Model of Misra et al for a Patient with Tuberculous Meningitis

 

Outcome: based on Barthel index at 3 months after initiation of therapy. A poor outcome was defined as death or a Barthel index < 12.

 

Parameters:

(1) age of the patient in years

(2) stage of tuberuclous meningitis

(3) focal weakness

(4) cranial nerve palsy (ophthalmoplegia, facial weakness, deafness, nasal speech, nasal regurgitation, etc)

(5) hydrocephalus

Parameter	Finding	Points
age of the patient in years	< 12 years of age	2
	>= 12 years of age	1
stage of TB meningitis	Stage I	3
	Stage II	2
	Stage III	1
focal weakness	absent	2
	present	1
cranial nerve palsy	absent	2
	present	1
hydrocephalus	absent	3
	obstructing	1
	communicating	2

 

X =

= (-2.72 * (points for age)) - (1.53 * (points for stage)) - (2.05 * (points for focal weakness)) + (2.63 * (points for cranial nerve palsy)) - (0.88 * (points for hydrocephalus)) + 7.15

 

where:

• Based on the discussion on page 59 I would have expected the regression coefficient for cranial nerve palsy to have a negative value. However, changing the coefficient resulted in absurd results.

• The output of the model as given above does not make sense. The one thing that did seem to give better performance was to reverse the point assignment for cranial nerve absent and present (1 vs 2 points).

 

probability of a poor outcome =

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