Data-driven HR, RR parameters might help reduce alarm fatigue

New research shows 55.6% fewer out-of-range vital sign measurements for CRA, RRT activation


Clinical question: Can alarm fatigue in pediatric inpatient settings be safely mitigated by modifying alarm limits with data-driven vital sign reference ranges?

Background: The management of patient alarms in the hospital is a significant safety issue, with the large majority of alarms (85%-99%) either false or not clinically significant. This leads to provider desensitization or alarm fatigue, which has been shown to contribute to adverse events.

In 2014, the Joint Commission made the issue of alarm system safety and alarm fatigue a priority for hospitals.1 Multiple studies have been published addressing alarm fatigue in hospitalized adult patients, but this issue is less well studied in pediatrics, including little guidance on optimizing alarm parameters. Widely used reference ranges and guides are based on limited evidence, primarily based on observational data in healthy outpatients or consensus data.

Dr. Carl Galloway
Dr. Carl Galloway

A 2013 study used vital sign data from hospitalized children to develop percentile curves for heart rate (HR) and respiratory rate (RR) and estimated that 54% of vital sign measurements in hospitalized children are out of range using currently accepted reference ranges.2

To safely decrease the number of out-of-range vital sign measurements resulting from current reference ranges, this study used data from non–critically ill hospitalized children to develop HR and RR percentile charts, and then performed retrospective safety analysis by evaluating effects of modifying the alarm limits on identification of cardiorespiratory arrests (CRA) and rapid response team (RRT) activations.


Study Design: Retrospective, cross-sectional study.

Setting: Single-site, 311-bed quaternary-care academic hospital, both general medical and surgical units.

Synopsis: Vital signs were extracted from the institution’s electronic health record (EHR) for all general medical and surgical patients discharged between Jan. 1, 3013, and May 3, 2014, excluding critically ill children and physiologically implausible vital signs. Two different sets were used, a training set (patients discharged between Jan. 1, 2013, and Dec. 31, 2013) and a validation set (Jan. 1, 2014-May 3, 2014). One HR and RR pair was randomly selected for each 4-hour interval during hospitalization, with a maximum of 10 HR and RR pairs per patient. Age-stratified percentiles were calculated using this data. The 5th and 95th percentile limits using the study data were compared with the 5th and 95th percentile values in the 2013 study, and the reference ranges currently in use at the institution (2004 National Institutes of Health ranges).2

The training set used 62,508 vital sign measurements for 7,202 patients to calculate percentiles for HR and RR among 14 different age groups. The validation set consisted of 82,993 vital sign measurements for 2,287 patients. Using the 5th and 95th percentiles for HR and RR resulted in 24,045 (55.6%) fewer out-of-range measurements in the validation set compared to NIH reference ranges (45% fewer HR values, 61% fewer RR values). This finding, as well as the vital sign percentile ranges, was consistent with the data published in the 2013 study.2

Data for all 148 out-of-ICU RRT and CRA events during the same time period were reviewed using manual chart review. Evaluating vital signs within the 12 hours preceding the events, 144 patients had out-of-range HR or RR measurements using NIH ranges. One hundred thirty-six (94.4%) of these 144 patients also had out-of-range measurements using the study-derived 5th and 95th percentile values.

Manual chart review of the remaining eight patients who had normal HR or RR demonstrated that the RRT or CRA interventions occurred for clinical indications that did no rely on HR or RR measurement (for example, desaturations, difficulty breathing, hematemesis), so the data-driven parameters did not miss any of these events.

Bottom line: In this retrospective study, using data-driven HR and RR parameters was at least as safe as the NIH-published reference ranges currently in use in this hospital. In addition to maintaining safety related to RRT and CRA events, use of the data-driven parameters resulted in 55.6% fewer out-of-range vital sign measurements in the studied population. This may reduce the frequency of false alarms and improve alarm fatigue, and should be studied prospectively in the future.

Citation: Goel VV, Poole SF, Longhurst CA, et al. Safety analysis of proposed data-driven physiologic alarm parameters for hospitalized children. J Hosp Med. 2016;11(12):817-823.

Dr. Galloway is a pediatric hospitalist at Sanford Children’s Hospital in Sioux Falls, S.D., assistant professor of pediatrics at the University of South Dakota Sanford School of Medicine, and vice chief of the division of hospital pediatrics at USD SSOM and Sanford Children’s Hospital.


1. The Joint Commission. Alarm system safety. Available at: Published December 11, 2013.

2. Bonafide CP, Brady PW, Keren R, Conway PH, Marsolo K, Daymont C. Development of heart and respiratory rate percentile curves for hospitalized children. Pediatrics. 2013;131(4):e1150-1157.