Journal Logo

Department: Innovations

Implementing real-time sepsis alerts using middleware and smartphone technology

Zimmermann, Melissa MS, BSN, RN-BC; Chung, You “Jay” MSN, RN-BC, CCRN-K; Fleming, Cara AGPCNP-BC, AOCNP; Garcia, Jericho MSN, RN-BC; Tayban, Yekaterina ACNP-BC, GNP-BC; Alvarez, Hector De Jesus; Connor, MaryAnn MSN, RN-BC, CPHIMS

Author Information
doi: 10.1097/01.CCN.0000654832.34404.99
  • Free

As clinical practice continues to evolve and improve, technology has become increasingly integrated into everyday clinical workflow. From alarms and alerts to point-of-care electronic clinical communication tools, the future of healthcare depends on the ability to implement technology to improve quality and safety of patient care.1 Adoption of electronic health records (EHRs) has been found to improve clinicians' performance by providing access to aggregated patient information such as lab results, nursing notes, and alerts.2 The use of decision support system technology, along with clinical reasoning, can help decrease errors and avoid delays in treatment.3 Real-time alerts and data to and from mobile platforms can lead to early diagnosis and detection, with the opportunity to improve quality of life and reduce healthcare costs.4

However, caution must be taken with these implementations to evaluate and address alert fatigue (often called alarm fatigue) and ensure alerts are clinically actionable and relevant.5 Development of alert algorithms through an EHR should help provide clinical decision support by supplying relevant information, at the time it is needed, to the correct clinician.6 An example of this includes the high-risk scenario of sepsis identification, with alerts triggered from the EHR.

Research has shown that sepsis alerts can help improve patient outcomes by assisting with early detection.7,8 A study by Dziadzko and colleagues compared the emergence of smartphones for sepsis alerts to EHR-based notifications and pagers to determine the best method of notification delivery.6 Due to technologic failures and barriers, sepsis smartphone alerts were unsuccessful, and clinicians continued to use pagers and EHR-based alerts.6 Continued research and development were identified as needs to better evaluate the efficacy of smartphone alerts in a clinical setting.6

Since May 2013, the New York State Department of Health (NYSDOH) has regulated that hospitals maintain sepsis protocols that use explicit algorithms and/or alert systems to assist in the early identification of patients with severe sepsis and septic shock.9,10 However, in January 2017, regulatory changes necessitated real-time, prospective identification of sepsis and standardized clinician documentation.11 This documentation was needed to record an initial assessment, and recognition of sepsis signs and symptoms as well as verifying reassessment of the patient's sepsis signs and symptoms within 6 hours of management.

At Memorial Sloan Kettering Cancer Center (MSKCC), the alert algorithm was initially set so that patients with three simultaneous abnormal vital signs, new (in a 24-hour period) altered mental status, or rigors in the presence of two abnormal vital signs triggered an alert. MSKCC has a unique population of oncology patients, and patient signs and symptoms or treatment adverse reactions could often be similar to sepsis indicators. Sepsis alerts were put in place to help with management of their complex healthcare needs and assist in identifying possible early sepsis. The nurse or patient-care technician, depending on who documented the vital signs/altered mental status, was then alerted via an instantaneous pop-up in the EHR to contact the responsible clinician to screen the patient for possible sepsis. This led to delays in clinician awareness of possible sepsis; among them was that the algorithm was designed based on nursing documentation workflows with limiting factors such as timing and availability of the documentation, as the alert was triggered by saving of the assessment or vital signs.

Critical care medicine leadership proposed a real-time clinical sepsis team composed of advanced practice providers (APPs), forming a Sepsis APP Team. Their role was to provide timely triage, medical management, and documentation for patients with sepsis. To be successful, this required a more direct delivery of sepsis alerts on hospitalized patients to the newly formed Sepsis APP Team, rather than wait for a busy nurse to page with abnormal findings.

Recognizing that the current process can delay activation of the Sepsis APP Team, the organization took advantage of the EHR functionality to send real-time email alerts to the Sepsis APP Team instead of waiting for the nurse to page the clinician when they receive an alert. However, during evaluation of the change it was identified that this still did not provide an urgent and apparent enough notification to the mobile clinical staff. A collaboration between critical care medicine and the nursing informatics department was initiated to develop a real-time alert to the Sepsis APP Team on a mobile device. This article focuses on this institution's successful implementation of sepsis alerts to a smartphone-based application for both alerts and secure clinical communication.

Methods

In 2015 a robust alert management program was put in place for secondary alerting of telemetry alerts to smartphones. This replaced an outdated pager-based system with new integrated technology.12 Special considerations were taken to evaluate alert fatigue, which was identified as one of the top health hazards by the ECRI Institute from 2012 to 2016.13-16 Using a Plan-Do-Study-Act methodology and collaboration between clinicians, vendors, nursing informatics, and technical resources led to an innovative use of smartphones. This was implemented using a middleware system, waveform smartphone application, and secured communication smartphone application to integrate with the current telemetry monitoring system in place at the hospital.12 A middleware system is software that connects different systems so that they can communicate with each other and exchange data.12 In this case, the middleware connected the telemetry system with the secure communication application and waveform application on the smartphone to send critical alerts for arrhythmias. The waveform application allowed clinicians to view patients' telemetry waveforms on the smartphone in real time.12 Best practices were put in place by establishing an Alarm Management Committee to evaluate the efficacy of both alarms and alerts, and working with end users for continuous improvement of secondary alerting and reduction of alert fatigue.12

To respond to the need for real-time sepsis alerts, the existing infrastructure for alerting using the middleware software and secure smartphone application was leveraged. A project team was created consisting of clinicians, nursing, nursing informatics, and information systems to plan for the configuration of automated smartphone sepsis alerts to support the Sepsis APP Team. This team worked together to capture requirements needed for the alert and develop a new workflow to support the receipt and response to that alert. Goals of the project included proper identification of roles and responsibilities of the Sepsis APP Team, creating an interface to receive sepsis alerts via a smartphone, developing a process to automatically assign sepsis alerts to the correct staff members, and providing sepsis expert clinicians with a real-time notification.

Figure
Figure:
Sepsis alert workflow

As EHR sepsis alerts were already in place and being routed to clinicians via the EHR and email notifications, the nursing informatics team worked with information systems to evaluate routing of the email sepsis alerts to the secured communication and alert-receiving smartphone application. Configuration was done in the current EHR sepsis alert scripting and the middleware system to accept the email alerts and parse the information into alert data that can be received by a smartphone. A new workflow was then established, and the prior workflow was left in place only for downtime procedures. (See Sepsis alert workflow.) Sepsis APPs logged into the smartphone application and assigned themselves to an additional role to receive sepsis alerts. Instead of relying on nursing staff to notify them of sepsis alerts or poor processing of email alerts, this supported a workflow to intuitively self-assign receiving sepsis alerts and allowed the Sepsis APPs to be notified in real time once the alert was triggered in the EHR. This enabled assessment and treatment of patients by the Sepsis APP at the bedside along with the primary team to screen, treat, and place sepsis order sets if appropriate.

Figure
Figure:
Compliance with sepsis bundle components pre/post intervention

Implementation

Before piloting, the APPs received sepsis bundle clinical and information technology training. Having a clinical champion for the project to help coordinate education and communicate the new workflow was integral to the success of the project. Multiple training sessions were held in the ICU by nursing informatics to train on the new sepsis alerts, hardware, and workflow considerations. Demonstration of the alerts was provided for hands-on training.

For hardware used, the Sepsis APP Team had the choice of using the shared hospital-owned MC40 Android devices, or to pilot alerts on their personal iOS/Android smartphones. MC40 Android devices were previously identified as the most reliable choice for receiving critical alerts based on the alert management program. The secure messaging smartphone application had three different modalities of software to be used on devices: an application for shared hospital-owned devices, an application for personal phones, and a web-based application.

Outcomes

Prior to the November 1, 2017, implementation of smartphone sepsis alerts, data showed compliance with NYSDOH components and comparison of pre- and postdata. (See Compliance with sepsis bundle components pre/post intervention.) Reports run in our middleware system since the go-live on November 1, 2017, show a 96% compliance with acknowledgment of alerts on the smartphones. While this shows improvement with compliance of the sepsis protocols to meet state requirements, the organization has not seen significant improvement in mortality and morbidity of patient outcomes related to the implementation of real-time smartphone sepsis alerts at this time.

In 2018, the Sepsis APP Team responded to 1,900 alerts; however, low specificity was noted. Only 27% of patients discharged with ICD-10 of severe sepsis or septic shock triggered an alert. In addition, of the 546 patient records submitted to NYSDOH with an ICD-10 of severe sepsis/septic shock, only a little over 50% triggered the automated alert. (See Comparison of sepsis alerts/no alerts to cases reported to NYSDOH.) Review of alert orders in comparison with sepsis alerts revealed no difference in mortality or initiation of treatment at time zero. (See Review of mortality comparison between alert orders.) Time zero is defined as the time of onset of sepsis signs and symptoms, which includes documented infection, new organ dysfunction, and two systemic inflammatory response syndrome (SIRS) criteria within 6 hours.17 This could be related to the overall complexity of the unique patient population of all oncology patients, and further studies are needed to evaluate the impact of real-time smartphone alerts in a clinical setting versus improvement of the alert system.

Figure
Figure:
Comparison of sepsis alerts/no alerts to cases reported to NYSDOH
Table
Table:
Review of mortality comparison between alert orders

Discussion

The success of the project was a result of multiple variables including collaboration between multiple disciplines, the willingness of Sepsis APPs to pilot the use of new alert-receiving technologies to their personal phones, and their vision to create a workforce to oversee the management of sepsis across the inpatient setting. The existing infrastructure and expertise from technical teams, including nursing informatics and information systems, were also key to the successful integration and delivery of smartphone sepsis alerts. Project planning included alert integration, testing, training, and clinical workflow analysis as well as decisions on how to send and resend alerts based on receiving and acknowledging alert information and escalating alerts when staff were unable to receive sepsis alerts.

The existing middleware system infrastructure allowed for adding a specific integration to receive real-time notification for sepsis alerts and allowed us to resource existing support teams to build and provide on-call support for alerting systems. Extensive alert testing was performed that included scenario-based testing for sepsis algorithms, testing of networking and operating software considerations, smartphone push notification testing to confirm reliability, and safe use of smartphones to receive sepsis alerts. There was a noted willingness from many team members to participate using their own personal phones to receive these alerts, which included staff volunteering to download security software and a clinical application used for secured communication and receiving alerts.

After the pilot began with alerts sent to personal devices, an issue was found with the application for personal phones and alert notifications. This issue was identified and investigated by the vendor, and a fix was tested and released. The issue stemmed from differences in how alerts were received between personal devices versus shared devices in how push notifications are handled on personal devices. This identified a risk for missed sepsis alerts based on current operating system settings or versioning of the smartphone application on personal devices. Using a shared device model offered more control over versioning and allowed for thorough testing of the application against the operating system before releasing to end users. In 2018, we moved toward iOS-shared hospital devices that allowed us to utilize the secure messaging application for shared devices as well as increase end-user satisfaction, who preferred the use of iOS over Android.

The Sepsis APP team has shown a strong dedication to providing the best care possible for patients with sepsis. Their willingness to support new technology was integral in supporting this implementation, and they continue to use the smartphone sepsis alerts to enhance their workflow. The team remains cognizant of how important these alerts can be in identifying possible patients with sepsis and remain engaged in reporting any issues to the technical team for further investigation. In addition to the smartphone sepsis alerts, improvement of compliance numbers can also be attributed to the sepsis team's dedication, senior abstractors, better documentation, and the implementation of a new database that could better capture data.

Despite increased compliance with the sepsis bundles, patient mortality outcomes remained the same. This could be attributed to the unique population of patients with cancer in the hospital, and more research is needed to see if smartphone sepsis alert implementation would have higher impact in a general patient population. Additionally, review of 2 years of data shows that while the EHR sepsis algorithm is sensitive, it is not specific and can generate false positives related to the complex disease process of oncology patients. There is opportunity for these data to be utilized to create greater specificity of alerts by incorporating trends of lab work results and other documented clinical factors.

Conclusions

Successful implementation of smartphone alerts has the potential to improve early response to sepsis identification and treatment. Leveraging existing infrastructure can help make the implementation more streamlined. Organizations should be cognizant that with any technology implementation, ongoing testing and troubleshooting is needed for technical considerations such as software compatibility and versioning on devices to ensure the safety and reliability of alerts. Other considerations include the availability of a multidisciplinary alarm management committee to assist with decision-making and prioritization, and the availability of clinical applications used for other secured communication purposes that could also be used for securely receiving alerts on hospital-provided and personal smartphones. Recreating an integrated system to receive sepsis alerts alone would be financially and resource-intensive. Other institutions should leverage available technologies and devices to send, integrate, and receive alerts. Focusing on best practices for implementing technologies includes close collaboration with clinical users to influence technology so it enhances their daily workflow as opposed to adding to possible alert fatigue. More studies are needed related to improved outcomes for smartphone mobile alerts as well as increased specificity of sepsis alerts to clinical providers.

REFERENCES

1. Gregory D, Buckner M. Point-of-care technology: integration for improved delivery of care. Crit Care Nurs Q. 2014;37(3):268–272.
2. Kim S, Lee KH, Hwang H, Yoo S. Analysis of the factors influencing healthcare professionals' adoption of mobile electronic medical record (EMR) using the unified theory of acceptance and use of technology (UTAUT) in a tertiary hospital. BMC Med Inform Decis Mak. 2016;16:1–12.
3. Piscotty RJ Jr, Kalisch B, Gracey-Thomas A. Impact of healthcare information technology on nursing practice. J Nurs Scholarsh. 2015;47(4):287–293.
4. Baig MM, Hosseini HG, Connolly MJ. Mobile healthcare applications: system design review, critical issues and challenges. Australas Phys Eng Sci Med. 2015;38(1):23–38.
5. Ruskin KJ, Hueske-Kraus D. Alarm fatigue: impacts on patient safety. Curr Opin Anaesthesiol. 2015;28(6):685–690.
6. Dziadzko MA, Harrison AM, Tiong IC, Pickering BW, Moreno Franco P, Herasevich V. Testing modes of computerized sepsis alert notification delivery systems. BMC Med Inform Decis Mak. 2016;16(1):156.
7. Guirgis FW, Jones L, Esma R, et al. Managing sepsis: electronic recognition, rapid response teams, and standardized care save lives. J Crit Care. 2017;40:296–302.
8. Hunter CL, Silvestri S, Stone A, et al. Prehospital sepsis alert notification decreases time to initiation of CMS sepsis core measures. Am J Emerg Med. 2019;37(1):114–117.
9. New York State. Title: Section 405.2—Governing body. 2017. https://regs.health.ny.gov/content/section-4052-governing-body.
10. New York State. Title: Section 405.4—Medical staff. 2018. https://regs.health.ny.gov/content/section-4054-medical-staff.
11. New York State Department of Health. NYSDOH Sepsis Webinar. 2016. https://ny.sepsis.ipro.org/files/NYSDOH_Sepsis_Webinar_12132016_Final.pptx.
12. Short K, Chung Y. Solving alarm fatigue with smartphone technology. Nurs Crit Care. 2018;13(3):43–47.
13. ECRI Institute. ECRI Institute announces its top 10 health technology hazards for 2012. 2011. www.marylandpatientsafety.org/html/education/2012/handouts/documents/Top 10 Technology Hazards for 2012 Article.pdf.
14. ECRI Institute. Top 10 health technology hazards for 2013. 2012. www.ecri.org/Resources/Whitepapers_and_reports/2013_Health_Devices_Top_10_Hazards.pdf.
    15. ECRI Institute. Top 10 health technology hazards for 2014. 2013. www.ecri.org/Resources/Whitepapers_and_reports/2014_Top_10_Hazards_Executive_Brief.pdf.
      16. ECRI Institute. Top 10 health technology hazards for 2015. 2014. www.ecri.org/Resources/Whitepapers_and_reports/Top_Ten_Technology_Hazards_2015.pdf.
      17. Mackay F, Roy A, Schorr C, Crabtree P, Puri N. CMS Sep-1 measure start time: do we agree? A comparison or clinicians versus quality staff. Crit Care Med. 2018;46(1):719.
      Wolters Kluwer Health, Inc. All rights reserved.