anesth analg 2011 schmid 78 83

7/31/2019 Anesth Analg 2011 Schmid 78 83

1/6

The Wolf Is Crying in the Operating Room: PatientMonitor and Anesthesia Workstation AlarmingPatterns During Cardiac SurgeryFelix Schmid, MD,* Matthias S. Goepfert, MD,* Daniela Kuhnt, Volker Eichhorn, MD,*Stefan Diedrichs, MD,* Hermann Reichenspurner, MD, PhD, Alwin E. Goetz, MD, PhD,*

and Daniel A. Reuter, MD, PhD*

BACKGROUND: Vital sign monitors and ventilator/anesthesia workstations are equipped withmultiple alarms to improve patient safety. A high number of false alarms can lead to a cryingwolf phenomenon with consecutively ignored critical situations. Systematic data on alarmpatterns and density in the perioperative phase are missing. Our objective of this study was tocharacterize the patterns of alarming of a commercially available patient monitor and aventilator/anesthesia workstation during elective cardiac surgery.METHODS: We performed a prospective, observational study in 25 consecutive elective cardiacsurgery patients. In all patients, identically xed alarm settings were used. All incoming patientdata and all alarms from the patient monitor and the anesthetic workstation were digitally recorded. Additionally, the anesthesia workplace was videotaped from 2 different angles to allowretrospective annotation and correlation of alarms with the clinical situation and assessment of

the anesthesiologists reaction to the alarms.RESULTS: Of the 8975 alarms, 7556 were hemodynamic alarms and 1419 were ventilatory alarms. For each procedure, 359 158 alarms were recorded, representing a mean density of alarms of 1.2/minute.CONCLUSION: Approximately 80% of the total 8975 alarms had no therapeutic consequences.Implementation of procedure-specic settings and optimization in artifact and technical alarmdetection could improve patient surveillance and safety. (Anesth Analg 2011;112:7883)

The use of alarming systems in patient monitoringdevices, such as ventilator/anesthesia workstations,is of paramount importance for patient safety. Thisaccounts for both perioperative anesthesia and monitoringin the intensive care unit (ICU). Inadequate use or failure to

respond to intraoperative alarms may result in patienthazard and undesirable outcomes. 1,2 The majority of alarms are so-called threshold alarms, i.e., a violation of apredefined threshold leads to an acoustic and/or opticalalarm. Therefore, it is crucial to set alarming thresholdscorrectly. Thresholds have to be tight enough to detectpotential deteriorations in vital functions as early as pos-sible. However, tight thresholds are naturally prone to ahigh number of false-positive alarms. Therefore, con-versely, they have to be set wide enough to account forphysiologic inter- and intrapatient variations. In addition,artifacts (e.g., patient movement or manipulation of sen-sors) may lead to false-positive alarms. Depending on the

surgical procedure, differing patterns and frequencies of alarms have been described. 3 For the situation in the

operating room (OR), the rate of false alarms was describedto even exceed the number of correct alarms, so that theactual function of the alarms was lost and they became adistraction. 3 Furthermore, studies in different adult andpediatric ICUs found false alarm rates ranging from 72% to

99%.3 6

The dangerous consequence is the crying wolf phenomenon, i.e., that because of the density of totalalarms and the high number of false alarms, correct andimportant alarms are ignored. 4 Thus, an important goal, inparticular for clinical situations with high-risk procedures,is to reduce false alarm rates to a minimum by an opti-mized setting of alarm thresholds. However, there are onlyfew data on quantity and quality of alarming in a complexperioperative setting.

Therefore, this study was performed to characterize thepatterns of alarms of a current patient monitor (Kappa XLT;Drager, Lubeck, Germany) and an anesthesia workstation(Zeus, Drager) during elective cardiac surgery with the use

of extracorporeal circulation (ECC). The objective was toquantify and to characterize all occurring alarms includingidentification of their origin during the entire perioperativephase. Furthermore, we sought to quantify the number of false-positive alarms produced by the monitoring system.

METHODS

The protocol of this observational study was approved andauthorized by the local ethics committee, the IRB, and theprivacy protection commissioner of the hospital. Afterproviding informed consent, 25 consecutive patients sched-uled for elective cardiac surgery (aortocoronary bypassgrafting and valve surgery) were included. Perioperativecare was given by an anesthesiologist who was informed of

From the Departments of *Anesthesiology, and Cardiovascular Surgery,University Medical Center Hamburg-Eppendorf, Hamburg; and HochschuleAnhalt, Kothen, Germany.Accepted for publication August 30, 2010.Supported by an unrestricted grant from Drager, Lubeck, Germany.FS and MSG contributed equally to this work.The authors report no conflicts of interest.Address correspondence and reprint requests to Daniel A. Reuter, MD,PhD, Department of Anesthesiology, Hamburg-Eppendorf UniversityHospital, Martinistr. 52, 20246 Hamburg, Germany. Address e-mail [email protected] 2010 International Anesthesia Research SocietyDOI: 10.1213/ANE.0b013e3181fcc504

78 www.anesthesia-analgesia.org January 2011 Volume 112 Number 1


2/6

being videotaped but blinded to the aims of the study. Allcaregiving anesthesiologists were staff anesthesiologists orresidents in their last year.

All patients were premedicated with 7.5 to 15 mg midazo-lam per os 1 hour before arrival in the OR. According to theinstitutional standards, induction of anesthesia was per-formed with sufentanil 0.7 g/kg and etomidate 0.15 mg/kg.Tracheal intubation was facilitated by pancuronium 0.1mg/kg. Patients lungs were mechanically ventilated in

volume-controlled mode (autoflow) with tidal volumes of 8mL/kg, and a ventilatory frequency of 8 to 12 breaths/min(Zeus, Drager). Anesthesiawas maintained by isoflurane 0.5%to 1.0% and by sufentanil up to a dose of 0.7 g/kg/h.

All patients were monitored with the same combina-tion of patient monitor (Kappa XLT, Drager) and anes-thesia workstation (Zeus, Drager). Monitoring includedelectrocardiogram and pulse oximetry. Furthermore, ar-terial blood pressure was measured invasively in 1 radialartery. Central venous pressure was continuously ob-tained with a central venous catheter in the right internal jugular vein. Bladder temperature was measured continu-ously. In addition, all ventilation data, i.e., respiratory rate,

airway pressures, tidal volume, and inspiratory and expi-ratory concentration of carbon dioxide and isoflurane, weremonitored by the anesthesia workstation. In the perioper-ative phase, all patients were repeatedly assessed by trans-esophageal echocardiography.

After initiation of monitoring and induction of anesthe-sia, patients were transferred to the OR and connected tothe anesthesia workstation and the patient monitor. Bothdevices were directly connected to a laptop computer fordigital data storage in dedicated full-disclosure files usingspecial software (MedLink, Nortis, Lubeck, Germany andeData TapeRec, Erasmus MC, Rotterdam, The Nether-lands). The data collection started at the moment of arrivalin the OR. The following data were recorded with asampling interval of 1 second: all numerical measurementsof the patient monitor and the anesthesia workstation, allreal-time waveforms of pressures and flow readings, andall alarm events that were generated by both devices. Inparallel, video recordings of the anesthesia workplace from2 different views were performed during the study (Fig. 1).Therefore, all reactions of the attending physician to up-coming alarms could be registered and annotated. All datafrom the patient monitor, the anesthesia workstation, andthe video cameras were time stamped, allowing an exactsynchronization of all stored data.

Fixed alarm settings were used for all patients (Table 1).The caregiving anesthesiologist was instructed not to

change this setting. During ECC, only the mean arterial blood pressure alarm and the body temperature alarmwere active; all other alarms were deactivated. Immedi-ately after ECC, all alarms were reactivated according tothe fixed alarm settings. Only the thresholds for thealarm heart rate were changed to 70 bpm (lower limit)and 110 bpm (upper limit), because of postoperativepacemaker use.

The stored information of each alarm consisted of the

alarm grade, the parameter causing the alarm, and the alarmmessage. The alarm grade indicated whether the alarm wasan advisory alarm (low priority), a serious alarm (mediumpriority), or a life-threatening alarm (high priority). Advisoryalarms indicated technical problems, such as a disconnec-tion of a sensor. Serious alarms were caused by a violationof thresholds. Life-threatening alarms were triggered only by arrhythmias such as ventricular tachycardia, ventricularfibrillation, or asystole. Furthermore, a so-called staticalarm for the invasive arterial blood pressure monitoringwas registered. A static blood pressure alarm is generatedif the height of the amplitude of the arterial pulse pressuremeasured invasively is 3 mm Hg. This is normally used torecognize a potential damping of the signal or a disconnec-tion of the arterial catheter. The principle of using aminimum amplitude to exclude technical problems with apulsating signal is not limited to only this monitor. It is alsoused frequently in pulse oximetry or capnometry to detectpatient disconnection.

After completion of data collection, all data were anno-tated by the same anesthesiologist. Every single alarm wasanalyzed on the basis of the numerical measurements, thedigitalized waveforms, and the respective video sequence.Based on this, all recorded alarms were categorized asfollows: technically true/technically false, clinicallyrelevant/not clinically relevant, and medical reaction:yes/no. Alarms were categorized as technically true if the

Figure 1. Views of the video cameras in the operating room inrelation to the anesthesia and surgical workplace.

Table 1. Standardized Fixed Alarm Settings

VariableHR ECG (beats/min)

Lower limitPre-ECC: 50

Post-ECC: 70

Upper limitPre-ECC: 90

Post-ECC: 110

Pulse Sp O2 (beats/min) 50 90SAP (mm Hg) 80 150MAP (mm Hg) 40 90DAP (mm Hg) 40 80CVP (mm Hg) 0 15LAP (mm Hg) 5 20Sp O2 (%) 93 100Tbladder (C) 35.0 38.5MV (L/min) 3 12RR (breaths/min) 6 20Paw (mbar) 5 30CO2 exp (mm Hg) 30 40CO2 insp (mm Hg) 5Isourane insp (%) 1.3 2.0Isourane exp (%) 0.5 1.3

ECC extracorporeal circulation; HR ECG heart rate electrocardiogram;pulse Sp O2 pulse rate measured by pulse oximetry; SAP systolic arterialpressure; MAP mean arterial pressure; DAP diastolic arterial pressure;CVP central venous pressure; LAP left atrial pressure; Sp O2 oxygensaturation measured by pulse oximetry; MV minute ventilation; RR respiratory rate; Paw airway pressure; CO 2 exp expiratory CO 2 content;CO2 insp inspiratory CO 2 content; isourane insp inspiratory isouraneconcentration; isourane exp expiratory isourane concentration; T bladder bladder temperature.

January 2011 Volume 112 Number 1 www.anesthesia-analgesia.org 79


3/6

measurement was technically correct (without artifacts)and showed a real threshold violation. Alarms were anno-tated as clinically relevant if there was a need for medicalintervention. Patient-related alarms were caused by vari-ables of the patients vital sign monitoring (without staticalarms [Fig. 2]). The annotating anesthesiologist recordedwhether or not a medical reaction was performed after analarm occurred. In uncertain situations, the particular videosequences were reanalyzed by a panel of 3 attendingphysicians.

Every onset of an alarm was counted and every alarmwas counted only once even though the sound continuedfor a longer time. If an alarm was silenced and appearedagain after the 2 minutes, it was counted again.

Da ta were anal yzed u si ng the sof tware R(http://www.r-project.org/) and Excel 2003 (Microsoft

Corp., Redmond, WA). Normal distribution of data wastested with the Kolmogorov-Smirnov test. Normally dis-tributed variables are expressed as mean SD, otherwiseas median (25th75th percentile).

RESULTS

Twenty-five patients (14 men, 11 women) were studied.The mean age was 67 11.4 years. Twelve patientsunderwent arterio-coronary bypass surgery, 11 valvularsurgery, and 2 patients had a combination of both. TheeuroSCORE at time of admission to the hospital was 4.7 3.7 The duration of surgery was 4.95 0.96 hours.

One hundred twenty-four hours of intraoperative moni-toring were recorded. In total, 8975 alarms were recorded.There were 7556 hemodynamic alarms and 1419 ventilatoryalarms. This accounted for 359 158 alarms per procedure,or 1.2 alarms per minute. The overall reaction time (timefrom occurrence to confirmation) to the alarms amountedto 4 43.67 seconds. In this observation, we found 6386serious and life-threatening alarms, which were furtheranalyzed. The remaining 2589 alarms belonged to thecategory advisory technical alarms ( n 836) or were static blood pressure alarms ( n 1753) during ECC (see below).Ninety-six percent of the serious and life-threateningalarms were caused by threshold violations (Table 2). Of those alarms, 4438 (70%) were valid, whereas 1948 (30%)were caused by artifacts and were not valid or relevant. Of the valid alarms, 1735 (39%) were classified as relevant, and2703 (61%) were not relevant.

In all patients, ECC was used. For analysis, all proce-dures were separated into 5 phases: pre-ECC, going onECC (last 15 minutes before start of ECC), during ECC,weaning from ECC (first 15 minutes after cessation of ECC), and post-ECC. During pre-ECC we found 2985alarms (33%) with a density of 1.4 alarms/min. Duringgoing on ECC, we observed 930 (10%), or 2.5alarms/min. In the phase during ECC, 3114 alarms (35%)with a density of 0.9 alarms/min were registered. Duringweaning from ECC 626 alarms (7%) with a density of 1.7

Active Alarms

n= 8975

Patient related alarms

n=6386

Other alarms

n=2589

Technically false

n=1948

Technically true

n=4438

Technical alarms

n=836

Static alarms

n=1753

l i n i c a

l l y n o

t r e

l e v a n

t

n =

1 9 4 8

P a

t i e n

t m o n

i t o r

n =

5 4 1

l i n i c a

l l y n o

t r e

l e v a n

t

n =

2 7 0 3

e s

t h e s i a w o r k s

t a t i o n

n =

2 9 5

C l i n i c a

l l y r e

l e v a n

t

n =

1 7 3 5

C C A

Figure 2. Differentiation of active alarms inpatient-related alarms and other alarms.

Patient-related alarms were caused by variables of the patient monitoring (e.g., invasive arterial bloodpressure and oxygen saturation). Patient-relatedalarms are further separated in technically true ortechnically false, and clinically relevant andnot relevant.

Table 2. Number of Threshold Violations, Sortedby Numbers of Total Violations

VariableLower limit

violationUpper Limit

violation Total %

MAP 739 946 1685 27.5HRECG 371 495 866 14.2SAP 554 117 671 11.0DAP 395 160 555 9.1CVP 240 290 530 8.7VT 0 448 448 7.3CO2 exp 284 20 304 5.0Pulse Sp O2 92 166 258 4.2Sp O2 120 0 120 2.0Paw 0 111 111 1.8LAP 64 18 82 1.3Tbladder 74 0 74 1.2Others 165 250 415 6.8Total 3098 3021 6119 100

MAP mean arterial pressure; HR ECG heart rate electrocardiogram; SAP systolic arterial pressure; DAP diastolic arterial pressure; CVP centralvenous pressure; V T tidal volume; CO 2 exp expiratory CO 2 content; pulseSp O2 pulse rate measured by pulse oximetry; Sp O2 oxygen saturationmeasured by pulse oximetry; Paw airway pressure; LAP left atrialpressure; T bladder bladder temperature; others inspiratory CO 2 content,inspiratory isourane concentration, expiratory isourane concentration, STdeviations (STI, STII, STIII, STV, STaVR, STaVL, STaAVF).

Monitoring Alarms in Cardiac Surgery

80 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA


4/6

alarms/min were found. Finally, during post-ECC, 1320alarms (15%) with a density of 1.5 alarms/min occurred. Inaddition, we found different patterns of alarms in thedifferent phases of the procedure. Therefore, the phasesgoing on ECC and weaning from ECC were character-ized by overly proportional numbers of heart rate, arrhyth-mia, blood pressure, central venous pressure, and end-tidalcarbon dioxide alarms. The tidal volume alarms werefound mainly in weaning from ECC. Patterns are illus-trated in Figures 3 and 4. Furthermore, the time, whichelapsed from onset of the hemodynamic alarms untilreaction, i.e., silencing or disappearance of the alarms(reaction time), was analyzed for each phase: reaction timeduring pre-ECC was 29.3 seconds (9 [323] seconds);going on ECC 20.9 seconds (5 [213] seconds); duringECC 12.0 seconds (4 [28] seconds); weaning from ECC33.7 seconds (10 [425] seconds); and during post-ECC45.6 seconds (9 [3.2530.75] seconds).

In this observation, we found 1648 static blood pressurealarms, accounting for 18% of all analyzed alarms. All weregenerated during ECC.

One thousand nine hundred forty-eight (30%) of theanalyzed alarms were technically false. Of those 1948alarms, 781 (40%) were caused by artifacts or manipula-tions of the patient monitoring system or its sensors ( n 727), the anesthesia workstation ( n 29), or the patient(n 25). The particular reasons for the artifacts were oftennot annotatable (72%). In 28%, the reason for the artifactscould be identified as blood draw or flushing of the arterialline (n 387), connection or disconnection of sensors orventilation ( n 96), or electrocautery ( n 58).

DISCUSSION

This study showed that in a standard perioperative settingin cardiac surgery, the patient monitor and the anesthesiaworkstation generated alarms with an overall density of 1.2alarms/min. Nearly 80% of these alarms had no therapeu-tic consequence. Specific patterns of alarming throughoutthe perioperative phase could be characterized, whichmight help to optimize alarm settings in the future.

Alarms are an essential part of safety in clinical anesthe-sia. However, alarms can only fulfill their function if theyindicate a dangerous patient condition correctly. Unneces-sary alarms may lead to an impairment of communicationand to a distraction from other tasks. 8 Several studies wereperformed investigating patterns of alarms in differentsituations in ICUs. 46,812 However, the situation of intra-operative care differs significantly from the situation in theICU. First, patients in the OR are predominantly anesthe-tized and their lungs mechanically ventilated. They are allunder permanent and direct surveillance of an anesthesi-ologist. Second, deteriorations in the patients conditionsoften happen faster in the OR because of the direct conse-quences of surgical manipulation. Third, many surgicalprocedures are frequently associated with specific maneu-vers, such as ECC in cardiac surgery. This results indifferent patterns of alarms in comparison to the ICUsetting. Thus far, this has not been investigated systemati-cally in a high-complex intraoperative setting, includingcomplete analysis of all raw data forming the basis of thealarms.

The present study also differs with regard to datacollection; in earlier studies, which were performed in the

alarms per phase - by type

2500

3000

3500

a l a r m s Technical

Static

Th h ld Hi h

0500

1000

1500

pre ECC going on ECC during ECC weaning from post ECC

n u m

b e r o

f a

hreshold High

Threshold Low

Arrhythmia

Apnea

ECCphase

Figure 3. Number of alarms per phase and furtherdifferentiated by type of alarms: 1. pre-extracorporeal circulation (ECC), 2. going onECC (last 15 minutes before start of ECC), 3.during ECC, 4. weaning from ECC (rst 15minutes after cessation of ECC), and 5.post-ECC.

Alarm density per phase by type-

2

2,5

3

3,5

a l a r m s / m

i n u

t e ]

Technical

Static

Threshold High

0

0,5

1

1,5

pre ECC going on ECC during ECC weaning fromECC

post ECC

a l a r m

d e n s

i t y

[ aThreshold Low

Arrhythmia Apnea

phase

Figure 4. Alarm density per phase (alarms perminute) shown for the different phases: 1. pre-extracorporeal circulation (ECC), 2. going onECC (last 15 minutes before start of ECC), 3.during ECC, 4. weaning from ECC (rst 15minutes after cessation of ECC), and 5.post-ECC.



5/6

ICU, only the numerical results of cardiorespiratory mea-surements were obtained automatically in varying timeintervals. 5,8,12 In a recent study, Gorges et al. 10 manuallyrecorded the time-stamped alarm information as well as thepresence of health care team members, using a pocketpersonal computer. In 2 studies, nurses recorded andclassified all alarms manually during their normal workingtime, 4,8 whereas in other studies, specially trained observ-

ers who were present at the bedside, but not in charge of patient care, documented and annotated alarm situa-tions. 5,8,10 However, both study designs offer potentialdisadvantages: If the data collection is done by the caregiv-ers themselves, there is the chance that in times with a highworkload not all alarms can be assessed as precisely asneeded. Conversely, if data collection is performed by extrainvestigators, the presence of those observers may result ina different behavior by the caregiving anesthesiologist.

Therefore, we decided for this investigation to recordand to extract all available measurements, alarm settings,and alarms of the patient monitor and the anesthesiaworkstation in a strictly automated manner with precise

time stamps at a high sample rate of 1 per second, and toperform annotations of all alarms on the basis of a recordedvideo of the complete anesthesia workplace. Siebig et al. 11

used a similar setup in their investigation in an ICU setting.Furthermore, this offered the opportunity to examine everyclinical situation as often as needed. This was particularlyhelpful when alarms occurred in quick succession. Inaddition, the video and monitoring data could be com-pared and evaluated more carefully without the pressure of the ongoing procedure.

The only disadvantage we see by annotating without a bedside observer is the missing opportunity to ask theattending anesthesiologist to verify certain situations, buteven this may have resulted in an additional bias.

In contrast to other studies, we used fixed alarm set-tings. 3,4,6,10,11 This was necessary for better comparabilityof the patients. A possible disadvantage may be a higheralarm rate in cases when the fixed settings are not optimalfor the individual patient, but the fixed setting we usedseemed to be adequate for most of the study population.From our point of view, predefined fixed alarm settings areuseful and important. However, fixed alarm settings must be tailored for the patient population (general, cardiacsurgery, pediatric, etc.), but also potentially tailored for theindividual patient, and must be made as user friendly aspossible.

We assessed the alarms with respect to validity, rel-evance, and medical reaction. Although comparable classi-fications were used in the above-mentioned studies in theintensive care environment, definitions of alarm classifica-tions were slightly different among the investigations. Wedefined alarms as technically true if the underlying mea-surement was technically correct and a real thresholdviolation was present. Alarms were relevant if there was aneed for medical reaction. In contrast, in the studies byKoski et al. 8 and Lawless, 4 alarms were defined as beingrelevant only if the patients condition was actively checkedor a treatment was administered. An alarm may sometimestrigger a potentially useful mental note without any notice-able reaction. This undoubtedly influenced our results;

nevertheless, the quantity of these reminders can only beestimated. In our observation, we found a reaction time(from occurrence to confirmation) of 4 43.67 seconds.However, it is possible that the attending anesthesiologistdid react without pressing the silence button or that thealarm was silenced without any further reaction. Last, it isimaginable that an alarm, as described above, leads to amental note without pressing the silence button.

Alarms occurring in situations in which a permanentand not self-correcting technical problem had to be solvedwere judged either as false or artifact alarms. 4,8 Incontrast, Chambrin et al. 12 also regarded technical alarmsas true-positive alarms if they were followed by an action.In agreement with Chambrin et al., we considered alarmscaused by persistent technical problems as necessary, be-cause correct measurement of physiologic variables must be ensured for patient safety. In current alarm systems,many of these alarms are already denoted as advisoryalarms to indicate the technical problem. This concept of differentiating technical and physiologic alarms could befurther refined by an intelligent alarm algorithm developed

on the basis of our annotated dataset.The annotations used in this study distinguish betweentechnically true and technically false alarms. A technicallyfalse alarm is given if measurements do not correctlyidentify the patients condition or if an advisory alarm isgiven without a technical problem. Tsien and Fackler 5

defined these technical alarms as false-positive alarms, but others subsumed technically as well as clinicallyirrelevant alarms as false-positive alarms. 4,8,12 Neverthe-less, it must be the aim for future improved alarmalgorithms to avoid all technically false alarms, becauseall of them represent false-positive alarms with regard totheir clinical consequence.

Alarms are often caused by manipulations performed bythe medical staff. To further characterize these alarms, wenoted whether staff members were directly working withthe patient, the monitoring system, or the anesthesia work-station at the occurrence of the alarm, because this oftenchanged the significance of the alarm. Some advisoryalarms or even technically true alarms may have beenannotated as not relevant in the presence of and withactive manipulation by staff members, but as relevant inthe absence of and without a manipulation of staff members. As in our study, Lawless 4 classified those alarms thatwere not clinically important and caused by staff manipu-lation as induced alarms.

In this observation, we found 1648 static blood pressurealarms, accounting for 18% of all analyzed alarms. All weregenerated during ECC. During ECC, a continuous nonpul-satile flow without pulse pressure is generated; therefore, astatic alarm of the arterial blood pressure measurement isgenerated continuously as long as any threshold alarm forthe blood pressure signal is enabled. This finding and thefact that the different phases of the procedure are charac-terized by different patterns of alarms (i.e., ventricularfibrillation during going on ECC) clearly underlines theneed for phase-adapted alarm settings, as for ECC.

False-negative alarms, i.e., the situation is alarm relevantand no alarm occurred, were not investigated. Althoughthe whole perioperative period was recorded on video,

Monitoring Alarms in Cardiac Surgery

82 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA


6/6

those situations of false-negative alarms cannot be identi-fied reliably by such a recording. It is retrospectively notpossible to reliably identify each medical action and itsrelation to an alarm that did not occur. However, theclinical experience concerning false-negative alarms is thatthey occur very rarely and then mostly because the alarmthresholds are set too wide. Studies that investigated thefrequency and reliability of ICU alarms support these

observations. In the study by Tsien and Fackler,5

not asingle false-negative alarm situation was recorded during298 hours of monitoring. Chambrin et al. 12 recorded 24false-negative cases during 1971 hours of care, but in thisstudy, a false-negative observation was identified evenwhen only alarm threshold settings were modified withouta prior audible alarm.

It is difficult to compare the results of our study withthose of other investigators, because most of the otherstudies were performed in ICU settings. 4 6,8,10,11 OnlySeagull and Sanderson 3 surveyed an intraoperative setting, but it included only 5 patients who underwent cardiacsurgery. The anesthesia workspace was not filmed and only

handwritten documentation of alarms was performed.However, this study examined a combination of a singlepatient monitor and an anesthesia workstation; the alarm-ing patterns of different devices may be tendentiouslysimilar to our results.

Based on the results of our study, several targets can beidentified for the achievement of an alarm reduction. First,the high rates of technical alarms need to be reduced bytechnical improvements with regard to the raw signals (forexample, reduction of oxygen saturation alarms by im-provement of artifact detection). Second, the predominantalarm type is threshold alarm, and alarm reductions based on robust signal extraction may be the mostuniversally applicable solutions. However, methods based on artificial intell igence such as machine learning,neural networks, fuzzy logic, and Bayesian networksmay also have a role in the future, although the require-ments in testing datasets, general applicability, and legalissues are highly challenging.

In conclusion, we found that during 25 consecutivecardiac surgical procedures, approximately 80% of the 8975alarms had no therapeutic consequences. The majority of static alarms were caused by ECC. The not-valid alarmswere mainly caused by manipulations (i.e., blood sampling,electrocautery). The implementation of phase-specific set-tings (e.g., an ECC setting), reminders for their proper use,and optimization in artifact or technical alarm detection

could improve patient surveillance and safety.13

REFERENCES

1. Beatty PC, Beatty SF. Anaesthetists intentions to violate safetyguidelines. Anaesthesia 2004;59:52840

2. Tinker JH, Dull DL, Caplan RA, Ward RJ, Cheney FW. Role of monitoring devices in prevention of anesthetic mishaps: aclosed claims analysis. Anesthesiology 1998;71:5416

3. Seagull F, Sanderson P. Anesthesia alarms in context: anobservational study. Hum Factors 2001;43:6678

4. Lawless ST. Crying wolf: false alarms in a pediatric intensivecare unit. Crit Care Med 1994;22:9815

5. Tsien CL, Fackler JC. Poor prognosis for existing monitors inthe intensive care unit. Crit Care Med 1997;25:6149

6. OCarroll TM. Survey of alarms in an intensive care unit.

Anaesthesia 1986;41:74247. Nashef SA, Roques F, Michel P, Gauducheau E, Lemeshow S,Salamon R. European system for cardiac operative risk evalu-ation (EuroSCORE). Eur J Cardiothorac Surg 1999;16:913

8. Koski EM, Makivirta A, Sukuvaara T, Kari A. Frequency andreliability of alarms in the monitoring of cardiac postoperativepatients. Int J Clin Monit Comput 1990;7:12933

9. Edworthy J, Hellier E. Fewer but better alarms will improvepatient safety. Qual Saf Health Care 2005;14:2125

10. Gorges M, Markewitz BA, Westenskow DR. Improving alarmperformance in the medical intensive care using delays andclinical context. Anesth Analg 2009;108:154652

11. Siebig S, Kuhls S, Imhoff M, Langgartner J, Reng M, Scholmer-ich J, Gather U, Wrede CE. Collection of annotated data in aclinical validation study for alarm algorithms in intensive care:a methodologic framework. J Crit Care 2009;25:12835

12. Chambrin MC, Ravaux P, Calvelo-Aros D, Jaborska A, ChopinC, Boniface B. Multicentric study of monitoring alarms in theadult intensive care unit (ICU): a descriptive analysis. IntensiveCare Med 1999;25:13606

13. Eden A, Pizov R, Toderis L, Kantor G, Perel A. The impact of an electronic reminder on the use of alarms after separationfrom cardiopulmonary bypass. Anesth Analg 2009;108:12038


anesth analg 2011 schmid 78 83

Documents