Information Technology Reference
In-Depth Information
Table 22. Predicted mortality versus actual mortality by hospital using all indices
Hospital
Actual Mortality
Predictive Mortality
Difference
Rank
1
1.23
4.00
2.77
8
2
0
0
0
10
3
4.46
7.14
2.68
9
4
0.79
5.93
5.14
6
5
1.14
7.95
6.81
4
6
0.57
6.29
5.72
5
7
3.53
14.84
11.31
1
8
2.22
8.89
6.67
3
9
1.28
11.54
10.26
2
10
3.13
6.25
3.12
7
Table 23. Rank comparisons between resource demand levels and APRDRG indices, and all three in-
dices combined
Hospital
APRDRG Rank
Resource Demand Rank
All Indices Combined
1
6
8
8
2
10
10
10
3
1
9
9
4
7
6
6
5
2
3
4
6
3
5
5
7
4
1
1
8
9
4
3
9
5
2
2
10
8
7
7
future trends
While resource utilization should be a valid measure of patient severity since it seems safe to assume
that patients with many co-morbidites will require more resources, there seems to be a problem since
the use of these measures consistently under-value the conditions of the most severe patients. Moreover,
the predicted models indicate that the mortality resource level is used primarily for prediction, and the
threshold value for the logistic regression is defined as level 5 of the mortality index. Perhaps this index
can be modified by adding more categories, and using an index with ten levels rather than 5.
Search WWH ::
Custom Search