Sepsis prediction and recognition systems: A Making Healthcare Safer IV rapid review

Background The Making Healthcare Safer III report in 2020 found moderate strength of evidence that sepsis prediction tools improve process measures, but the impact on clinical outcomes was unclear. This rapid review aimed to determine the effectiveness of recent sepsis prediction and recognition systems. Methods We searched PubMed and the Cochrane Library for systematic reviews and primary studies published from January 2018 through August 2023. We included reviews and studies of sepsis prediction and recognition interventions reporting measures of clinical process (e.g. timeliness of diagnosis), patient outcomes (e.g. mortality), implementation (e.g. use of system recommendations), or costs. Results We found seven systematic reviews and eight original studies. The studies involved multicomponent interventions in pediatric and adult populations in pre-hospital, emergency department, intensive care unit (ICU), and hospital ward settings. The interventions improved clinical process measures in the neonatal population (low strength of evidence), but not in the adult population. In neonates, the evidence was insufficient about length of stay or mortality outcomes. For adults, interventions did not impact hospital length of stay or mortality outcomes. Sepsis alert systems struggled with poor predictive value, alert overload, delays, mistrust, and software limitations. Such challenges can be overcome with frequent communication, iterative improvement, clinician involvement, and training with test versions. Conclusion Recent studies do not show that sepsis prediction and recognition interventions are effective at reducing mortality or length of stay or improving clinical processes in adults, but may improve clinical process outcomes in neonatal ICUs. Several implementation barriers and facilitators were identified.