Central monitoring of all patients with sepsis helps ensure those patients are getting the timeliest and most appropriate care, saving both lives and costs.
Perhaps equally important, the sepsis surveillance technology used to monitor known sepsis patients is also a critical tool for early identification of and actionable support for patients at risk for the condition.
Avoid false sepsis alerts and delayed treatments
As sepsis coordinators know all too well, early identification of true sepsis has long been a challenge. Traditional EHR sepsis alerts are based on a general one-size-fits-all Systemic Inflammatory Response Syndrome (SIRS) score to flag sepsis. In turn, surveillance of at-risk patients tends to be extremely sensitive but with poor specificity, leading to disturbing numbers of false alerts.
Unfortunately, research has demonstrated that once the false positive alerts rate passes 30 percent, clinicians turn a deaf ear to alerts overall. Other research has found that clinicians ignore EHR safety notifications 49-96 percent of the time and that alert fatigue contributes to clinician burnout, dissatisfaction, and turnover.
To address alert fatigue concerns while simultaneously improving hospitals’ ability to respond to sepsis, advanced sepsis surveillance systems deploy validated artificial intelligence (AI) with natural language processing (NLP). These systems reliably power earlier identification of clinical concerns that can quickly lead to sepsis while dramatically improving alert accuracy. Accurate and timely alerts build clinician trust and, therefore, better compliance with CMS treatment bundles and each hospital’s unique treatment protocols.
How do these systems achieve results that until now have remained elusive? They use NLP to parse through unstructured clinical notes and clinician-informed algorithms to appropriately ignore confounding co-morbidities, conditions, and even medications that can mimic SIRS and early sepsis. Essentially, the system monitors and evaluates patients as clinicians do every day, except the AI-driven surveillance system monitors every relevant patient on an ongoing basis, achieving coverage and incorporating data points that extend well beyond human capacity.
The evidence for the efficacy of such systems is building. A study published in 2020 in the Journal of Patient Safety reviewed six studies that provided evidence that patient monitoring systems reducing sepsis-related mortality, including in one study in which the risk of death was nearly 50 percent lower.
Applying AI to hidden sepsis hotbeds
This monitoring of at-risk patients is especially important in areas of the hospital known as sepsis hotbeds. These are areas where clinicians may not be paying as close attention to early warning signals as they do in the intensive care unit and emergency department (ED), where most hospitals have already implemented effective sepsis detection and treatment programs.
Two other care settings are particularly ripe for sepsis surveillance that reliably captures the early warning signs. The first is patients whose ED course is complete and who have technically been admitted to a ward but are “boarding” in the ED while waiting for an open bed. This group represents about one-third of the cases that present in the ED. And when these patients develop sepsis while waiting for a bed, it’s not uncommon for them to crash within six hours because their diagnosis and treatment are often delayed due to thinly stretched resources and the failure to screen patients for sepsis during this transitional period.
The other area where patients can slip through the cracks is the inpatient medical-surgical ward, which accounts for roughly 25 percent of hospital-developed sepsis. Studies have shown that in these wards, sepsis goes undetected for more extended periods and, once detected, is not always treated with a timely and/or appropriate intervention. This is partly because medical-surgical units do not typically have the staffing for intensive involvement with a single patient and may not have had adequate sepsis awareness and protocols training.
The good news is that better sepsis surveillance in these two key areas is well within reach. Bringing together the aforementioned AI-informed sepsis surveillance technology with readily accessible sepsis-specific order sets, detailed policies and procedures tailored to each of these settings, ongoing analysis of performance metrics, and the sharing of those metrics with all key personnel can yield significant improvements in sepsis-related mortality, readmissions, and length-of-stay.
One example: an article published in the Journal of the American Informatics Association (JAMIA) in January 2017 concluded, “A program consisting of change management and electronic surveillance with highly sensitive and specific rates decision support delivered to the point of care resulted in a significant reduction in deaths from sepsis.” In that study, a surveillance system that achieved 95 percent sensitivity and 82 percent specificity helped reduce sepsis mortality by 53 percent and sepsis-related 30-day readmissions by 30 percent.
Join the battle to fight sepsis
Sepsis continues to confound and frustrate many hospitals and remains one of the world’s most prolific killers. In 2020, a study in the medical journal The Lancet revealed that sepsis accounts for 1 in 5 deaths globally and is the most common cause of deaths in the hospital in the United States.
This must change, and the opportunity exists today to reduce sepsis-related hospital deaths significantly. To learn more about other challenges hospitals face to improve sepsis outcomes and address those challenges, download our guide, titled: Best Practices For Improving Sepsis Care and Outcomes.