Julian-Jimenez et al reported a model for predicting bacteremia in a patient in the Emergency Department. The authors are from multiple institutions in Spain.
Patient selection: patient in the Emergency Department
Parameters:
(1) serum procalcitonin in ng/mL
(2) respiratory rate in breaths per minute
(3) body temperature in °C
(4) Charlson comorbidity index (CCI)
(5) WBC count per microliter
(6) tremors/shivering/chills
(7) platelet count per microliter
|
Parameters
|
Findings
|
Points
|
Beta-Coefficient
|
|
procalcitonin
|
< 0.51 ng/mL
|
0
|
0
|
|
|
>= 0.51 ng/mL
|
4
|
4.377
|
|
respiratory rate
|
< 22 bpm
|
0
|
0
|
|
|
>= 22 bpm
|
1
|
0.889
|
|
temperature
|
<= 38.3 °C
|
0
|
0
|
|
|
> 38.3°C
|
1
|
0.789
|
|
Charlson index
|
< 3
|
0
|
0
|
|
|
>= 3
|
1
|
0.690
|
|
WBC count
|
<= 12,000 per µL
|
0
|
0
|
|
|
> 12,000 per µL
|
1
|
0.661
|
|
chills
|
no
|
0
|
0
|
|
|
yes
|
1
|
0.458
|
|
platelet count
|
>= 150,000 per µL
|
0
|
0
|
|
|
< 150,000 per µL
|
1
|
0.453
|
total score =
= SUM(points for all of the parameters)
value of X =
= SUM(beta-coefficients for all of the parameters) - 6.027
probability of bacteremia =
= 1 / (1 + EXP((-1) * X))
Interpretation:
• minimum score: 0
• maximum score: 10
|
Total Score
|
Percent Bacteremia (observed)
|
|
0
|
0%
|
|
1
|
0%
|
|
2
|
0.7%
|
|
3
|
1.9%
|
|
4
|
6.8%
|
|
5
|
17.6%
|
|
6
|
43.7%
|
|
7
|
53%
|
|
8
|
70%
|
|
9
|
79%
|
|
10
|
94%
|
Performance:
• The area under the ROC curve is 0.92.
