The blood level of C-reactive protein (CRP), a marker indicating inflammation in the body, is not accurate enough alone to diagnose late-onset infection in newborn infants.
Late-onset neonatal infection, occurring more than three days after birth, is potentially serious and is relatively common. Tests measuring the blood level of CRP are widely used by physicians to guide their decision on whether or not to start antibiotic treatment for suspected infection.
This NIHR-funded review found 20 studies, including 1,615 hospitalised infants, comparing the use of CRP to bacterial culture of the blood in children with suspected infection. For every 1,000 babies, assuming 40% have an infection, 152 cases of infection would be missed, and 156 would receive unnecessary antibiotics if the decision was just based on CRP.
The findings cast doubt on the use of CRP for this purpose.
Why was this study needed?
Infections in the first three days of life are often associated with maternal infection or complicated childbirth and children at risk are often given antibiotics until given the all clear.
Late-onset neonatal infection is common, especially in children needing intensive care, with prematurity, low birth weight and the need for a central venous catheter. Life-threatening general infection (sepsis) can rapidly develop. Late-onset infection occurs in seven out of every 1,000 births and is responsible for 10% of all newborn deaths in the UK.
Ideally, infections need to be diagnosed early with certainty before antibiotics are prescribed, as widespread use of antibiotics can contribute to antimicrobial resistance. However, early diagnosis can be challenging as bacterial culture typically takes between 24 to 48 hours.
This UK Cochrane review aimed to assess the evidence for the accuracy of C-reactive protein (CRP) measurement when detecting late-onset infection in newborns.
What did this study do?
This systematic review and meta-analysis pooled 20 cohort and cross-sectional studies of 1,615 infants in hospital with suspected infection. The serum CRP threshold level for a ‘positive’ result in the included studies was between 5mg/L and 10mg/L. Diagnosis of infection confirmed using microbial blood culture was the reference standard. Sixteen of the 20 studies were conducted in high-income countries in Europe (16 of the 20). None was based in the UK.
Due to variation between the studies, it was not possible to perform separate analyses according to age, type of infection or infective organism. The risk of bias in the studies was low, and the evidence was of moderate quality so the results should be reliable.
What did it find?
- C-reactive protein levels are only accurate approximately 6 times out of 10 in correctly identifying late-onset infection.
- At the average (median) reported specificity of 0.74, when 74% of newborns without infection are correctly identified as free of infection, the sensitivity was 0.62, meaning that 62% of infections are correctly diagnosed (95% confidence interval 0.65 to 0.84).
- When these values were applied to a hypothetical cohort of 1,000 newborns with suspected late-onset neonatal infection, if the chance of being infected was 40% then 248 babies would be correctly diagnosed (true positives), 152 cases of infection would be missed (false negative) and 156 non-infected newborns would be incorrectly classified as having infection and might receive treatment unnecessarily (false positive). The remaining 444 would test negative and not have a late-onset infection.
What does current guidance say on this issue?
At present, there are no relevant UK guidelines available covering the assessment of late-onset neonatal infection.
The 2012 NICE guideline on neonatal infection (early onset): antibiotics for prevention and treatment is currently under review and will be extended to cover late-onset neonatal infection. The expected publication date is August 2020.
What are the implications?
This review casts doubt on the value of CRP testing for the purpose of identifying late-onset infection in newborns.
Clinicians should be reminded of the high number of false positives (unnecessarily treated) and false negatives (missed cases) and are therefore not advised to use CRP as a rule in or out test. There may be a place of CRP alongside other tests and if used in conjunction with clinical signs and symptoms.
It is possible that newer biomarkers such as procalcitonin could prove to be more reliable and further research is likely in this important area.
Citation and Funding
Brown JVE, Meader N, Cleminson J, McGuire W. C-reactive protein for diagnosing late-onset infection in newborn infants. Cochrane Database Syst Rev. 2019;1:CD012126.
This project was funded by the NIHR Cochrane Programme Grant (16/114/03) and the Vermont Oxford Network, USA.
NICE. Neonatal infection. QS75. London: National Institute for Health and Care Excellence; 2014.
NICE. Neonatal infection (early-onset): antibiotics for prevention and treatment. CG149. London: National Institute for Health and Care Excellence; 2012.
Produced by the University of Southampton and Bazian on behalf of NIHR through the NIHR Dissemination Centre