Fernandez et al reported a model that can help to distinguish dengue fever from other febrile illnesses. The authors are from McMaster University in Hamilton, Ontario.
Patient selection: febrile illness in a tropical or subtropical country
Parameters:
(1) petechiae
(2) retro-ocular pain
(3) gingival bleeding
(4) epistaxis
(5) skin pallor
Parameter |
Finding |
Points |
petechiae |
no |
0 |
|
yes |
0.718 |
retro-ocular pain |
no |
0 |
|
yes |
0.516 |
gingival bleeding |
no |
0 |
|
yes |
1.316 |
epistaxis |
no |
0 |
|
yes |
-0.474 |
skin pallor |
no |
0 |
|
yes |
-0.535 |
value of X =
= SUM(points for all of the parameters) + 0.694
probability of dengue fever =
= 1 / (1 + EXP((-1) * Y))
Alternatively
Parameter |
Finding |
Points |
petechiae |
no |
0 |
|
yes |
7 |
retro-ocular pain |
no |
0 |
|
yes |
5 |
gingival bleeding |
no |
0 |
|
yes |
13 |
epistaxis |
no |
0 |
|
yes |
-5 |
skin pallor |
no |
0 |
|
yes |
-5 |
total score =
= SUM(points for all of the parameters)
Interpretation:
• minimum score: -10
• maximum score: 25
Performance:
• The area under the AROC curve for the logistic regression model is 0.65.
Specialty: Infectious Diseases