Serial interleukin 6 measurements in the early diagnosis of neonatal sepsis.

The objective of the present study was to evaluate serial interleukin 6 (IL6) levels in the early diagnosis of neonatal sepsis. Subjects included 255 neonates from the Neonatal Unit of Johannesburg Hospital evaluated for suspected sepsis between February and May 1998. All infants had IL6, full blood count (FBC), C reactive protein (CRP) and blood cultures done at presentation. CRP and IL6 were repeated after 24 h. Infants were categorized into groups according to the likelihood of infection on the basis of clinical presentation, CRP, FBC and culture results, i.e., group 1 (no infection) to group 4 (definite infection). IL6 was compared between the groups by the U-test of Mann-Whitney; stepwise logistic regression was done to establish the best predictors of infection, sensitivity, specificity, positive and negative predictive values were determined. The initial IL6 level was significantly raised in those infants with possible infection [880.67 pg/ml (2966.04), p value 0.0104], probable infection [422.62pg/ml (4077.7), p value 0.0021] and definite infection [11164.39pg/ml (24139.77), p value 0.0000] as compared to those infants without infection [58.65 (182.4)]. The best predictors of infection were the combination of the initial IL6 value and CRP value after 24 h (goodness of fit 97.7 per cent). An initial IL6 value below 20 pg/ml gave a negative predictive value of 90.18 per cent. It is concluded that an IL6 value done at the time of presentation of signs and symptoms suggestive of infection is useful in the early diagnosis of neonatal sepsis. In particular, an initial IL6 value below 20 pg/ml may allow antibiotics to be withheld in a number of infants evaluated for sepsis. There is no benefit in serial determination of IL6 in the diagnosis of neonatal sepsis.