TABLE 3.

Diagnostic utility of CRP and cytokine measurements in predicting sepsis in VLBW infantsa

StudyInfant populationNo. of patientsNo. of culture- proven sepsis eventsTime of testingTestSensitivity (%)Specificity (%)PPV (%)NPV (%)
Franz et al., 2001 (145)Preterm and term709 (387 preterm)14BirthIL-8NRNR
    22-26 wk93 (a)55 (a)
    27-29 wk89 (a)56 (a)
    30-32 wk83 (a)77 (a)
Kuster et al., 1998 (272)VLBW12521>2 daysIL-1ra100 (a)92 (a)NRNR
IL-689 (b)83 (b)
CRP50 (b)93 (b)
Ng et al., 1997 (340)VLBW6845>3 daysCRP + IL-698 (c)91 (c)90 (c)98 (c)
Chan and Ho, 1997 (77)VLBW7030>3 daysCRP48 (d)79 (d)74 (d)55 (d)
Wagle et al., 1994 (487)<29 wk gestation123361-87 daysCRP90 (a)81 (a)48 (a)98 (a)
Seibert et al., 1990 (426)23-31 wk gestation1258BirthCRP63 (a)70 (a)13 (a)96 (a)
67 (b)82 (b)60 (b)86 (b)
8523>3 daysCRP57 (a)61 (a)30 (a)82 (a)
58 (b)90 (b)93 (b)49 (b)
  • a Selected studies are summarized with sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) calculated for CRP and cytokine levels in VLBW infants. Tests were compared to different definitions of infection, which included culture-proven plus clinical sepsis (a), culture-proven sepsis (b), septicemia plus NEC plus microbiologically confirmed focal infection (c) or sepsis-like syndrome with positive blood, CSF, or joint fluid culture (d). Serial testing was performed in all studies except one (Chan et al.), but the number and timing of tests varied for the other studies. Positive values were defined as CRP ≥ 1.0 to 1.5 mg/dl, IL-6 ≥ 25 to 31 pg/ml, IL-8 ≥ 70 pg/ml, and IL-1 receptor antagonist (IL-1ra) ≥ 12,000 pg/ml. NR, not reported.