Validation of C-reactive protein in the early diagnosis of neonatal sepsis in a Tertiary Care Hospital in Kenya

Abstract

Neonatal sepsis continues to be a major cause of morbidity and mortality in developing countries. Inadequacy of microbiological laboratory facilities compromise ability to discriminate between neonates who need and those who do not need early commencement of antibiotics for sepsis. This has led to overuse of antibiotics, emergence of multidrug resistant organisms and inefficient use of scarce resources. Serum C-reactive protein levels have been shown to be a sensitive and specific indicator of sepsis, with high predictive values and overall accuracy. Research question: What is the diagnostic utility of serum C-Reactive protein level determination in the early diagnosis of neonatal sepsis at Kenyatta National Hospital? Objectives: To evaluate utility of C-Reactive Proteins in the early diagnosis of neonatal sepsis in a tertiary care Newborn Unit in Kenya. Methods: A hospital based cross-sectional study, was carried out at the Newborn Unit, Kenyatta National Hospital. All neonates with suspected sepsis based on either perinatal risk factors (preterm labor, premature rupture of membranes >24 hrs, chorioamnionitis, intrapartum fever) and suspicious clinical findings (apnoea, tachypnoea, difficulty in breathing, diarrhea, vomiting, abdominal distension, lethargy, irritability, seizures, tremor, hypotonia, hypertonia, sclerema, petechiae, bradycardia, tachycardia, temperature instability) were eligible for inclusion. Infants with a history of meconium aspiration, tissue injury, perinatal asphyxia, severe hepatocellular involvement and those without informed consent from parent or guardian were excluded. Samples for complete blood counts with differentials, blood cultures and serum C-reactive protein tests were taken before commencement of antibiotic treatment or before change to second line antibiotics. Repeat C-reactive protein tests were done on a sub-group of study patients, especially those showing a poor clinical response to treatment, after an interval of 48 hours. Cerebrospinal fluid specimens were collected as indicated and processed using standard bacteriological techniques. Stool cultures were done in cases of diarrhea. Chest and abdominal Xrays were also performed as indicated. Infants were classified into Proven Sepsis (bacteria isolated from blood, Cerebrospinal fluid), Probable Sepsis (clinical and laboratory findings consistent with bacterial infection but without a positive culture) and No Sepsis (signs/symptoms subsided within 24 hours without specific treatment and no radiological/hematological or microbiological findings attributable to sepsis). C-reactive protein test was not used in the categorization of the babies. Timing of infection was defined as "early-onset" if within 48 hours of life and "late-onset" if after 48 hours. All data from this study was analyzed using SPSS software version 9.0. Results: Of the 310 suspected infants with sepsis, 168 were within the first two days after birth (early-onset) and 142 were late-onset. There were 27 early-onset and 56 lateonset episodes of Proven Sepsis compared to 37 early-onset and 57 late-onset episodes of Probable Sepsis. Using the standard recommended C-reactive protein cut-off value of 5 mg/dl, a sensitivity of 95.2% in Proven Septic episodes and 98.9% for Probable Septic episodes were noted. In Proven Sepsis, a specificity of 85.3%, Positive Predictive Value of 80.6%, and a Negative Predictive Value of 96.5% were noted. In the Probable Sepsis, a specificity of 83.3%, Positive Predictive Value of 80.9% and a Negative Predictive Value of 99.1 % were noted. The overall accuracy in Proven Sepsis was 96.5%, and in Probable Sepsis was noted to be 99.1 %. A further sub-analysis showed a low Positive Predictive Value of less than 68% in early-onset episodes, compared to late-onset episodes where the Positive Predictive Values were more than 93%. Repeat C-reactive protein tests showed a ten-fold increase in serum CRP levels in 22(75.9%) babies with proven/probable infection compared to initial CRP values; but CRP samples were noted to be low or reducing in 7(100%) babies showing signs of improvement clinically. Using a Receiver Operator Characteristic curve, the optimal cut-off point was found to be 5 mg/dl which was in-keeping with the standard recommendation. Conclusions: 1. Serum CRP was an accurate indicator of neonatal sepsis. 2. The sensitivity, specificity, predictive values and overall accuracy were better fulfilled in late-onset episodes, than for early-onset episodes. 3. The standard recommended CRP cut-off of 5 is appropriate for local use. Recommendations: CRP should be routinely used for diagnosis of sepsis using 5as the cut-off point.