Catastrophic healthcare expenditure and coping strategies among patients attending cancer treatment services in Addis Ababa, Ethiopia

Hospital-based cross-sectional study design was employed. The study was conducted in Addis Ababa, the capital city of Ethiopia (the second-most populous country in Africa) and headquarters of African Union. Addis Ababa has 13 public hospitals, 32 private hospitals, and 93 health centers. From ten cancer diagnostic and treatment centers, about 150–200 new cases of cancer are registered monthly. Each month, about 1200 cancer patients visit these health facilities either for follow up or treatment. From the public hospitals, Tikur Anbessa Specialized Hospital (TASH) is the biggest teaching and referral public hospital with the highest number of cases. The study was conducted in TASH and three private hospitals (Hallelujah, Bete-Zata and Leghar General Hospitals) between January and March 2018.

The source population was all cancer patients attending treatment service in Addis Ababa healthcare facilities. Histological and/or pathologically confirmed cancer patients, patients who took at least one of the anticancer therapy options (chemotherapy, radiotherapy or hormonal/others), patients with a regular follow up for the last 1 year preceding the interview date, and with no co-morbidity and were volunteer to participate, constituted the study participants.

The sample size was determined using a single proportion population formula [17], and a total of 404 participants were recruited from one government hospital (TASH) and three private hospitals. TASH was included as it is the sole oncology referral and radiotherapy center in the country and it has also the highest number of cancer patients. Private hospitals were also selected based on their voluntariness to be enrolled and patient load. Participants were recruited using a convenience sampling technique and the number of participants to be drawn from each hospital was decided based on proportion to patient load.

A structured questionnaire (S1 Table 1 Data collection instrument) was developed from the WHO SAGE (Study of Global Ageing and Adult Health) study [18] and other relevant literatures [15, 19, 20]. This English version of the questionnaire was translated into Amharic (national language) and back-translated to English to ensure consistency. The questionnaire was pretested before the commencement of the data collection process. Appropriate modification and validation process were made based on the result of the pretest.

The data collection tool was consisted of six parts: (1) socio-demographic characteristics (age. Gender, occupation, marital status, education level and household size); (2) medical information (such as type of cancer, time of first diagnosis, treatment initiated, type of treatments taken and history of visit to a private health facility); (3 and 4) expenditures on the outpatient and inpatient services (including consultation, investigation, medicine and other relevant costs); (5) household essential consumption and income (weekly food and others spending, monthly house rent, cloths, transport and other charges, annual education payment, durable materials e.g., television, phone, furniture, vehicles, ceremonies and others spending, overall household yearly expenditure and income, and patient income if available); and (6) households financial situation outlook (rate of financial burden, coping mechanisms taken and its amount). Besides, patient medical charts were reviewed to collect relevant clinical information, investigations and treatments taken which also helped in estimating patient treatment/diagnostic expenditures.

Data was collected in a face-to-face exit interview and daily supervision was made by the principal investigator to ensure its completeness and consistency.

The magnitude of CHE was estimated using Wagstaff and Van Doorslaer [4] approach. When previous one-year patient households’ OOP expenditure for cancer care exceeded 10% of total annual household income, it was considered catastrophic. The overall, outpatient and inpatient cancer diagnostic and treatment service expenditures of the last 12 months were estimated and presented as average expenditure per patient. Therefore, these households incurred a catastrophic expenditure assumed as catastrophic payment headcounts (H_cat).

Healthcare service OOP expenditure is a direct payment made to healthcare providers on receiving a service excluding prepayment, reimbursement, and other sources of payment mechanisms [4, 5, 21]. However, households can use different coping strategies to solve their financial burden. Coping mechanisms may include any means of income obtained from household member and other savings including Eqqub and Iddir, any financial support from relatives, religious organization and other sources which are non-refundable. Income derived from selling household assets like property, livestock, jewellery, and other household items; and any types of borrowing from financial institutions or individuals could be sources of finance to cover any medical and non-medical expenses. These coping mechanisms could be rearranged as; savings, financial support, selling assets, and borrowings [7, 20]. Households had used different coping strategies, which can be classified into two categories: expenditure covered by only the households themselves and by other mechanisms to estimate the CHE level of cancer care.

Households’ subjective rate of self-reported financial burden was assessed by asking respondents to rate their current household economic situation compared to the past; as very good, good, medium/similar, bad and worse. Later, they were reclassified as “manageable” for very good, good and medium/similar and “unmanageable” for bad and worse.

Expenditure was estimated as all costs spent for cancer care before the interview time in the last year of medical service upon possible probing approaches to minimize recall bias. This expenditure includes both medical and non-medical expenses. Medical expenses were all expenditures related to consultation, investigation, medicine, bed and traditional medicine. On the other hand, non-medical cost includes transportation, food and other accommodation expenditures associated with the patient care. Expenses were also categorized as outpatient expenditures (consultation, investigation, medicine, and other spending related to outpatient care), and inpatient expenditures (consultation, investigation, medicine, bed and other expenses associated with inpatient care). There are controversies on estimating cost data, especially when the data have a skewed nature. Some studies reported “mean” as a reasonable choice, although it could be affected by the extreme values [22, 23]. However, others studies preferred to report using the median since it is not affected by skewness, although it only shows the position of the distribution [24]. Hence, we used both mean and median measurements to provide a full picture of the estimation, which might ease health policy decision making and resource allocation.

Household income and expenditure were measured based on respondents’ self-reported daily or monthly income and expenditure. Participants with in-kind income were approached for their type of income and amount. The in-kind income was then changed to monetary terms based on their current value to the local market. Besides, all expenditures were measured, and costs were reported as an expenditure (US$) per patient (1US$ equivalent to 23.41 Ethiopian Birr).

Socio-demographic, clinical and economic variables including type of health facility (private and public), kind of cancer (breast cancer, colorectal cancer, cervical cancer, nasopharyngeal cancer (NPC) and others), cycles of chemotherapy taken (less than or equal to three cycles, four to six cycles, more than six cycles and on other treatment options), history of a visit to a private health facility, gender, marital status,residence (from Addis Ababa and outside of Addis Ababa), the level of education, occupation, income, and expenditure were the variables considered as covariates in the analysis.

Data were coded, entered in to, and analyzed using STATA version 14 (https://​www.​stata.​com/​stata14/​). Descriptive statistics such as mean, median, standard deviation (SD) and Inter-Quartile Range (IQR) were used to compute socio-demographic, clinical and economic characteristics. Hosmer-Lemeshow test was employed for the goodness of fit test of the logistic regression model. A multivariable logistic regression model was used to assess the relationship between CHE and potential explanatory variables. The significance level was set at 5 and 95% confidence interval (CI).

A total of 404 participants were approached, and 352 study participants were interviewed, with a response rate of 87.1%. The majority (73.3%) of the patients were females, and 285 (81.0%) were married. Most (94%) of the respondents were from public hospitals (Table 1). More than three-fourth (85.2%) of the study participants were in the productive age category, and their mean (±SD) age was 48 (±13.2) years (ranged 19 to 87 years).

More than one-third (36.9%) of the participants were breast cancer patients, followed by cervical cancer 58 (16.5%) and colorectal cancer 46 (13.1%). The majority (89.0%) of patients have taken chemotherapy on their treatment course, and 155 (44.0%) of them were on 4–6 cycles of chemotherapy. All patients have used supportive treatments. The majority (63.4%) of respondents had a history of visit at private health facilities during the course of the disease (Table 2).

The average overall expenditure per patient in the last one-year was estimated to be $2366 (SD: $4262), median: $1709 (IQR: $1153–2424). The inpatient services accounted for two-thirds of the total expenditure with a mean cost of $1584 (SD: $4002), and a median of $1067 (IQR: $641–1580) per patient. The remaining expenditure was spent on outpatient service, with a mean of $782 (SD: $1468) and a median of $557 (IQR: $256–940).

The overall mean expenditure of patients with colorectal cancer who attended private hospitals, who had a history of visit at private health facilities, and those living out of Addis Ababa was higher than their counterparts (Table 3).

The direct mean and median medical expenditures per patient in the last year was estimated to be $1978 (SD: $3554) and $1394 (IQR: $917–1982), respectively. On the other hand, the patients had a mean and median non-medical expenditure of $388 (SD: $993) and $222 (IQR: $122–461), respectively. The medical spending constituted the highest expenditure (83.6%), taking a large share of the overall expenditure. The mean cost on traditional medicines was also estimated to be $7 (SD: $65).

The average headcount adult equivalent income and expenditure was found to be $1821 (median: $1220) and $1111 (median: $949) per year, respectively. The mean annual unadjusted household income and expenditure was estimated to be $4997 (median: $3076) and $3006 (median: $2563) per year, respectively. The incidence of incurring CHE was 74.4% (95%, CI: 69.6–78.7). Patients in the fourth income quintile had faced higher CHE. Regarding the expenditure quintile, patients in the middle quintile had encountered higher catastrophic cancer care expenditure (Table 4).

Based on participant’s self-report, it was estimated that 69% (95% CI: 64.0–73.7) of patients’ households faced with an unmanageable financial burden which could lead them to a financial crisis.

The multivariable logistic regression analysis revealed that taking greater than six cycles of chemotherapy (AOR: 3.64; 95% CI: 1.11–11.92), and age (AOR: 1.03; 95% CI: 1.01–1.06) were factors statistically associated with CHE among cancer patients. In this regard, the tendency to encounter CHE of cancer care was associated with older age and an increased number of cycles of chemotherapy (Table 4).

The majority of households had used their household savings (85.5%) for healthcare payments. However, a considerable percentage (43%) of households have received substantial financial support from their relatives, religious and other non-governmental organizations to cover the financial burden imposed on the household. The remaining expenditure was covered by selling assets and borrowings, as illustrated in Table 5.