Guillamet et al developed a clinical prediction rule for evaluating a hospitalized patient with severe sepsis or septic shock. This can help to identify a patient who may require aggressive antifungal therapy. The authors are from the University of New Mexico, St Louis College of Pharmacy and Washington University in St Louis.
Patient selection: severe sepsis, septic shock
Parameters:
(1) antibiotic therapy within past 30 days
(2) central vein catheter
(3) location prior to admission
(4) total parenteral nutrition (TPN)
(5) mechanical ventilation
(6) presumed source of sepsis
|
Parameter
|
Finding
|
Points
|
|
recent antibiotic therapy
|
no
|
0
|
|
|
yes
|
2
|
|
central vein catheter
|
no
|
0
|
|
|
yes
|
2
|
|
location prior to admission
|
outside hospital
|
1
|
|
|
nursing home
|
2
|
|
|
other
|
0
|
|
total parenteral nutrition
|
no
|
0
|
|
|
yes
|
2
|
|
mechanical ventilation
|
no
|
0
|
|
|
yes
|
1
|
|
presumed source
|
lung
|
-6
|
|
|
other
|
0
|
score =
= SUM(points for all 6 parameters)
Interpretation:
• minimum score: -6
• maximum score 9
• The higher the score the greater the risk of candidemia.
• The cutoff selected was >=3.
Performance:
• At the cutoff >= 3, the sensitivity was 88%, specificity 56% and area under the ROC 0.72.
Alternatively a logitic regression model can be used:
X =
= (0.93 if TPN) + (0.88 if previous antibiotics) + (0.61 if from outside hospital) + (0.81 if from nursing home) + (0.52 if mechanical ventilation) + (0.8 if central line) - (3.22 if presumed source lung) - 3.61
probability of candidemia =
= 1 / (1 + EXP((-1) * X))
