O0491 Oct.4: 4:56pm - Implementation of an Automated Sepsis Screening Tool: A Cost Analysis

Background: Implementation of screening tools for sepsis has improved outcomes, but the cost impact is unknown.

Objective: To determine the difference in median medical costs per comparable affected patient encounter before and after implementation of our institution's automated interruptive sepsis screening tool.

Methods: We conducted a retrospective cost comparison of patient encounters in a large, freestanding pediatric emergency department (ED) before and after implementing an automated interruptive alert for sepsis screening. We identified two initial cohorts of encounters, from which we derived propensity matched comparison groups. One cohort included all encounters that triggered the screening alert in 2018. The other cohort included all encounters in 2012 that would have triggered the screening alert had it been active at the time, based on documented vital signs and physical assessment. The two cohorts were not directly comparable because the full automated screening process used in 2018 utilized multiple exclusions for minor complaints and relied on current data in the problem list and past medical history, which could not be applied retrospectively to the 2012 cohort. As a result, we performed propensity-score matching based on a sepsis score derived from triage data and comorbidities identified from the hospital’s Pediatric Health Information System (PHIS) data, creating two comparison groups with similar acuity and comorbidities. For each matched group, we determined the median total medical cost for patient encounters using PHIS cost data and adjusted for inflation using the Consumer Price Index for medical care.

Results: In 2018, 1680 patient encounters triggered the automated interruptive alert for sepsis screening, and 161 (9.5%) of these had sepsis ICD codes. Based solely on vital signs and physical assessment and without exclusions for minor complaints, 3518 patient encounters in 2012 would have triggered our institution's sepsis screening; 114 (3.2%) of these had sepsis ICD codes. There were 1416 encounters in each of the propensity matched comparison groups. Seventy patient encounters (4.9%) in the 2012 group and 132 (9.3%) in the 2018 group had sepsis ICD codes. The median cost of care per encounter decreased by 21.2%, from US$6,454 (interquartile range [IQR] $968-$21,697) in 2012 to $5,084 (IQR $802-$16,618) in 2018. The median cost per encounter for those with a sepsis ICD codes decreased by 51.1%, from $58,685 (IQR $$32,224-$134,895) in 2012 to $28,672 (IQR $16,796-$60,657) in 2018.

Conclusion: We demonstrated a reduction in medical costs per comparable patient encounter with implementation of an automated interruptive alert for sepsis screening which was more substantial for children diagnosed with sepsis. As the use of automated interruptive alerts becomes more common, evaluations of their effect on medical costs are warranted.