Study design, period and area
An institution based cross sectional study design was employed. The study was conducted from July 1-30, 2023, in the capital city of Ethiopia, which is Addis Ababa. It was conducted at three selected public hospitals in Addis Ababa; Eka Kotebe General Hospital, St. Paul Hospital Millennium Medical College and Yekatit 12 Hospital Medical College.
Study population
The study population included all selected epilepsy patients in the follow-up unit of Eka Kotebe General Hospital, St. Paul Hospital Millennium Medical College and Yekatit 12 Hospital Medical College from July 1-30, 2023, who fulfilled the inclusion criteria of the study.
Patients aged 18 years and above with a diagnosis of epilepsy and who were receiving treatment for at least 6 months were included in the study. Patients who were critically ill and unable to communicate during the time of data collection were excluded.
Sample size determination
The sample size was calculated using Single population proportion formula with 95% confidence level, a proportion of 37% from a previous study held at the University of Gondar Comprehensive Specialized Hospital, Northwest Ethiopia (5), a 5% margin of error , a 10% non-response rate and a design effect of 1.5. The final sample size was 591.
Sampling procedure
Among the 12 public hospitals in Addis Ababa three (Eka Kotebe General Hospital, St. Paul Hospital Millennium Medical College and Yekatit 12 Hospital Medical College) were selected by using simple random sampling (lottery) method. The data were collected over a one month period and before the data were collected the total number of patients who visited the epilepsy OPDs of the three hospitals for the previous 12 months was taken from patient records and the average number of patients per month was subsequently calculated. After that the required sample size was proportionally allocated for each hospital.
Where n= the number of epilepsy patients in the follow-up unit of each hospital in the one month period
Ni= final sample size of the study
N= Total number of epilepsy patients in the follow up unit of Eka Kotebe General Hospital, St. Paul Hospital Millennium Medical College and Yekatit 12 Hospital Medical College over a one month period
To select study participants from the three hospitals, a systematic random sampling method was used. The k value (sampling interval) was calculated by dividing the total study population who had followed up for one month before the data collection period (1250) by the total sample size (591). The sampling interval ( k ) is 1250/591=2. The first study participant was selected by lottery method from each hospital separately and the next study participants were selected at a regular interval (every 2) individual at the outpatient department.
Operational definitions
Depression: was assessed by the Patient Health Questionnaire-9 which has been validated and extensively used in Ethiopia(17). In this study a PHQ-9 total score of ≥10 was considered as having depression.
Substance use: was measured using the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST), a structured interview developed by the World Health Organization (WHO) (18).
Perceived stigma: was assessed by the kilifi stigma scale (KSS) which was developed and validated in kilifi, Kenya and adopted in Ethiopia(19). A score above the 66th percentile of the data measured by the kilifi stigma scale of epilepsy indicates the presence of perceived stigma(20).
Social support: was measured with the Oslo-3 items social support scale (Oslo SSS). A score of 3–8 is considered “poor social support”, a score of 9–11 is considered “moderate social support” and a score of 12–14 is considered “strong social support(21).
Data collection procedures, and quality assurance
The data were collected using interviewer administered structured questionnaire and a data abstraction tool. The questionnaire had four parts. The first part contains sociodemographic data , the second part contains clinical and substance related factors, the third part contains psychosocial factors and last part contains a patient health questionnaire-9 to assess depression. The data abstraction format was designed to collect data from patient charts about EEG reports.
The questionnaire and data abstraction tool were pretested prior to the actual data collection on 5% of the final sample size(randomly selected charts and patients ) at Eka Kotebe General Hospital, St. Paul Hospital Millennium Medical College and Yekatit 12 Hospital Medical College and necessary amendments to the data collection tool were made. The study units for the pretest were not included in the final data collection. To guarantee that all participants could comprehend it, the questionnaire and data abstraction tool were translated into Amharic and then back translated into English to maintain consistency.
Depression was assessed via the Patient Health Questionnaire-9 and the internal consistency of the Cronbach’s alpha of the PHQ-9 in this study was 0.857. Perceived stigma was measured by kilifi stigma scale with a Cronbach’s alpha of 0.774. Social support was assessed by Oslo-3 items social support scale and its internal consistency of the Cronbach’s alpha in the current study was 0.934.
Four psychiatric nurses and three supervisors received training from the principal investigator on the basics of the questionnaire and data abstraction tool on June 29, 2023, for one day. Data were collected from July 1-30, 2023.The supervisors reviewed the process of data collection on daily basis to verify its completeness, correctness, and consistency by checking the questionnaire and data abstraction tool.
Data management and data analysis
The collected data were coded and entered into Epi Info, cleaned, stored and exported into SPSS. The data were analyzed using SPSS version 27.0 software. A descriptive statistical method was used to generate frequencies and percentages. Categorical variables were summarized using proportions and numerical variables were checked for normality and summarized with a mean value (± standard deviation). If normality failed (skewed data), the data were summarized as median( inter quartile range).
A binary logistic regression model was used to assess the associations between independent variables and depression among epileptic patients. Univariate analysis was used and independent variables with p-value of ≤ 25% were selected for multivariable binary logistic regression analysis. AORs were calculated and independent variables with p-value of ≤ 0.05 were considered significantly associated with depression among epileptic patients.
The final model fitness was checked using the Hosmer- Lemshow goodness of fit test. The adequacy of the final model was assessed using goodness of fit test and the final model fitted the data well (Pearson x 2 =13.997, p-value = 0.082)