Diagnostic Value of Clinical Presentation, Parental Concern, and Clinician’s Non-Analytical Reasoning in Identifying Serious Bacterial Infections in Febrile Children

Fever is one of the main reasons for visits to paediatric emergency departments (ED). Although in most cases the underlying cause is self-limiting viral infections, 4 to 25 % of children visiting ED with fever develop serious bacterial infections (SBI), which are significant causes of childhood mortality, even in developed countries. Due to high number of patients visiting ED with febrile illness, rapid discrimination between children with and without possible SBI is challenging.This study aimed to improve early recognition of SBI in children who present to ED by assessing the diagnostic value of clinical signs at presentation, clinician’s non-analytical reasoning, defined as “gut feeling” of serious illness and “sense of reassurance”, and parental concern of different / more severe illness. Based on these variables, derivation and external validation of two clinical prediction models (CPMs) for SBI was performed, and the performance of a CPM based on clinical variables alone was compared to a model integrating clinical features together with variables of non-analytical reasoning. The models were derived from a dataset of 517 febrile patients presenting to the ED of Children’s Clinical University Hospital (CCUH) in Riga, and externally validated in a dataset of 188 patients prospectively enrolled in six regional hospitals in Latvia.While the prognostic value of clinician’s “gut feeling” as an independent variable for diagnosing SBI was limited, “sense of reassurance” was significantly predictive of absence of SBI, and the performance of the CPM 2 integrating the non-analytical variables with clinical features was superior in both derivation (Receiver Operating Characteristic curve (ROC) Area Under Curve (AUC) 0.783, 95 % confidence interval (CI) 0.727–0.839) and validation cohorts (ROC AUC 0.752, 95 % CI 0.674–0.830), when compared to the performance of the CPM 1, which was based solely on clinical variables (ROC AUC in derivation population 0.738, 95 % CI 0.688–0.788, in validation population 0.677, 95 % CI 0.586–0.767). Both CPMs had moderate ability to predict SBI in febrile children presenting to ED and acceptable performance in the validation cohort. A scoring system based on the superior prediction model was created to distinguish between patients with high or low risk of SBI, as well as to identify patients in diagnostic “grey area”, in which the severity of manifestations of SBI and mild infections overlapped. Contrary to studies in primary care performed in other European countries, parental concern was not significantly predictive of SBI. Elements of fever-related anxiety were identified as factors influencig the level of parental concern and urging parents to present to healthcare early. A qualitative interview study including 34 parents of patients enrolled in derivation cohort revealed existing misconceptions regarding the possible negative effects of fever, which often were a result of unfulfilled educational and emotional needs when caring for a febrile child. This study suggests that educational intervention is necessary to reduce “fever phobia” in parents and to improve the diagnostic reliability of parental concern.