Information Technology Reference
In-Depth Information
Table 15. Procedures by hospital
Table of DSHOSPID by PR1
HOSPID
PR1(Principal procedure)
Total
Frequency
Row Pct
Col Pct
3611
3612
3613
3614
3615
3616
3619
1
16
4.92
14.81
37
11.38
12.85
47
14.46
11.69
24
7.38
9.09
193
59.38
51.47
8
2.46
28.57
0
0.00
0.00
325
2
3
27.27
2.78
4
36.36
1.39
1
9.09
0.25
0
0.00
0.00
2
18.18
0.53
1
9.09
3.57
0
0.00
0.00
11
3
5
4.46
4.63
10
8.93
3.47
19
16.96
4.73
7
6.25
2.65
69
61.61
18.40
2
1.79
7.14
0
0.00
0.00
112
4
22
8.70
20.37
59
23.32
20.49
82
32.41
20.40
75
29.64
28.41
15
5.93
4.00
0
0.00
0.00
0
0.00
0.00
253
5
2
2.27
1.85
14
15.91
4.86
34
38.64
8.46
29
32.95
10.98
9
10.23
2.40
0
0.00
0.00
0
0.00
0.00
88
6
5
2.86
4.63
23
13.14
7.99
51
29.14
12.69
36
20.57
13.64
58
33.14
15.47
2
1.14
7.14
0
0.00
0.00
175
7
21
7.42
19.44
79
27.92
27.43
95
33.57
23.63
60
21.20
22.73
15
5.30
4.00
12
4.24
42.86
1
0.35
100.00
283
8
7
15.56
6.48
11
24.44
3.82
17
37.78
4.23
9
20.00
3.41
1
2.22
0.27
0
0.00
0.00
0
0.00
0.00
45
9
15
19.23
13.89
23
29.49
7.99
23
29.49
5.72
7
8.97
2.65
9
11.54
2.40
1
1.28
3.57
0
0.00
0.00
78
10
12
12.50
11.11
28
29.17
9.72
33
34.38
8.21
17
17.71
6.44
4
4.17
1.07
2
2.08
7.14
0
0.00
0.00
96
Total
108
288
402
264
375
28
1
1466
Note that the difference between the predicted mortality and the actual value is considerable, both by
procedure and by hospital. Therefore, the ability of this model to rank hospitals is highly questionable.
There is a considerable difference between the average and the maximum values of outcomes (Table
18). In particular, at least one patient stayed 150 days or more for 3611, 3612, 3613, and 3614. How
should these outliers be considered when ranking quality? We use the following code to find the kernel
density estimation functions.
proc sort
data=nis.cardiovascular out=work.cardiovascular2;
by pr1;
proc kde
data=work.cardiovascular2;
univar los/gridl=
0
gridu=
15
out=nis.kdecardlos;
Search WWH ::
Custom Search