Using ordinal logistic regression to evaluate the performance of laser-Doppler predictions of burn-healing time

Table 3 Demographic and observational predictors of healing based on 581 burn areas, where se* is the bootstrapped standard error of the estimated parameter.

Parameter	$\hat{β}$	se	se*	p-value 1	p-value 2
α ₁	1.209	0.267	0.532	.041	0.023
α ₂ - α ₁	1.008	0.095	0.122	< .001	< .001
Burn site 2 (extremity)	0.912	0.225	0.361	.018	0.012
Burn site 3 (torso)	0.370	0.250	0.300	.21	0.217
Burn site 4 (face)	1.732	0.469	1.18	.023	0.141
Gender (female)	-0.696	0.197	0.389	.084	0.073
Race (black)	0.168	0.587	2.34	.888	0.943
Age (yrs)	-0.0084	0.0041	0.0072	.236	0.239
(Age - 35)²	-0.0006	0.0002	0.00034	.106	0.103
%TBSA	-0.0104	0.0088	0.043	.736	0.762
Burn cause: flame	-0.589	0.219	0.370	.098	0.111
Burn cause: chemical	-0.847	0.503	1.97	.351	0.668
Burn cause: flash	-0.385	0.436	0.772	.617	0.618
Burn cause: electrical	-2.175	0.964	8.49	.687	0.798
Burn cause: contact	-1.507	0.511	1.36	.143	0.266

The two bootstrap p-values from the two-sided bootstrap hypothesis test are from the bootstrap percentile and bootstrap parametric tests described in the text. The more reliable p-values derived from permutation tests are given in the text.

ISSN: 1471-2288