The Use of Electronic Alert System for Early Sepsis Identification

Purpose: Sepsis is one of the leading causes of mortality worldwide.  Sepsis-related hospitalization rates are also on the rise, with impact on hospital length of stay and the utilization of patient care resources. In 2008, the Surviving Sepsis Campaign International Guidelines for Management of Severe Sepsis and Septic Shock (SCC guidelines) were developed to raise sepsis awareness and to increase early and appropriate interventions to decrease mortality rate.  Despite strong evidence of the benefits of early recognition and interventions, dissemination of the guidelines into practice has been inconsistent, incomplete and slow.  Challenges identified are knowledge deficit, failure to recognize sepsis early and timely bundle management.

The primary purpose of the project was to translate the SCC guidelines into practice and to determine the positive predictive value (PPV) of the alert.  The secondary purpose was to determine whether clinicians implemented the three hour bundle interventions as recommended in the SCC guidelines once made aware of possible sepsis.

Scope: A fully integrated healthcare system which consists of two full-service hospitals did not have a standardized program to identify and care for the patients with sepsis, severe sepsis, and septic shock.  Therefore, a pilot project was approved to develop and implement an automatically triggered alert system in the electronic medical record to identify patients before they become septic.  Six different pilot units consisting of emergency centers, rehabilitation and medical surgical units participated in the study. 

Method: The project team consisted of multidisciplinary members who defined the physiologic parameters for the Sepsis Best Practice Alert (BPA) and the lock out time for the alerts.  The team developed clinical algorithms and education modules prior to implementation of the pilot project. All nursing staff received on-line educational modules on sepsis, while physicians were notified through e-mails and at service meetings.  Nurse clinicians collected data manually during the four-week pilot program.   Data were collected to identify the PPV of the alert and the turn-around-time (TAT) for the 3 hour bundles which include the time of: blood culture draws, antibiotic administration, lactate level collection, and administration of fluid resuscitation boluses.

Results:  Results suggest a low PPV on non-Emergency Center (EC) units and all interventions for the 3 hour bundles did not meet the three hours completion goal.  This study demonstrated that the Sepsis BPA, which used only two physiologic parameters, was too sensitive and thus could lead to alarm fatigue. It also confirmed that the presence of the alert did not necessarily improve the providers’ compliance to the SSC guidelines. 

Recommendations:  First, organizational policies should ensure the presence and full support of the Information Technology (IT) department.  The absence of an automated report led to the utilization of manual chart abstraction.  The inter-rater reliability of the auditors could have affected the results of the data collected.  This contributed to the limitation of evaluating the effectiveness of the project. Secondly, the evaluation of both the PPV and the compliance rate to SSC guidelines will need to be refined for more meaningful data.  Future evaluation should include organizational impact as measured by decrease in length of stay and decrease in sepsis mortality rate.

Conclusion: The project provided an opportunity to assess, develop, and implement a pilot study that benefits both patients and organization.  Although the early notification through the Sepsis BPA is beneficial, it lacks structural support for providers to implement the SSC guidelines.  Further discussions are underway to evaluate the possibility of implementing the recommendations prior to a full system roll-out.  The evaluation process lacks rigor, as there was no pre-implementation data available to conduct a pre and post assessment of the interventions.  The lack of IT-generated reports limited capture of process outcomes data only.  The findings of the Sepsis BPA merits further study that can address the barriers identified.