Early onset neonatal sepsis currently has an incidence in North America among term infants of about 0.5 to 1 per thousand live births. But 15% of newborn infants have risk factors and they end up receiving antibiotics if current guidelines are followed. The group from Kaiser Permanente have developed an algorithm that could markedly reduce sepsis workups and antibiotic utilisation, they have tested this in their own population and showed that if their calculator was used many fewer babies required sepsis evaluations, and all babies who actually developed sepsis were evaluated and treated quickly.

The calculator they created is now available on-line.To use the calculator you need some data from the mothers course during labour and a physical examination  of the baby.

This new publication (Warren S, et al. Impact of neonatal early-onset sepsis calculator on antibiotic use within two tertiary healthcare centers. J Perinatol. 2016) is from an independent group who looked at all the babies who had a septic workup and antibiotic treatment in two hospitals, of the 205 babies most (92%) required evaluations and antibiotics following the CDC/AAP guidelines, but if the calculator had been used only 23% of them would have got antibiotics. They had no cases of actual culture-positive sepsis, seven patients were defined as having culture negative sepsis, and all were treated according to both CDC guidelines and the sepsis calculator. Although it seems a very reasonable approach, I think that fear of liability will prevent many people from using this calculator until the official guidelines change. They certainly need to change, when only 1 infant benefits from antibiotics for every 150 treated, the other 149 are only having adverse impacts of our current standards, including long-term effects on the intestinal microbiome. The consequences of missing a case, on the other hand, are important, and we need more data about the safety of using the calculator, preferably from a big enough sample to include some positive culture babies.