View as PDF version
Early warning scoring systems to identify patients at risk of clinical deterioration make a lot of sense in theory. But keep in mind that a good early warning scoring system is only as good as what you do with any alerts generated. The literature on early warning scoring systems and rapid response teams has shown mixed and inconsistent results in terms of actual patient outcomes. Our many columns on both those issues (listed below) have demonstrated our ambiguity and uncertainty about the value of these systems. But a new study moves the needle toward adoption of a new integrated program.
Clinicians and researchers at Kaiser Permanente Northern California (KNPC), a 21-hospital system, previously developed an automated early warning system based on data obtained from the electronic medical record (EMR) (Kipnis 2016). The scoring system they derived, called the Advanced Alert Monitor (AAM) compared favorably to 2 other early warning scores (NEWS and eCART) in predicting events (unplanned transfers to the intensive care unit from a general med/surg floor). Predictors of events included laboratory tests, individual vital signs, neurologic status, severity of illness and longitudinal indexes of coexisting conditions, care directives, and health services indicators (e.g., length of stay). A score is generated and a threshold level was ascertained for prediction of events. For all three scores, about half of alerts occurred within 12 hours of the event, and almost two thirds within 24 hours of the event. The AAM system is monitored remotely by nurses who then communicate alerts to rapid-response teams at hospitals. An AAM score of 5 (alert threshold) indicates a 12-hour risk of clinical deterioration of 8% or more. At this threshold, the model generates one new alert per day per 35 patients. Notably, the alerts provide 12-hour warnings and do not require an immediate response from clinicians.
After a successful pilot at 2 of the KNPC hospitals, the program was subsequently deployed in a staggered fashion at the other 19 hospitals in the KPNC system between August 1, 2016, and February 28, 2019 (Escobar 2020). To assess the impact, they looked at adult patients admitted to those hospitals after the intervention. A comparison cohort that included all the patients who had been admitted to any of the study hospitals in the 1 year before the introduction of the intervention in the first hospital was used as a historical control. A nontarget population included all the patients whose condition did not reach the alert threshold.
Patients in the intervention cohort, as compared with those in the comparison cohort, had a lower incidence of ICU admission (17.7% vs. 20.9%), a shorter length of stay among survivors (6.5 days vs. 7.2 days), and lower mortality within 30 days after an event reaching the alert threshold (15.8% vs. 20.4%).
They also did adjusted analyses, which estimated an absolute difference of 3.8 percentage points in mortality within 30 days after an event reaching the alert threshold between the intervention cohort and the comparison cohort. That difference translated into 3.0 deaths avoided per 1000 eligible patients or to 520 deaths per year over the 3.5-year study period among approximately 153,000 annual hospitalizations. The intervention was also associated with a lower incidence of ICU admission, a higher percentage of patients with a favorable status 30 days after the alert, a shorter length of stay, and longer survival. Results were reasonably similar across the hospitals in the system.
There are several advantages the AAM system coupled with this specific response system have over prior early warning scoring systems and rapid response teams. The AAM is fully automated, takes advantage of detailed EHR data, and does not require an immediate response by hospital staff. The authors point out that these factors facilitated its incorporation into a rapid-response system that uses remote monitoring, thus shielding providers from alert fatigue. That you have almost a 12-hour window in which to respond is a major advantage over systems that immediately trigger a Code Blue or other rapid response.
Though this was not a randomized, controlled trial (RCT) it had the advantage of being assessed in a very large population and the results seemed consistent across hospitals.
We love the concept of a system that relies on readily available data and works relatively unseen in the background and does not disrupt clinical workflows. Yet we lament somewhat that clinical judgement is not part of this system. Our February 2015 What's New in the Patient Safety World column Detecting Clinical Deterioration: Dont Neglect Clinical Impression reminded us not to neglect the value of the clinical impression a nurse or physician has about the patients status. In that column we noted a study that added the question How likely is this patient to suffer a cardiac arrest or require emergent transfer to the ICU in the next 24 hours? improved the predictive value of at least one early warning system tool (Patel 2015).
Our latter concern aside, this work by the clinicians and researchers at KPNC makes one of the strongest cases weve ever seen for an integrated response system tied to an early warning scoring system. Kudos to KPNC!
Some of our other columns on MEWS or recognition of clinical deterioration:
Our other columns on rapid response teams:
Kipnis P, Turk BJ, Wulf DA, et al. Development and validation of an electronic medical record-based alert score for detection of inpatient deterioration outside the ICU. J Biomed Inform 2016; 64: 10-19
Escobar GJ, Liu VX, Schuler A, et al. Automated Identification of Adults at Risk for In-Hospital Clinical Deterioration. N Engl J Med 2020; 383: 1951-1960
Patel AR, Zadravecz FJ, Young RS, et al. The Value of Clinical Judgment in the Detection of Clinical Deterioration. JAMA Intern Med 2015; 175(3): 456-458 Published online January 05, 2015
Print PDF version