Sepsis is a life-threatening condition that has doubled in incidence over the past decade, and timely aggressive medical intervention has been shown to save lives. Rural sepsis patients have a 38% higher mortality rate, possibly attributable to delays in early sepsis care. Rural emergency department (ED)-based provider-to-provider telemedicine has been proposed to standardize care and support local clinicians in rural hospitals. The goal of this multicenter observational comparative effectiveness study is to measure the association between tele-ED use and clinical outcomes in a cohort of rural sepsis patients.

Sepsis is a life-threatening condition that has doubled in incidence over the past decade, and timely aggressive medical intervention has been shown to save lives. Rural sepsis patients have a 38% higher mortality rate, possibly attributable to delays in early sepsis care. This effect persists even among patients who are transferred between hospitals and who bypass rural hospitals. With 17% of all hospital deaths attributable to sepsis and 19% of Americans living in rural areas, there is a critical need to identify strategies to reduce the disparities in outcomes between rural and urban sepsis care.

Rural ED-based telemedicine has been proposed to standardize care and support local clinicians in rural hospitals. Telemedicine networks provide a real-time, high-definition on-demand video connection between a rural hospital and a tertiary hub 24 h daily. Based in Sioux Falls, South Dakota, Avera eCare is a tele-emergency network that serves as a hub for a 140-hospital network that spans 12 rural Midwestern states. It is the largest rural ED-based telehealth network in North America, and a network the investigators have studied previously.

Our central hypothesis is that telemedicine will improve clinical outcomes through improved adherence with Surviving Sepsis Campaign (SSC) guidelines. Using comparative effectiveness methods and a patient-centered outcomes research (PCOR) approach, this study will test the hypotheses with the following specific aims:

1. To measure the association between rural ED-based telemedicine use, guideline adherence, and clinical outcomes using an observational cohort comparative effectiveness research study. Rural clinicians choose whether individual sepsis patients will be treated with telemedicine-supplemented care. Medical records will be analyzed from patients with severe sepsis who present to 25 rural hospital EDs that are part of a telemedicine network to estimate the effect of telemedicine on changing early SSC guideline adherence. Guideline adherence has been studied extensively as an outcome of sepsis implementation studies. The study will also analyze the impact of telemedicine on clinical outcomes, such as mechanical ventilation, hospital length-of-stay, and survival, using mediation analysis in a propensity-matched cohort design. Our working hypothesis is that telemedicine consultation will improve SSC guideline adherence and will reduce delays in care, leading to improved clinical outcomes.
2. To measure the effect of ED-based telemedicine on guideline adherence among patients who have telemedicine available but not used. In addition to the effect of use for individual patient care, telemedicine interactions may provide ongoing training for providers and nurses and influence care even for patients for whom telemedicine is not used. This effect may result from a learning effect in which local providers adopt practices they observe in telemedicine-consulted patients. The investigators will use hospital fixed-effects models to measure this association. The working hypothesis is that guideline adherence will increase after telemedicine adoption even in non-telemedicine patients, and adherence will be associated with the number of prior telemedicine sepsis encounters (dose-response).

The rationale for this research is that dissemination and implementation of best practices through rural networks remains difficult, but telemedicine offers one potential solution. Sepsis is an ideal model to study the effect of telemedicine because it differs from other acute care conditions treated in rural hospitals (e.g., trauma, myocardial infarction) in that early treatment provided in rural hospitals may be more important than rapid transfer to tertiary centers. Focusing on telemedicine in rural sepsis care will serve as a powerful model for examining strategies for disseminating innovations across rural networks.

Inclusion Criteria:

• Adults (age 18 years or older)
• Arrive at participating emergency department between August 1, 2016 and June 30, 2019
• Hospital diagnosis of infection and organ failure
• Identification of infection in the emergency department
• Presence of organ failure in the emergency department (SOFA score of at least 2)
• Presence of systemic inflammatory response syndrome (SIRS) in the emergency department

Exclusion Criteria:

• None

This study will include all adult (age≥18 years) sepsis patients who presented to a participating rural ED between August 1, 2016 and June 30, 2019. Because of poor sensitivity in diagnosis code-based definitions of sepsis, we elected to use a multi-step definition requiring (1) hospital diagnosis of both infection and organ failure, (2) identification of infection in the ED, (3) presence of organ failure in the ED, and (4) presence of systemic inflammatory response syndrome (SIRS) criteria in the ED. To identify hospital diagnosis of infection and organ failure, we used the Fleischmann-Struzek approximation of sepsis using International Classification of Diseases, 10th edition, Clinical Modification (ICD-10-CM) or an explicit sepsis code (R65.20 or R65.21).

Aggregate data will be shared with investigators who make a written request to the study team. These data will include aggregate effect sizes, but will not be patient-level data. Because of the sparsely populated region where this study is being conducted, hospital-identifiable data would be sufficient to identify individuals in this region, so fully de-identifying this data set would make it unusable for subsequent independent analyses (e.g., would require removing age, sex, hospital, transfer distance, transport times, comorbidities). The study team will, however, collaborate with other investigators to conduct additional analysis, maintaining the security of the data set.