[C-reactive protein concentrations during initial (empiric) treatment of neonatal sepsis]

Neonatal septicaemia is characterized by high mortality so that intravenous antibiotics must be administered on clinical suspicion. Initial antibiotic therapy, before the results of microbiological evaluation, is based on empirical data in regard to sensitivity of prevalent bacterial strains. In the past years, aetiological causes of neonatal sepsis have been changed with an increased bacterial resistance to the usual combination of initial antibiotics.

We compared changes of serum C-reactive protein (CRP) concentrations in neonates with proven neonatal sepsis in response to initial antibiotic therapy (inappropriate or appropriate). Our hypothesis was that changes of CRP levels during the first 48 hours of treatment and before microbiologic results could be useful in evaluation of effectiveness of empiric antibiotics.

Our prospective study included all neonates with suspected sepsis referred to our Intensive Care Unit from January 1992 to December 1996. Neonates received ampicillin and gentamycin/or amikacin (during the first week of life), while infants older than 7 days of life were given combination of cloxacillin and aminoglycozides. In patients with late neonatal sepsis who also had meninigitis, cloxacillin was substituted with ampicillin. Microbiological identification was performed with routine bacteriological methods. Susceptibility of isolated bacterial strains to antibiotics was performed by Kirby-Bauer disc-diffusion method. Serum concentration of CRP was measured by immunoturbidimetry (Turbox CRP, Orion Diagnostica) and CRP concentration higher than 20 mg/l was regarded as elevated. Blood sampling for CRP measurements were taken before the treatment (CRP0), and during the first (CRP1) and second (CRP2) day of empiric therapy. The interval between sampling was from 12 to 24 hours.

A total of 1520 neonates were evaluated during this study period and 47 patients fulfilled criteria for final analysis. In 14 of 47 patients initial antibiotic treatment was inappropriate. The most frequent resistant strains was KI. pneumoniae (6) followed with St. aureus (4), E. coli (2) and Pseudomonas (2). During initial evaluation six patients had concomitant meningitis while two had concomitant septic arthritis and two necrotizing enterocolitis, respectively. Seven (50%) of 14 patients with non-adequate initial treatment died. In 33 cases of adequately treated septicaemia the course was uncomplicated and no lethal outcome was observed. In the first group of 14 patients who received inappropriate treatment serum CRP concentations (mg/L; mean and +/- SD) were: CRP0 = 107.5 +/- 65.6; CRP1 = 155.3 +/- 75.7; CRP2 = 209.1 +/- 67.0, while in 33 repeated samples of the 33 patients in the second group who received adequate treatment the following results were recorded: CRP0 = 124.0 +/- 78.1; CRP1 = 133.8 +/- 63.5; CRP2 = 94.6 +/- 46.4. Increase in serum CRP concentration in the first group during the first 48 hours of initial non-adequate therapy was significantly higher (p = 0.015, two way ANOVA) than in the second group with appropriate treatment. During the first 24 hours of treatment increase in serum concentration of CRP was registered in 12 (85.7%) of 14 measurements in patients with non-adequate therapy and in 19 (56.7%) of 33 measurements in patients with adequate therapy. In the first group during the second day of treatment, in 11 (78.6%) of 14 cases an increase in serum CRP concentration was recorded while in 3 (14.3%) cases CRP concentration decreased. In 31 (91.2%) of 34 measurements in patients with adequate treatment CRP concentration decreased during the second day of treatment and in only 3 (8.8%) cases CRP concentration increased. With an increase in serum concentration of CRP more than 10 mg/L in the second day of antibiotic treatment, probability of non-adequate antibiotic therapy (positive predictive value) was estimated to be 77.0%. Any recorded decrease of serum CRP concentration may confirm appropriate choice of antibiotics during the second day of treatment with probability of 93.3% (negative predictive value).

Measurement of serum CRP concentration is useful for diagnosis of severe neonatal sepsis. In our study all isolated bacterial strains were comparable in their ability to activate systemic inflammatory response and CRP production. It is known that serial CRP measurements during neonatal sepsis are useful in making decision to cease antibiotic treatment. The highest serum CRP concentrations were detected during the first day of illness but, in some cases even with appropriate treatment, CRP peak levels due to sustained pro-inflammatory action of interleukin-6 production could be detected even 24 hours after treatment was started. Our study showed that in patients with non-adequate initial antibiotic therapy of neonatal sepsis serum CRP concentrations increase further during the second day of treatment. By contrast, the use of appropriate antibiotic therapy in the same time period followed the significant decrease of serum CRP levels in our patients. Therefore, increase of CRP level during initial treatment, especially during the second day of treatment of neonatal sepsis should be taken as indication for replacement of initial antibiotics, even before sensitivity of microbiological causes to given antibiotics is known.