Study design and data collection
In this cross-sectional study, information on breast cancer was obtained through face-to-face interviews and completing a questionnaire by 138 women with this disease in Urmia, Iran, in 2019. This information covered the demographic data of patients and heads of the household, household expenses and income, clinical data on the disease, and funding sources for the disease management. Samples were randomly selected from a list of registered patients in the private and public specialized centers of Urmia University of Medical Sciences. To minimize the number of missed cases, at least three times at two-week intervals, contact was made with any respondent who was not ready for the interview for any reason, and an interview was conducted with her. The inclusion criteria were having at least 25 years of age, passing at least one month since the definitive diagnosis of the disease, and been living in Iran for the past year.
Measuring catastrophic breast cancer expenditures
If a household is forced to cut their living subsistence over a while due to OOP payments for health care services, the costs are catastrophic. Technically, health expenditure is considered catastrophic when OOP health expenditures exceed a specific household capacity ratio to pay subsistence. This threshold is set by the World Health Organization (WHO) at 40%, but other standards, such as 20% and 10%, have been introduced in various related studies. Household capacity to pay is obtained by subtracting the average monthly cost of subsistence from the average monthly effective income (total consumption expenditure) of the household over the past year. In other words, the capacity to pay is defined as non-subsistence expenses. Based on the WHO recommendation, we consider subsistence expenses to be equal to household food expenditures. [11] Food expenses make up a large portion of household expenses, have a low income elasticity, and are strongly influenced by household members [12]. Household food expenditures include the costs incurred by the household in purchasing foodstuffs and the financial value of the food products produced and consumed by the household itself. Food expenses incurred by households in restaurants and hotels are excluded from these calculations. Using the variable of household expenditures has been that better than income can show the purchasing power of households, especially in developing countries [13]. Given that household expenditures can have an unbalanced distribution during different months of the year, we asked people about the average monthly household expenditure over the past year, which included all of following items; food, beverages, recreation, education services, hotel and restaurant, clothing and footwear, cigarettes and tobacco, house and shop rent, housing, water, fuels (gas, electricity, and other possible fuels), transportation، communication، household appliances، furniture، health, and financial value of any consumption of household products (agricultural, services, industrial, etc.)
To considering the diminishing marginal utility of remained household consumption expenditure after subtracting the Out-of-pocket (OOP) health payment from effective income for different SES quintiles, we used three levels of non-food expenditure thresholds; 10%, 20%, and 40%. OOP payments refer to the spending made by households at the point they receive breast cancer services, which cover the periodical visits, diagnostic services, hospitalization care, treatment services (drugs, chemotherapy, radiotherapy, surgery, and others), rehabilitation services, home cares provided by the physician, nurse, and household members, and finally receive informal care from traditional therapists.
Measuring impoverishing due to breast cancer spending
The impoverishing payment is measured by the proportion of households that falls below the absolute poverty line after breast cancer spending is subtracted from total household consumption [14]. Such households can still not meet their basic food needs, even if they spend all their remaining expenses on them. According to a report by the Iranian Ministry of Labor, the monthly cost of a subsistence basket for each person was 11393940 Rials (84.40 PPP $US) in 2019 [15]. Therefore, by multiplying this value by the standardized household dimension, each household's absolute poverty line is calculated. If the total household expenditure was less than their equivalent absolute poverty line, that household was considered poor. If the total remaining household expenditure goes below this estimated line due to the cost of treating breast cancer, that household will incur impoverishing spending. Also, we asked the participants the potentially unmet health needs of patients due to the cost of treating breast cancer and the sources of funding for the disease.
Statistical analysis
Households were disaggregated into SES quintiles based on their monthly consumption expenditures, where SES classifications were standardized by Adult Equivalent (AE) values of consumption expenditures. For this, we applied the formula presented by Cirto and Michael [16] as follow:
AE = (A + aK) θ
Where A = number of adults (aged over 18 years), K = number of children, α = cost coefficient for children, and θ = degree of economies of scale. The recommended values for α and θ for developing countries are 0.4 and 1.0, respectively [16].
For statistical comparisons of the incidence of catastrophic and impoverishing spending on breast cancer treatment between different sociodemographic subgroups, we applied the independent t-test and one-way ANOVA respectively for two and more than two separate subgroups. In this analysis, the p-values which are equal or below 0.05 are considered to be significant.
To identify the factors affecting on catastrophic and impoverishing health expenditures for breast cancer treatment, we performed the two separate multivariate logistic regression models with the following dependent variables:
- Disease characteristics (cancer type, disease duration, treatment types).
- Patient characteristics (age, educational level, place of living, marital status, insurance status, household size).
- Head of household characteristics (age, educational level, marital status).
Those variables that had multicollinearity problems were omitted from the models. We estimated and reported the odds ratio, 95% confidence interval (CI), and p-value statistics for each variable. Normality distribution of the variables was checked by One-Sample Kolmogorov–Smirnov test. All statistical analyses were performed by STATA version 15.0 software (Stata Corp, College Station, TX).