Design
Using data from 2008 to 2013, this study performed two analyzes, one of which was a pooled time series cross-section analysis (N*T). Another one was a longitudinal analysis that was used to know a time series.
Data source
This study used KHP data from 2008 to 2013 co-organized by the Korea Institute for Health and Social Affairs (KIHASA) and NHI Corporation. KHP data has been nationally representative for decision making about healthcare policy. The KHP is using the extraction framework for the entire 90% of the 2005 census to maintain the national scale for the survey. KHP subjects were selected according to the probability-proportional stratified sampling method and surveyed repeatedly for the same variables annually since 2008. The baseline sample was 7,866 households, 24,616 household members. However, in 2012, 5,856 households and 17,417 household members remained due to panel attrition such as death or rejection of investigation. Accordingly, approximately 2,500 households were extracted as additional samples nationwide and included in the survey starting from the 8th survey in 2013.
The KHP survey primarily surveys about 500 variables. Demographic and sociological characteristics such as households and household members' assets and income per year, family, health security type, registration and type of disability et al.; Health care characteristics such as diagnosed disease, utilization of emergency room, outpatient, and hospitalization and copays per case, medication costs, medical use satisfaction, complementary medical use and costs et al.; additional investigations such as health behaviors and quality of life et al.; private insurance information such as private insurance subscription and types, receipt of private insurance et al. are being investigated. In addition, KIHASA and NHI Corporation generated the total household income decile using the primary data. That was, this was the total household income per year divided by the square root of the number of household members [4], which are classified as the 1st (minimum) and the 10th (maximum). In addition, it applied sampling weight considering the attrition rate of panel data. For this study, IRB formally approved the use of KHP data (KIHASA 2016-01).
Data collection
The panel data were collected annually by trained interviewer visiting house and face-to-face interviewing subjects. To reduce the recall bias about medical records, subjects recorded the OOP expenditure on the medical household account book with the reason for healthcare utilization immediately after visiting the clinics or hospitals. The collection of receipts, the importance of quick records, and how to fill out a medical household account book are continuously educated through guide materials and counseling every year. In addition, the visited interviewer reviewed receipts and confirmed to determine if the records of medical households were accurate. In some cases, to improve accuracy, the actual hospital clam data was checked in a cross. Since it is data over a long period of time, the actual cost was recorded without applying the consumer price index so that even the price fluctuation of the medical service could be known. The collected data will be released to the researchers approximately three years after confirming that it is complete data after undergoing verification of data entry editing, imputation, building weight, variance estimation, and trial data conference finally.
Using variables
This study used the variables of the raw data as possible. For example, gender, survey year, type of health security system (NHI or MA), type of disability, number of comorbidities, visits to the care center: emergency- room (ER), out-patient department (OPD) and inpatient, OOP expenditure, and total household income decile et al.
Comorbidity means any chronic disease diagnosed by medical doctors and had over the past year such as hypertension, diabetes, hyperlipidemia, arthritis, tuberculosis, ischemic heart disease, cerebrovascular disease and others. The various chronic disease presented by the subjects were checked by a trained interviewer and entered into a standardized disease code.
To date, medical expenditure has been analyzed the cost of insurance coverage billed by each hospital. OOP expenditure for each individual was few studied because insurance claims data do not represent individual OOPs. It is very useful to know the gross amount of healthcare expenditure. Therefore, KHP has investigated annual OOP expenditure to decide health policy. OOP expenditures are non-insured benefits or copayments. The copayment is that each person must pay after excluding insured coverage for ER, admission, OPD visits, and prescription drug purchases. Therefore, the copayment of dialysis, transplantation, and conservation care of ESRD were included in OOP expenditure. OOP expenditures were analyzed separately by using them in the ER, admission and OPD visits. Also, to know OOP expenditure of the drug, drug costs were added to the OOP expenditure of ER, admission and OPD visits, then referred to as personal total OOP medical expenditure. The newly created variable for this study using raw data was total household income per year; SE and CHE; a type of healthcare service: dialysis, KT and conservative care.
Originally KHP data generated total household income per year by adding gross earned income and gross asset income in household. Household gross earned income is the sum of all household members' earned income: months worked. Total asset income was a sum of real estate and property income, financial income, social insurance, private insurance, government subsidies, private subsidies, and other income. In this study, total household income per year was adjusted for household size according to the OECD's square root index method [18].
KHP data investigated monthly average living expenses, which excludes savings. For the SE, only the food cost can be applied to apply the extreme poverty line. But when applying the wide poverty line, including food consumption can be applied [5]. In other words, SE was not standardized in all countries. Korea's Ministry of Health and Welfare announced that SE is the minimum cost necessary to maintain a healthy and cultural life. Therefore, in this study, SE was defined as the cost of living after excluding saving and then the consumption equivalence scale was applied to adjust the size of the household [18]. There are several equivalent methods, but we used the OECD square root index. This is a method of calculating equalized personal income by dividing household income by the square root of the number of household members.
In this study, the household capacity to pay was created by subtracting the SE from the total household income, which was adjusted for household size. Then, if it exceeded 40% of the household capacity to pay, it was defined as CHE. In addition, if the household capacity to pay was zero or a negative value, the person was defined as the medical poor [5].
Type of healthcare service such as dialysis, KT and conservative care was a newly created variable for this study. In KHP data, since the type of healthcare service or fee for service of each disease was not investigated. Therefore, we classified the type of healthcare service using disability type. It was classified as dialysis in case of dialysis-disabled by law; KT in case of KT-disabled by law; conservative service in case of absence of disability by law. This is because dialysis patients and undergoing KT are enrolled as Grade 2 and 5 kidney-disabled respectively under the Disabled Welfare Act in Korea.
Study subjects
Of the 111,869 KHP subjects from 2008 to 2013, 305 (0.28%) were diagnosed with ESRD (N18-N19 according to Korea Classification of Diseases-6 code) by medical doctors. When looking at 305 subjects by year, there were 34 in 2008, 47 in 2009, 56 in 2010, 60 in 2011, 52 in 2012, and 56 in 2013.
Statistical analysis
This study conducted cross-sectional analysis and panel analysis. In pooled time series cross-section analysis, chi-square and t-tests were performed to compare demographic characteristics, CHE and the medical poor ratios between NHI and MA. In addition, ANCOVA confirmed total OOP expenditures due to ER, admission and OPD visits, and prescription between NHI and MA. At this time, gender, age, type of healthcare service, and comorbidities were used as covariables. Every OOP expenditure (South Korean Won, KRW) converted to USD ($) based on the exchange rate on July 1, 2008 (1$=1,050.89 KRW) [19].
In the panel analysis, the total OOP expenditure trend of subjects for 6 years from 2008 to 2013 were identified. We built three models; a saturated model with an unstructured covariance matrix, a saturated model with a compound symmetry covariance matrix, and the main effects model with a compound symmetry covariance matrix. Then, the final results were presented by the main Effects model, which had the lowest Akaike’s Information Criterion and Bayesian Information Criterion.
The statistical test was done after excluding missing data of each variable using SAS 9.4 (SAS Institute Inc., Cary, NC, USA). P values of less than 0.05 were regarded as statistically significant.
Definitions of terms
NHI and MA among Health Security system
In Korea, the health security system has NHI and MA. NHI is an obligatory system for the national people. The people have to pay an insurance fee to NHI Cooperation and then can receive medical benefits if they need it. Their copayment of individuals has been from 20% up to 60% of the total medical fee for each medical service [10, 11]. They pay copayment in admission about 20% of the total fee, and also copayment in OPD visits about 30% ~60% of total fee [11]. In this study, subjects with NHI were 237.
Meanwhile, MA is one of public healthcare assistance program which supports the lowest income group or the person with an incapacity for maintaining their life. Their medical expenses are covered by the national tax and local tax under government responsibility. MA consists of type 1 and types 2. MA type 1 includes persons who are poor and the incapacity of working. MA subjects do not pay any copayment in admission and do pay copayment ($0.95 or $1.93) at only OPD visits [11]. MA Type 2 includes people who are poor but can work, different from MA Type 1. There were 68 subjects with MA in this study, which consisted of type 1 (n = 55) and type 2 (n = 3), but we did not classify them separately, because of a small number of subjects.
Table 1. Representative programs of Korean Health Security System
Criteria
|
NHI
|
MA
|
System
|
Social insurance
|
Public assistance (Type 1, Type 2)
|
Subject
|
All people except MA
(Mandatory subscription)
|
Selecting people who have difficulty living
|
Finance method
|
Premiums and Treasury
|
Tax
|
Operation& management
|
Nation & public corporation
|
National & local government
|
Insurance premium burden
|
Proportion of burden according to income level
|
Non premium
|
Insurance benefits
|
Uniformity
|
Promiscuity
|
Population (2020)
|
97.2%
|
2.8%
|
Copayment
|
Admission
|
20% of total medical expenses
|
None
|
OPD
|
60%~30% of total medical expenses
|
About $ 0.95~ $ 1.93
Type 1: 5% of special equipment service fee
Type 2: 15% of special equipment service fee & 15% of medical benefit costs when using secondary or tertiary medical institutions
|