Patient Safety Tip of the Week

September 8, 2015

TREWScore for Early Recognition of Sepsis



The Surviving Sepsis Campaign has focused for several years on reducing the morbidity and mortality from sepsis. Early recognition and aggressive treatment of sepsis are key to survival in patients with sepsis. But recognizing patients at risk for sepsis and septic shock earlier has always been challenging.


Now researchers at Johns Hopkins (Henry 2015) used 27 routinely available physiological and laboratory data from intensive care unit patients to develop the TREWScore (Targeted Real-time Early Warning Score), an early warning score that predicts which patients are at risk for septic shock. The TREWScore identified patients before the onset of septic shock with a sensitivity of 0.85 and specificity of 0.67 and identified patients a median of 28.2 hours before onset of septic shock. Of those identified, two-thirds were identified before any sepsis-related organ dysfunction. Compared to using the MEWS (modified early warning score), the TREWScore demonstrated a 58.6% increase in the number of patients identified before sepsis-related organ failure. The authors conclude that continuous sampling of data from the electronic health records and calculation of TREWScore may allow clinicians to identify patients at risk for septic shock and provide earlier interventions that would prevent or mitigate the associated morbidity and mortality.


This is not the first effort to use information technology to help identify sepsis earlier. In our March 15, 2011 Patient Safety Tip of the Week “Early Warnings for Sepsis” we noted that investigators at Barnes-Jewish Hospital in St. Louis had published preliminary results of a system of real-time computerized alerts for possible sepsis in non-ICU patients (Sawyer 2011). Through review of prior cases the same group of investigators had derived and validated a real-time computerized prediction tool (Thiel 2010). That tool contains both information from vital signs monitoring and a variety of laboratory parameters. When the prediction tool identified a patient on the medicine ward with possible sepsis, an alert was sent automatically (via text page) to the charge nurse on that ward. That nurse would then assess the patient and notify the covering physician, who would decide on any further course of action. Their results showed that patients in the intervention group were more likely to have an increased rate of interventions (such as antibiotic escalation, fluid therapy, oxygen, cultures and other diagnostic tests, etc.) within 12 hours than the nonintervention group. Though both groups had the same rate of transfer to the ICU, those in the intervention group were transferred earlier on average. However, there was no difference in mortality or total hospital length of stay between the two groups, though this pilot study was underpowered to show any such difference.


And earlier this year we noted a study by Umsheid and colleagues that demonstrated that an early warning and response system for sepsis resulted in a significant increase in early sepsis care, ICU transfer, and sepsis documentation (Umscheid 2014). There was also a trend toward decreased sepsis mortality and increased discharge to home that did not reach statistical significance. The system was based on laboratory values and vital signs from the electronic health record monitored in real time. If a patient had ≥4 predefined abnormalities at any single time, the provider, nurse, and rapid response coordinator were notified and performed an immediate bedside patient evaluation.


And two studies presented in abstract form at specialty society meetings this year also demonstrated encouraging results with IT-based systems for early sepsis detection.

One was presented at the HIMSS annual meeting in April (Terry 2015) and discussed an electronic alerting system, POC Advisor, developed by Huntsville Hospital (in Alabama) in conjunction with Wolters-Kluwer. The system collects data from the EMR, lab, nursing notes, and patient monitors and generates alerts for nurses. The nurses then contact the responsible physician and a sepsis protocol is begun. Sensitivity of the tool was 94-98% and specificity 96-99%. During use of the tool mortality fell from 9% to 4.2%. However, in addition to use of the tool, the hospital developed staff education, sepsis protocols, standardized order sets, and sepsis teams. They did note, however, that sepsis deaths did not fall on other floors involved in the study that did not use the alert tool.


The other was presented at the Society of Critical Care Medicine Critical Care Conference in January (Melville 2015) and was used to identify sepsis in trauma patients. It used 4 readily available measures to create a predictive score. Twice-daily measures of the WBC count, heart rate, respiratory rate, and temperature were each given a score of 0 to 4. A score of 4 or higher led to a nurse alerting the clinician, who subsequently identified whether infection was present and initiated therapies. The tool had a sensitivity of 92.5%, specificity of 97.4%, positive predictive value of 73.5% and negative predictive value of 99.4%. Furthermore, the researchers found that use of the tool was associated with a reduction in the 30-day ICU mortality rate (from 13% down to 8%).


It’s not clear how the latter 2 tools compare to the Hopkins-developed tool in terms of how soon at-risk patients are identified. And it is quite likely that components of the sepsis programs in addition to use of the early detection tools played a role in the sepsis mortality improvements. Nevertheless, the ability of these new sepsis prediction tools to identify likely cases earlier is very encouraging.


And there are commercially available surveillance programs. The Cerner Sepsis Biosurveillance Program was recently implemented at Dignity Health Sierra Nevada Memorial Hospital (Cooke 2015).


These sepsis early warning systems are arriving just in time. CMS (The Centers for Medicare & Medicaid Services) has notified hospitals participating in the inpatient quality reporting program that data collection of the Severe Sepsis and Septic Shock: Management Bundle measure (NQF #0500) will begin with discharges on or after Oct. 1, 2015 (QualityNet 2015). Key elements are as follows, with the timeframe for the first three elements being within 3 hours, and the latter four elements (septic shock) within 6 hours:



In our April 1, 2014 Patient Safety Tip of the Week “Expensive Aspects of Sepsis Protocol Debunked” we discussed the results of the ProCESS trial (The ProCESS Investigators 2014). The ProCESS trial involved 31 medical centers and over 1300 patients with sepsis. The patients were randomized to one of 3 treatment arms. One arm received the full early goal-directed therapy (EGDT) protocol, which included the invasive monitoring plus guidelines for vasopressors and blood transfusions. The second arm also received management by a protocol but it did not require the invasive monitoring, vasopressors/inotropes, or blood transfusions. The third arm was the “usual care” group where the clinicians basically decided how to manage the patient. The primary end-point of the trial, mortality at 60 days, did not differ across the 3 trial arms. There were also no differences in secondary outcomes such as 90-day mortality, one-year mortality, and need for organ support.


In the accompanying editorial (Lilly 2014) Craig Lilly, MD noted that guidelines such as those included in state legislation and those endorsed by the National Quality Forum (NQF) needed to be updated now to remove the requirement for central hemodynamic monitoring. It’s pretty clear now that the expense and potential unintended consequences of such monitoring are no longer necessary.


And, indeed, the Surviving Sepsis Campaign has responded “With publication of 3 trials that do not demonstrate superiority of required use of a central venous catheter (CVC) to monitor central venous pressure (CVP) and central venous oxygen saturation (ScvO2) in all patients with septic shock who have received timely antibiotics and fluid resuscitation compared with controls or in all patients with lactate >4 mmol/L, the SSC Executive​ Committee has revised the improvement bundles.”


Also timely was a review in the New England Journal of Medicine about what’s new in management of septic shock, what is evidence-based and what is uncertain (Seymour 2015). Seymour and Rosengart discuss in detail the studies that led to removal of the need for protocolized invasive early goal-directed therapy (EGDT) from management guidelines. They provide a good discussion on use of focused ultrasound for assessing central hemodynamics, markers of tissue injury, the types of fluids used in resuscitation, the colloid vs. crystalloid issue, and which vasopressors are recommended. They also noted that fluid overload is common in management of septic shock and note uncertainty about the best ways to handle that. They also note that, while measurement of serum lactate is universally recommended in sepsis management, questions remain regarding lactate thresholds and frequency of determination.


In our April 1, 2014 Patient Safety Tip of the Week “Expensive Aspects of Sepsis Protocol Debunked” we did get on our soapbox to express our opinion that some of the perceived improvements in sepsis morbidity and mortality over the last decade may actually have been artifacts of changes in hospital coding practices. The pneumonia patient happily walking up and down the ward hallway pushing his IV pole might now be labelled as having sepsis because he also happens to meet 2 of the 4 SIRS criteria. Hardly what we would have envisioned as a septic patient years ago!


Nevertheless, we still recognize the importance of early recognition, timely antibiotics and adequate fluid resuscitation to reduce morbidity and mortality from sepsis. Hopefully, as more hospitals develop and adopt these computerized tools to identify sepsis earlier we will further reduce the morbidity and mortality from sepsis.







Surviving Sepsis Campaign. Website.



Henry KE, Hager DN, Pronovost PJ, Saria S. A targeted real-time early warning score (TREWScore) for septic shock. Science Translational Medicine 2015; 299(7): 299ra122; 05 Aug 2015



Sawyer AM., Deal EN, Labelle AJ, et al. Implementation of a real-time computerized sepsis alert in nonintensive care unit patients. Critical Care Medicine 2011; 39(3): 469-473



Thiel SW, Rosini JM, Shannon W, et al. Early prediction of septic shock in hospitalized patients. J Hosp Med 2010; 5(1): 19-25



Umscheid CA, Betesh J, VanZandbergen C, et al. Development, implementation, and impact of an automated early warning and response system for sepsis. J Hosp Med 2014; Article first published online: 26 SEP 2014



Terry K. Clinical Decision Support May Help Reduce Sepsis Mortality. Medscape Conference News 2015; May 20, 2015

Healthcare Information and Management Systems Society (HIMSS) Annual Conference and Exhibition: Presentation 109. Presented April 14, 2015



Melville N. Sepsis Screening Tool Spots Subtle Signs, Saves Lives. Medscape Medical News January 23, 2015

Society of Critical Care Medicine (SCCM) 44th Critical Care Congress: Abstract 8. Presented January 18, 2015



Cooke G. New tool at SNMH alerts doctors to potential sepsis cases. The Union (Western Nevada County, CA) 2015; September 1, 2015



QualityNet. Specifications Manual, Version 5.0a. Discharges 10/01/2015 to 06/30/2016. Sepsis Bundle Project (Sep) National Hospital Inpatient Quality Measures (updated 5/29/2015)



The ProCESS Investigators. A Randomized Trial of Protocol-Based Care for Early Septic Shock. N Engl J Med 2014; 370(18): 1683-93 published online March 18, 2014



Lilly CM. The ProCESS Trial - A New Era of Sepsis Management. N Engl J Med 2014; 370(18): 1750-1751 published online March 18, 2014



Surviving Sepsis Campaign. SSC Six-Hour Bundle Revised.



Seymour CW, Rosengart MR. Septic Shock. Advances in Diagnosis and Treatment. JAMA 2015; 314(7): 708-717






Print “PDF version










Tip of the Week Archive


What’s New in the Patient Safety World Archive