Threshold Vital Sign Abnormalities as Triggers for Rapid Response Activation

Abstract & Commentary

By David J. Pierson, MD, Professor Emeritus, Pulmonary and Critical Care Medicine, University of Washington, Seattle.

This article originally appeared in the January 2013 issue of Critical Care Alert. It was peer reviewed by William Thompson, MD. Dr. Thompson is Associate Professor of Medicine, University of Washington, Seattle. Drs. Pierson and Thompson report no financial relationships relevant to this field of study.

Synopsis: This study shows that as hospitals adopt electronic workflows, automatic triggering of a rapid response system based solely on changes in vital signs could place a tremendous burden on the system.

Source: Fagan K, et al. Vital sign abnormalities, rapid response, and adverse outcomes in hospitalized patients. Am J Med Qual 2012; Feb 28. [Epub ahead of print.]

Fagan and colleagues at Denver Health Medical Center examined electronic data collected on all adult patients who generated a medical surgical acute-care room charge during a recent 6-month period and remained on the ward for at least 24 hours. At the authors’ institution, all vital signs are entered into an electronic database. Rapid-response activation (RRA) occurs whenever any of the following threshold vital sign abnormalities (TVSAs) is entered: respiratory rate < 8 or > 28 breaths/min, heart rate < 50 or > 120 beats/min, systolic blood pressure < 90 mmHg, diastolic blood pressure > 110 mmHg, oxygen saturation by pulse oximetry < 90% despite supplemental oxygen, or temperature > 39° C. The investigators examined data from patients for whom an RRA occurred, comparing them to non-RRA patients who had at least one TVSA and also to patients in whom neither of these occurred. Outcomes sought were in-hospital mortality, non-ICU cardiopulmonary arrest, and unexpected ICU transfer.

During the study period, there were 9074 adult patient discharges. A total of 2018 of these were non-index hospitalizations for those patients (only the first admission for a given patient during the study interval was used), and 728 admissions were for less than 24 hours. An additional 2485 admissions did not generate a medical-surgical acute-care room charge, leaving 3843 hospitalizations (22,126 acute-care hospital days; 545,773 electronically documented vital signs) that constituted the study population. RRA occurred in 120 patients (3.1%), of whom 114 (95%) had a TVSA. An additional 1111 patients (29%) had at least one TVSA but no RRA, and 2612 patients (68%) had neither of these.

Patients for whom an RRA occurred were more likely to be female and have longer hospital lengths of stay than patients without an RRA. They were more likely to be transferred unexpectedly to the ICU (20.8% vs. 7.3% for patients with TVSA but no RRA, and 1.7% for patients with neither; P < 0.01), and had a non-significant trend toward being more likely to experience a cardiopulmonary arrest during the hospitalization (1.7% vs. 0.2% vs. 0.3%, respectively; P = 0.07). Patients for whom an RRA occurred were more likely to have one or more of the sought-after adverse events than patients in either of the other groups. However, only 2.5% of TVSA recorded during the hospitalization triggered an RRA (120 RRAs for 4739 TVSAs), with a low systolic blood pressure and an elevated heart rate being the first and second most common TVSAs. Overall, one of every 20 recorded vital sign episodes contained a TVSA by the institution’s threshold criteria, and about the same proportion of patient-days included at least two TVSAs in a single day.

Considering the entire patient population, the sensitivity of any TVSA for predicting an adverse event during hospitalization was 70%. However, the occurrence of any TVSA was the least specific of all characteristics examined (70%), with a positive predictive value (PPV) of only 9.2%. Four of the individual vital sign thresholds had PPVs < 10%, while an elevated respiratory rate had the highest PPV at 46%. The fact that only 120 RRAs were called, in the face of 4739 TVSAs that occurred during the same period, supports the assumption that the floor nurses exercised judgment and used additional clinical information in deciding whether the TVSA should trigger an RRA. The authors point out that if their institution were to automate the activation of their rapid response system based on their current thresholds, “we would immediately overwhelm our current resources to respond.”


Although it is hard to argue with the concept that identifying, assessing, and intervening with non-ICU patients who experience acute clinical deterioration should improve both clinical and administrative outcomes, demonstration of this assumed benefit and the establishment of the right triggering mechanisms for rapid response have proven to be challenging. This study shows that electronically detecting vital sign changes by themselves is not the answer in terms of system efficiency and personnel costs. Ward patients whose vital signs were never recorded in the abnormal range that would qualify for a TVSA had an exceedingly low rate of adverse events. However, this study also demonstrates that if generally accepted vital sign abnormalities were used as the only criteria for triggering a rapid response system, the great majority of such calls would be false alarms and the logistics of responding to them all would be prohibitive. As hospitals become more reliant on electronic databases and automate more of their operations, the crucial importance of the bedside nurse’s clinical skills in the operation of a rapid response system must not be left out of the equation.