Clinical Decision Support for Early Recognition of Sepsis

This is a review of a 2014 article by Amland, R.C. and Hahn-Cover, K.E. entitled; Creating Clinical Decision Support for Early Recognition of Sepsis

Introduction

According to the authors, clinicians frequently have a difficult time in distinguishing the inflamatory response of sepsis from other concurrent infectious disease processes the patient may be experiencing. When an inflamatory response to an infection is uncontrolled it is termed sepsis and can lead to a systemic inflamatory response syndrome (SIRS). Failure to detect and treat this early can lead to patient death. A multicenter retrospective study - using data from 6200 adult hospitalizations spanning 2012 through 2013- was carried out to determine if implementation of computerized early warning tools for sepsis would improve patient outcomes by early detection and treatment of the inflamatory response. They employed a cloud-based computerized sepsis clinical decision support system (CDS) for this purpose. The authors also sought to understand the epidemiology of sepsis, and identify opportunities for quality improvement.

Methodology

This multicenter retrospective cohort study was performed at 5 different medical centers in 4 distinct geographic regions across the United States and included a mixture of trauma, community and specialty hospitals. Each of the centers had an enterprise EHR system. Objective diagnostic tests for infection were performed on the patient cohort. The cloud-based [[1]]Sepsis CDS computerized system ran in real time and was able to differentiate patients with SIRS or severe SIRS by analyzing criteria documented and then abstracted from the EHR. The cloud-based sepsis CDS had the ability to run in live surveillance mode (notification delivered directly to provider) or in silent mode (cases flagged for manual validation of alerts received). Surveillance mode steps include:

1. Patient’s initial clinical findings documented in the EHR system
2. SIRS and severe SIRS criteria trigger different CDS definitions.
3. An alert is activated when there is a match in clinical criteria for sepsis
4. Time criteria captured.
5. Notification delivered to providers for action.

For this quality improvement study, the Sepsis CDS ran in silent mode with encounters flagged but without provider notification. EHR charts for patients flagged by the sepsis CDS were examined manually to understand the relationship and timing between alerts and clinical indications of SIRS.

Results

• 13% (817) of 6200 hospitalizations examined generated an alert from the cloud-based sepsis CDS.
• 1.5% (80 patients) of the 5383 patients without an alert were diagnosed as having sepsis and as a result false negative in the CDS.
• In 25% of patient with alerts there was variation in the causative organism of the sepsis according to geographical location.
• ~ 60% of patients had a first activated alert (the most important alert) of SIRS versus severe SIRS alert within a median arrival time at the hospital of 3.5 hours.
• The CDS was able to identify and provide alerts for patients where SIRS or sepsis was not even suspected or recognized by the physician.

Discussion

The authors note a number of limitations with the study such as the fact that it was a multicenter study with geographical and or hospital location variations in SIRS definitions and distribution of infectious organisms associated with sepsis. The CDS was able to effectively identify several different sepsis patterns in the cohort studied.

Conclusion

The study provides an in-depth look into the use of a cloud based CDS system that is able to abstract / analyze pertinent clinical data from different EHR systems for sepsis and generate an applicable alert to the provider. Time is of the essence in order to prevent sepsis from progressing to a point where treatment is non-effective adversely impacting patient outcomes. The CDS described here is able to generate the first alert in about 3.5hours of presentation and appears dependent on the length of time providers take to complete the sepsis diagnostic assessments. Review of metrics about the time to generate an alert versus the time stamps from the diagnostic assessments may help to inform process improvement initiatives in centers where there are workflow related delays in diagnostic testing and provision of results. The Volume of patients seen by the center may also impact the ability to perform these tests in a timely fashion. Information security and confidentiality may be potential issues to consider when using a cloud based system to trigger CDS alerts to multiple facilities.

Related Articles

Validation of a Screening Tool for the Early Identification of Sepsis

Automated electronic medical record sepsis detection in the emergency department