of suicidal behaviour among cancer patients visiting the oncology outpatient unit in Mekelle oncologic clinics, Tigray, Ethiopia: a cross−sectional study

Cancer is a group of diseases characterized by the uncontrolled growth and spread of abnormal cells. According to estimates from the International Agency for Research on Cancer, there will be 17.0 million new cancer cases in 2018 worldwide. Depression, age, sex, divorced, and hopelessness are of most factors can patient with cancer result in suicidal behaviour. The purpose of this study is to identify and associated factors of suicidal behaviour among cancer patients in Mekelle, Ethiopia. The cross-sectional study design was employed with a total of 345 study subjects in Mekelle, Ethiopia. Suicidal behaviour was measured by the Suicidal Behavior Questionnaire-Revised (SBQ-R) scale. Bivariate and multiple logistic regression analyses were performed to determine between the explanatory and outcome variables.

A study done in Colorado, Western United States showed the magnitude of suicidal behaviour ranged greatly from 0.7-46.3%. Commonly identi ed risk factors for suicidal behaviour were included age, sex, and disease/treatment-related characteristics, as well as psychological constructs including depression, anxiety, hopelessness, existential distress, and social support [12].
A study done in Taiwan indicated that the rst month of cancer diagnosis was associated behaviour of suicide with in the rst month (AOR = 3.47, 95% CI 52.60-4.62) and the sixth month following a cancer diagnosis (AOR = 1.53, 95% CI 51.11-2.12) [13].
A study was done in South Korea showed that the risk of suicide attempts was signi cantly higher in participants diagnosed with cancer before 45 years of age compared with those diagnosed at 45-64 years (AOR = 3.81, 95% CI 1.07 − 13.60, P = 0.039), and the higher risk of suicide attempts with borderline signi cance was found in those for whom more than 10 years had passed since diagnosis compared with those for whom the diagnosis was made only 2-10 years ago (AOR = 3.38, 95% CI 1.98 − 11.70, P = 0.055) [10].
Cancer patients are at high risk for suicide, particularly when they are informed about the cancer diagnosis or hospitalized for cancer treatment. Therefore, oncology tertiary healthcare settings like Ayder comprehensive specialized hospital may represent an ideal setting to identify and treat suicidality in cancer patients in Mekelle, Ethiopia.

Methodology Study areas and period
This study was conducted in 2019 at Mekelle, oncology clinics, Ayder Comprehensive Specialized Hospital. According to the Ethiopian central statistical agency report; the total population in 2012 Ethiopian Calendar has been 420,350, which is located in Tigray regional state 783 km away from Addis Ababa, the capital city of Ethiopia. Oncologic services are given by oncologists and general practitioners. More than 36 health professionals work in an oncologic clinic, out of them, 4 were oncologists. The current ow of oncologic patients on average was 1500 patients per month in Ayder Comprehensive Specialized Hospital, which has 25 beds for inpatient services. The study period was from April 10 to June 10, 2019.

Study design
An institutional-based cross−sectional study was conducted.

Source population and study population
Source population All oncologic outpatient unit visitors were in Mekelle, oncologic clinics, Tigray, Ethiopia.

Study population
All oncologic sampled outpatient unit visitors were in Mekelle, oncologic clinics, Tigray, Ethiopia.

Eligibility criteria
Inclusion criteria Oncologic outpatient unit visitors, patients who had diagnosed known cancer and aged 18 years and above, patients attending treatment at the oncology clinic during the data collection period were included in the study.

Exclusion criteria
Patients who were unable to communicate, severe pain, unable to sign a verbal informed consent and those who had decision incapacity were not be included in the study.
Sample size and sampling procedure Sample size The sample size was calculated using a single proportion formula and since there is no study published in Ethiopia show the prevalence so that we used 50%. Other assumptions made during the sample size calculation are 5% marginal error (d) and a con dence interval of 95% (z α/2 = 1.96). Based on these assumptions, the sample size calculated as follows: n=Z 2 pq ̸ d2 n = (1.96) 2 (0.50.5) ̸ (0.05) 2 = 384. But during that study, 39 respondents missed from the study though we tried a 10% non-response rate, no participants participated. So the nal sample size was 345.

Sampling technique and procedure
The systematic sampling technique was employed. The study participants were proportionally taken. Since the ratio of 1500 samples to 345 was 4, we took every 4 samples were selected.

Data collection procedure
Face to face interview method using a structured questionnaire was used in this study to identify the magnitude and associated factors of suicidal behaviour such as (1) sociodemographic information, (2) biopsychosocial and (3) suicidal behaviour.
The SBQ-R is a self-report measure of suicidal behaviour. This shortened version of the SBQ consists of four questions to assess suicidal behaviour history, current suicide status and self-appraisal and expectancies about the future likelihood of engaging in suicidality. The magnitude of overall suicidal behaviour (as de ned by SBQ-R a total score ≥ 8 for adult clinical population) and the total score ranges from 3 to 18.
The Oslo 3-items social support scale (OSSS-3) was used to measure the strength of social support. The scores range from 3-14. A score ranging between 3 and 8 is classi ed as poor support, a score between 9 and 11 as moderate support, and a score between 12 and 14 as strong support. These three items were considered to be the best predictors of mental health, covering different elds of social support.
Data were collected by seven B.Sc oncology professionals having previous experience of data collection. The principal investigator checked the lled questionnaires for consistency and completeness each day. The questionnaire was translated from English to Tigrigna language by English teacher and retranslated to English by another expert in English. This primary version was made to compare with the original English version to resolve inconsistencies and then the data collectors, who are Tigrigna native speakers, collected data in the Tigrigna questionnaire.

Study variables
Dependent variables Suicidal behaviour.

Page 7/22
Socio-demographic, medical and psychiatry illnesses and substance.

Operational de nitions
Suicidal behaviour: after the diagnosis of cancer, the patient experience of having recurrent thought of death, ideation, intention, and plan to kill oneself and having organized plan; and based on SBQ-R scale scores ≥8 from 18, the total scale scores range from 3 to 18 [16]. The pre-test was conducted on a sample of 10% (34 samples) of the total study population in Quiha General Hospital before one month of data collection and a common understanding was reached between the data collectors. The pre-test questionnaires were not included in the analysis as part of the main study. Data collection was collected within 50 working days. Regular supervision by the principal investigator was done. During data collection, lled questionnaires' were checked for completeness and consistency daily.

Data analysis procedure
After data collection, lled questionnaires were coded. The data were entered using Epi data version 4.2 statistical software to minimize error that occurs during data entry and exported to SPSS; and analyzed using SPSS version 25. Data cleaning was performed to check for frequencies, accuracy, and consistencies and missed values and variables. The nding of this study was presented using text and tables form the result of frequencies and crosstabs.
A bivariate logistic regression model analysis was done to see the association between the risk factors and outcome variables. Multivariable logistic regression analysis was employed by selecting only variables with a P-value < 0.25 in the bivariate analysis. The odds ratio with 95% CI was used to measure the strength between dependent and independent variables at P-value < 0.05 to determine the level of statistical signi cance. Variables with the P value less than 0.05 in multivariate regression were considered to be potential predictors for suicidal behaviour.   Psychiatric illness that comorbid with cancer includes PTSD (post-traumatic stress disorder), OCD (obsessive-compulsive disorder) and BDD (body dysmorphic disorder).
Other A includes weakness, sepsis, traumatic brain injury (TBI) Other B includes PTSD, OCD and BDD.
Other C includes brain cancer, uterine cancer Psychiatric illness that comorbid with cancer includes PTSD (post-traumatic stress disorder), OCD (obsessive-compulsive disorder) and BDD (body dysmorphic disorder).
Other A includes weakness, sepsis, traumatic brain injury (TBI) Other B includes PTSD, OCD and BDD.
Other C includes brain cancer, uterine cancer Psychiatric illness that comorbid with cancer includes PTSD (post-traumatic stress disorder), OCD (obsessive-compulsive disorder) and BDD (body dysmorphic disorder).
Other A includes weakness, sepsis, traumatic brain injury (TBI) Other B includes PTSD, OCD and BDD.
Other C includes brain cancer, uterine cancer Substance Regarding the frequency of taking the substance, (10.5%%) participants that had no suicidal behaviour and (15.9%) participants that had suicidal behaviour smoked a cigarette (Fig. 1).

Comorbid medical and psychiatric illness
From all respondents participated in study 345 (20%) of them had suicidal behaviour. Out of those participants with comorbid physical illness had 37.7% suicidal behaviour. The magnitude of suicidal behaviour among cancer patient with attempt suicide previously was 76.8%.
The magnitude of suicidal behaviour was higher in respondents with stomach cancer at 46.4%. Regarding the history of psychiatric disorder, schizophrenia 42.9%, major depressive disorder 28.6%) and bipolar I disorder 14.3% had suicidal behaviour.
The distributions of suicidal behaviour among patients who had a family history of suicide attempt 15.9% and the stage of cancer participants with local stage were 42.0% (Table 3).  Respondents with poor social support (55.1%) and Moderate social support (39.1%) had a higher prevalence of suicidal behaviour than with strong social support 5.8% (Fig. 1).

Suicidal behaviour Assessment (SBQ-R)
This study found that (80.8%) never lifetime suicidal ideation, intention and attempt and (68.1%) suicidal participants had suicide attempt history. The Frequency suicidal ideation in the past year more than half of the participants reported never and one-fourth participants once a time in the past year (Table 3).

Bivariate regression of analysis of suicidal behaviour
In the bi-variable logistic regression analysis, variables such as diagnosis, educational status, psychiatric comorbidity and alcohol use, hospital admission, current treatment, family history of suicide and previous suicidal attempt were a candidate for multiple logistic regression with (P − value < 0.25) ( Table 4 ). This study showed that the magnitude of suicide behaviour was 20%.this lower than a study done in Colorado (46.3%) [12], this discrepancy due to sample size and tool. Whereas it is higher than studies conducted in USA (0.07%), Canada (15.7%), China (15.3%), and Korea (11.4%) [7,8,9,10]. This might be due to sample size, settings, scale, infrastructure and domestic, and community-based study in the USA.
In the study, cancer like cervical cancer was 3.3%, which is lower than a study done in USA [7]; the magnitude of cervical cancer was 78.1% and this discrepancy might be due to community-based study variables were found to be statistically signi cant at the level of P < 0.05 and one variable as protective factors.  Current treatment regimen (surgery and palliative care) were signi cantly protective factors for suicidal behaviour and this result is contrasted with a study done in China [9), which were associated factors. It might be due to tool, settings and sample size of the population.
No comorbid physical illness and suicide attempt were signi cant factors of suicidal behaviour but not in other studies.

Conclusion
No comorbid physical illness, suicide attempt and current treatment regimen were signi cant factors of suicidal behaviour.
Recommendation i. Oncologic professionals should assess patient suicidal risk assessment routinely and should put the diagnosis with suicidal if the client is suicidal so that every professional focuses on management besides the medication.
ii. Oncologic professionals should assess patient whether he/she has a comorbid physical illness or not to prevent the complication of cancer together with suicide which in turn refused to take proper management because of hopelessness.
iii. Educate the family/caregivers of suicidal patients with previous attempts, suicidal thought and intention to have closely followed up.
iv. It is also recommended researchers to conduct further research studies.

Limitation Of The Study
Inspite of providing valuable baseline data, there are also some limitations encountered: Social interest bias as the data was collected by a face to face an interviewer-administered approach. So the respondents might reply in favour of others that are either over-reporting or underreporting because of the illness severity.
Recall bias; there might be forgetfulness and there was not cross-check about the information they gave us because treatment and/or healthcare service.
In this study, only adult oncologic patients were included, so it is di cult to generalize all oncologic patients because children and adolescents oncologic patients are not included in the study.

Abbreviations
Availability of data and materials Permission letter was obtained from Mekelle University to oncologic clinics, and nally, the letter was distributed to oncologic professionals, data collectors, who work in the oncologic clinics. Verbal consent was attained from each participant before starting data collection. Study participant had the right to withdraw from the study at any time and information was recorded anonymously.

Consent for publication
Consent of publication for this research is Mekelle University, College of health science, oncology and psychiatry department and participants.