Table 4 Univariate and multivariate analyses/logistic regression for late neurological toxicities

From: Early toxicity predicts long-term survival in high-grade glioma

  

Univariate analysis

Multivariate analysis b

Variable

Comparison

OR (95% CI) a

P -value

OR (95% CI) a

P -value

Age

<50

—

 

—

—

 

⩾50

0.95 (0.6, 1.5)

0.81

—

—

Zubrod

0

—

 

—

—

 

1, 2, 3

0.93 (0.6, 1.4)

0.75

—

—

Surgery

Partial/total resection

—

 

—

—

 

Biopsy

1.09 (0.7, 1.8)

0.73

—

—

Neurological dysfunction

None/minor

—

 

—

—

 

Moderate/severe

0.76 (0.5, 1.2)

0.26

—

—

Mental status

Normal function

—

 

—

—

 

Minor/gross confusion

0.86 (0.5, 1.5)

0.62

—

—

Twice-daily RT

No

—

 

—

—

 

Yes

0.37 (0.2, 0.6)

0.0001

0.42 (0.2, 0.7)

0.002

Histology

AA

—

 

—

—

 

GBM

1.76 (0.7, 4.4)

0.23

—

—

Chemotherapy

No

—

 

—

—

 

Yes

0.64 (0.4, 1.1)

0.09

—

—

BED

⩽120

—

 

—

—

 

>120

2.83 (1.7, 4.8)

<0.0001

1.98 (1.1, 3.4)

0.016

Acute CNS

No

—

 

—

—

 

Yes

2.11 (1.1, 4.2)

0.03

2.40 (1.2, 4.8)

0.014

  1. Abbreviations: AA=anaplastic astrocytoma; BED=biological-equivalent dose; CI=confidence interval; CNS=central nervous system; GBM=glioblastoma; OR=odds ratio; RT=radiation therapy.
  2. aOdds ratio: The odds ratio of 1 indicates no difference between the two subgroups, less than 1 indicates a protective effect.
  3. bMultivariate model derived from stepwise selection.
  4. Variable(s) not included in final model: chemotherapy (dropped out during the stepwise selection process).
  5. Variable(s) dropped from modelling as not significant with reference variables: age, Zubrod performance status, surgery, neurologic function, mental status, histology (not significant during the univariate setting).
Back to article page