Study setting and period
An institutional-based cross-sectional study design was conducted among adult diabetes patients attending at University of Gondar Comprehensive Specialized Hospital (UoGCSH) chronic illness follow up clinic from March 28 to May 15, 2021. The hospital is situated in the town of Gondar,738 km from Addis Ababa, Ethiopia’s capital. It is a teaching and research center and serves as a referral hospital for more than 7 million people in the northwest part of the country. The hospital has 28 wards, 15 different outpatient services, and 960 beds. During the study year, over 4,000 patients with diabetes received medical treatment and follow-up care at the chronic disease ward [38].
Study population and sampling procedure
The population was all adult patients diagnosed with diabetic disease who had been attending a diabetic follow up at UoGCSH for more than one year were a source of population, all adult diabetic patients who were attending this hospital for more than one year and had follow up care during the data collection period were the study population, whereas patients enrolled in community-based health insurances and fee waiver package users were excluded from the study. Additionally, diabetic patients who returned to the hospital a second time or more during the data collection period were also excluded from the study.
Sample size was determined using the single population proportion formula \(\:n=\frac{{\left({z}_{a/2}\right)}^{2}P\left(1-p\right)}{{d}^{2}}\) by considering the following assumptions. The proportion (p = 59.6%) [39], 95% confidence interval, 5% margin of error and 10% non-response rate. Therefore, the final calculated sample size was 407. A systematic random sampling method was used to recruit the study participants. The patient chart and the appointment register were used as a sampling frame. Based on average daily patient flow, the K interval was calculated. Accordingly, the first patient was randomly selected using the patient cards during the day of the hospital visit, and the next patients were selected every 4th interval until the required sample size was reached.
Study variables
The outcome variable for this study was catastrophic out-of-pocket health expenditure.
Socio-demographic and economic factors (age, sex, religion, ethnicity, educational status, occupational status, marital status, family size, place of residence, distance from a hospital, means of transport, and wealth status) and clinical characteristics of respondents (duration of illness, complications of DM, comorbidity illness, frequency of visits, type of treatment service, availability of medical supplies, history of admission, and duration of admission) were considered predictor variables for this study.
Data collection procedures
Data were collected using a semi-structured questionnaire administered by an interviewer. It was developed through reviewed of previously published articles [39, 40]. The English version questionnaire is presented as Supplementary file (S1 file). The questionnaire was first prepared in English and translated to Amharic, then back to English. It consisted of demographic and socioeconomic characteristics, clinical characteristics, expenditure of DM (direct medical expenditure, and direct non-medical expenditure), coping strategies, household expenditure, and wealth status for rural and urban households.
Three BSc graduate health officers for data collectors and one MSc holder for supervisor were recruited, and a one-day training was given about the objective of the study, the definition of terms, issues of confidentiality and privacy, data collection techniques before actual work.
Pre-testing was done on 21 patients at Felege Hiwot Comprehensive Specialized Hospital, found in Bahir Dar City, then the necessary modifications were made based on the pre-test findings. The exit interview was conducted after the patients finished their treatment sessions. During the data collection process, supervisors and the principal investigator were checked the data accuracy, consistency, and completeness of the data on a daily basis.
Measurement
Measurement of out-of-pocket health expenditure
Out-of-pocket health expenditures refer to all expenses incurred by diabetes patients for diabetes treatment and follow-up over the past 12 months. These OOP health expenditures include both direct medical costs and non-direct medical costs. Expenses such as consultations, diagnostic tests, and treatment services were categorized as direct medical expenditures. On the other hand, costs related to transportation, food, and lodging for both the patient and their companion were classified as direct nonmedical out-of-pocket expenditures.
The OOP health expenditures were estimated using a micro-costing approach with a bottom-up technique from the patient perspective. The annual household expenditure and OOP medical expenditure of adult diabetes patients were reported as average per patient. The unit of measurement for OOP medical expenditure was the Ethiopian Birr, which was then converted into USD ($1 USD equivalent to 34.9505 ETB) based on the average annual exchange rate for 2020. Household expenditures were measured based on respondents’ self-reported expenditure data.
Measurement of magnitude of catastrophic out-of-pocket health expenditure
The magnitude of catastrophic OOP health expenditure was measured using Wagstaff and van Doorslaer’s approach at a 40% threshold as a share of annual household non-food expenditure. This approach measures catastrophic health expenditure using catastrophic headcount, catastrophic overshoot, and mean positive overshoot (MPO) at various thresholds of household non-food expenditure [41].
Catastrophic headcount: is the proportion of households incurring catastrophic OOP expenditures for diabetes payments that exceed the 40% threshold of annual household non-food expenditure [42]. The headcount estimate is calculated using Eq. 1.
$$\:Headcount\:\left(H\right)=\:\frac{1}{N}{\sum\:}_{i=1}^{N}{E}_{i}\:$$
(1)
Where: \(\:H\) = head count ratio,\(\:{E}_{i}\) = an indicator function, \(\:{E}_{i}=1\) if \(\:\frac{DM{x}_{i}}{{x}_{h}}>z\) (i.e., the out-of-pocket expenditure for diabetes care exceeds a 40% threshold), \(\:{E}_{i}=0\), Otherwise, \(\:DM{x}_{i}\) = out-of-pocket expenditure for diabetes care, \(\:{x}_{h}\) = total household expenditure and \(\:z\) = threshold level (in this case 40%). The catastrophic headcount did not include data on the degree (intensity) of catastrophic expenses (by how much household OOP payments exceed the catastrophic threshold). Intensity is thus determined by catastrophic overshoot and mean positive overshoot (MPO) [41].
Catastrophic overshoot: the mean value of household OOP expenditure for the treatment of diabetes, as a percentage of total household expenditure, exceeding a 40% threshold of annual household non-food expenditure [43]. The overshoot is calculated using Eq. 2.
$$\:Headcount\left(O\right)=\:\frac{1}{N}{\sum\:}_{i=1}^{N}{O}_{i}$$
(2)
Where: \(\:O\)= \(\:{E}_{i,\:}\)\(\:1\) if \(\:\frac{DM{x}_{i}}{{x}_{h}}>z,\:\)\(\:{E}_{i}=0\), Otherwise, \(\:DM{x}_{i}\) = out-of-pocket payment for diabetes care, \(\:{x}_{h}\)= total health expenditure.
Mean positive overshoot: is defined as the mean level by which out-of-pocket expenditure on diabetes treatment, by households reporting catastrophic health expenditure, exceeded the 40% threshold of annual household non-food expenditure [44]. It was calculated using Eq. 3.
$$\:MPO=\frac{\text{O}}{\text{H}}$$
(3)
Where, H = Catastrophic headcount O = Catastrophic overshoot.
Statistical analysis
Data were cleaned, coded, and entered into Epi-data version 4.6 and exported to STATA Software Version-16 for analysis. Descriptive statistics such as frequency, mean with standard deviation, and median with interquartile range were calculated. The wealth index was constructed into five categories using principal component analysis. A bi-variable logistic regression model was used to select variables for the multivariable logistic regression analysis, considering those with a p-value of less than 0.2. Independent variables associated with catastrophic OOP health expenditure were identified based on the adjusted odds ratio (AOR), 95% confidence interval (CI), and a p-value of less than 0.05. In the final model, variables with a p-value < 0.05 were considered significant independent factors associated with catastrophic OOP health expenditure.
link