Abstract
Hospital Paediatrics
Procalcitonin (PCT), a serum inflammatory biomarker, has recently been incorporated into several clinical decision tools to identify febrile infants at low risk for serious bacterial infection (SBI). These include the Pediatric Emergency Care Applied Research Network (PECARN) tool, the “Step-by-Step” approach, and the “Laboratory-Score.” Our institution is one of a few in Canada to incorporate serum PCT routinely, allowing us to complete these clinical decision tools. Thus, the objectives of this study were to externally validate and compare these tools in a Canadian pediatric population, indirectly assessing the utility of serum PCT in clinical practice.
The primary outcomes were to derive the sensitivity, specificity, and negative predictive value (NPV) of each stratification tool in predicting SBI.
We retrospectively reviewed the medical records of all infants less than 90 days of age presenting to our emergency departments between April 2016 and October 2019 with fever without a source, who had sufficient investigations to apply one (or more) of the above clinical decision tools.
We applied the PECARN tool to 51 cases, and had sufficient data to apply the Step-by-Step and Lab Score criteria to 43 of these patients. Seventeen of the 51 patients (33%) were identified to have a SBI. The PECARN and Step-by-Step tools both had NPV of 100%; both were sensitive enough to detect all patients with SBI. They had poor specificity (0.47 and 0.55 respectively). These two tools were in agreement in 38 of 43 (88%) cases. Though the Laboratory-Score had the highest positive predictive value (0.88) and specificity (0.85), it failed to identify 3 of 16 true cases of SBI and had a suboptimal sensitivity of 0.81.
The ability to identify febrile infants at low risk for SBI in a reliable way would have significant clinical potential to change practice. Given the strong NPV of both the PECARN and Step-by-Step tools, we conclude that their use, incorporating the measurement of serum PCT, may be of use in reducing pediatric hospitalization, use of empiric broad-spectrum antibiotics, and investigations such as lumbar punctures, in these low-risk patients. This study had a small sample size. We look forward to analyzing a larger population of febrile infants, particularly in infants of a chronologic age (28-90 days) more amenable to clinical practice change.
