1. How Can We Safely Reduce 50% of Patient Monitor Alarms in the Surgical Intensive Care Unit? Peter F. Hu, PhD1,2, Li-Chien Lee, MS1, Hsiao-Chi Li, PhD1, Shiming Yang, PhD1, Samuel Galvagno, MD1, Samuel Tisherman, MD2, Peter Rock, MD1 1. Department of Anesthesiology, University of Maryland School of Medicine. 2. Department of Surgery, University of Maryland School of Medicine.
Alarm fatigue
Results
There were 426,647 alarms recorded during the 4-month study period resulting in 148 alarms per bed per day in the 24 bed SICU. In the four industry pre-defined alarm classifications, the majority of the alarms were classified as “C1: and System Warning” (66,300, 15.5%); “C2: Patient Advisory” alarms (n = 245,779, 57.6%); “C3: Patient Warning” (98,024, 23%). Only 3.9% were in the “C4: Patient Crisis” alarm category. The top ten alarm events in each of the four alarm classification are listed in Table 1. In each of the above alarm classifications 25% of alarms were less than 28 seconds (C1), 4 seconds (C2), 2 seconds (C3), and 4 seconds (C4). 50% of alarms were less than 322 seconds (C1), 10 seconds (C2), 13 seconds (C3), and 12 seconds (C4). Changing from the current default alarm threshold settings of SpO2 low (≤90%) to SpO2≤88% (Figure 1), and tachycardia (HR ≥ 130 bpm to HR≥135 bpm could reduce alarms by 41%, and 40% (Figure 2).
“We didn’t hear it”
“We couldn’t tell what was alarming”
“We just got too busy”
“Things were just crazy that day”
“The place is like a casino with so many
bells and whistles”
Background
Alarm fatigue has been recognized as a critical patient safety concern in the modern hospital setting. Better characterization of alarm types and thresholds may reduce the burden of alarms and improve staff responsiveness. We tested the hypothesis that a significant reduction in the number of monitor alarms could be achieved by instituting a short delay (seconds) in activating the alarm to eliminate brief, transitory alarms, and by changing specific vital signs (VS) alarm limits based on analysis of the patient monitor alarms.
Alarm N 2s 4s 6s 8s
10s
1min
Levels
Categories
3 (C1)
System Warning
66300 (15.5%)
6.26%
10.97%
13.76%
15.61%
17.11%
31.61%
5 (C2)
Patient Advisory
(57.6%)
26.55%
34.97%
39.98%
44.04%
47.45%
73.23%
6 (C3)
Patient Warning
98024 (23.0%)
24.70%
34.35%
41.65%
47.57%
52.38%
85.51%
7 (C4)
Patient Crisis
16544 (3.9%)
18.78%
28.93%
35.87%
41.36%
45.90%
81.10%
All
All Alarms
(100.0%)
22.67%
30.87%
36.13%
40.33%
43.81%
69.89%
Methods
We retrospectively analyzed patient VS in a 24-bed Surgical Intensive Care Unit (SICU) and collected alarm data between October 12, 2015, and February 15, 2016, from networked patient VS monitors (GE Solar) using the BedMasterEX (Excel Medical LLC, FL) system. Alarm VS name, four industry defended alarm classifications, duration, and frequency were recorded and analyzed. Most alarms were found to be brief and transitory lasting just a few seconds. Specific duration (seconds) was analyzed to achieve 25% and 50% alarm reduction. To reduce individual VS alarms, different alarm limit settings were compared with the default settings of hypoxia (SpO2 low ≤ 90%,) and tachycardia (heart rate: HR, HR high ≥ 130 bpm).
SpO2 Low
SpO2% ≤ 90%  88%
41%
Fig 1: SpO2 low limit change from the default setting (SPO2 ≤ 90%) and its associated alarm reduction ( frequency n, change % and duration in hours)
Table 2: Durations of alarms in each category
Tachycardia
HR ≥ 130  ≥ 135 bpm
40%
Alarm (Top 10)
Total N 1 2 3 4 5 6 7 8 9 10
Level
Category
(C1)
System Warning
66300 (15.5%)
SPO2 PROBE
NO ECG
CONNECT PROBE
NBP MAX TIME
SENSOR
ARRHY SUSPEND
SPO2 SENSOR
NBP FAIL
RR LEADS FAIL
NBP OVER PRES
33.4%
23.6%
16.2%
14.7%
5.8%
4.6%
1.0%
0.4%
(C2)
Patient Advisory
245779
(57.6%)
ART S LO
PVC
CHECK ADAPTER
ART S HI
NBP S LO
ART M LO
CO2 RSP HI
NBP S HI
ART D HI
ART M HI
25.7%
15.7%
13.6%
13.3%
6.2%
4.7%
4.2%
3.5%
3.4%
2.7%
(C3)
Patient Warning
98024
(23.0%)
SPO2 LO
ART DISCONN
V TACH
VT > 2
NO BREATH
FEM2 DISCONN
HR HI
93.4%
2.8%
1.8%
1.1%
0.9%
0.0%
(C4)
Patient Crisis
16544
(3.9%)
LEADS FAIL
HR LO
BRADY 
V TACH 
ASYSTOLE 
V BRADY 
VFIB/VTAC 
27.1%
25.0%
14.9%
12.4% 
11.8% 
6.5% 
1.4% 
0.6% 
0.3% 
Next Step: Hospital Wide Alarm Dashboard
Target: Reduce 30% alarms in 12 Month
Fig 2: HR high limit change from the default setting (HRSPO2 ≥ 90%) and its associated alarm reduction ( frequency n, change % and duration in hours)
Triple-redundant BedMaster®
VS System
Near 100% complete VS and alarm collection
Conclusion
Alarm fatigue from physiologic alarms in SICU is well recognized but a safe solution to safely reduce alarms has not been established. Our study suggests that by delaying all alarms for 4 seconds we could reduce 25% of the total alarms. By lowering alarm thresholds of SpO2 LO by 2% and increasing the tachycardiac threshold by 5 bpm could reduce an additional 40% of alarms in SICU. Further study is needed to determine what impact such changes would have upon the safety of patients being cared for in the SICU.
4-month data from 24 SICU beds
Table 1: Top 10 most frequent alarms in each category
Supported by University of Maryland School of Medicine, Department of Anesthesiology and University of Maryland Medical Center.