Our definitions and measures
By using the WB approach to evaluate the degree to which the NRCMS alleviates medical impoverishment, we find two kinds of individuals in regard to the change in their impoverishment statuses after encountering medical payment. For example, assume that the poverty line is at 5,000 Yuan per annum, while person A’s income is 4,000 Yuan and his/her medical expenses are 10,000 Yuan. Therefore, A belongs to the poverty population, as his/her income is already below the poverty line. That is, his/her medical expenses do not change him/her from non-poor to poor or vice versa. According to the NRCMS principle (i.e., sharing the cost of catastrophic illness), A is a typical person the NRCMS targets. Second, assume person B’s income is 5,010 Yuan, which becomes 4,990 Yuan after the payment of medical expenses of 20 Yuan. Based on the WB approach, B does not belong to the poverty population before payment, but becomes poor after the payment of medical expenses (i.e., B is driven into medical impoverishment). However, as one of the key features of the NRCMS is to cover the costs of catastrophic medical care as opposed to day-to-day costs, B may not be among the NRCMS’s target population. The WB approach thus calculates the change in poverty status after the payment of medical expenses from an economics perspective, but it is not specific enough to evaluate the effects of the NRCMS.
On the other hand, the NRCMS focuses on the prevention of medical impoverishment associated with costs of catastrophic medical care. Therefore, rural residents’ health-care expenses beyond a certain level of their income need to be considered. The target population of the NRCMS should include not only those who fall below the poverty line after incurring medical expenses but also those below the poverty line who would then fall deeper into poverty. To measure medical impoverishment, we assume that the household is the basic economic unit that contracts with the government for farming and other economic activities in rural China. Medical impoverishment is thus defined as a household’s inability to pay OOP medical expenses above a certain level of its discretionary income.
A number of absolute and relative measures are used to define the poverty line. According to the goals of NRCMS, we use 50 % of the per capita income of rural residents (I) in the sample area as the individual poverty line (Li) conservatively [
18] as the Organization for Economic Cooperation and Development (OECD) used [
19]; the household poverty line (Lh) is thus equal to the individual poverty line multiplied by the number of individuals in the household (Nh), as shown below:
$$ Li=\frac{1}{2}\times I $$
(1)
Ability to pay (ATP) is introduced to distinguish persons such as A and B, mentioned previously. ATP, from the individual perspective, is defined as the amount remaining after nondiscretionary consumption has been subtracted from annual per capita income [
20]. Two assumptions usually hold in China: (1) almost everyone can pay the entire amount remaining after nondiscretionary consumption for medical expenses; and (2) the household is always the basic economic unit in rural China, not only for contracting with the government but also when someone in the household incurs medical bills. Therefore, we choose to use the ATP of a household, which is the individual’s ATP multiplied by the number of persons in the household, as shown below:
$$ \mathrm{R}=\mathrm{N}\mathrm{h}\times \left(\mathrm{I}-\mathrm{L}\mathrm{i}\right)=\mathrm{N}\mathrm{h}\times \mathrm{I}-\mathrm{L}\mathrm{h} $$
(3)
where R represents average household ATP, Li represents the individual poverty line, I represents per capita income, Nh indicates the number of persons in a household, and Lh indicates the household poverty line.
Three measures are used to assess the level of medical impoverishment. First, the rate of medical impoverishment, reflecting the overall level of poverty among a population, is defined as
where U indicates the poverty headcount ratio, P represents the number of households below the household poverty line (i.e., households with medical expenses [Mi] greater than the maximum annual average household ATP [R] and an income level below the poverty line after paying medical bills [Lh]), and N indicates the number of households in the population.
The second is the relative poverty gap, which is used to reflect the depth of poverty among the poverty sub-population. This is defined as the ratio of total medical care expenses exceeding total ATP to total ATP for medically impoverished households. The specific formula is as follows:
$$ S=\frac{{\displaystyle {\sum}_{i=1}^P\left(Mi-Ri\right)}}{{\displaystyle {\sum}_{i=1}^PRi}} $$
(5)
where S is the relative poverty gap, Mi represents the i-th household’s OOP medical expenses, Ri indicates the i-th household’s ATP, and P indicates the number of medically impoverished households.
Third, we calculate the average poverty gap to measure the overall poverty scope and depth among the entire population. This is defined as the ratio of the total ATP of households in poverty to the total ATP of the entire population. The specific formula is as follows:
$$ S\hbox{'}=\frac{{\displaystyle {\sum}_{i=1}^P\left(Mi-Ri\right)}}{{\displaystyle {\sum}_{i=1}^NRi}} $$
(6)
where S’ is the average poverty gap, Mi represents the i-th household’s OOP medical expenses, Ri indicates the i-th household’s ATP, and N is the number of households in the entire population.
Finally, we use two indicators to measure the degree to which the NRCMS alleviates medical impoverishment. The first is the change in the poverty headcount ratio, which is calculated as
$$ \Delta \mathrm{U}={U}_{pre\_ reimbursement-}{U}_{post\_ reimbursement} $$
(7)
$$ \Delta \mathrm{U}\%=\frac{\Delta U}{U_{pre\_ reimbursement}}\times 100\% $$
(8)
In this calculation, Upre_reimbursement is the poverty headcount ratio before receiving NRCMS reimbursement and Upost_reimbursement represents the poverty headcount ratio after the reimbursement, while ΔU % is the percentage change in the poverty headcount ratio after the reimbursement.
The second indicator is the difference between the poverty gaps before and after NRCMS reimbursement:
$$ \Delta \mathrm{S}=\frac{{\displaystyle {\sum}_{i=1}^P\left(M{i}_{\mathrm{pre}\_ reimbursement}-Ri\right)}}{{\displaystyle {\sum}_{i=1}^PRi}}-\frac{{\displaystyle {\sum}_{i=1}^P\left(M{i}_{post\_ reimbursement}-Ri\right)}}{{\displaystyle {\sum}_{i=1}^PRi}} $$
(9)
$$ \Delta {\mathrm{S}}^{\prime }=\frac{{\displaystyle {\sum}_{i=1}^P\left(M{i}_{pre\_ reimbursement}-Ri\right)}}{{\displaystyle {\sum}_{i=1}^NRi}}-\frac{{\displaystyle {\sum}_{i=1}^P\left(M{i}_{post\_ reimbursement}-Ri\right)}}{{\displaystyle {\sum}_{i=1}^NRi}} $$
(10)
$$ \Delta \mathrm{S}\%=\Delta \mathrm{S}\div \left[\frac{{\displaystyle {\sum}_{i=1}^P\left(M{i}_{\mathrm{pre}\_ reimbursement}-Ri\right)}}{{\displaystyle {\sum}_{i=1}^PRi}}\right]\times 100\% $$
(11)
$$ \Delta {\mathrm{S}}^{\prime}\%=\Delta {\mathrm{S}}^{\prime}\div \left[\frac{{\displaystyle {\sum}_{i=1}^P\left(M{i}_{\mathrm{pre}\_ reimbursement}-Ri\right)}}{{\displaystyle {\sum}_{i=1}^NRi}}\right]\times 100\% $$
(12)
In these calculations, Mipre_reimbursement represents OOP medical expenses before NRCMS reimbursement; Mipost_reimbursement represents OOP medical expenses after the reimbursement; and ΔS ' represent the absolute differences in the relative and average poverty gaps, respectively, between the pre- and post-reimbursement levels; and ΔS % and ΔS ' % represent changes in the percentages of the relative and average poverty gaps, respectively, between the pre- and post- reimbursement levels.
Case study
We used a cross-sectional study to illustrate our approach and compare the results with those of the WB method. Household health-care utilization and expenditure surveys were conducted in January 2008 in eight counties in Yanbian Korean Autonomous Prefecture, Jilin province, China. The surveys gathered individual and household demographics, socioeconomic characteristics, two-week morbidity, six-month chronic illness prevalence, health service needs and utilization, and health-care expenses [
21].
We used the probability proportionate to size sampling method to randomly select villages in each of the eight counties. Around 10 % of the villages in each county were selected. In each village, we first selected a random household from a list of household register numbers. The next household was the one that was closest to the first household in terms of walking steps. If more than one household was equally close to the first household, the one on the left (for a person stepping out of the house) was selected. The process continued until a sufficient number of households had been surveyed. About 10 % of the households in a village were selected. A total of 6,135 individuals in 1,987 households were surveyed.
Further, we used Myer’s Index to test the representative character of our survey data in terms of age structure based on China’s 2000 Census data [
22]. We obtained a Myer’s Index of 3.38, which showed that our data represented the age distribution of national population well. A Myer’s Index greater than 60 indicates significant differences in age structures between one sample and another.
This study was approved (IRB#08–03–0130) by the Medical Research Ethics Committee, School of Public Health, Fudan University (IRB00002408&FWA00002399). A verbal consent was obtained from all the respondents before conducting the interviews.