No hospital wants to see avoidable patient deaths, but unfortunately, sometimes they do occur.1 To reduce the likelihood of an avoidable death and to enhance quality of care, many hospitals employ early warning systems (EWS) that use scores based on vital signs, such as pulse, temperature, oxygen saturation level, blood pressure, and respiration rate.2 If the score indicates a patient is declining, a rapid response team of nurses and other clinicians can be deployed to assess the patient and, if necessary, move him or her into the intensive care unit (ICU).
This approach seems reasonable and has no doubt saved lives, but a fair evaluation of such EWS shows that they have had disappointing impact on outcomes.3
To truly stay on top of patient conditions and get ahead of any emergencies, a broader, more comprehensive suite of variables that reveal subtle physiological alterations should be considered, says Michael Rothman, PhD, the co-founder and chief science officer of Charlotte, North Carolina-based PeraHealth.
What should the target be for an EWS?
The original EWS (MEWS) was developed in 1999 in the United Kingdom with the goal of preventing cardiac arrest in medical-surgical units. However, few patients in these units experience cardiac arrest — perhaps two-tenths of one percent.3 If one wants to address significant avoidable in-hospital mortality, one must look elsewhere. In fact, there is another, important target. On the order of 20% of hospitalized patients experience a major complication.3 Most of these are addressed on a timely basis by physicians and nurses, but not all. Those patients with complications, which are not attended to on a timely basis are a significant target.
Why do vital-sign-based EWS fail?
Rothman claims that vital-sign-based EWS sometimes fail for two reasons. First, such systems generate an alert during decompensation, when vitals are “crashing”, when it’s often unfortunately too late to affect outcomes. The crisis has already occurred with the resulting negative impact.3 Second, vital sign data is input into the hospital electronic medical record by nurses. When there is a significant deviation in these measurements, the nurse often already knows about it. Therefore, any “warning” generated by such a system is after the fact and provides no new information.
But there is another side to decompensation, and that is compensation. “When the body is decompensating, feedback systems in the body are trying to keep key operating parameters within a reasonable range,” he said. As a result, significant changes in vital signs occur late in the deterioration process and make it difficult for clinicians to intervene. Other data, including nursing notes and lab values are needed to provide information that allows clinicians to intervene.
Trifecta of value
According to Rothman, an EWS can work well only when what he refers to as the “trifecta of value” is met. First, the information has to be correct. “If it says you’re sick, you have to be sick,” Rothman said. “That’s the easy part.”
Second, the information must be delivered in a timely manner. In other words, it’s not useful if a rapid response team finds out a patient is in trouble 10 minutes before he or she collapses, which is one of EWS weaknesses. And third, the information revealed has to be new. That turns out to be the difficult part.
Putting wide-ranging data to work
Meeting all the conditions of a successful EWS is difficult, Rothman acknowledges. But there are ways to track patients’ conditions using typically overlooked data. To become attuned to nuances of deterioration and enable staff to intervene before a patient experiences sepsis or another emergency, Rothman shares the concept of an index that hospitals can use to gage the severity of a patient’s condition. The system is based upon electronic data and algorithms to capture a range of physiological data points. Using this information, a scoring index is generated that can be deployed facility wide. The bottom line? Clinicians could potentially be informed sooner than without the index.
The data Rothman references typically is collected during a nursing assessment. This involves a full-body, structured evaluation of a dozen or so physiological systems that determine if the patient has “met” or “not met” a minimum standard. Should any of the components of a given system assessment not be met, the patient would have “not met” the standard, which is then noted as abnormal and entered into the electronic medical record. “Think of these as ‘functional failures.’ The patient stops eating, is confused, becomes incontinent, develops edema, has trouble walking, etc.” says Rothman.
As the goal is to catch these patient signals in real time, it’s important to cast a “wide net” and include not only vital signs and nursing assessment data but cardiac rhythm data and laboratory test results. When all of this information together is captured, a more helpful picture of a patient’s underlying condition emerges. Using an algorithm to process the data, such an index can then produce a score indicating the patient’s relative level of illness. If data is entered regularly, a pattern of decline should be noticed and appropriate clinical steps can then be taken.
Optimizing electronic systems
In order to utilize the index Rothman describes, a hospital must already be using an electronic medical record platform, into which the scoring system can be integrated. Seeing an index score could then be as easy as opening a patient’s chart and clicking on the appropriate tab. Graphs then provide a visual representation of a patient’s trajectory and can be customized to a target length of time, including a patient’s entire hospital stay.
To simplify things for busy clinicians responsible for more than one patient, the index may also be set to display an array of graphs for multiple patients. This can visually highlight the most vulnerable patients according to warning level and index score, alerting doctors and nurses as to who should be prioritized at any given time either by displaying a warning when a patient’s chart is opened or by generating a text that can be received on a mobile phone.
“In order to see the compensation process in action, you need to be sensitive to the functional fails that are captured in the nursing assessments,” said Rothman, who added that an electronic system capturing these data points could potentially reduce in-hospital mortality by 20% to 30% simply by making doctors and nurses aware of problems that might otherwise slip under the radar.4 “I think of this as curating the most valuable resource —clinicians’ attention.”
For further reading on the index created by Dr. Rothman, please click here.
2. Institute for Healthcare Improvement. Early Warning Systems: Scorecards That Save Lives. http://www.ihi.org/resources/Pages/ImprovementStories/EarlyWarningSystemsScorecardsThatSaveLives.aspx. Accessed June 15, 2020.
3. Rothman MJ. The Emperor Has No Clothes. Crit Care Med. 2019;47(1):129-130. doi:10.1097/CCM.0000000000003505
4. Rothman MJ, Rothman SI, Beals J 4th. Development and validation of a continuous measure of patient condition using the Electronic Medical Record. J Biomed Inform. 2013;46(5):837-848. doi:10.1016/j.jbi.2013.06.011
Access related articles, videos, webinars at PRECISELY, by GE Healthcare and Roche.