1.1 Source of Materials
The first batch of 30 county-level clinical key specialty construction projects was implemented in 2014, the construction cycle of 3 years, including 2 infectious disease department, 2 general surgery departments, 4 orthopedic departments, 5 neurology departments, 6 pediatric departments, 4 cardiovascular departments, 2 nephrology departments, 1 obstetric department, 1 gynecology department, 1 critical care unit, 1 neurosurgery and 1 oncology department. The questionnaire was designed consistent with construction objectives, the survey covered the basic conditions of specialty, medical technical team, medical service ability, medical quality and the capacity of scientific research and teaching, which were in the form of quantitative and qualitative indicators. The questionnaire was distributed among the 30 specialties, when the construction cycle is completed in 2017, with recovery rate of 100%. The data from the questionnaire survey results were used then in the TOPSIS analysis and linear correlation analysis.
1.2 Methods
According to the results of the questionnaire survey research, combining with the literature research, the index system was established, as well as the weight of each indicator. And the TOPSIS method was used to evaluate the capacity of the first batch of county-level clinical key specialty medical services in Henan Province. The medical service capacity was classified and grouped by rank and ratio (RSR). Simple linear regression analysis and multiple linear regression analysis were carried out to filter and determine the influencing factors of medical service ability by SPSS 19.0.
1.2.1 Construction of the indicator system Based on the method of literature analysis and selection, after reference to the relevant literature [2-3], the indicators related to medical service capacity in the questionnaire entered the index system, the Medical Service Ability Evaluation Index System was established, which is composed of 11 indicators that can reflect the capacity of medical services, including diagnosis and treatment ability, efficiency, innovation ability and radiation capacity, as detailed in Table 1. Among them, the high-excellent indicator is the higher the better, and the low-excellent indicator is the lower the better. The weight in the index system is based on the system of evaluation index of county-level clinical key specialty in the existing study.
Table 1 Medical Service Ability Evaluation Index System
Level 1 Indicator
|
Level 2 Indicator
|
Weight(wj)
|
Code
|
Character
|
Diagnosis and Treatment Capacity
|
Annual Outpatient visits
|
0.032381
|
X1
|
High-excellent
|
Annual Discharge
|
0.032381
|
X2
|
High-excellent
|
Percentage of Critical Cases of Difficult patients
|
0.069316
|
X3
|
High-excellent
|
Cure Rate for Major Diseases
|
0.0843448
|
X4
|
High-excellent
|
Success Rate of Difficult and Serious Rescue
|
0.174113
|
X5
|
High-excellent
|
Work Efficiency
|
Average Inpatient Day
|
0.149739
|
X6
|
Low-excellent
|
Bed Utilization Rate
|
0.045344
|
X7
|
High-excellent
|
Bed Turnovers
|
0.082418
|
X8
|
High-excellent
|
Innovation Capability
|
Number of New Technologies in the County and above
|
0.054995
|
X9
|
High-excellent
|
Annual Number of Patients Treated with The New Technology
|
0.110006
|
X10
|
High-excellent
|
Radiation Ability
|
Percentage of Out-of-County Patients
|
0.016500
|
X11
|
High-excellent
|
1.2.2 Method of TOPSIS (Technique for Order Preference by Similarity to an Ideal Solution) It’s a method to find the best and worst solution sands in the space of the limited scheme of positive and negative ideal solution, and to sort and evaluate the advantages as well as disadvantages by evaluating the relative distance between the object and the optimal scheme [4]. Among the high-excellent indicators, the maximum data of the 30 evaluation objects is the optimal, while in the low excellent index, the minimum data of the 30 evaluation objects is the optimal. The data set composed of the optimal data of 11 evaluation indicators is the optimal scheme, and the worst scheme works the same way with the optimal scheme. The method has no strict restrictions on the type of data distribution, sample content, index correlation and quantity, and is often used for the overall evaluation of medical institutions, and can make full use of the original data information [5].
(1) Same trend transformation Due to the different nature of the indicators, some the higher the better, some the lower the better, it need to go through appropriate transformation so that all indexes show the same property. In this paper, the low optimal index X6 (Average Inpatient Day) was converted into high optimal index (X6’) by reciprocal method. As shown in Table 2.
Table 2 Original Medical Service Capacity Data of the First Batch of 30 County Clinical Key Specialty
Specialty
|
X1
|
X2
|
X3
|
X4
|
X5
|
X6
|
X6‘
|
X7
|
X8
|
X9
|
X10
|
X11
|
A1
|
0.72
|
0.3
|
34.4
|
82.9
|
99
|
7.45
|
0.134228188
|
95
|
23.08
|
2
|
69
|
19
|
A2
|
0.6433
|
0.2464
|
28
|
94.1
|
86.31
|
11.55
|
0.086580087
|
97.8
|
28.5
|
5
|
324
|
20.5
|
A3
|
2.4158
|
0.6015
|
4.53
|
83.3
|
34.83
|
10.6
|
0.094339623
|
123
|
41
|
2
|
109
|
3.5
|
A4
|
0.7
|
0.85
|
29.24
|
94.76
|
81.25
|
8
|
0.125
|
109
|
5.6
|
2
|
63
|
5
|
A5
|
1.886
|
0.4616
|
21
|
92.9
|
78.6
|
9.8
|
0.102040816
|
97.91
|
36.45
|
2
|
23
|
12.5
|
A6
|
0.1527
|
0.1508
|
16.58
|
70.94
|
58.33
|
14.2
|
0.070422535
|
104.52
|
36.12
|
5
|
320
|
1
|
A7
|
0.1527
|
0.0803
|
45.9
|
77.9
|
79.5
|
6
|
0.166666667
|
74.3
|
47.2
|
1
|
184
|
10.1
|
A8
|
1.198
|
0.1471
|
36.15
|
95.4
|
92.31
|
9.2
|
0.108695652
|
86.9
|
31.74
|
1
|
119
|
6.5
|
A9
|
28.6857
|
3.2496
|
5.4
|
87.3
|
97
|
4.7
|
0.212765957
|
96
|
7
|
2
|
150
|
11
|
A10
|
3.5454
|
0.1956
|
22
|
89.2
|
81.4
|
10.2
|
0.098039216
|
97.6
|
2.6
|
1
|
447
|
1.5
|
A11
|
0.278
|
0.1583
|
11.99
|
88.7
|
86.42
|
16.3
|
0.061349693
|
86
|
38.7
|
6
|
372
|
16
|
A12
|
2.6412
|
0.6454
|
40
|
77
|
56.8
|
9
|
0.111111111
|
138
|
60
|
1
|
3141
|
15
|
A13
|
0.416
|
0.14
|
2.1
|
69.4
|
0
|
23.4
|
0.042735043
|
100
|
32
|
3
|
38
|
8.1
|
A14
|
1.3878
|
0.3773
|
16.61
|
94.28
|
91.23
|
6.8
|
0.147058824
|
11.37
|
48.46
|
1
|
1065
|
7.6
|
A15
|
7.8252
|
0.717
|
8.663
|
97
|
90.4
|
8.1
|
0.12345679
|
85.4
|
45.4
|
3
|
3560
|
27.78
|
A16
|
3.807
|
0.705
|
6.7
|
87.81
|
86.66
|
12.3
|
0.081300813
|
96.82
|
20.22
|
9
|
3069
|
5
|
A17
|
2.3035
|
0.2247
|
3
|
87.81
|
86.53
|
11.7
|
0.085470085
|
90
|
14
|
7
|
123
|
5
|
A18
|
2.1385
|
0.3429
|
26
|
87.4
|
29.85
|
8.9
|
0.112359551
|
93.2
|
30
|
5
|
3020
|
41
|
A19
|
1.7596
|
0.3869
|
18
|
19.57
|
90.4
|
6.9
|
0.144927536
|
118
|
4.6
|
5
|
308
|
3.7
|
A20
|
1.8724
|
0.2044
|
12
|
95.52
|
98.81
|
8.7
|
0.114942529
|
99
|
35.12
|
5
|
472
|
11.7
|
A21
|
5.6586
|
0.7128
|
1.16
|
91.88
|
99.52
|
4.7
|
0.212765957
|
89
|
59.5
|
3
|
17827
|
10.17
|
A22
|
4.5559
|
0.125
|
1.16
|
91.22
|
90.4
|
4.7
|
0.212765957
|
90
|
22
|
5
|
520
|
17.8
|
A23
|
16.6421
|
0.9112
|
53
|
83.74
|
98
|
5.5
|
0.181818182
|
94.8
|
75.9
|
3
|
1508
|
8.1
|
A24
|
1.8842
|
0.6771
|
3.3
|
90.41
|
70
|
7.1
|
0.14084507
|
90.4
|
46
|
1
|
33
|
10.79
|
A25
|
1.8842
|
0.52
|
3.53
|
84.85
|
97.66
|
7.1
|
0.14084507
|
94.8
|
36.3
|
5
|
559
|
14.4
|
A26
|
2
|
0.803
|
17
|
91.4
|
96.33
|
8.5
|
0.117647059
|
92
|
41
|
5
|
68
|
5
|
A27
|
1.7641
|
0.1601
|
25.5
|
89.44
|
98.6
|
7.2
|
0.138888889
|
88.2
|
4.1
|
5
|
49
|
10.7
|
A28
|
14.0788
|
10.983
|
35.1
|
98.6
|
93.53
|
6.4
|
0.15625
|
92.5
|
72.4
|
5
|
324
|
12.86
|
A29
|
3.8506
|
0.1218
|
35
|
94.2
|
95.13
|
11.1
|
0.09009009
|
92.2
|
30.45
|
6
|
465
|
7
|
A30
|
5.0285
|
0.5167
|
5.91
|
98.5
|
97.33
|
4.9
|
0.204081633
|
87
|
60.5
|
3
|
1540
|
3.5
|
(2) Nondimensionalize the index The original data matrix with the same trend was normalized to eliminate the influence of indicator measurement units, and the normalized matrix Z was established, the naturalization process is carried out according to the following formula.(see Formula 1 in the Supplemental Files.)
The normalized matrix Z is:
Z=
Determine the positive and negative ideal solutions:
According to the normalized matrix Z, the positive ideal solution Z+ ( the optimal vector) and the negative ideal solution Z- (the worst vector) were calculated:
Positive Ideal Solution:Z+=(0.0675,0.0256,0.3063,0.1608,0.1225,0.1504,0.2633,0.2767,0.0437,0.1641,0.1991)
Negative Ideal Solution:Z-=(0.0355,0.0440, 0.1272,0.1969,0.1968,0.1990,0.0217,0.2235,0.0437,0.0556,0.1009)
(3) Calculate the Ci value of Euclidean distance and relative proximity Calculate the Euclidean distance Di+ and Di- of the evaluation object index value and the positive and negative ideal solutions according the following formula: (see Formula 2 in the Supplementary Files)
Based on the calculated Euclidean distance, the Ci value of the index value of the evaluation object, which was relatively close to the positive ideal solution and negative ideal solution, was calculated by the following formula. The Ci value is taken as the evaluation result of the medical service capacity of the key clinical specialties at the county level, which is between 0 and 1, and the specialty with the largest Ci is the best. (see Formula 3 in the Supplementary Files)
1.2.3 Rank Sum Ratio (RSR) The RSR refers to the average of the rank of a row or column. The basic idea is to obtain the statistical RSR by rank transformation in n evaluation objects and m evaluation indexes matrix. Then, parameter statistical analysis is used to study the distribution of RSR, and the RSR value is used to grade the evaluation objects directly.
1.2.4 Analysis of Influence Factors The Ci value of TOPSIS was used as the dependent variable for the analysis of influencing factors, and 13 variables were selected from the indicators in the questionnaire as the independent variables for the analysis, which is as shown in Table 3. Simple linear regression was used to screen the influencing factors of medical service capability. The variables considered as influencing factors in the simple linear regression model were incorporated into the multiple linear regression model to determine the influencing factors, and the test level was a=0.05.
Table 3 Influencing Factors
Level 1 Variable
|
Level 2 Variable
|
Code
|
The Basic Information of The Hospital
|
population in the county
|
V1
|
Specialist basic conditions
|
construction period investment total
|
V2
|
net usable area of each bed
|
V3
|
specialist actual open number of beds
|
V4
|
doctor to nurse ratio
|
V5
|
patient to nurse ratio
|
V6
|
specialist talent team
|
proportion of the average annual number of expert visits
|
V7
|
number of technical promotion project training courses
|
V8
|
number of technical promotion trainees
|
V9
|
number of sent trainees
|
V10
|
medical quality
|
proportion of drugs
|
V11
|
coincidence rate of main clinical diagnosis and pathological diagnosis
|
V12
|
scientific research
|
number of papers in core or national journals
|
V13
|