Skip to main content

Table 2 Association between plasma procalcitonin level at admission and all-cause 50-day mortality in multivariable logistic regression analysis

From: Usefulness of procalcitonin at admission as a risk-stratifying biomarker for 50-day in-hospital mortality among patients with community-acquired bloodstream infection: an observational cohort study

Covariate

Odds ratio

95% CI

P value*

Percent of cases correctly classified

AUROCa

(95% CI)

Model #1: Plasma PCT, ROC-defined threshold

81%

0.721 (0.677–0.762)

 Procalcitonin > 4.24 ng/mL

2.58

1.57 to 4.25

0.0002

 Age (years)

1.02

1.01 to 1.03

0.0007

 Pathogen genus, Streptococcus

0.52

0.20 to 1.30

0.16

 Pathogen genus, Staphylococcus

1.59

0.91 to 2.78

0.10

 Pathogen genus, Klebsiella

1.92

0.81 to 4.54

0.14

Model #2: Plasma PCT quartiles (continuous)

80%

0.711 (0.669–0.754)

 Procalcitonin, quartiles (continuous)

1.47

1.17 to 1.85

0.001

 Age (years)

1.02

1.01 to 1.03

0.0007

 Pathogen genus, Streptococcus

0.52

0.21 to 1.31

0.17

 Pathogen genus, Staphylococcus

1.55

0.89 to 2.70

0.12

 Pathogen genus, Klebsiella

1.93

0.82 to 4.55

0.13

Model #3: Plasma PCT, 4th vs. 1st to 3rd quartiles

81%

0.702 (0.658–0.744)

 Procalcitonin, 4th quartile

2.12

1.26 to 3.54

0.004

 Age (years)

1.02

1.01 to 1.03

0.0004

 Pathogen genus, Streptococcus

0.49

0.20 to 1.24

0.13

 Pathogen genus, Staphylococcus

1.54

0.89 to 2.69

0.13

 Pathogen genus, Klebsiella

2.07

0.88 to 4.88

0.10

Model #4: Plasma PCT, 4th vs. 1st quartile

80%

0.723 (0.660–0.780)

 Procalcitonin, 4th quartile

3.30

1.53 to 7.12

0.002

 Age (years)

1.01

1.00 to 1.02

0.08

 Pathogen genus, Streptococcus

0.47

0.13 to 1.72

0.25

 Pathogen genus, Staphylococcus

1.35

0.62 to 2.97

0.45

 Pathogen genus, Klebsiella

1.93

0.48 to 7.78

0.36

  1. Note. AUROC area under the receiver operating characteristic curve, PCT procalcitonin
  2. *Logistic regression using the forced entry method
  3. aAUROC of the prognostic indices generated by the logistic regression model to discriminate between positive and negative cases