All estimates were based on data from public hospitals and Insurance Services. These estimates included economic data, lost labor productivity due to illness, and health indicators. In this study we applied the COI method. This method was developed in more detail by Rice et al. This method shows the social burden of the disease as a quantitative (monetary) value, and since it easily calculates the cost of human capital (lost productivity), it is used more today than other methods. We measured direct and indirect costs of prostate cancer, during the one year after the diagnosis(15-21). We estimated the cost of illness from three perspectives: society, the financer (e.g., insurance services), and the patient. This approach may help health planners at different levels to use the results of this study.
We used a prevalence-based approach to estimate the costs. This approach uses the costs of an illness in a given period, regardless of when the disease occurred. This approached was used since it is known that the incidence-based approach underestimates the costs for cancers with a high survival rate. In COI, the structure of human capital is usually targeted by investment based on formal and informal education formulated in the form of skills, knowledge, competence and personal experience. Thus, the individual’s lost productivity related to disease is examined socially (22-25). A bottom-up approach was used in cost estimation due to its higher precision. This approach consists of two stages: in the first stage, health data are measured and quantified, and in the second stage, the cost per unit of service used in production or consumption for specific medical services as well as health care is determined(13).
The participants in the study were selected from the Ghazi hospital in Tabriz and Labbafinejad Hospital in Tehran, Iran in 2017. The selection method was simple random selection by selection of consecutive patients of the cancer centers. At this stage, a questionnaire was used for two time periods (2 months). Questionnaire (direct and indirect medical and non-medical expenses) was completed for patients who received care or referred for follow-up.
Costing
We divided direct costs (DC) in our study into medical (DMC) and non-medical costs (DNMC). Direct medical costs were defined as the sum of the costs of diagnostic services (DSC) (such as imaging and cystoscopy), laboratory and pathology services (LC), therapy costs (TC) (including radiotherapy, hormone therapy, surgery and medicine ), in-patient costs (IPC), rehabilitation costs (RC), out-patient costs (OPC) such as physician visits, emergency costs (EC) and costs of medical devices (MDC). Direct non-medical costs included the costs of transportation (TPC), accommodation (AC), and food (FC) attributable to the disease.
Indirect costs included "absenteeism" (AL), defined as the loss of productivity due to the absence from the workforce (which also equals the income loss to the patient and his companion) as calculated by the human capital method and the loss of productivity due to the “premature death” (mortality cost) (MC). Only the income loss and mortality costs of patients below the age of 65 were included in the calculation for indirect costs. The income loss for all patients' companions was included in the calculation regardless of the patients' age. The daily wage of each patient and his companion was multiplied by hospitalization days to calculate the income loss (Appendix 1).
The productivity loss due to premature mortality was calculated by first subtracting the age of the patient from 65, multiplying it by the yearly income of the patient, and multiplying the resultant number by the one-year mortality rate for prostate cancer in Iran (19% as found in the earlier studies(26). For the patients with missing information regarding their income, we used the minimum wage in the country for the year of the study. We used a discount rate of 5.8%, as found in two studies, to be the discount rate for the Iranian population to calculate the loss across the years(27, 28).
We calculated the costs from the patient's perspective (PPC) by adding up the out-of-pocket direct costs and the income loss to the patient and his companion.
The costs from the society's perspective (SPC) were calculated by summing the real value of the direct costs, absenteeism, and mortality costs and multiplying the sum by the prevalence of the disease in the country (P) as reported by the Global Cancer Observatory(29).
The cost of illness for the financer (FPC) was defined as the difference between the real direct costs (both medical and non-medical) and the amount of money paid for them, out of pocket, by the patient.
The bills given to the patients from different providers (e.g., hospitals, labs), in general, included the real costs (before insurance) of each service. Where the receipts did not include these costs, or the patients could only provide us with the data on the out-of-pocket payments, we have calculated a ratio (R) for each subtype of health payment (for example, diagnostic procedure costs, laboratory costs) to calculate the total costs. The ratio was calculated by dividing the total costs to out of pocket costs, where both were available and non-zero, for each patient, and then taking the mean of the results. We then multiplied the out-of-pocket payments by the ratio to estimate the total costs.
We have gathered all cost data in Iranian Rials (IRR) and converted it to the purchasing power parity (PPP) US dollars (US$-2018) based on the conversion factor for the year of the study calculated by the World Bank.
Sensitivity analysis
We used a probabilistic approach for sensitivity analysis in our study. Gamma distributions were built based on each cost subcategory (all subcategories of direct medical costs, direct non-medical costs, absenteeism, and mortality cost). Then one thousand bootstraps with 297 samples (same number as participants) were drawn from these distributions (each sample having a randomly assigned expenditure on each subcategory). We then took the mean of the expenditures for each subcategory within every bootstrap and made a dataset with the results for the bootstraps. Eventually, we calculated the 5th and 95th percentiles and medians, interquartile ranges (IQR), means, and standard errors (SE) for different cost categories in the resultant dataset.
Financial risk protection (Catastrophic and impoverishing health expenditures)
We used the methods proposed by the world health organization (WHO) to calculate the indicators for financial risk protection (catastrophic and impoverishing health expenditures)(30). According to this standard, households who spent ≥40% capacity to pay for health expenditures were exposed to catastrophic health expenditure(31). As a brief explanation, we calculated a relative poverty line for Iran using the Iran’s Household Expenditure and Income Survey (HEIS). Then, we used this poverty line to determine if costs attributable to prostate cancer caused the patient's household to fall below the poverty line. To determine the prevalence of catastrophic health expenditure among the patients’ households, we used thresholds of 40 percent relative to the capacity to pay of each household, calculated as the money available to a household after accounting for its food needs(31). We also calculated the proportion of the individuals (as the patients and their family members) incurring catastrophic health expenditure using the United Nations Sustainable Development Goals approach (SDGs; indicator 3.8.2), which considers health expenditures above 10% and 25% of total household expenditure or income as catastrophic(32). The proportion of households already under the poverty line was also determined as the health expenditure due to prostate cancer pushed such households further into poverty. Only direct medical and non-medical costs were used in these calculations.
We used a logistic regression model to estimate the determinants of catastrophic and impoverishing health expenditures in our study participants. The variables used in the model were age, marital status, level of education, household size, and the insurance coverage of the participants.