Acad Emerg Med. 2021 Apr 19. doi: 10.1111/acem.14265. Online ahead of print.
ABSTRACT
OBJECTIVES: Effective sepsis resuscitation depends on useful criteria for prompt identification of eligible patients. These criteria should reliably predict a discharge diagnosis of sepsis, ensuring that interventions are triggered for those who need it while avoiding potentially harmful interventions in those who do not. We sought to determine the proportion of patients meeting sepsis criteria in the emergency department (ED) that was ultimately diagnosed with sepsis, and to quantify the subset of non-septic patients with risk factors for harm from fluid resuscitation.
METHODS: This retrospective cohort study of adult ED patients at a tertiary academic medical center included vital signs and laboratory results from the first six hours, plus administration of intravenous antibiotics, to determine if patients met 2016 Sepsis-3 consensus criteria. If these patients also had hypotension and lactic acidosis, we categorized them as Sepsis-3 plus shock. We used discharge ICD-9 codes to determine if patients were ultimately diagnosed with sepsis.
RESULTS: Over eight years, 3,121 ED patients met 2016 Sepsis-3 criteria in the first six hours. Of these, only 25% and 48% met explicit and implicit criteria for a discharge diagnosis of sepsis. Of 1,032 patients with Sepsis-3 plus shock, 48% and 62% met explicit and implicit criteria. Overall, 60-75% of ED patients meeting Sepsis-3 criteria +/- shock did not receive a sepsis discharge diagnosis. At least one risk factor for harm from large-volume fluid resuscitation was identified among 19-36% of patients meeting sepsis criteria in the ED but not ultimately diagnosed with sepsis at discharge.
CONCLUSIONS: Most patients meeting sepsis criteria in the ED were not diagnosed with sepsis at discharge. Urgent treatment bundles triggered by consensus criteria in the early phase of ED care may be administered to several patients without sepsis, potentially exposing some to interventions of uncertain benefit and possible harm.
PMID:33872430 | DOI:10.1111/acem.14265