Three Tips for Reducing Sepsis Risk with Healthcare IT

There is no better time for the health IT community to come together under one umbrella to raise national awareness of the benefits information technology can bring to the US health system. National Health IT (NHIT) Week is a nationwide awareness week focused on the value of health IT. Each year, NHIT Week Partners educate industry and policy stakeholders on the value of health IT for the US healthcare system. Every Tuesday leading up to NHIT Week, our valued partners will share their voice and experience on how they demonstrate the value of health IT.

Nurse Christine Chan regularly receives clinical advisories from the EHR. They are designed to notify her about steps she needs to take for patients with specific conditions and to highlight potential risks like drug interactions. In addition to the numerous advisories the EHR already delivers, her hospital recently added new notifications to identify patients at risk for sepsis.

Nurse Chan likes the idea of the sepsis risk notifications, but there are two problems: First, nearly half of them are false-positives. Second, they get mixed in with the routine pop-ups she's become accustomed to briefly skimming and clicking through, and that she only sees when she is using the EHR. Before long, she starts to ignore the sepsis notifications altogether. This puts patients at risk.

With the right healthcare technology, Nurse Chan can receive only actionable notifications with patient context, helping reduce interruption fatigue and improve patient safety.

Sepsis Risk Intervention Can Reduce Mortality Rates

Sepsis, according to The Sepsis Alliance, is “the body's overwhelming and life-threatening response to infection that can lead to tissue damage, organ failure, and death.” It is a leading cause of death in the U.S., where 750,000 people die each year from it. 1. With $24B spent annually, it is the costliest medical condition to treat in this country. 2 But strong evidence shows that early, tailored intervention can significantly reduce the likelihood of sepsis-related complications and death. A recent study has shown that deploying a comprehensive sepsis program can reduce mortality rates by more than 50% and 30-day readmissions by 30%. 3 And a growing body of research indicates the most statistically significant variable in sepsis treatment is time.

Provide Actionable Information Through Sensitive, Specific Systems

Many hospitals deploy automated systems as part of a sepsis program to identify patient risk so the appropriate therapies can be initiated quickly. These systems typically use an algorithm that combines patient data from the EHR such as demographics, vital signs, medications and lab values.

As Nurse Chan experienced, many systems are highly sensitive, but they're not specific enough. That is, they may detect common risk factors but don't account for the complex comorbidities that can alter a patient's condition, thereby triggering false positives. For example, factors related to liver disease, heart failure, and chemotherapy treatment can commonly trigger false-positive alerts for sepsis.

Even if a system is sensitive and specific, it might not provide actionable information to the right person, in the right way, quickly enough to initiate the right response. This can lead to ineffective care and preventable patient complications.

The Three-Part Solution: Make Alerts Accurate, Reliable, and Actionable

An effective sepsis surveillance and clinical decision support system needs to be accurate (highly sensitive and specific). It needs to be reliable, reaching the appropriate caregiver with the right information, at the right time, in the right way. And it needs to be actionable, providing clear instructions about what to do next.

1. Make Alerts Accurate: To improve accuracy, the system needs to be highly specific. The algorithm used to detect sepsis, in addition to including the core stats from the EHR, should be designed to include nursing and provider documentation. If the documents are free-texted, natural language processing (NLP) should be used to extract medical conditions. The documentation provides information necessary to understand the unique combination of conditions and other treatments that apply to each patient. Some systems also include a much broader library of physician-authored rules that have been proven to more accurately detect sepsis risk. Administrators can finely tune these algorithms to reflect common attributes for a particular patient population, such as patients in the respiratory unit. Perhaps more importantly, systems should not send alerts when the care team is already aware of the diagnosis and is already following treatment advice. Alerts should only be delivered when they provide new information to the staff, or when appropriate treatment steps have not been completed. Sending only actionable alerts will significantly reduce alarm fatigue.

2. Make Alerts Reliable: Move sepsis alerts out of the routine flow of EHR notifications where they are likely to get lost in the shuffle. Instead, deliver them to caregivers on their mobile devices at the point of care, in the same way you deliver other critical alerts and alarms. Using a common method for presenting and delivering sepsis risk alerts enables caregivers to trust alert accuracy. If you don’t use an alerting solution, you are leaving sepsis detection to chance because caregivers may not check the EHR for long periods of time. Proactive alerts can accelerate the time to respond and initiate therapies, which is of critical importance for deteriorating patients. Build in escalation rules so that if the first recipient of an alert is not available, it is sent to another caregiver who can act upon it. Require the final recipient to acknowledge the alert. This allows certainty that the message was received and creates an audit trail.

3. Make Alerts Actionable: Don't just tell the nurse there's a septic patient in room 101. Provide more detailed information about the level of their condition (such as severe sepsis), why the alert was triggered (for example, hypotension), and what to do next. The protocol at many hospitals is for the alert recipient to contact the patient’s physician to review the factors leading to the sepsis alert and confirm the appropriate order sets. With some systems, once the recipient has accepted the initial alert, a second alert can automatically be sent to another group, such as the rapid response team. This further reduces the time it takes to mobilize the resources needed for patient intervention. If possible, you should also send reminders to notify the care team of any missing steps in the order set.

Vocera partners with providers of sepsis and other clinical surveillance solutions to deliver accurate, reliable and actionable alerts. Deploying the right workflows and technology, driving care team engagement, and managing performance improvement against a defined set of goals are keys to a successful sepsis program. Do it right, and you can significantly improve patient outcomes.


Related links: Rapid Sepsis Response at Halifax Health


Want to learn more about how we can help you improve patient outcomes through faster sepsis risk response? Contact us.



1. Benchmarking the incidence and mortality of severe sepsis in the United States,

2. National Inpatient Hospital Costs: The Most Expensive Conditions by Payer, 2013,

3. Evaluating the impact of a computerized surveillance algorithm and decision support system on sepsis mortality,