Extent and associated factors of suicidal behavior among postpartum mothers attending public health centers in Addis Ababa, Ethiopia | BMC Psychiatry

The questionnaire was initially in English and was translated into Amharic (local working language). The questionnaire includes socio-demographic characteristics (age, education, marital status, grew up with, employment, relative wealth and age at marriage), psychosocial characteristics (verbal abuse, physical abuse, history of rape, social support, substance abuse user husband, sexually unfaithful husband , alcohol consumption, smoking, khat chewing) and clinical characteristics (parity, pregnancy, history of abortion, infant health status, history of depression) [5, 9, 22].

Suicidal behavior was measured using the Suicidal Screening Tool that is part of the International Neuropsychiatric Mini-Interview, and it was defined based on criteria from the Diagnostic and Statistical Manual of Mental Disorders-IV and includes the following questions. In the past month: Have you wished you were dead? Have you thought about killing yourself? Did you want to hurt yourself? Have you attempted suicide? Have you ever attempted suicide? Have you planned how to kill yourself?

If the mother answers “yes” to any of the six questions, she is considered at risk for suicidal behavior [23]. This tool is short and structured and has good validity and reliability in screening for suicidal behavior [24].

The Oslo-3 Social Support Scale was used to assess social support [25], which assesses the number of close confidants, the amount of care reported by others, and the perceived ease of asking neighbors for help. Substance use was measured using the Alcohol, Smoking and Drug Abuse Screening Test [26]. Abuse was assessed using an Abuse Rating Scale [27]which is a useful tool for assessing lifespan and current violence among mothers in clinical settings.

The mothers’ current wealth was estimated by asking them how wealthy they thought they were compared to other people in the community. [28]. They were also asked to compare their personal income to that of others, categorizing it as lower than that of others, similar to that of others, or higher than that of others.

Operational definitions

Social support – If the mother had close confidants, the sense of interest from others and the relationship with neighbors with the accessibility of practical help is considered social support.

Substance use – Recurrent consumption of alcohol, chewing khat and smoking by the mother is considered substance use.

Abuse – If the mother is manipulated and controlled by someone emotionally, physically without her interest is considered abuse.

Sexually Unfaithful Husband – To know about the sexually unfaithful husband, mothers are asked about the history of cheating on her husband and if true then the husband is labeled as sexually unfaithful.

Data quality assurance and analysis

The data collection instruments were pretested on 5% of non-study participants who met the inclusion criteria to check accuracy of responses, clarity of language, appropriateness of data collection tools, estimate necessary time and the necessary modifications were considered according to it before the actual data collection. To confirm that the questions are internally consistent, we validated by Cronbach’s Alpha test and pilot tests. We did Cronbach’s Alpha test for all questions and the result was 72%, which indicates excellent internal consistency in the answers..

Five nurses and a supervisor who were not employees of the selected health centers were assigned as data collectors and they were trained for one day on information about the purpose of the research, eligible study subjects , data collection tools and procedures, and interview methods. After data collection, the completed data was entered into Epi data 4.6 and exported to SPSS version 25 statistical software for further analysis. All variables with a p-value ≤0.25 were taken in the multivariable logistic regression model to control for all possible confounders. A logistic regression model with adjusted odds ratio (AOR), 95% confidence interval (CI), and p-value ≤ 0.05 was used to identify predictors of the outcome variable.

Hosmer and Lomeshow’s goodness-of-fit test was used to assess the overall goodness of the model with the fitted data. In addition, the receiver operating characteristic curve (ROC curve), which is produced by plotting the true positive rate against the false positive rate of the model at different thresholds, was used to verify the discriminating power of the model, and an area under the curve value (AUC) greater than 70% was used to categorize the model, as it can adequately classify individuals with and without the results.

Binary Logistic Regression Assumption

Hosmer and Lome show that the multi-logistic regression test revealed that the model constituting these variables is good at predicting the outcome variable, i.e., the magnitude of suicidal behavior (P-value=0.530). Indeed, the discrimination power of the model was verified by the operating curve of the receiver and the value of the area under the curve (AUC value = 0.730) showed that the model is excellent for discriminating individuals with suicidal behavior. (true positives) and individuals without suicidal behavior (Fig. 1).

Fig. 1

Graph of the ROC curve showing the discriminatory power of the model towards the extent of suicidal behavior among postpartum mothers attending Addis Ababa Public Health Centers, Addis Ababa, Ethiopia, 2021