Electronic Application of a Severe Sepsis Screening Tool and Management Bundle (eASSIST-M)

Project: Application of a Severe Sepsis Electronic Health Record Integrated Screening Tool and Management Bundle (eASSIST-M) Study

Study Type: Prospective Cohort Outcomes Study

Background: Early recognition of severe sepsis is critical for the institution of goal- directed therapy and for improving patient outcomes. Barriers to early recognition include the lack of standardized tools to identify children with severe sepsis. The investigators will study the potential impact of applying a novel Pediatric Severe Sepsis Screening Tool (PSSST) integrated with the Electronic Health Record (EHR) to facilitate earlier detection and effective management of severe sepsis.

Population: Patients between the ages 1 month and 18 years admitted to the hospital or presenting to the Emergency Department (ED) with clinical signs of Systemic Inflammatory Response Syndrome (SIRS).

Primary Hypothesis: Prospective application of a PSSST electronically integrated with the EHR can accurately diagnose pediatric patients with early signs of severe sepsis.

Primary Outcome Measure: Proportion of correctly diagnosed patients with severe sepsis among those presenting with SIRS will be assessed to prospectively validate the sensitivity, specificity, positive and negative predictive values of the PSSST tool.

Secondary Hypothesis 1: Application of the PSSST will reduce lag-time for the administration of a goal-directed sepsis therapy bundle by 50%.

Outcome Measure: Time delay from diagnostic identification to critical therapeutic intervention. Sepsis Recognition Lag Time will be defined as the time elapsed from the first electronic alert to the first therapeutic intervention. Therapeutic interventions will be defined as Cardiopulmonary Resuscitation (CPR), fluid resuscitation, initiation of cardiotonic agents, or request for transfer to the ICU. Since this is an intent-to-diagnose study, outcomes defined above will be compared with those for the same period in the preceding year.

Secondary Hypothesis 2: Application of a severe sepsis management bundle will reduce mortality and morbidity.

Outcome Measure: Proportion of deaths in the study population due to severe sepsis. Morbidity will be defined as the number of ventilator days, days on vasopressor or inotropic agents, days of Extracorporeal Membrane Oxygenation (ECMO) support, need for tracheostomy, need for Gastrostomy tube.

Proposed Study Design:

The investigators will conduct a prospective study where patients within the ED and inpatient units are electronically screened using the PSSST during a 3-year study period. Patients will be screened using a novel electronic tracking tool designed locally, using pre-defined severe sepsis variables and validated on patients with severe sepsis. Patients screened as positive for severe sepsis using the electronic tracking tool will be treated prospectively with a standardized severe sepsis management bundle. Data will be collected prospectively on all patients admitted to the ED and inpatient units. For the ED and each inpatient unit, the outcomes defined above will be compared with those for the same period in the preceding year.

Covariates: Demographic, clinical characteristics and sepsis-specific factors that could potentially influence the effect of the alerts on critical intervention.

Statistical Analysis:

To assess the effectiveness of this tool, we will compare the proportions of the population with each variable of interest between the intervention and prior periods.

1. To assess the efficacy of the PSSST, we will compare the changes over time in the proportion of patients diagnosed severe sepsis adjusting for key covariates of interest.
2. To assess the effectiveness of the PSSST in reducing treatment delays, we will compare changes in the Sepsis Recognition Lag Time between the intervention and control periods, while adjusting for key covariates of interest.
3. To assess the efficacy of the PSSST in reducing mortality, we will compare the changes in the age and risk adjusted mortality rates over time between the intervention and control periods.

The investigators will use a robust variance estimate for all analyses to account for the clustering of patients within units. We will also use propensity scores to control for differences in patient characteristics and diagnostic categories.