Retrospective Evaluation of Diurnal Patterns in Clinical Alarm Rates at a Children’s Hospital

Cardiorespiratory (CR) and pulse oximetry (POx) monitoring have improved patient care since their introduction at the bedside. However, ubiquitous implementation and reliance on continuous monitoring has led to overuse, alarm fatigue, delayed response times, and missed patient decompensation, despite limited effectiveness data outside of high-risk populations.1–4 In 1 year, as many as 8 million alarms were reported at a single children’s hospital, equating to more than 94 alarms per patient per day.5 Diurnal bedside alarm patterns have been described in adult health care settings, highlighting times of the day, referred to as alarm floods, in which significant numbers of alarms are experienced in a short period of time.6,7 Whether diurnal patterns in alarms and alarm floods exist at children’s hospitals remains unknown. Uncovering when alarm burden is most significant and when alarm flood is more likely to occur may assist in identifying times at greatest risk for alarm fatigue and missed patient decompensation, as well as highlighting opportunities for alarm stewardship improvement efforts. Therefore, we sought to describe alarm rate trends by hour of the day and compared daytime vs nighttime alarm rates.We conducted a retrospective study of clinical alarm rates at a freestanding children’s hospital from January through December 2023. We included 14 alarm types (eg, ventricular tachycardia), categorized into 6 alarm causes (eg, arrhythmia), arising from 3 alarm sources (eg, CR lead). We included data from 8 medical/surgical units (MSU), 1 pediatric intensive care unit (PICU), and 1 cardiac intensive care unit (CICU). Alarm rates per patient-day per hour of day were calculated. Hour-of-day alarm rates were stratified by unit type (ie, MSU, PICU, and CICU), alarm source, and cause. Alarms were also categorized as daytime (0700–1859) vs nighttime (1900–0659) events. Poisson regression determined P values for comparisons. Additional methodology can be found in the Supplemental Methods.There were 3 274 872 alarms across 54 558 patient-days. The overall rate was 60 alarms per patient-day (95% CI, 60.0–60.1), with variation between unit types (MSU 57.6 vs PICU 62.8 vs CICU 78.1; P < .001). Alarm rates varied by hour of day (range, 40.3 at 0100 to 85.8 at 2000). Major peaks occurred at 0800 and 2000 and minor peaks at 0000, 0400, 1200, and 1600 (Figure A). Similar peaks occurred across nearly all unit types, alarm sources, and causes. Most prominent peaks were seen for the “POx” source and “technical” cause (Figure B and C). An alarm occurring when a POx probe becomes disconnected from the monitor is an example of a technical alarm.Overall, daytime hours had significantly higher rates of alarms per patient day than nighttime hours (66.8 [95% CI, 66.7–66.9] vs 53.2 [95% CI, 53.1–53.3]; P < .001). Across all unit types, alarm sources, and causes, daytime alarm rates were also significantly higher than nighttime rates, except for “low heart rate,” which had a significantly higher nighttime vs daytime alarm rate (6.4 [95% CI, 6.3–6.4] vs 3.7 [95% CI, 3.7–3.7]; P < .001; Table).We found diurnal trends in clinical alarms at our institution in 2023, with higher rates occurring during daytime vs nighttime hours for nearly all characteristics studied. Alarm rates peaked at 0800 and 2000 (around times of patient evaluations after nursing unit shift changes at 0700 and 1900). Lesser peaks occurred at 4-hour intervals, which happened to coincide with our institutional vital sign measurement schedule. While our study reflects data from only 1 pediatric institution, similar alarm peaks have been encountered in adult settings6,7 suggesting that twice-daily alarm floods, occurring around times of unit shift changes, are likely to be a phenomenon encountered at other health care institutions. These trends require deeper exploration, such as understanding differences in patterns across unit types. Finding interventions that blunt alarm peaks may serve as a starting point for alarm stewardship efforts. Using wireless POx probes is one intervention that may serve as a levee against alarm floods arising from technical POx alarms.8 Further studies are needed to better understand alarm rate nuances and drivers of diurnal patterns.The authorship team thanks our institution’s Managing Clinical Alarms Committee for its continual support of our research endeavors.