Table 4.

Logistic regression analysis

	OR (95% CI)	P-value
I. Univariate model
Ethnicity	0.34 (0.17-0.71)	0.004
Arrival hospital after 2.5 hours of symptom onset	0.033 (0.005-0.21)	< 0.001
Age	0.99 (0.98-1.0)	0.29
Gender	1.8 (1.1-2.9)	0.02
Previous stroke and diabetes	0.48 (0.14-1.6)	0.24
Blood pressure higher than 185/110	0.68 (0.35-1.3)	0.25
NIHSS < 5	8.1 (4.2-16)	< 0.001
NIHSS > 25	4.9 (0.30-79)	0.26
Oral anticoagulation use	0.15 (0.035-0.61)	0.008
II. Bivariate model (ethnicity is added to each model)
Arrival hospital after 2.5 hours of symptom onset	0.60 (0.24-1.50)*	0.28
Age	0.30 (0.14-0.63)*	0.001
Gender	0.36 (0.17-0.74)	0.006
Previous stroke and diabetes	0.34 (0.17-0.71)	0.004
Blood pressure higher than 185/110	0.33 (0.15-0.71)*	0.004
NIHSS < 5	0.36 (0.16-0.78)	0.01
NIHSS > 25	0.34 (0.16-0.73)	0.006
Oral anticoagulation use	0.31 (0.15-0.64)*	0.002
III. Multivariate model	0.38 (0.13-1.16)	0.09

The univariate model (I) shows the odds ratio for receiving thrombolysis for the variables separately. Next, the association with ethnicity is analysed in the bivariate model (II), using the same preselected variables. Finally, in the multivariate model (III) the variables which significantly changed the odds ratio (age; blood pressure higher than 185/110; oral anticoagulation use; arrival at the hospital after 2.5 hours of symptom onset) were entered in a model together with ethnicity.

OR = odds ratio; CI = confidence intervals; NIHSS = National institute of health stroke scale; * indicates a 5% or greater change in odds ratio