Evaluation of a Comprehensive Algorithm for PICU Patients With New Fever or Instability: Association of Clinical Decision Support With Testing Practices

Abstract

Objectives: Previously, we implemented a comprehensive decision support tool, a "New Fever Algorithm," to support the evaluation of PICU patients with new fever or instability. This tool was associated with a decline in culture rates without safety concerns. We assessed the impact of the algorithm on testing practices by identifying the proportion of cultures pre- vs. post-implementation that were discordant with algorithm guidance and may have been avoidable.

Design: Retrospective evaluation 12 months pre- vs. post-quality improvement intervention.

Setting: Single-center academic PICU and pediatric cardiac ICU.

Subjects: All admitted patients.

Interventions: Implementing the "New Fever Algorithm" in July 2020.

Measurements and main results: Patient medical records were reviewed to categorize indications for all blood, respiratory, and urine cultures. Among cultures obtained for new fever or new clinical instability, we assessed specific testing patterns that were discordant from the algorithm's guidance such as blood cultures obtained without documented concern for sepsis without initiation of antibiotics, respiratory cultures without respiratory symptoms, urine cultures without a urinalysis or pyuria, and pan-cultures (concurrent blood, respiratory, and urine cultures). Among 2827 cultures, 1950 (69%) were obtained for new fever or instability. The proportion of peripheral blood cultures obtained without clinical concern for sepsis declined from 18.6% to 10.4% ( p < 0.0007). Respiratory cultures without respiratory symptoms declined from 41.5% to 27.4% ( p = 0.01). Urine cultures without a urinalysis did not decline (from 27.6% to 25.1%). Urine cultures without pyuria declined from 83.0% to 73.7% ( p = 0.04). Pan-cultures declined from 22.4% to 10.6% ( p < 0.0001). Overall, algorithm-discordant testing declined from 39% to 30% ( p < 0.0001).

Conclusions: The majority of cultures obtained were for new fever or instability and introduction of the "New Fever Algorithm" was associated with reductions in algorithm-discordant testing practices and pan-cultures. There remain opportunities for improvement and additional strategies are warranted to optimize testing practices for in this complex patient population.