EMR Sepsis Surveillance – Achieving Optimal Sepsis Sensitivity & Specificity

EMR Sepsis Surveillance – Achieving Optimal Sepsis Sensitivity & Specificity

This article is a continuation of our sepsis surveillance series. Read the first article in the series, “Say Goodbye to HAI’s: Reducing Hospital Acquired Infections,” to learn about sepsis surveillance in MEDITECH.

Sepsis Kills More People Than AIDS, Breast Cancer, & Stroke – Combined

Sepsis accounts for half of hospital deaths, according to research published in the Journal of the American Medical Association. It’s the leading cause of readmissions and more than $20B is spent on it annually.  According to the Centers for Disease Control and Prevention (CDC), sepsis is a medical emergency. Every year there are at least 1.7 million cases of sepsis and 270,000 deaths, the CDC says.

Sepsis Hospital Acquired Condition Infographic

Sepsis starts with any kind of infection or any time germs enter the bloodstream. The human body tries to fight the infection, but fails. The organs shut down, one by one. But what’s unique about sepsis, compared to cancer, for instance, is that there is a known cure—usually antibiotics and fluids are suffice. Therein lies the challenge in the medical community, though, since the issue is about detection rather than treatment.

Sepsis Detection & Prevention Saves Lives

About 69 percent of the 95 providers polled by KLAS say hospital surveillance technology for sepsis detection have led to improved patient safety outcomes, with some reporting a 50 percent drop in mortality. Other benefits included shorter lengths of stay, fewer readmissions, and lower treatment costs. Best practices include setting up an order set for each detection of sepsis, so that once patients are identified, they start treatment immediately.  This is critical, as studies have found that there is a 7.6% drop in chance of survival each hour until antimicrobials are begun.

Sepsis Detection Approaches

EMR Sepsis Detection

Sepsis is a rapidly progressing disease with higher costs and worse outcomes as the disease advances along the sepsis spectrum. A major challenge in the early identification of sepsis is that the symptoms associated with sepsis are the same as many other conditions and diseases. This makes it particularly difficult to distinguish septic patients from other high-acuity patients in the hospital setting. For the purposes of clinical identification, most sepsis detection solutions start with determining if a patient meets systemic inflammatory response syndrome (SIRS) criteria:

Table 1 – Clinical Elements of SIRS

Temp >38°C (100.4°F) or < 36°C (96.8°F)
Heart rate > 90
Respiration rate > 20 or PaCO₂ < 32 mm Hg
WBC > 12,000/mm³, < 4,000/mm³, or > 10% bands

Other detection algorithms in use include q SOFA (quick Sequential [sepsis-related] Organ Failure Assessment),  Modified Early Warning Score (MEWS),  Cerner’s St John Sepsis Agent, and the Rothman Index (RI) The latter is often viewed as superioer, as it includes all the clinical elements used in both SIRS and qSOFA, but also incorporate additional vital signs and labs, as well as a full range of body-system nursing assessments which are known to be leading indicators of deterioration.

Table 2 – Comparison of Sepsis Detection Tools

False Positives # Inputs Nursing Burden Frequency of Calculation Real-time Alerts on Score Value & Trend
Rothman Index Low 26 None continuous Yes Yes
SIRS High 4 High per-shift No No
qSOFA High 3 High per-shift No No
Cerner St. John High 5-9 Medium varies No No
MEWS High 5 High 3-5x daily No No

Paramount to a successful sepsis detection solution is performance with regard to sensitivity and specificity. Sensitivity is the ability of a test to recognize true positives, while specificity measures number of true negatives correctly identified. In a screening test for a potentially life-threatening disease, such as severe sepsis, high sensitivity would be valued over high specificity so that cases are not missed.

Additional Effort Required to Implement & Integrate Sepsis Surveillance into EMR Workflows; Alert Fatigue an Issue

The majority of healthcare delivery organizations (HDOs) look to leverage sepsis detection capabilities native to their EMR. Depending on the vendor, sepsis detection manifests itself within the EMR through specific modules, frameworks or tools, notification and alerting, surveillance dashboards, and in some cases, custom solutions through extensibility offered by the vendor. That said, in some cases, users report the need for significant in-house effort for implementation and integration into workflows. This according to a KLAS report from last year,  “Sepsis 2017: Which Vendors Can Help?,” which identified opportunity for EMR users to focus on sepsis prevention through adoption of internal or external solutions. In addition, EMR users report alert fatigue with the system, potentially due to sensitivity issues.

End User Perspective

We’re still in the process of pre-deployment with Wolters Kluwer POC Advisor. The issue with sepsis is it’s a sensitivity and specificity challenge. You can set up an alert that monitors blood pressure, heart rate, temperature, and things like that, but the sensitivity can be very poor, so alerts are firing way too often. As much as we tried to refine alerts on our own, we were still only having a sensitivity and specificity with our sepsis alerts in our EMR in the upper 60s. So, as you can imagine, a little over 30% of the time, when the alert fires, it’s not sepsis, and a little over 30% of the time, when there is sepsis, the alerts not picking it up. What we really liked is some of the early published data around the Wolters Kluwer POC Advisor where they’re in the low 90s, both sensitivity and specificity, so alert fatigue is reduced and you’re picking up more sepsis cases earlier. We’re combing that with a technology from a company called Hiteks that has a NLP engine where they can pull non-structured data out of the EHR to also feed the POC Advisor engine, so we fully expect it will be much more accurate.

“For decades, the medical community has struggled to improve sepsis outcomes, in part because the condition’s initial symptoms mimic several common illnesses, but also because today’s EHRs cannot support the enterprise-level surveillance required to impact sepsis rates. POC Advisor will help the Rush team overcome the challenges inherent in early identification, enabling timelier and more accurate diagnoses for faster and more effective treatments. This will, in turn, reduce the severity of sepsis and the number of associated deaths.”

-Brian D Patty, VP, CMIO, Clinical Information Systems, Rush University Medical Center, via CHIME CIO Interview Series & Wolters Kluwer Press Release

Improving Sepsis Detection, Prevention & Surveillance

According to a study published in the Public Library of Science, “Creating an Automated Trigger for Sepsis Clinical Decision Support at Emergency Department Triage Using Machine Learning,” utilizing free text drastically improves the discriminatory ability of identifying infection when compared to using only structured data such as vital signs and demographic information. Further, the results of a study, “Effect of a Machine Learning-Based Severe Sepsis Prediction Algorithm on Patient Survival and Hospital Length of Stay,” as published in the British Medical Journal, demonstrated average length of stay decrease from 13.0 days in the control to 10.3 days in the experimental group, with in-hospital mortality decreasing 12.4 percentage points when using machine learning algorithm. This study was the first randomized, controlled trial of a sepsis surveillance system to demonstrate statistically significant differences in length of stay and in-hospital mortality.

Accurate triggering of clinical decision support will become increasingly important as clinical decision support becomes more integrated into EMRs. Since decision support has the potential to interrupt the clinical workflow, every attempt should be made to ensure that all eligible patients receive the decision support (sensitivity), and that non-eligible patients are not mistakenly targeted (specificity) leading to alert fatigue.

Please contact us to receive more resources regarding sepsis detection, prevention & surveillance or for help in implementing within your EMR platform.

Facebook Twitter Email

+ There are no comments

Add yours

This site uses Akismet to reduce spam. Learn how your comment data is processed.