Patient characteristics:
(1) transplanted from 1984 to 1990
(2) single organ kidney transplants
(3) negative for diabetes mellitus prior to transplant
(4) followed at least 18 months after transplant to identify late onset cases
Parameters in the index as identified by logistic regression:
(1) age in years
(2) family history of diabetes mellitus
(3) glucose instability (intolerance) on days 4 to 7 post-transplant (serum glucose > 140 mg/dL)
(4) presence of HLA A26
|
Parameter
|
Finding
|
Points
|
|
age of the patient in years
|
|
(age)
|
|
family history of diabetes mellitus
|
no
|
0
|
|
|
yes
|
1
|
|
glucose instability on day 4-7
|
no
|
0
|
|
|
yes
|
1
|
|
presence of HLA A26
|
no
|
0
|
|
|
yes
|
1
|
|
Parameter
|
Odds Ratio
|
95% CI
|
p value
|
|
age
|
1.0928
|
1.0476-1.1399
|
0.0001
|
|
family history
|
4.4396
|
1.8571-10.6136
|
0.0008
|
|
glucose instability
|
3.9772
|
1.4244-11.1047
|
0.0084
|
|
HLA A26
|
6.2069
|
1.2012-32.0728
|
0.0293
|
from Table 3, page 335
index =
= (0.0887 * (age in years)) + (1.4906 * (points for family history)) + (1.3806 * (points for glucose instability)) + (1.8257 * (points for HLA A26)) – 7.3571
Although I could not find this specifically stated in the paper, I assume that the index was used in the probability equation:
probability of new onset diabetes mellitus =
= 1 / (1 + EXP((-1) * (index)))
Performance (Table 4, page 336):
• sensitivity 96%, specificity 71.9%
• positive predictive value 42.4%, negative predictive value 98.9%
• diagnostic efficiency: 76%
