SIMIS payment
Table 3 showed the descriptive statistics of serious illness patients in Jinzhai county.
The serious illness patients were often transferred to other hospitals in different regions. The average number of annul hospitalization was 3.64. The annual average medical expenses for serious illness patients were CNY 96,100, annual average NRCMS payment was CNY 48,000, annual average SIMIS payment was CNY 7,600, annual average inside medical insurance self-payment was CNY 26,200, annual average outside medical insurance self-payment was CNY 14300. The average annual expenses of inpatients at County was CNY 86,500,City was CNY 91,100, Province was CNY 90,500, Outside province was CNY 97,600, Cross regional was CNY 97,400. The proportion of NRCMS payment to the total medical expenses was 49.90%, proportion of SIMIS payment to the total medical expenses was 7.92%, inside medical insurance self-payment to the total medical expenses was 27.26%, outside medical insurance self-payment to the total medical expenses was 14.92%.
Table3 Annual average inpatients expenses of seriously illness patients in Jinzhai County from 2013 to 2016
Items
|
Total
|
County
|
City
|
Province
|
Outside province
|
Cross regional
|
Hospitalization
|
proportion(%)
|
100
|
2.08
|
5.61
|
12.23
|
27.57
|
52.51
|
Annual per capita hospital stays
|
3.64(2.72)
|
1.73(1.65)
|
2.26(1.96)
|
2.12(1.65)
|
2.58(2.41)
|
4.77(2.66)
|
Medical expenses(CNY Ten thousand)
|
NRCMS payment
|
4.80(3.01)
|
5.11(3.01)
|
5.09(3.09)
|
4.83(2.83)
|
4.33(3.38)
|
4.99(2.79)
|
SIMIS payment
|
0.76(0.11)
|
0.55(1.40)
|
0.60(0.83)
|
0.57(0.63)
|
0.87(1.50)
|
0.76(0.96)
|
Inside medical insurance
self-payment
|
2.62(1.17)
|
2.33(1.16)
|
2.35(0.90)
|
2.36(0.72)
|
2.78(1.28)
|
2.61(1.01)
|
Outside medical insurance self-payment
|
1.43(1.37)
|
0.66(0.57)
|
1.07(0.42)
|
1.31(1.19)
|
1.77(2.36)
|
1.37(1.98)
|
Total
|
9.61(5.77)
|
8.65(4.58)
|
9.11(4.67)
|
9.05(4.62)
|
9.76(7.59)
|
9.74(4.97)
|
Payment proportion (%)
|
NRCMS payment
|
49.90(10.46)
|
59.08(16.25)
|
55.89(9.09)
|
53.34(9.37)
|
44.41(10.86)
|
51.30(9.21)
|
SIMIS payment
|
7.92(5.49)
|
6.36(5.00)
|
6.57(5.20)
|
6.03(4.80)
|
8.96(5.93)
|
7.85(5.32)
|
Inside medical insurance self-payment
|
27.26(10.85)
|
26.94(18.49)
|
25.80(9.35)
|
26.07(9.95)
|
28.52(11.64)
|
26.79(10.36)
|
Outside medical insurance Self-payment
|
14.92(11.06)
|
7.63(7.76)
|
11.73(9.22)
|
14.56(10.76)
|
18.11(11.50)
|
14.06(10.92)
|
Note: the values inside and outside the brackets represent the mean and standard deviation respectively.
Analysis of RD
SIMIS effect
Table 4 showed the estimated effect of policy treatment. In the absence of covariates, the lwald estimate of the actual medical insurance payment proportion (AMIPP) at the breakpoint was 0.025, which was significant at the level of 1%. Figure 2 showed that the reimbursement on the right side of the breakpoint was slightly higher than that on the left side, indicating that the implementation of SIMIS has improved the actual payment level, about 2.5%. The lwald estimate of the inside medical insurance payment proportion (IMIPP) was 0.020, Figure3 showed that the reimbursement on the right side of the breakpoint was slightly higher than that on the left side, indicating that the implementation of SIMIS had improved the actual payment level to some extent, about 2%, but slightly lower than the actual payment level of medical insurance. The lwald estimate of inside medical insurance self-payment proportion (IMSPP) was -0.006, and it's not significant, Figure 3 showed that the level of self-payment inside medical insurance was almost the same on the left and right sides of the breakpoint, indicating that the implementation of SIMIS has little impact on the level of self-payment inside medical insurance. The lwald estimate of outside medical insurance self-payment proportion (OMISPP) was -0.016, which was significant at the level of 5%, Figure 3 showed that the level of self-payment outside medical insurance on the right side of the breakpoint was slightly lower than that on the left side of the breakpoint, indicating that the implementation of SIMIS reduced the level of self-payment outside medical insurance to a certain extent, about 1.6%. After adding covariates, the results of RD were consistent.
Table 4 RD results of SIMIS
lwald
|
AMIPP (H=0.6581)
|
IMIPP
(H=0.5345)
|
IMSPP
(H=0.5949)
|
OMISPP
(H=1.079)
|
Not add Covariate
|
0.025***(0.010)
|
0.020*(0.011)
|
-0.006(0.009)
|
-0.016**(0.006)
|
Add covariates
|
0.022***(0.008)
|
0.014**(0.006)
|
-0.006(0.009)
|
-0.011**(0.005)
|
Note: * * * * represents significant at the level of 1%, * * represents significant at the level of 5%, and * represents significant at the level of 10%. H represents the optimal bandwidth. The value in brackets is SD of lwald.
Validity test
Independent variable continuity test
McGrary (2008) method was used to test whether the probability density function of independent variable was continuous at the breakpoint. The results showed θ=-0.054, standard error was 0.11, so the assumption of continuity of independent variable density function at the breakpoint could be accepted. Figure 5 showed that the confidence intervals of the estimated values of the independent variable probability density functions on both sides of the breakpoint were mostly overlapped, indicating that patients were randomly assigned on both sides of the breakpoint, and there was no endogenous grouping problem.
Covariate continuity test
Table 5 showed that the estimated value lwald was not significant for Age, Gender, Hospital stay,
Total expenses, Gender and Inpatient type, which showed that the effect of SIMIS had no impact on covariates, and the policy effect could be attributed to the implementation of SIMIS.
Table 5 Covariate continuity test results
lwald
|
AMIPR
H=0.6581
|
IMIPR
H=0.5345
|
IMSPR
H=0.5949
|
OMISPR
H=1.079
|
Age
|
1.690(14.516)
|
3.008(16.197)
|
2.355(15.300)
|
-2.93(11.38)
|
Total expenses
|
606.16(1971.54)
|
1209.01(2135.78)
|
968.03(2045.23)
|
-907.36(1581.57.)
|
Hospital stay
|
0.943(2.897)
|
2.082(3.241)
|
1.433(3.058)
|
2.176(2.225)
|
Inpatient type
|
-0.041(0.083)
|
-0.033(0.094)
|
-0.037(0.089)
|
-0.052(0.066)
|
Gender
|
0.035(0.039)
|
0.031(0.043)
|
0.0358(0.041)
|
0.035(0.030)
|
Note: The dependent variables are age, total expenses, hospital stay, inpatient type and gender. The independent variables are serious illness expenses. The value outside the bracket is the estimated value of lwald, and the value inside the bracket is the estimated value of standard error.