Skip to main content
Annals of Intensive Care

Table 2 Effects of colloid infusion on 90-day mortality in Cox regression analysis

From: Fluid therapy and outcome: a prospective observational study in 65 German intensive care units between 2010 and 2011

Odds/hazard ratio estimates

Unadjusted/crude mortality (population eligible for multivariable analysis)a

Adjustment for baseline variables

Full model, adjusted for evolution variables

Effect

OR

95% CI

HR

95% CI

HR

95% CI

Colloid infusion

2.027

1.697

2.422

1.666

1.405

1.976

1.003

0.980

1.027

Gender, male

   

1.141

0.974

1.337

0.798

0.679

0.936

Predicted risk of death 90–99%

   

6.003

3.773

9.551

4.505

2.781

7.298

Predicted risk of death 80–89%

   

3.266

2.046

5.212

2.630

1.637

4.226

Predicted risk of death 70–79%

   

3.018

1.905

4.780

2.250

1.412

3.586

Predicted risk of death 60–69%

   

2.893

1.870

4.474

2.184

1.403

3.401

Predicted risk of death 50–59%

   

2.171

1.365

3.453

1.718

1.070

2.757

Predicted risk of death 40–49%

   

2.073

1.353

3.176

1.732

1.128

2.660

Predicted risk of death 30–39%

   

1.648

1.064

2.552

1.484

0.957

2.303

Predicted risk of death 20–29%

   

1.777

1.154

2.736

1.611

1.045

2.482

Predicted risk of death 10–19%

   

1.833

1.189

2.826

1.759

1.140

2.712

Chronic kidney disease

   

2.149

1.812

2.549

1.960

1.645

2.336

Severe sepsis on admission

   

2.017

1.648

2.469

1.284

1.025

1.608

Vasopressor equivalent > 0.6 mg/h

      

1.019

1.002

1.035

SOFA score AUC

      

1.030

1.024

1.036

Cumulative red blood cell products (per litre)

      

0.977

0.958

0.997

Cumulative other blood products (per litre)

      

1.002

0.968

1.037

Cumulative fluid balance (per litre)

      

1.014

1.008

1.020

Daily crystalloid infusion (per litre)

      

1.121

1.067

1.178

  1. Risk of mortality was estimated from severity scores as detailed in the methods section. Gender, risk of mortality categories, vasopressor requirement, chronic kidney disease and severe sepsis on admission were used as binary variables (yes/no), whereas the hazard ratios (HRs) for fluid-based variables and Sequential Organ Failure Assessment (SOFA) score describe the excess risk per litre of fluid and per SOFA area under the curve (AUC), respectively
  2. CI confidence interval
  3. aTo show the effect of the modelling, we showed the unadjusted odds ratio for the population that was eligible for the multivariable analysis. Crude OR for all patients in the registry is shown in the text
Back to article page