In this study, the vitals trajectory algorithm will run on all adult patients presenting to the emergency department (ED) across the Emory Healthcare hospitals. Patients will be enrolled and eligible for randomization once they meet suspicion of infection criteria. Suspicion of infection is defined broadly as the ordering of blood cultures in the emergency department. Group D patients with a blood culture order in whom clinicians initiate a normal saline order will be randomized within the electronic health record to either usual care or the intervention arm. In the intervention arm, if a clinician orders normal saline in a patient classified as Group D, there will be an EHR alert to change the order to balanced crystalloids (i.e., Lactated Ringer or Plasma-Lyte solution). The alert will announce that preliminary data suggests a mortality benefit from balanced crystalloids in this patient and will ask the clinician whether there is a strong clinical indication to use normal saline.
The intervention is intended to start as early in the hospitalization as possible in the emergency department, as early crystalloid resuscitation choices may have effects on outcomes. The primary analysis is an intention-to-treat analysis (i.e., all patients randomized to intervention versus usual care regardless of whether the intervention resulted in a change in fluid choice). The intervention will be continued only for patients admitted from the emergency department to the intensive care unit. For patients in the intervention arm who are admitted to the ICU, the alert will fire every time normal saline is ordered till ICU discharge, death, or for up to 30 days of hospitalization. The intervention is necessarily unblinded when a clinician receives an alert when ordering normal saline on study patients.
