Skip to main content

Table 3 Performance of each biomarker in predicting infection

From: Accuracy of pancreatic stone protein for the diagnosis of infection in hospitalized adults: a systematic review and individual patient level meta-analysis

Biomarker (n =)

With/without infection

Youden’s index on predicted probabilities

Biomarker’ cut-off values for Youden’s index

AUC [95%CI) (SE)

Sensitivity (95%CI)

Specificity (95%CI)

PPV (95%CI)

NPV (95%CI)

PLR (95%CI)

PLR (95%CI)

PSP [ng/ml] n = 631

369/262

0.49

44.18

0.81 [0.78, 0.85] (0.017)

0.66 [0.61, 0.71]

0.83 [0.78, 0.88]

0.85 [0.81, 0.89]

0.63 [0.58, 0.68]

3.95 [2.99, 5.23]

0.40 [0.35, 0.47]

CRP [mg/ml] n = 629

369/260

0.39

99.05

0.77 [0.73, 0.80] (0.019)

0.70 [0.66, 0.75]

0.69 [0.63, 0.74]

0.76 [0.72, 0.81]

0.62 [0.56, 0.67]

2.35 [1.92, 2.87]

0.44 [0.38, 0.53]

PCT [ng/ml] n = 527

175/352

0.44

0.20

0.78 [0.74, 0.82] (0.022)

0.83 [0.79, 0.87]

0.62 [0.54, 0.69]

0.81 [0.78, 0.85]

0.64 [0.56, 0.71]

2.14 [1.77, 2.60]

0.27 [0.21, 0.36]

  1. Using paired DeLong test on 527 patients for whom all three biomarkers are available, the p values for the null hypothesis ‘true difference in AUC is 0’ are 0.010 and 0.044 for comparisons PSP versus PCT and PSP versus CRP respectively (without adjustment for multiple testing)
  2. PPV and NPV stand for positive, respectively, negative predicting value, when PLR and NLR for positive, respectively, negative likelihood ratio