• C-Reactive protein should NOT be checked as a routine part of a septic screen in neonates.
• It may be helpful in excluding infection in some cases of sepsis if normal levels are obtained 24-48 hours after the onset of symptoms. It should therefore be considered in cases in which diagnosis of sepsis is unclear, and where negative cultures may not give sufficient reassurance to discontinue antibiotics. Common examples may include when maternal intrapartum IV antibiotics were given, and in late onset sepsis where lumbar puncture specimens were unobtainable or blood-stained.
• A level of greater than 10mg/litre is abnormal
  • Less than 10 is normal and makes sepsis much less likely (see Background and Applications).
• Judicious use of CRP to enable earlier discontinuation of antibiotics in some instances may be cost-effective, decrease the possibility of resistance and in some cases may decrease duration of hospital stay.

Sample

• The laboratory requires 0.10-0.15 ml of SERUM, therefore 0.3-0.5ml (depending on haematocrit) of whole blood will be required, in a lithium heparin (green top) tube.

Background

• CRP is an acute phase protein synthesised in the liver in response to inflammatory cytokines. Its levels may increase up to a thousand-fold during an acute phase response. Its short half-life (19hours) means levels should fall rapidly after elimination of the (often microbial) source. ¹
• There is generally a delay of up to 24 hours between onset of symptoms of infection and a rise in serum CRP. Sensitivity of the test at presentation is only 40% - that is, 60% of subsequently proven sepsis episodes will have a normal initial CRP (compared to 80% sensitivity of immature to total neutrophil ratio). ²
• With this in mind CRP level should not be checked at birth or at initial presentation with symptoms suggestive of sepsis.

Applications

• The value of CRP is in its reliability in the 24-48 hours after birth/onset of infection.
• In early onset sepsis (before 72 hours old), a single CRP 24 hours into illness has a 93% sensitivity for "probable" sepsis. ³  Two measures 24 hours apart are even better. A more intuitive measure is the Bayesian ratio (likelihood ratio) - for a single normal CRP the ratio for probable sepsis was 0.10. In effect this means that having obtained a normal CRP one can assume probable sepsis to be 10 times less likely. For 2 measures the ratios is 0.03, making probable sepsis 30 times less likely.
• In late onset sepsis the reliability of the test is similarly high. A single CRP has a sensitivity of 85%, with likelihood ratio of 0.19 (sepsis 5 times less likely given normal result), and 2 separate measures have a likelihood ratio 0.07 (sepsis 14 times less likely).
• The positive predictive value for a raised CRP is poor so a positive test is poorly predictive of sepsis. False positives arise from intraventricular haemorrhage, meconium aspiration, NEC, pneumothorax, surgery and immunisation.
• Reassuringly, false positives don’t tend to occur with birth asphyxia, prolonged rupture of membranes, respiratory distress syndrome and viral illnesses (except invasive HSV).
• False negative results generally only occur early in infective episodes, though also in UTI. False negatives don’t appear to occur in either infection with coagulase negative staphylococcus ² or childhood meningitis, though there have been few studies focussed on neonatal meningitis. ^4,5
• A level of 10mg/litre has consistently been shown to be the most reliable cut-off to indicate sepsis. There is a link between the level of elevation of the CRP and the risk of sepsis, with positive predictive value steadily increasing up to CRP >100.
• Judicious use of CRP to enable earlier discontinuation of antibiotics in such instances would be cost-effective, decrease the possibility of resistance and in some cases may decrease duration of stay.

References