This is  a nicely done study, with interesting data, but I am not sure I agree 100% with the conclusions.

Lots of newborn babies get antibiotics for suspected infection, they are usually stopped at 48 hours if the cultures are negative, which is much the most frequent outcome. New recommendations are to stop the antibiotics at 36 hours if the cultures are negative, and the baby is without signs of sepsis.

The authors of this study measured CRP with a bedside machine at 18 hours and then kept the results masked from the caregivers. The idea was to see if using a negative CRP was sufficiently predictive that antibiotics could be stopped earlier, after about 18 hours. There were preterm, late preterm and term babies in the study, 1202 of them. Which makes it quite big. There were 16 babies who actually had positive cultures. Which shows how many babies are unnecessarily exposed to antibiotics. In addition there were 107 babies who are called possible sepsis, as their mothers had received antibiotics and they were treated for more than 72hours. Most of the possible sepsis babies were probably not infected, but some may have been.

The CRP threshold that was used was 10 mg/L. Of the babies with an elevated CRP, only 14% had possible or proven sepsis, and under 3% had proven sepsis. Demonstrating once again the low specificity of CRP. There was only one infant with definite early onset sepsis who had a negative CRP, but another 43 who may have been infected, using their definition. Interestingly the negative CRP may be more useful for the late preterm and term baby, as all the babies more 34 weeks gestation who were asymptomatic at 18 hours of age and who had a CRP less than 10 were uninfected except one, whose mother had a positive blood culture.

The reason I say in the title of this post that I am not yet convinced is that it seems to me that there is a great risk that there will be unnecessary prolongation of antibiotic courses in uninfected infants if this becomes a routine. I am also not sure about the definition of possible sepsis, it is a real dilemma whether to start and or continue antibiotics in such infants. Many mothers are treated for fever, which becomes frequent the longer an  epidural is in place, many of the mothers don’t have a blood culture done, or other accurate investigations for serious infections.

Anyway overall this confirms what I thought about CRP, that they are very non-specific, but fairly sensitive. If you just focus on proven sepsis in infants who were 35 weeks or more, the sensitivity was 100%. I think it is likely that many of the ‘possible sepsis’ babies did not have sepsis, which is why there was little inflammation. They can be most helpful then in deciding to stop antibiotics (or sometimes making you feel more comfortable about a decision not to start them).

This study also makes me re-question the advice that I wrote about before from NICE, who suggest a CRP at baseline and then a repeat at 12 to 18 hours. I think this study suggests to me that the initial CRP would be a waste of time, if  you are going to do one, and you decide that a low CRP will change your management decision, then you should probably just do one, and do it at 18 hours after starting treatment.