E BRIEF REPORT

Ventilator Alarms in Intensive Care Units: Frequency,

Duration, Priority, and Relationship to Ventilator
Parameters
Maria M. Cvach, DNP, RN, FAAN,* Jacqueline E. Stokes, MS, MEd, RRT,†
Sajid H. Manzoor, BS, MEd, RRT,† Patrick O. Brooks, BS, RRT,† Timothy S. Burger, RRT,†
Allan Gottschalk, MD, PhD,† and Aliaksei Pustavoitau, MD, MHS†

Ventilator alarms have long been presumed to contribute substantially to the overall alarm bur-
den in the intensive care unit. In a prospective observational study, we determined that each
ventilator triggered an alarm cascade of up to 8 separate notifications once every 6 minutes.
In 1 intensive care unit with different ventilator manufacturers, the distribution of high-priority
alarms was manufacturer dependent with 8.6% of alarms from 1 type and 89.8% of alarms from
another type of ventilator. Alarm limits were not a function of patient-specific ventilator settings.
(Anesth Analg XXX;XXX:00–00)

P
roviders in intensive care units (ICUs) are exposed METHODS
to frequent alarms, many of which may not require This manuscript adheres to the Strengthening the Report­
action by a provider.1,2 Excess alarm exposure may ing of Observational Studies in Epidemiology (STROBE)
affect patient safety,3,4 disturb sleep,5 and are consistently guidelines. The Johns Hopkins Medicine Institutional
listed among the top ventilator-associated events reported Review Board deemed this project as quality improve-
to the Food and Drug Administration.6,7 Despite their pre- ment and waived the requirement for informed consent.
sumed importance, the frequency of ventilator alarms has We conducted a prospective observational study in March
not been systematically quantified.6,8 2017 in the cardiovascular surgical ICU (CVSICU), medi-
The number of times a device signals an alarm condi- cal ICU (MICU), and neurocritical care unit (NCCU) at an
tion may not fully represent the alarm burden. For example, academic tertiary care medical center in an urban setting
in Johns Hopkins Hospital (JHH) adult ICUs, ventilator (JHH, Baltimore, MD).
alarms are communicated to providers in multiple ways We evaluated Puritan Bennett 840 (PB840; Medtronic,
(Supplemental Digital Content, Figure 1, http://links.lww. Minneapolis, MN) ventilators in the MICU, NCCU, and
com/AA/C583), and a single alarm may trigger a cascade CVSICU and Hamilton G5 (G5; Hamilton Medical, Reno,
of up to 8 notifications. Although a primary notification is NV) ventilators in the CVSICU. During the study period,
conveyed through visual and auditory signals generated each ICU used additional ventilators not included in our
by the ventilator to staff in close proximity, supplemental study. Once randomly selected, all study ventilators stayed
alarm notifications are sent through the nurse call system in the same ICU for the study duration.
(Telligence; GE Healthcare, Chicago, IL) and to Wi-Fi phones To collect ventilator parameters and alarm data, we used
through the hospital’s middleware system (Connexall; 18 device-to-network bridge Capsule axons (Qualcomm
GlobeStar Systems Inc, Toronto, ON, Canada). Life, Inc, San Diego, CA) connected to the study ventila-
To determine the frequency of ventilator alarms at JHH, tors, of which 17 consistently provided data. Ventilator
we evaluated the frequency, duration, and type of ventila- parameters and alarm settings data were transmitted
tor alarms in 3 adult ICUs and examined factors influencing from the ventilator to a dedicated server every 10 seconds
alarm settings. or less for PB840, and less frequently for G5 ventilators.
Ventilator parameters included respiratory rate, ventila-
tion mode, tidal volume, minute volume, spontaneous
From the *Department of Integrated Healthcare Delivery, Johns Hopkins minute volume, spontaneous expired tidal volume, spon-
Health System, Baltimore, Maryland; and †Department of Anesthesiology
and Critical Care Medicine, The Johns Hopkins Hospital, Baltimore,
taneous ventilation type, and spontaneous respiratory
Maryland. rate. Alarm settings included inspiratory pressure upper
Accepted for publication August 17, 2018. and lower alarm limit, respiratory rate upper alarm limit,
Funding: Supported, in part, by a grant from the Association for the and minute volume upper and lower alarm limit. We col-
Advancement of Medical Instrumentation (AAMI). lected alarm codes according to the specific alarm label and
The authors declare no conflicts of interest. priority as defined by manufacturer; ventilator type; ICU;
Supplemental digital content is available for this article. Direct URL citations and time, date, and duration of alarm. Ventilator alarms
appear in the printed text and are provided in the HTML and PDF versions of
this article on the journal’s website (www.anesthesia-analgesia.org). included apnea, high and low inspiratory pressure, high
Reprints will not be available from the authors. and low minute volume, expired minute volume, loss of
Address correspondence to Maria M. Cvach, DNP, RN, FAAN, Department positive end-expiratory pressure, patient disconnect, cir-
of Integrated Healthcare Delivery, Johns Hopkins Health System, Room 631, cuit occlusion, high airway pressure, high respiratory rate,
1830 Bldg, 1830 E Monument St, Baltimore, MD 21287. Address e-mail to
[email protected]. high spontaneous inspired time, high tidal volume, inspi-
Copyright © 2018 International Anesthesia Research Society ration too long, low expired mandatory and spontaneous
DOI: 10.1213/ANE.0000000000003801 tidal volume, low oxygen supply pressure, and volume

XXX 2018 • Volume XXX • Number XXX www.anesthesia-analgesia.org 1

Copyright © 2018 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
 EE Brief Report

Table. Set and Measured Ventilator Parameters, Alarm Settings, and Alarms Across Adult Intensive
Care Units
CVSICU MICU NCCU
Ventilator Parameters, Settings, and Alarmsa Overall PB840 G5 PB840 PB840
Set parameters
Ventilation mode, %
  A/C 46.9 0.5 - 65.9 45.3
  SIMV-PS 7.3 41.1 - <0.01 1.1
  SPONT-PS 42.4 46.7 - 34.1 52.6
  SPONT-VS 2.4 11.8 - 0 1.1
  SIMV 0.6 - 59.3 - -
  SPONT 0.1 - 12.3 - -
  Other 0.3 - 28.4 - -
Tidal volume, L 0.41 (0.08) 0.53 (0.05)b 0.53 (0.05)b 0.38 (0.06) 0.39 (0.08)
Set respiratory rate, breaths/min 11.70 (11.48) 7.79 (9.75) - 14.63 (11.53) 9.69 (11.17)
Pressure support, cm H2O 3.54 (4.60) 4.77 (2.25) 5.18 (3.18) 3.45 (5.88) 3.02 (3.25)
Measured parameters
Tidal volume, L 0.44 (0.20) 0.53 (0.26) 0.55 (0.13) 0.41 (0.17) 0.42 (0.18)
Spontaneous tidal volume, L 0.24 (0.29) 0.46 (0.30) 0.47 (0.17) 0.15 (0.25) 0.25 (0.28)
Spontaneous respiratory rate, breaths/min 5.69 (9.10) - 5.69 (9.10) - -
Respiratory rate, breaths/min 21.87 (9.04) 19.79 (8.96) 21.03 (5.70) 24.23 (8.83) 19.80 (8.81)
Minute volume, L/min 9.70 (4.18) 11.79 (6.11) 11.18 (2.64) 9.84 (3.61) 8.38 (3.29)
Spontaneous minute volume, L/min 4.32 (5.75) 8.58 (8.05) - 2.96 (4.77) 4.10 (4.53)
Alarm settings
Inspiratory pressure upper alarm limit, cm H2O 49.13 (6.41) 46.54 (5.39)b 46.44 (5.92)b 52.23 (6.04) 46.35 (5.41)
Respiratory rate upper alarm limit, breaths/min 48.27 (4.19) 48.59 (3.71) 44.34 (8.44) 49.98 (1.28) 46.02 (5.36)
Minute volume upper alarm limit, L/min 18.70 (3.16) - 18.70 (3.16) - -
Minute volume lower alarm limit, L/min 3.07 (0.72) 3.75 (0.99) 3.10 (0.30) 2.95 (0.70) 2.91 (0.28)
Alarms
Total alarms, n 10,905 1263 522 4374 4746
Total ventilator-hours, n 1555 214 92 649 600
Duration of alarms, s 20.4 (34.2) 30.7 (58.0) 26.5 (45.8) 17.2 (29.9) 20.0 (26.3)
Alarms with duration ≤15 s, % 60.0 50.9 57.5 64.2 58.8
Alarms with duration >15 s, % 40.0 49.1 42.5 35.8 41.2
Type of alarm, %
Apnea 1.7 0.6 2.9 2.1 1.3
High inspiratory pressure 34.2 26.5 47.3 35.3c 33.8c
High minute volume 4.0 13.4 2.7 4.2 1.6
High respiratory rate 17.8 11.5c 9.8c 20.4 18.0
High tidal volume 3.0 2.9 - 1.6 4.3
Low expired mandatory tidal volume 12.9 11.7c - 11.5c 15.9
Low expired spontaneous tidal volume 8.6 9.3c - 7.6 10.2c
Low minute volume 9.3 5.3 12.8 10.2c 9.1c
Patient disconnect 3.8 4.4 23.9 2.4c 2.7c
Otherd 4.9 14.3 0.6 4.7 3.1
Alarm priority, %
Low 75.5 82.1 0.0 80.7 77.1
Medium 10.9 9.3 10.2 10.2 12.0
High 13.7 8.6 89.8 9.1 10.8
Abbreviations: ANOVA, analysis of variance; A/C, assist control; CVSICU, cardiovascular surgical ICU; G5, Hamilton G5 ventilator; ICU, intensive care unit; MICU,
medical ICU; NCCU, neurosciences ICU; PB840, Puritan Bennett 840 ventilator; SD, standard deviation; SIMV, synchronized intermittent mandatory ventilation;
SIMV-PS, SIMV with pressure support; SPONT, spontaneous; SPONT-PS, SPONT with pressure support; SPONT-VS, SPONT with volume support.
a
All data are presented as mean (SD) unless otherwise specified. ANOVA for continuous variables and χ2 test for categorical variables were used to compare
PB840 ventilator parameters and settings between the ICUs; whereas t test for continuous variables and χ2 test for categorical variables were used to compare
ventilator parameters and settings between the 2 ventilator types (PB840 and G5) in CVSICU.
b
Nonsignificant differences when comparing ventilator parameters and settings between the 2 ventilator types (PB840 and G5) in CVSICU. All other comparisons
concerning set and measured ventilator parameters, and alarm settings are statistically significant (P < .001).
c
Nonsignificant differences between the 2 groups. All other comparisons concerning types of alarms are statistically significant (P < .05).
d
Other alarms included (a) for PB840 ventilator: circuit occlusion, expired minute volume, high airway pressure, inspiration too long, low inspiratory pressure, low
oxygen supply pressure, and volume not delivered (for volume support ventilation); and (b) for G5 ventilator: loss of positive end-expiratory pressure.

not delivered (for volume support ventilation mode only). ventilator, alarm indicators include high, medium, and
We excluded alarms that occurred during setup; the PB840 low priority, operator message, and technical fault.9 The
high spontaneous inspiratory time alarm (which occurs PB840 ventilator alarms are designated high, medium, and
during noninvasive ventilation); the G5 failure to cycle low priority or device fault. Priority is either preassigned
and operator messages (which appear on the ventilator as or based on the duration of the measured parameter devia-
visual alarms), and technical or device fault. Alarm prior- tion from the alarm setting.10 However, once an alarm is
ity is predefined by ventilator manufacturers. For the G5 generated, its priority remains fixed, although additional

2   
www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA
Copyright © 2018 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
 Ventilator Alarms in Intensive Care Units

alarms of higher priority may be generated if the alarm

condition remains unresolved.

Statistical Analyses
We counted all alarm notifications occurring at time zero
as the initial notification cascade, and notifications at the
15-second mark were tallied as an additional notification
cascade (Supplemental Digital Content, Figure 1, http://
links.lww.com/AA/C583). Notifications to the “buddy
nurse” and to the charge nurse were not included in the
analyses because it was not possible to verify whether
acknowledgments actually occurred along the notifica-
tion timeline. We determined the number of alarms per
ventilator-hour (1 hour of ventilation on a single ventila-
tor). Next, we assessed how many alarms reached respi-
ratory therapists (RTs) and nurses via Wi-Fi phones by
comparing proportions of the alarms that persisted for >15
seconds across ICUs and by ventilator types. Ventilator
parameters, settings, and alarms (1) across the CVSICU,
MICU, and NCCU, (2) between the 2 ventilator types in
the CVSICU were compared by Student t test for continu-
ous variables and χ2 test for categorical variables. To com-
pare alarm distributions during the 24-hour period, we
used a goodness-of-fit test to a uniform distribution and
one determined from the average across all units. Finally,
we performed linear regression of: (1) set and measured
respiratory rate on respiratory rate upper alarm limit and
(2) minute volume on minute volume lower alarm limit.
All analyses were performed with Stata 14 version 14.2
(StataCorp LLC, College Station, TX) and Mathematica 10
(Wolfram Research, Champaign, IL).

RESULTS
We collected ventilator parameter settings from March 2 to
March 16, 2017, and ventilator alarm data from March 2 to
March 19, 2017. Data were acquired from at most 6 ventila-
tors in the CVSICU, 5 in the MICU, and 6 in the NCCU. The
average (range) number of ventilators used per day on the
3 units during the study period was 5 (2–8) in the CVSICU,
10 (5–16) in the MICU, and 6 (3–9) in the NCCU. These data
are summarized in Table and demonstrate different patterns
for each ICU and ventilator type (P < .001 for comparisons
further specified in Table and its legend).
A total of 10,905 ventilator alarms were initiated over
1555 ventilator-hours in the 3 study ICUs. The mean (stan-
dard deviation) number of alarms per ventilator-hour was
6 (3) in the CVSICU, 7 (4) in the MICU, 8 (2) in the NCCU,
and 7 (4) overall. In the CVSICU, the mean (standard devia-
tion) number of alarms per ventilator-hour by ventilator type
was 8 (2) for the PB840 and 6 (5) for the G5. All alarms in the
CVSICU and NCCU (with no delay in notification) and 2.5%
of alarms in the MICU (with a 60-second delay in notifica- Figure. Distribution of alarms by duration and priority (red = high,
tion) resulted in a cascade of supplemental alarm notifica- orange = medium, yellow = low) across adult ICUs and different ven-
tilator types. A, All ICUs and ventilators in total. B, Alarms from the
tions through the nurse call system. Overall, ventilator alarm Puritan Bennett 840 (PB840) ventilator in the CVSICU. C, Alarms from
conditions did not resolve within 15 seconds 40.0% of the the Hamilton G5 ventilator in the CVSICU. D, Alarms from the PB840 in
time, leading to 2.8 additional Wi-Fi phone notifications per the MICU. E, Alarms from the PB840 in the NCCU. Alarms unresolved
ventilator-hour to RTs and nurses through our alarm notifica- after 15 s triggered additional notification to respiratory therapists and
nurses via Wi-Fi phone in 40% of instances. Note also the difference
tion middleware. The distribution of alarms by duration and in the alarm priority between the 2 ventilator types in the CVSICU.
priority across adult ICUs and ventilator types is given in CVSICU indicates cardiovascular surgical ICU; ICU, intensive care unit;
Figure. Alarms were randomly distributed over the 24-hour MICU, medical ICU; NCCU, neurosciences critical care unit.

XXX 2018 • Volume XXX • Number XXX www.anesthesia-analgesia.org 3

Copyright © 2018 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
 EE Brief Report

day (P > .99) when compared to a uniform alarm distribu- Name: Jacqueline E. Stokes, MS, MEd, RRT.
tion (Supplemental Digital Content, Figure 2, http://links. Contribution: This author helped conduct the study, and write the
manuscript, Methods, and Discussion section.
lww.com/AA/C583), except for CVSICU (P < .001). Only the Name: Sajid H. Manzoor, BS, MEd, RRT.
CVSICU exhibited a temporal pattern of alarm distribution Contribution: This author helped conduct the study, and write the
that differed from the average over all units (P < .001). manuscript, Methods, and Discussion section.
Alarm data across the different ICUs and ventilator Name: Patrick O. Brooks, BS, RRT.
types are summarized in Table. Overall, the 3 most common Contribution: This author helped conduct the study, and write the
manuscript and Methods section.
alarms were high inspiratory pressure (34.2%), high respira- Name: Timothy S. Burger, RRT.
tory rate (17.8%), and low expired mandatory tidal volume Contribution: This author helped conduct the study, and write the
(12.9%). Although only 8.6% of PB840 ventilator alarms in manuscript, Methods, and Discussion.
the CVSICU were classified as high priority (range, 8.6%– Name: Allan Gottschalk, MD, PhD.
10.8% among all ICUs), G5 alarms for the same patient Contribution: This author helped with statistical analysis and
helped write the manuscript, Results, and Discussion.
population (Figure) were classified predominantly as high Name: Aliaksei Pustavoitau, MD, MHS.
priority (89.8%). Such difference is attributed to differences Contribution: This author was the principal investigator and
in manufacturer-specific alarm definitions of priorities helped with study conception, conduct, statistical analysis, and
rather than differences in the underlying population. writing/editing of the manuscript.
This manuscript was handled by: Avery Tung, MD, FCCM.
Ventilator alarm limits were not clinically meaningfully
adjusted based on ventilator parameters (eg, for an increase REFERENCES
in measured respiratory rate of 10 breaths/min, respira- 1. Keller JP Jr. Clinical alarm hazards: a “top ten” health technol-
tory rate upper alarm limit is adjusted by ≈1 breath/min) ogy safety concern. J Electrocardiol. 2012;45:588–591.
(Supplemental Digital Content, Table 1, http://links.lww. 2. The Joint Commission. Sentinel event alert issue 50: medical
com/AA/C583). device alarm safety in hospitals. Available at: https://www.
jointcommission.org/sea_issue_50/. Accessed October 30, 2017.
3. Basner M, Babisch W, Davis A, et al. Auditory and non-audi-
DISCUSSION tory effects of noise on health. Lancet. 2014;383:1325–1332.
This study supports the long-held assumption that the con- 4. Konkani A, Oakley B. Noise in hospital intensive care units–a
tribution of ventilator alarms to the overall alarm burden in critical review of a critical topic. J Crit Care. 2012;27:522.e1–522.
e9.
the ICU is substantial. Overall, we observed 7.0 initial alarm 5. Elbaz M, Léger D, Sauvet F, et al. Sound level intensity severely
cascades per ventilator-hour. Forty percent of those, or 2.8 disrupts sleep in ventilated ICU patients throughout a 24-h
per ventilator-hour, triggered additional notification cascades period: a preliminary 24-h study of sleep stages and associated
of RTs and nurses via the Wi-Fi phone system, increasing sound levels. Ann Intensive Care. 2017;7:25.
6. Pham JC, Williams TL, Sparnon EM, Cillie TK, Scharen HF,
the total to 9.8 notification cascades per ventilator-hour or
Marella WM. Ventilator-related adverse events: a taxonomy
approximately 1 every 6 minutes for each ventilator in use. and findings from 3 incident reporting systems. Respir Care.
These data reinforce previous reports of high overall alarm 2016;61:621–631.
prevalence in ICUs2,11,12 and the operating room.13,14 High 7. Love LC, Millin CJ, Kerns CD. Take precautions with audible
inspiratory pressure, high respiratory rate, and low expired alarms on ventilators. Nursing. 2011;41:65.
8. Carlucci A, Mattei A, Rossi V, Paracchini E, Raineri SM,
mandatory tidal volume were the most frequently triggered Gregoretti C. Ventilator settings to avoid nuisance alarms dur-
alarms in all 3 ICUs. Despite accepted definitions of alarm pri- ing mouthpiece ventilation. Respir Care. 2016;61:462–467.
ority,15 in the CVSICU, we found that different ventilator man- 9. Hamilton Medical, Inc. Hamilton-G5 Operator’s Manual. Reno,
ufacturers prioritized alarms differently, so that high priority NV: Hamilton Medical, Inc; 2016.
alarms accounted for only 8.6% of PB840 alarms, but 89.8% 10. Puritan Bennett Company. Puritan BennettTM 800 Series
Ventilator System Operator’s and Technical Manual. Mansfield,
of G5 alarms. Finally, we found that ventilator alarm settings MA: Covidien LLC; 2011. Available at: http://www.medtronic.
were not meaningfully adjusted with ventilator settings. E com/content/dam/covidien/library/us/en/product/
acute-care-ventilation/PB840_Technical_Reference_Manual_
EN_10067720D00.pdf. Accessed October 30, 2017.
ACKNOWLEDGMENTS
11. Chambrin MC, Ravaux P, Calvelo-Aros D, Jaborska A, Chopin
The authors thank the Association for the Advancement of C, Boniface B. Multicentric study of monitoring alarms in the
Medical Instrumentation Foundation for its contribution to adult intensive care unit (ICU): a descriptive analysis. Intensive
funding a portion of this project; Qualcomm Life, Inc, San Care Med. 1999;25:1360–1366.
Diego, CA, for loaning us 18 Capsule Axons, which were 12. Siebig S, Kuhls S, Imhoff M, Gather U, Schölmerich J, Wrede
used to collect patient and device data electronically; Yih- CE. Intensive care unit alarms–how many do we need? Crit
Jang Chang, MS, and Robert Frank, CET, who provided Care Med. 2010;38:451–456.
information technology support to organize data collected 13. Schmid F, Goepfert MS, Kuhnt D, et al. The wolf is crying in
from the Capsule Axons; and Rachel Reynolds, BS, who the operating room: patient monitor and anesthesia worksta-
provided support in analyzing ventilator alarm data. tion alarming patterns during cardiac surgery. Anesth Analg.
2011;112:78–83.
14. de Man FR, Greuters S, Boer C, Veerman DP, Loer SA. Intra-
DISCLOSURES operative monitoring–many alarms with minor impact.
Name: Maria M. Cvach, DNP, RN, FAAN. Anaesthesia. 2013;68:804–810.
Contribution: This author helped conceive the idea; conduct 15. American Association for Reparatory Care (AARP). Consensus
the study; write the Introduction, Methods, Discussion, and conference on the essentials of mechanical ventilators. Respir
Conclusion; and edit the manuscript. Care. 1992;37:998–1025.

4   
www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA
Copyright © 2018 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.