Table 3

Multivariable Cox proportional hazards model for the identification of independent variables associated with in-hospital death

Variable

Start model

Final model

HR

95% CI

p value

HR

95% CI

p value

Δ ln(lactate) T1 to T2*

0.20

0.05 to 0.79

0.022

0.20

0.05 to 0.76

0.018

SBP T1 _{per mmHg}

1.00

0.99 to 1.01

0.56

Not in model

0.87

Heart rate T1 _{per beat/minute}

1.01

0.99 to 1.02

0.47

Not in model

0.66

GCS T1 _{per unit}

0.93

0.87 to 0.99

0.034

0.93

0.88 to 0.99

0.022

The variables were simultaneously entered in the model (start model). A backward elimination method was used to construct the final model.

* Δ ln(lactate) T1 to T2: for every 63% decrease (100*(1-(1/e)) = 63%) of the lactate level at T2 relative to the level at T1, the hazard of death decreased by 80% (100 (1-HR)) in the final model (95% CI = 24 to 95%). e = 2.71828, GCS = Glasgow Coma Scale, HR = hazard ratio, ln = natural logarithm, SBP = systolic blood pressure, T1 = on arrival of the ambulance on the scene, T2 = just before or on arrival at the emergency department.

Jansen et al. Critical Care 2008 12:R160   doi:10.1186/cc7159