The early diagnosis of neonatal sepsis: an evaluation of leucocyte counts and acute phase reactants
Philip, Alistair Gordon Sleigh
Several laboratory tests suggested as being helpful in the diagnosis of neonatal sepsis were evaluated individually and in combination. The tests chosen were considered to be rapid, simple and inexpensive, so that they could be generally available.There were 524 babies evaluated for sepsis in the first week after birth and 56 babies evaluated between 8 and 60 days of age. The reasons for investigation were varied, being primarily "risk factors" (e.g., prolonged rupture of membranes, maternal fever, etc.) in the first two days and "clinical factors" (e.g., apnoea, lethargy, abdominal distension, etc.) beyond that time.Of the babies investigated during the first week, 41 proved to have sepsis and in another 34 it was "very probable" that infection was present. Among the older infants, 12 had sepsis, 2 had necrotising enterocolítis and 5 had "very probable" infection.No single test can be considered entirely satisfactory, but the single most useful test seems to be the immature /total neutrophil ratio. Unlike some other tests, it does not seem to be influenced by birth weight or gestational age. When the immature /total neutrophil ratio (or fraction) was 5 0.2, 34/41 infants with sepsis were detected in the first week (sensitivity = 83 %), and 7/12 (58 %) between 8 and 60 days, but positive predictive accuracy (number of proved cases with a positive test /total positive tests) was only 24% and 40% respectively, In contrast, a combination of tests designated as a "sepsis screen" (two or more of five diagnostic findings: wbc < 5.0 x 109 /Q, I/T ratio 5 0.2, ESR > 15 mm /h, latex C- reactive protein positive, and latex haptoglobin positive) detected 38/41 infants with sepsis in the first week (sensitivity = 93 %) and 10/12 (83 %) between 8 and 60 days. Positive predictive accuracy for the sepsis screen was 38% (0 -7 days) and 43% (8 -60 days). When "infection" (sepsis plus "very probable ") was evaluated, sensitivity of the sepsis screen was 897 and positive predictive accuracy was 68% for infants aged 0 -7 days. The most useful pair of tests in the first week after birth was wbc < 5.0 x 109 /Q and I/T ratio 5 0.2 which detected 15/41 infants with sepsis (37 %) with a positive predictive accuracy of 68% for sepsis and 86% for "infection." Between 8 and 60 days, the best pair of tests was latex CRP positive and I/T ratio 5 0.2 with 6/12 (50 %) infants with sepsis detected and positive predictive accuracy of 60% for sepsis and 80% for "infection."In the second half of the study, the tests were used at the bedside. Antibiotic use decreased in those babies who subsequently proved not to be infected. In the first half of the study, only 14% of the babies evaluated did not receive antibiotics, but in the second half of the study, this number increased to 47 %. This reduced the ratio of babies treated with antibiotics to babies with proven sepsis from 11:1 to 6.6:1.Although these tests are non -specific indicators of infection, they are sufficiently simple and inexpensive to be used in almost any hospital where babies are delivered. There is some evidence (which needs to be confirmed) that acute phase proteins are valuable, not only in diagnosis, but also in following the course of the illness. This should allow more rational decisions to be made both in starting and in stopping antibiotics.