Skip to main content
Annals of Intensive Care

Table 3 Secondary analysis with common matching variables included as covariates—conditional logistic regression for differential treatment effects of IRP A and IRP B on mortality

From: Pharmacogenomic biomarkers do not predict response to drotrecogin alfa in patients with severe sepsis

Factor/effect

Estimate

SE

Odds Ratio (OR) estimate

OR 95% CI

P valuec

Lower

Upper

Analysis for IRP A

      

IRP A*Treatment Interactiona

−0.34

0.304

0.710

0.391

1.289

0.26

IRP A+ : DrotAA versus NonDrotAA

0.08

0.246

1.087

0.671

1.761

0.74

IRP A−: DrotAA versus NonDrotAA

0.43

0.146

1.532

1.151

2.039

< 0.01

DrotAA: IRP A+ versus IRP A−

−0.32

0.259

0.725

0.436

1.205

0.22

NonDrotAA: IRP A+ versus IRP A−

0.02

0.150

1.022

0.761

1.372

0.89

AIM Panel PCsb

     

0.13

Common covariates used in matching

     

0.74

 Age (per year)

−0.02

0.024

0.976

0.930

1.023

0.30

 APACHE II (per point)d

0.05

0.080

1.054

0.901

1.233

0.51

 SAPS II (per point)d

0.01

0.041

1.006

0.929

1.091

0.88

 Respiratory dysfunction (Yes vs. No)

0.86

0.778

2.371

0.515

10.90

0.27

Analysis for IRP B

IRP B*Treatment Interactiona

−0.10

0.325

0.908

0.480

1.719

0.77

IRP B+: DrotAA versus NonDrotAA

0.22

0.278

1.251

0.726

2.155

0.42

IRP B−: DrotAA versus NonDrotAA

0.32

0.138

1.377

1.050

1.806

0.02

DrotAA: IRP B+ vs IRP B−

−0.19

0.285

0.826

0.472

1.443

0.50

NonDrotAA: IRP B+ vs IRP B−

−0.10

0.157

0.909

0.668

1.236

0.54

AIM panel PCsb

     

0.15

Common covariates used in matching

     

0.75

 Age (per year)

−0.02

0.024

0.976

0.931

1.024

0.32

 APACHE II (per point)d

0.04

0.080

1.045

0.894

1.223

0.58

 SAPS II (per point)d

−0.00

0.041

0.997

0.919

1.081

0.94

 Respiratory dysfunction (Yes vs. No)

0.12

0.935

1.127

0.180

7.050

0.90

  1. Analysis for IRP A involved 376 discordant matched sets from a total of 637 matched sets
  2. Analysis for IRP B involved 372 discordant matched sets from a total of 634 matched sets
  3. aThe interaction odds ratio is a ratio of odds ratios
  4. bA total of 10 AIM Panel PCs were included which account for 33.9% of the variance in the AIM Panel data for the Matched-INDICATED Primary Analysis Population based on all cohorts
  5. cP values from conditional logistic regression partial likelihood ratio tests for the IRP*Treatment Interaction and the combined AIM Panel PCs; all other P values are from Wald Chi-square tests
  6. dSites with APACHE II scores were analyzed with 0’s for SAPS II scores, and vice versa
Back to article page