medical decision support medical decision support
Sign In

Models of Thao et al for Predicting Mortality in a Patient with Tuberculous Meningitis: HIV-Positive Patient

Description

Thao et al reported models for predicting mortality in a patient with tuberculous meningitis. One model predicts mortality for HIV-positive patients and a second predicts mortality for HIV-negative patients. The authors are from Oxford University Clinical Research Unit, University of Oxford, University of Cambridge, Hospital for Tropical Diseases in Ho Chi Min City, Liverpool School of Tropical Medicine and Pham Ngoc Thach Hospital in Ho Chi Min City.


Patient selection: HIV-positive adult with tuberculous meningitis in Vietnam

 

Outcome: 9-month mortality

 

Parameters:

(1) weight in kg (from 75 to 25 kgs)

(2) MRC grade for tuberculous meningitis

(3) CSF lymphocyte count per microliter (from 1 to 500)

(4) serum sodium in mmol/L (from 100 to 155)

(5) absolute CD4 count in peripheral blood per microliter (from 1 to 400)

 

points for weight =

= MAX(0,150 - (2 * weight))

 

MRC Grade

Points

1

0

2

40

3

90

 

 

Serum Sodium

Points

100 to 130

(0.04812 * ((sodium)^2)) - (14.07 * (sodium)) + 1014

130 to 155

MAX(0,(1.769 * (sodium)) - 231.1)

 

 

CSF lymphocytes

Points

1 to 30

(-0.002684 * ((count)^3)) + (0.1637 * ((count)^2)) - (3.4 * (count)) + 46.77

30 to 100

(-0.000007806 * ((count)^3)) + (0.002576 * ((count)^2)) - (0.3481 * (count)) + 28.01

100 to 500

(-0.0000001961 * ((count)^3)) + (0.0002248 * ((count)^2)) - (0.102 * (count)) + 19.29

> 500

0

 

 

CD4 Count

Points

1 to 30

(-0.002087 * ((count)^3)) + (0.1336 * ((count)^2)) - (3.115 * (count)) + 53.96

30 to 150

(0.0008062 * ((count)^2)) - (0.2657 * (count)) + 31.22

150 to 400

(0.00007249 * ((count)^2)) - (0.07691 * (count)) + 19.2

> 400

0

 

score =

= SUM(points for all 5 parameters)

 

Interpretation:

• minimum score: 0

• maximum score: 370.4

 

Score

Probability of 9-Month Mortality

< 94.4

< 5%

94.4 to 354

X = (0.0000454 * ((score)^2)) + (0.001086 * (score)) - 3.34;

probability = 1 / (1 + EXP((-1) * X))

> 354

> 95%

 

Performance:

• The area under the ROC curve is 0.78.


To read more or access our algorithms and calculators, please log in or register.

Specialty