Cabell Huntington Hospital Case Study | Dascena

Facilitating septic patient care by reducing alarm fatigue

Learn how a machine learning system helped improve outcomes through earlier notification and higher accuracy than a native EHR rules-based system.

Cabell Huntington Hospital Outcomes


reduction in mortality


reduction in length of stay

Sepsis Care Improvement at Cabell Huntington Hospital

Cabell Huntington Hospital (CHH) is a regional referral center in Huntington, West Virginia, and it is also home to the Hoops Family Children’s Hospital and the Edwards Comprehensive Cancer Center. CHH cares for patients from more than twenty-six counties in West Virginia, eastern Kentucky and southern Ohio. In August 2017, CHH adopted InSight in the emergency department and the medical, surgical, and burn intensive care units.


Choosing a Sepsis Surveillance System

Prior to and during the initial InSight implementation phase, CHH clinicians relied on modified Systemic Inflammatory Response Syndrome (SIRS) alerts supplied by the St. John Sepsis Agent in the Cerner EHR. Patients triggered these alerts when they met a specified number of SIRS criteria and/or possessed abnormal lab values indicative of organ dysfunction. Although this rules-based system captured all patients with out-of-range vital signs and lab results, it alerted too frequently and often produced false alerts, as many patients suffered other conditions with similar symptoms. Clinicians became increasingly indifferent to these noisy alerts.

CHH decided to introduce InSight to the bedside to deliver more accurate sepsis warnings. To this end, CHH compared InSight’s performance with that of the pre-existing sepsis detection system. On average, InSight alerted about three times more accurately, ten times less frequently, and two hours earlier than the pre-existing system. These metrics suggest that InSight’s heightened alert specificity and more optimal alert timing may have promoted a greater clinician response rate in the critical early intervention period.



In the ten-month period following InSight deployment, CHH observed a 48.8% reduction in average sepsis-related length of stay, which fell from 5.12 days (pre-implementation) to 2.62 days (post-implementation). In parallel, CHH observed a 39.3% reduction in sepsis-related in-hospital mortality, which fell from 2.42% to 1.47%. Baseline outcomes calculated for the pre-implementation period (May 1, 2017 to July 31, 2017) were compared against InSight outcomes calculated for the implementation period (August 1, 2017 to May 1, 2018). The inclusion criteria limited “sepsis-related” analyses to those performed on adult patients meeting at least two SIRS criteria during their stay in the units where InSight was deployed (n = 8,161).


Interested in learning more?