- Email: [email protected]
Automated external defibrillators and in-hospital cardiac arrest: Patient survival and device performance at an Australian teaching hospital
Resuscitation 82 (2011) 1537–1542
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Automated external defibrillators and in-hospital cardiac arrest: Patient survival and device performance at an Australian teaching hospital夽 Roger J. Smith ∗ , Bernadette B. Hickey a , John D. Santamaria b St Vincent’s Hospital Melbourne, Intensive Care, Inpatient Services Building, Level 1, St Vincent’s Hospital Melbourne, 41 Victoria Parade, Fitzroy, VIC 3065, Australia
a r t i c l e
i n f o
Article history: Received 17 April 2011 Accepted 21 June 2011
Keywords: Automated external defibrillator In-hospital cardiac arrest Cardiopulmonary resuscitation Defibrillation
a b s t r a c t Aims: To evaluate the effect of automated external defibrillators (AEDs) on patient survival and to describe the performance of AEDs after in-hospital cardiac arrest. Methods: Prospectively collected data were analysed for cardiac arrests in the general patient care areas of a teaching hospital during the 3 years before and the 3 years after the deployment of AEDs. The association between availability of an AED and survival to hospital discharge was assessed using multivariate logistic regression. AED performance during automated management of the initial rhythms was assessed using information captured by the AEDs. Results: There were 84 cardiac arrests in the AED period and 82 in the pre-AED period. Patient and event characteristics were similar in each period. The initial rhythm was shockable in 16% of cases. Return of spontaneous circulation was higher in the AED period (54% vs. 35%, P = 0.02) but the proportion of hospital survivors in each period was similar (22% vs. 19%, P = 0.56). The adjusted odds ratio for hospital survival when an AED was available was 1.22 (95% CI 0.53–2.84, P = 0.64). An AED was applied in 77/84 (92%) possible cases. Median interruption to chest compressions was 12 s (inter-quartile range 12–13). An automated shock was delivered in 8/13 (62%) possible cases. Conclusions: Availability of AEDs was not independently associated with hospital survival. Shockable presenting rhythms were not common and, in keeping with the manufacturer’s specifications, the AEDs did not shock all potentially shockable rhythms. The hands-off time associated with automated rhythm management was considerable. © 2011 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Automated external defibrillators (AEDs) can analyse the cardiac rhythm, charge automatically if a shockable rhythm (i.e. ventricular fibrillation [VF] or ventricular tachycardia [VT]) is recognised and provide the operator with audible and/or visual prompts for the safe delivery of an electrical shock.1 According to the recently updated guidelines of the Australian Resuscitation Council, the use of AEDs as a component of managing in-hospital cardiac arrest is acceptable.2–6 The benefits to patients of using AEDs during cardiac arrests in certain out-of-hospital settings have been demonstrated.7,8 However, there has never been a randomised controlled trial of AEDs
夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2011.06.025. ∗ Corresponding author. Tel.: +61 3 9288 3972/2211; fax: +61 3 9288 4487. E-mail addresses: [email protected] (R.J. Smith), [email protected] (B.B. Hickey), [email protected] (J.D. Santamaria). a Tel.: +61 3 9288 4576/2211; fax: +61 3 9288 4487. b Tel.: +61 3 9288 4488; fax: +61 3 9288 4487. 0300-9572/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2011.06.025
for in-hospital cardiac arrest and recent observational studies have cast doubt on the effectiveness of AEDs in hospitals. A single-centre study found that replacing manual defibrillators with AEDs made no difference to survival to hospital discharge after in-hospital cardiac arrest when the initial rhythm was shockable (31% vs. 29%, P = 0.80), and survival to hospital discharge was significantly reduced if the rhythm was not shockable (15% vs. 23%, P = 0.04).9 A study of almost 12,000 patients from over 200 hospitals found AED use during in-hospital cardiac arrest was independently associated with a reduction in survival to hospital discharge (16.3% vs. 19.3%; adjusted rate ratio 0.85, P < 0.001).10 The accompanying editorial advocated a cautious approach to introducing AEDs into the hospital setting.11 There are factors that may count against the use of AEDs for in-hospital cardiac arrest. Automated rhythm management is associated with longer interruptions to chest compressions compared to non-automated management12,13 and, in out-of-hospital settings where AEDs have improved patient outcomes, the initial cardiac arrest rhythm was shockable in more than half the cases and the response times of advanced life support providers were relatively long,7,8 but in the hospital setting the initial rhythm is shockable in only about one out of five cases and the
1538
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
response times of advanced life support providers are relatively fast.9,10,14,15 We have previously reported that there was no change in survival to hospital discharge after in-hospital cardiac arrest in the first year following the deployment of AEDs to the general patient care areas of our hospital.15 We now present cardiac arrest data for the 3 years before and the 3 years after the AED deployment. The use and performance of AEDs and the effect on patient outcomes is assessed.
2. Methods Following Human Research Ethics Committee approval, the study was conducted at St Vincent’s Hospital Melbourne, Australia, a university-affiliated hospital which had approximately 300 acute and 80 subacute inpatient beds for adults. Each year at the hospital there were over 40,000 admitted patient episodes and over 30,000 emergency department presentations. There were two types of medical emergency response at the hospital: a ‘Respond MET’ was available for inpatients that were displaying serious (but non-arrest) signs and symptoms and a ‘Respond Blue’ was available to assist patients suffering cardiac arrest, respiratory arrest or a threatened airway. The Medical Emergency Team (MET) consisted of a medical registrar, intensive care registrar and a senior intensive care nurse. The Respond Blue team consisted of the MET personnel, an anaesthetic registrar and, for subacute patient areas, an emergency registrar. Australian Resuscitation Council algorithms for basic life support and advanced life support were followed. In May 2007, 18 AEDs (Heartstart FR2+, model M3860A, Philips Medical Systems, Seattle, Washington, USA) were purchased. These were stand-alone devices, not manual defibrillators with AED capability. The AEDs permitted biphasic waveform defibrillation at a fixed energy level of 150 J, had an ECG display that allowed basic rhythm interpretation and had manual override capability to permit manual defibrillation. The AEDs had a reported sensitivity (proportion of shockable rhythms correctly identified) of ≥67% for VT and ≥87% for VF and a reported specificity (proportion of nonshockable rhythms correctly identified) of ≥97% for sinus rhythm, ≥92% for asystole and ≥88% for other rhythms.16 The AEDs were programmed to perform automated rhythm analysis when first applied to a patient and thereafter at 2-min intervals. For shockable rhythms, the AEDs gave a single automated shock (i.e. not ‘stacked’ shocks) for each attempt at defibrillation. The AEDs were also programmed so that, for the first automated rhythm analysis, if a shockable rhythm was detected an automated shock would not be given if the device determined the rhythm characteristics were such that Return of Spontaneous Circulation (ROSC) after an attempt at defibrillation was unlikely, a feature the manufacturer called ‘Smart CPR’. 16 Software was used that enabled collection of electrocardiographic (ECG) and event data (e.g. the time of application of electrode pads and shock advice) stored by the AEDs (Event Review Pro 3.5.1, Philips Medical Systems, Seattle, Washington, USA). On 8 November 2007, the AEDs were deployed to the general patient care areas of the hospital including dialysis, medical imaging, the rehabilitation ward and the acute inpatient wards. AEDs completely replaced manual defibrillators in these areas. Little continuous cardiac monitoring was available in these areas and most first responders could not perform manual defibrillation. Manual defibrillators were retained in the emergency department, operating theatres, intensive care, coronary care, the cardiology and cardiothoracic wards and the cardiac catheterisation laboratories. In the 3 months before the AED deployment, a staff education programme was undertaken and this provided over 80% of clinical
staff with instruction in the use of AEDs. The use of AEDs was then incorporated into the hospital’s usual resuscitation training programmes. Prospective, Utstein-style17 data were gathered by intensive care staff in relation to all cardiac arrests that occurred at the hospital and were associated with a medical emergency call. Data were checked against a log of medical emergency calls from the hospital paging system to ensure that all cases of cardiac arrest associated with an emergency call were captured. Data collection was overseen by the medical director of intensive care (JDS). Cardiac arrest was deemed to have occurred when a patient was treated with chest compressions or electrical defibrillation. We analysed the data for cardiac arrests that occurred in areas where AEDs were deployed, for the 3 years before and the 3 years after the deployment, i.e. the 6-year period from 8 November 2004 to 7 November 2010. The patient’s primary treating unit at the time of arrest was classified as either medical or surgical. Medical units included general medicine, gastroenterology, haematology, neurology, renal and stroke. Surgical units included breast and endocrine, colorectal, hepatobiliary, orthopaedics, neurosurgery and vascular. Predicted hospital mortality for acute inpatients was calculated using a validated technique called the Hospital Outcome Prediction Equation, which is based on age, gender, hospital admission source, hospital admission urgency and hospital admission diagnosis.18 The ‘handsoff’ time with initial use of the AEDs was calculated from the time the device recorded that electrode pads were connected until the device began timing the 2-min period for delivery of chest compressions and rescue breathing. This incorporates the time taken for automated rhythm analysis and, when necessary, charging the AED and discharging a shock. Statistical analysis was performed using the statistical software programme Stata, Version 11.0 (StataCorp, College Station, Texas, USA). Categorical variables were reported as counts and proportions. Normally distributed continuous variables were presented as mean and standard deviation, while non-normal variables were presented as median and inter-quartile range. Differences in ROSC and survival to hospital discharge between the pre-AED and AED periods were compared using Pearson’s chi-squared. Reported Pvalues were two-sided and values <0.05 were taken to signify statistical significance. Multivariate logistic regression was used to determine the adjusted odds ratio of survival to hospital discharge after cardiac arrest when an AED was available. A stepwise backward elimination procedure was used to select variables.19 The following pre-specified variables that were considered to be potential confounders were included and then, beginning with the least important explanatory variable, omitted in a stepwise fashion if their P-value was greater than 0.20: age, gender, predicted mortality, primary treating unit (medical or surgical) at the time of arrest, the shock eligibility of the initial rhythm (shockable or not shockable), and whether the onset of arrest was witnessed or not witnessed. A second regression model, using the same methodology, was constructed that tested the association between use (rather than availability) of an AED and hospital survival.
3. Results 3.1. Numbers of cardiac arrests and patients There were 166 cardiac arrests in the AED areas, 82 during the 3 years before the AED deployment (pre-AED period) and 84 during the 3 years after the deployment (AED period). The 166 cardiac arrests involved 162 different patients: during their hospital stay 158 patients had 1 arrest and; 4 patients had 2 arrests.
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542 Table 1 Characteristics of patients at the time of cardiac arrest. Characteristic
Pre-AED perioda (n = 82)
AED periodb (n = 84)
Age in years—median (IQR) Male Primary treating unit medical Predicted risk of death (HOPE)—median (IQR)
73 (61–81) 50 (61%) 58 (71%)
74 (63–82) 51 (61%) 52 (62%)
8.8% (4.5–20.2%)
8.3% (3.5–17.2%)
AED: automated external defibrillator. IQR: inter-quartile range. HOPE: hospital outcome prediction equation. Values are number and (percentage), unless specified. a 8 November 2004–7 November 2007. b 8 November 2007–7 November 2010.
1539
Table 4 Univariate logistic regression models of survival to hospital discharge for prespecified patient and event characteristics. Variable
Odds ratio
95% confidence interval
P
Age in years Male Predicted risk of death (HOPE) Primary treating unit: surgical Witnessed arrest Shockable initial rhythm AED available AED applied
1.01 1.18 2.69 2.17 3.08 4.61 1.26 1.08
0.98–1.05 0.54–2.61 0.13–55.32 1.00–4.73 1.29–7.33 1.87–11.35 0.58–2.71 0.50–2.33
0.40 0.68 0.52 0.05 0.01 0.001 0.56 0.84
HOPE: hospital outcome prediction equation. AED: automated external defibrillator. Some patients (4 out of 162) had more than one cardiac arrest during their hospital stay: the models were constructed on variables at the time of the first cardiac arrest during the hospital stay.
Table 2 Characteristics of cardiac arrest events. Characteristic
Pre-AED perioda (n = 82)
AED periodb (n = 84)
AED applied Witnessed Location Acute inpatient ward Medical imaging Other AED area Initial rhythm Asystole PEA VT or VF Not known
0 47 (57%)
77 (92%) 47 (56%)
64 (78%) 9 (11%) 9 (11%)
77 (92%) 5 (6%) 2 (2%)
32 (39%) 35 (43%) 13 (16%) 2 (2%)
25 (30%) 46 (55%) 13 (15%) 0
AED: automated external defibrillator. PEA: pulseless electrical activity. VT: pulseless ventricular tachycardia. VF: ventricular fibrillation. Values are number and (percentage). a 8 November 2004–7 November 2007. b 8 November 2007–7 November 2010.
3.2. Patient characteristics Table 1 shows the characteristics of patients at the time of cardiac arrest for the periods before and after the deployment of AEDs. Patients in the pre-AED period were similar to those in the AED period with respect to age (median 73 years [61–81] vs. median 74 [63–82]), the proportion of males (61% vs. 61%) and the proportion who were under the care of a medical unit (71% vs. 62%). The groups were also similar in predicted mortality (median 8.8% [4.5–20.2%] vs. median 8.3% [3.5–17.2%]).
3.3. Event characteristics Table 2 shows the characteristics of cardiac arrest events during the pre-AED and AED periods. The proportion of cases where the onset of the cardiac arrest was witnessed was very similar in the pre-AED and AED periods (57% vs. 56%). Cases in the pre-AED and AED periods were also similar in relation to the locations of the events, with an acute inpatient ward being the most common site (78% vs. 92%). The initial rhythm was shockable in 16% of cases during the pre-AED period and in 15% during the AED period. 3.4. ROSC and hospital survival Table 3 shows ROSC for cardiac arrest cases in the pre-AED and AED periods. The proportion with ROSC was higher in the AED period for all categories of initial rhythm (asystole, pulseless electrical activity, and VT or VF). When the data for all rhythm categories were considered together, the proportion with ROSC in the AED period was significantly higher (35% vs. 54%, P = 0.02). Table 3 also shows survival to hospital discharge for patients in the pre-AED and AED periods. Hospital survival was similar in each period for each category of initial rhythm. When the data for all rhythm categories were considered together, the proportion of hospital survivors was very similar in the pre-AED and AED periods (19% vs. 22%, P = 0.56). 3.5. Adjusted hospital survival and AEDs Table 4 shows univariate logistic regression models of hospital survival for patient and event variables including availability of an
Table 3 Return of spontaneous circulation and survival to hospital discharge by initial cardiac arrest rhythm. Outcome
Pre-AED perioda
AED periodb
Return of spontaneous circulation All rhythms Asystole PEA VT or VF Hospital survivalc All rhythms Asystole PEA VT or VF
(n = 82 cardiac arrests) 29/82 (35%) 7/32 (22%) 16/35 (46%) 6/13 (46%) (n = 81 patients) 15/81 (19%) 3/32 (9%) 7/34 (21%) 5/13 (38%)
(n = 84 cardiac arrests) 45/84 (54%) 10/25 (40%) 25/46 (54%) 10/13 (77%) (n = 81 patients) 18/81 (22%) 2/25 (8%) 9/43 (21%) 7/13 (54%)
P 0.02 0.14 0.44 0.11 0.56 0.86 0.97 0.43
AED: automated external defibrillator. PEA: pulseless electrical activity. VT: pulseless ventricular tachycardia. VF: ventricular fibrillation. Values are number/total number and (percentage). P-Values are for differences between patients in the pre-AED and AED periods assessed by Pearson’s chi-squared. a 8 November 2004–7 November 2007. b 8 November 2007–7 November 2010. c Some patients (4 out of 162) had more than one cardiac arrest during their hospital stay: hospital survival is shown for only the initial rhythm of the first cardiac arrest during the hospital stay.
1540
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
Table 5 Multivariate logistic regression models of survival to hospital discharge when an AED was available and when an AED was applied; N = 160. Variable
Odds ratio
AED available Primary treating unit: surgical Witnessed arrest Shockable initial rhythm Hosmer–Lemeshow P = 0.39. Area under ROC curve = 0.74 AED applied Primary treating unit: surgical Witnessed arrest Shockable initial rhythm Hosmer–Lemeshow P = 0.21. Area under ROC curve = 0.75
1.22 3.26 3.93 5.75
95% confidence interval 0.53–2.84 1.34–7.94 1.51–10.20 2.15–15.39
P 0.64 0.009 0.005 <0.001
1.05 3.29 3.89 5.71
0.45–2.44 1.36–7.99 1.50–10.10 2.14–15.24
0.91 0.008 0.005 0.001
AED: automated external defibrillator. ROC: receiver operating characteristic. Some patients (4 out of 162) had more than one cardiac arrest during their hospital stay: the models were constructed on variables at the time of the first cardiac arrest during the hospital stay. Two cases for which the initial rhythm was not known were excluded.
AED and use of an AED. The multivariate regression models of survival to hospital discharge are shown in Table 5. After controlling for the shock eligibility of the initial rhythm, the patient’s primary treating unit and whether the onset of cardiac arrest was witnessed, the odds ratio for hospital survival when an AED was available was 1.22 (95% CI 0.53–2.84, P = 0.64). After controlling for the shock eligibility of the initial rhythm, the patient’s primary treating unit and whether the onset of cardiac arrest was witnessed, the odds ratio for hospital survival when an AED was applied was 1.05 (95% CI 0.45–2.44, P = 0.91).
3.6. Utilisation of AEDs AEDs were applied in 77 of 84 (92%) cases where an AED was available. Of the 7 cases where an AED was available but not used: in 5 cases advanced life support providers who were capable of performing manual defibrillation were already present at the onset of cardiac arrest; in 1 case ROSC occurred less than one minute after commencing chest compressions, making the AED redundant and; in 1 case it is not known why the AED was not applied.
3.7. Hands-off time with AEDs ECG and event data stored by the AEDs were examined in relation to the initial use of the AEDs for 72 of 77 (94%) cases where AEDs were applied. Of the 5 cases where this data was not examined: in 3 cases it was lost during device maintenance; in 1 case the AED was immediately placed into manual mode so no automated rhythm analysis was performed and; in 1 case the data was uninterpretable. In cases where an automated shock was delivered, the median time from connection of electrode pads to resumption of chest compressions was 23 s (22–24 s). In cases where an automated shock was not given, the median time from connection of electrode pads to resumption of chest compressions was 12 s (12–12 s); for 15% of these patients this interval was 20 s or longer.
3.8. Shock eligibility and automated shock delivery No automated shocks were observed for initial rhythms that were not shockable. An automated shock was delivered in 8 of 13 (62%) cases where the initial rhythm was potentially shockable. Of the 5 cases where no automated shock was given: in 1 case the arrhythmia self-reverted while the device was charging causing the device to ‘dump’ the charge; in 2 cases the arrhythmia was VT and; in 2 cases the rhythm was low amplitude VF. Fig. 1 shows the initial ECG and event data recorded by the AED for one of these cases of VT.
4. Discussion AEDs replaced manual defibrillators in the general patient care areas of a teaching hospital. Most first responders in these areas could not perform manual defibrillation. We analysed prospectively collected cardiac arrest data for the 3 years before and the 3 years after the AED deployment. The analysis included ECG and event data captured by the AEDs. Patient and event characteristics were similar in the periods before and after the AED deployment. ROSC was higher for cases in the AED period but this did not translate into an improvement in survival to hospital discharge. After adjusting for potential confounders, neither the availability of an AED or the use of an AED was associated with survival to hospital discharge. The proportion of patients who survived to hospital discharge was 20% and, while this may appear disappointing, it is similar to rates reported by others.9,10,14 Using a validated technique to determine predicted hospital mortality risk based on hospital admission characteristics, we found the cardiac arrest patients had a median risk of 8.4% and more than a quarter of the patients had a risk greater than 18%. This compares to a representative cohort of acute hospital inpatients in Victoria, Australia, which had a predicted hospital mortality risk under 3%.18 A high level of underlying morbidity is probably an important factor in the poor outcomes observed after in-hospital cardiac arrest. The primary purpose of AEDs is to shorten the time to defibrillation, but the proportion of in-hospital cardiac arrest patients who can potentially benefit from prompt defibrillation is small. Of the cardiac arrest cases where an AED was available, in only 15% was the initial rhythm potentially shockable. Furthermore, the AEDs in this study performed in a manner in keeping with the manufacturer’s specifications and did not shock all potentially shockable rhythms. We did not measure the response times of advanced life support providers to the scenes of the cardiac arrests but estimate it was less than 5 min for most cases and in some cases advanced life support providers were present at the onset of the arrest. Cases certainly occurred where patients received automated defibrillation before the arrival of advanced life support providers, but the extent to which the time to defibrillation was shortened by use of the AED may have only been marginal. We measured the interruption to chest compressions that occurred during automated management of the initial rhythm by examining the ECG and event data captured by the AEDs. In cases where an automated shock was delivered, the median hands-off time was 23 s and, for cases where an automated shock was not given, the median was 12 s. Previous work has found that AEDs are associated with longer pauses in chest compressions than manual defibrillators.12,13 Our study has limitations beyond its single-centre, before-andafter design. First, the relatively small number of patients means
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
1541
Fig. 1. Electrocardiograph and event data recorded by an automated external defibrillator (AED) for a broad QRS complex tachycardia (presumed ventricular tachycardia) in the setting of cardiac arrest: ‘no shock’ was advised by the AED.
a clinically important association between AED use and survival to hospital discharge may be present but we were unable to detect this association. Second, the cardiac arrest events occurred over a 6year period. Although survival to hospital discharge was adjusted for potential confounders, it is difficult to control for the evolution in care over such a period. For example, we have observed an increase in the intensity of medical emergency calls at the hospital as well as a reduction in the hospital’s mortality and cardiac arrest rates.20 Third, for patients in the AED period ECG data was available from the AED to assist with classification of the initial rhythm, where as ECG data of the initial arrest rhythm was rarely available for patients in the pre-AED period. This might have created a bias in the classification of the initial rhythm. If such a bias occurred we believe it is more likely to have affected the categorisation of the non-shockable rhythms into asystole or pulseless electrical activity, rather than the classification of rhythms as shockable or non-shockable. Finally, we only calculated the hands-off time for patients managed with an AED so we cannot comment on the hands-off time when an AED was not used. Further to this, we only estimated the hands-off time attributable to the AEDs and did not investigate human factors that may also have affected the hands-off time.
5. Conclusions The availability of AEDs was not independently associated with survival to hospital discharge after in-hospital cardiac arrest in
the general patient care areas of an Australian teaching hospital. Only a small proportion of patients had an initial cardiac arrest rhythm that was shockable and, in keeping with the manufacturer’s specifications, the AEDs did not shock all potentially shockable rhythms. No inappropriate shocks were observed. The hands-off time associated with automated rhythm management was considerable. Further work is needed to define the role of AEDs in the hospital setting. Conflict of interest statement No conflicts of interest to declare. Acknowledgements We thank the staff of St Vincent’s Hospital Melbourne for their commitment to the pursuit of optimal resuscitation outcomes. We are very grateful to Mr David Reid for his assistance calculating predicted mortality using the HOPE method. References 1. Australian Resuscitation Council. Foreword to guidelines: glossary of terms. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/). 2. Australian Resuscitation Council. Guideline 7: automated external defibrillation in basic life support. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/).
1542
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
3. Australian Resuscitation Council. Guideline 8: cardiopulmonary resuscitation. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/). 4. Australian Resuscitation Council. Guideline 11.4: electrical therapy for adult advance life support. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/). 5. Australian Resuscitation Council. Summary of major changes to ALS guidelines. Australian Resuscitation Council Guidelines. Melbourne: ARC; November 2010. (Accessed 17 March 2011, at http://www.resus.org.au/ files/summary of major changes to als guidelinesmar2011.pdf). 6. Australian Resuscitation Council. Summary of major changes to BLS guidelines. Australian Resuscitation Council Guidelines. Melbourne: ARC; November 2010. Available at http://www.resus.org.au/files/summary-of-majorchangestobls-guidelines-nov10.pdf. 7. Valenzuela TD, Roe DJ, Nichol G, et al. Outcomes of rapid defibrillation by security officers after cardiac arrest in casinos. N Engl J Med 2000;343:1206–9. 8. Hallstrom AP, Ornato JP, Weisfeldt M, et al. Public-access defibrillation and survival after out-of-hospital cardiac arrest. N Engl J Med 2004;351:637–46. 9. Forcina MS, Farhat AY, O’Neil WW, Haines DE. Cardiac arrest survival after implementation of automated external defibrillator technology in the in-hospital setting. Crit Care Med 2009;37:1229–36. 10. Chan PS, Krumholz HM, Spertus JA, et al. Automated external defibrillators and survival after in-hospital cardiac arrest. JAMA 2010;304:2129–36. 11. Haines DE. Automated external defibrillators and the law of unintended consequences. JAMA 2010;304:2178–9. 12. Kramer-Johansen J, Edelson DP, Abella BS, et al. Pauses in chest compression and inappropriate shocks: a comparison of manual and semi-automatic defibrillation attempts. Resuscitation 2007;73:212–20.
13. Pytte M, Pedersen TE, Ottem J, et al. Comparison of the hands-off time during CPR with manual and semi-automatic defibrillation in a manikin model. Resuscitation 2007;73:131–6. 14. Peberdy MA, Kaye W, Ornato JP, et al. Cardiopulmonary resuscitation of adults in the hospital: a report of 14720 cardiac arrests from the National Registry of Cardiopulmonary Resuscitation. Resuscitation 2003;58:297– 308. 15. Smith RJ, Hickey BB, Santamaria JD. Automated external defibrillators and survival after in-hospital cardiac arrest: early experience at an Australian teaching hospital. Crit Care Resusc 2009;11:261–5. 16. Philips Medical Systems. Heartstart FR2+ defibrillator: instructions for use, M3860A, M3861A. 12th ed. Philips Electronics North America Corp.; 2006. 17. Cummins RO, Chamberlain D, Hazinski MF, et al. Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation: the in-hospital ‘Utstein style’. A statement for healthcare professionals from the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa. Circulation 1997;95: 2213–39. 18. Duke GJ, Graco M, Santamaria J, Shann F. Validation of the hospital outcome prediction equation (HOPE) model for monitoring clinical performance. Intern Med J 2009;39:283–9. 19. Kirkwood BR, Sterne JAC. Essential medical statistics. 2nd ed. Melbourne: Blackwell Science Ltd.; 2003. 20. Santamaria J, Tobin A, Holmes J. Changing cardiac arrest and hospital mortality rates through a medical emergency team takes time and constant review. Crit Care Med 2010;38:445–50.
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Automated external defibrillators and in-hospital cardiac arrest: Patient survival and device performance at an Australian teaching hospital夽 Roger J. Smith ∗ , Bernadette B. Hickey a , John D. Santamaria b St Vincent’s Hospital Melbourne, Intensive Care, Inpatient Services Building, Level 1, St Vincent’s Hospital Melbourne, 41 Victoria Parade, Fitzroy, VIC 3065, Australia
a r t i c l e
i n f o
Article history: Received 17 April 2011 Accepted 21 June 2011
Keywords: Automated external defibrillator In-hospital cardiac arrest Cardiopulmonary resuscitation Defibrillation
a b s t r a c t Aims: To evaluate the effect of automated external defibrillators (AEDs) on patient survival and to describe the performance of AEDs after in-hospital cardiac arrest. Methods: Prospectively collected data were analysed for cardiac arrests in the general patient care areas of a teaching hospital during the 3 years before and the 3 years after the deployment of AEDs. The association between availability of an AED and survival to hospital discharge was assessed using multivariate logistic regression. AED performance during automated management of the initial rhythms was assessed using information captured by the AEDs. Results: There were 84 cardiac arrests in the AED period and 82 in the pre-AED period. Patient and event characteristics were similar in each period. The initial rhythm was shockable in 16% of cases. Return of spontaneous circulation was higher in the AED period (54% vs. 35%, P = 0.02) but the proportion of hospital survivors in each period was similar (22% vs. 19%, P = 0.56). The adjusted odds ratio for hospital survival when an AED was available was 1.22 (95% CI 0.53–2.84, P = 0.64). An AED was applied in 77/84 (92%) possible cases. Median interruption to chest compressions was 12 s (inter-quartile range 12–13). An automated shock was delivered in 8/13 (62%) possible cases. Conclusions: Availability of AEDs was not independently associated with hospital survival. Shockable presenting rhythms were not common and, in keeping with the manufacturer’s specifications, the AEDs did not shock all potentially shockable rhythms. The hands-off time associated with automated rhythm management was considerable. © 2011 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Automated external defibrillators (AEDs) can analyse the cardiac rhythm, charge automatically if a shockable rhythm (i.e. ventricular fibrillation [VF] or ventricular tachycardia [VT]) is recognised and provide the operator with audible and/or visual prompts for the safe delivery of an electrical shock.1 According to the recently updated guidelines of the Australian Resuscitation Council, the use of AEDs as a component of managing in-hospital cardiac arrest is acceptable.2–6 The benefits to patients of using AEDs during cardiac arrests in certain out-of-hospital settings have been demonstrated.7,8 However, there has never been a randomised controlled trial of AEDs
夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2011.06.025. ∗ Corresponding author. Tel.: +61 3 9288 3972/2211; fax: +61 3 9288 4487. E-mail addresses: [email protected] (R.J. Smith), [email protected] (B.B. Hickey), [email protected] (J.D. Santamaria). a Tel.: +61 3 9288 4576/2211; fax: +61 3 9288 4487. b Tel.: +61 3 9288 4488; fax: +61 3 9288 4487. 0300-9572/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2011.06.025
for in-hospital cardiac arrest and recent observational studies have cast doubt on the effectiveness of AEDs in hospitals. A single-centre study found that replacing manual defibrillators with AEDs made no difference to survival to hospital discharge after in-hospital cardiac arrest when the initial rhythm was shockable (31% vs. 29%, P = 0.80), and survival to hospital discharge was significantly reduced if the rhythm was not shockable (15% vs. 23%, P = 0.04).9 A study of almost 12,000 patients from over 200 hospitals found AED use during in-hospital cardiac arrest was independently associated with a reduction in survival to hospital discharge (16.3% vs. 19.3%; adjusted rate ratio 0.85, P < 0.001).10 The accompanying editorial advocated a cautious approach to introducing AEDs into the hospital setting.11 There are factors that may count against the use of AEDs for in-hospital cardiac arrest. Automated rhythm management is associated with longer interruptions to chest compressions compared to non-automated management12,13 and, in out-of-hospital settings where AEDs have improved patient outcomes, the initial cardiac arrest rhythm was shockable in more than half the cases and the response times of advanced life support providers were relatively long,7,8 but in the hospital setting the initial rhythm is shockable in only about one out of five cases and the
1538
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
response times of advanced life support providers are relatively fast.9,10,14,15 We have previously reported that there was no change in survival to hospital discharge after in-hospital cardiac arrest in the first year following the deployment of AEDs to the general patient care areas of our hospital.15 We now present cardiac arrest data for the 3 years before and the 3 years after the AED deployment. The use and performance of AEDs and the effect on patient outcomes is assessed.
2. Methods Following Human Research Ethics Committee approval, the study was conducted at St Vincent’s Hospital Melbourne, Australia, a university-affiliated hospital which had approximately 300 acute and 80 subacute inpatient beds for adults. Each year at the hospital there were over 40,000 admitted patient episodes and over 30,000 emergency department presentations. There were two types of medical emergency response at the hospital: a ‘Respond MET’ was available for inpatients that were displaying serious (but non-arrest) signs and symptoms and a ‘Respond Blue’ was available to assist patients suffering cardiac arrest, respiratory arrest or a threatened airway. The Medical Emergency Team (MET) consisted of a medical registrar, intensive care registrar and a senior intensive care nurse. The Respond Blue team consisted of the MET personnel, an anaesthetic registrar and, for subacute patient areas, an emergency registrar. Australian Resuscitation Council algorithms for basic life support and advanced life support were followed. In May 2007, 18 AEDs (Heartstart FR2+, model M3860A, Philips Medical Systems, Seattle, Washington, USA) were purchased. These were stand-alone devices, not manual defibrillators with AED capability. The AEDs permitted biphasic waveform defibrillation at a fixed energy level of 150 J, had an ECG display that allowed basic rhythm interpretation and had manual override capability to permit manual defibrillation. The AEDs had a reported sensitivity (proportion of shockable rhythms correctly identified) of ≥67% for VT and ≥87% for VF and a reported specificity (proportion of nonshockable rhythms correctly identified) of ≥97% for sinus rhythm, ≥92% for asystole and ≥88% for other rhythms.16 The AEDs were programmed to perform automated rhythm analysis when first applied to a patient and thereafter at 2-min intervals. For shockable rhythms, the AEDs gave a single automated shock (i.e. not ‘stacked’ shocks) for each attempt at defibrillation. The AEDs were also programmed so that, for the first automated rhythm analysis, if a shockable rhythm was detected an automated shock would not be given if the device determined the rhythm characteristics were such that Return of Spontaneous Circulation (ROSC) after an attempt at defibrillation was unlikely, a feature the manufacturer called ‘Smart CPR’. 16 Software was used that enabled collection of electrocardiographic (ECG) and event data (e.g. the time of application of electrode pads and shock advice) stored by the AEDs (Event Review Pro 3.5.1, Philips Medical Systems, Seattle, Washington, USA). On 8 November 2007, the AEDs were deployed to the general patient care areas of the hospital including dialysis, medical imaging, the rehabilitation ward and the acute inpatient wards. AEDs completely replaced manual defibrillators in these areas. Little continuous cardiac monitoring was available in these areas and most first responders could not perform manual defibrillation. Manual defibrillators were retained in the emergency department, operating theatres, intensive care, coronary care, the cardiology and cardiothoracic wards and the cardiac catheterisation laboratories. In the 3 months before the AED deployment, a staff education programme was undertaken and this provided over 80% of clinical
staff with instruction in the use of AEDs. The use of AEDs was then incorporated into the hospital’s usual resuscitation training programmes. Prospective, Utstein-style17 data were gathered by intensive care staff in relation to all cardiac arrests that occurred at the hospital and were associated with a medical emergency call. Data were checked against a log of medical emergency calls from the hospital paging system to ensure that all cases of cardiac arrest associated with an emergency call were captured. Data collection was overseen by the medical director of intensive care (JDS). Cardiac arrest was deemed to have occurred when a patient was treated with chest compressions or electrical defibrillation. We analysed the data for cardiac arrests that occurred in areas where AEDs were deployed, for the 3 years before and the 3 years after the deployment, i.e. the 6-year period from 8 November 2004 to 7 November 2010. The patient’s primary treating unit at the time of arrest was classified as either medical or surgical. Medical units included general medicine, gastroenterology, haematology, neurology, renal and stroke. Surgical units included breast and endocrine, colorectal, hepatobiliary, orthopaedics, neurosurgery and vascular. Predicted hospital mortality for acute inpatients was calculated using a validated technique called the Hospital Outcome Prediction Equation, which is based on age, gender, hospital admission source, hospital admission urgency and hospital admission diagnosis.18 The ‘handsoff’ time with initial use of the AEDs was calculated from the time the device recorded that electrode pads were connected until the device began timing the 2-min period for delivery of chest compressions and rescue breathing. This incorporates the time taken for automated rhythm analysis and, when necessary, charging the AED and discharging a shock. Statistical analysis was performed using the statistical software programme Stata, Version 11.0 (StataCorp, College Station, Texas, USA). Categorical variables were reported as counts and proportions. Normally distributed continuous variables were presented as mean and standard deviation, while non-normal variables were presented as median and inter-quartile range. Differences in ROSC and survival to hospital discharge between the pre-AED and AED periods were compared using Pearson’s chi-squared. Reported Pvalues were two-sided and values <0.05 were taken to signify statistical significance. Multivariate logistic regression was used to determine the adjusted odds ratio of survival to hospital discharge after cardiac arrest when an AED was available. A stepwise backward elimination procedure was used to select variables.19 The following pre-specified variables that were considered to be potential confounders were included and then, beginning with the least important explanatory variable, omitted in a stepwise fashion if their P-value was greater than 0.20: age, gender, predicted mortality, primary treating unit (medical or surgical) at the time of arrest, the shock eligibility of the initial rhythm (shockable or not shockable), and whether the onset of arrest was witnessed or not witnessed. A second regression model, using the same methodology, was constructed that tested the association between use (rather than availability) of an AED and hospital survival.
3. Results 3.1. Numbers of cardiac arrests and patients There were 166 cardiac arrests in the AED areas, 82 during the 3 years before the AED deployment (pre-AED period) and 84 during the 3 years after the deployment (AED period). The 166 cardiac arrests involved 162 different patients: during their hospital stay 158 patients had 1 arrest and; 4 patients had 2 arrests.
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542 Table 1 Characteristics of patients at the time of cardiac arrest. Characteristic
Pre-AED perioda (n = 82)
AED periodb (n = 84)
Age in years—median (IQR) Male Primary treating unit medical Predicted risk of death (HOPE)—median (IQR)
73 (61–81) 50 (61%) 58 (71%)
74 (63–82) 51 (61%) 52 (62%)
8.8% (4.5–20.2%)
8.3% (3.5–17.2%)
AED: automated external defibrillator. IQR: inter-quartile range. HOPE: hospital outcome prediction equation. Values are number and (percentage), unless specified. a 8 November 2004–7 November 2007. b 8 November 2007–7 November 2010.
1539
Table 4 Univariate logistic regression models of survival to hospital discharge for prespecified patient and event characteristics. Variable
Odds ratio
95% confidence interval
P
Age in years Male Predicted risk of death (HOPE) Primary treating unit: surgical Witnessed arrest Shockable initial rhythm AED available AED applied
1.01 1.18 2.69 2.17 3.08 4.61 1.26 1.08
0.98–1.05 0.54–2.61 0.13–55.32 1.00–4.73 1.29–7.33 1.87–11.35 0.58–2.71 0.50–2.33
0.40 0.68 0.52 0.05 0.01 0.001 0.56 0.84
HOPE: hospital outcome prediction equation. AED: automated external defibrillator. Some patients (4 out of 162) had more than one cardiac arrest during their hospital stay: the models were constructed on variables at the time of the first cardiac arrest during the hospital stay.
Table 2 Characteristics of cardiac arrest events. Characteristic
Pre-AED perioda (n = 82)
AED periodb (n = 84)
AED applied Witnessed Location Acute inpatient ward Medical imaging Other AED area Initial rhythm Asystole PEA VT or VF Not known
0 47 (57%)
77 (92%) 47 (56%)
64 (78%) 9 (11%) 9 (11%)
77 (92%) 5 (6%) 2 (2%)
32 (39%) 35 (43%) 13 (16%) 2 (2%)
25 (30%) 46 (55%) 13 (15%) 0
AED: automated external defibrillator. PEA: pulseless electrical activity. VT: pulseless ventricular tachycardia. VF: ventricular fibrillation. Values are number and (percentage). a 8 November 2004–7 November 2007. b 8 November 2007–7 November 2010.
3.2. Patient characteristics Table 1 shows the characteristics of patients at the time of cardiac arrest for the periods before and after the deployment of AEDs. Patients in the pre-AED period were similar to those in the AED period with respect to age (median 73 years [61–81] vs. median 74 [63–82]), the proportion of males (61% vs. 61%) and the proportion who were under the care of a medical unit (71% vs. 62%). The groups were also similar in predicted mortality (median 8.8% [4.5–20.2%] vs. median 8.3% [3.5–17.2%]).
3.3. Event characteristics Table 2 shows the characteristics of cardiac arrest events during the pre-AED and AED periods. The proportion of cases where the onset of the cardiac arrest was witnessed was very similar in the pre-AED and AED periods (57% vs. 56%). Cases in the pre-AED and AED periods were also similar in relation to the locations of the events, with an acute inpatient ward being the most common site (78% vs. 92%). The initial rhythm was shockable in 16% of cases during the pre-AED period and in 15% during the AED period. 3.4. ROSC and hospital survival Table 3 shows ROSC for cardiac arrest cases in the pre-AED and AED periods. The proportion with ROSC was higher in the AED period for all categories of initial rhythm (asystole, pulseless electrical activity, and VT or VF). When the data for all rhythm categories were considered together, the proportion with ROSC in the AED period was significantly higher (35% vs. 54%, P = 0.02). Table 3 also shows survival to hospital discharge for patients in the pre-AED and AED periods. Hospital survival was similar in each period for each category of initial rhythm. When the data for all rhythm categories were considered together, the proportion of hospital survivors was very similar in the pre-AED and AED periods (19% vs. 22%, P = 0.56). 3.5. Adjusted hospital survival and AEDs Table 4 shows univariate logistic regression models of hospital survival for patient and event variables including availability of an
Table 3 Return of spontaneous circulation and survival to hospital discharge by initial cardiac arrest rhythm. Outcome
Pre-AED perioda
AED periodb
Return of spontaneous circulation All rhythms Asystole PEA VT or VF Hospital survivalc All rhythms Asystole PEA VT or VF
(n = 82 cardiac arrests) 29/82 (35%) 7/32 (22%) 16/35 (46%) 6/13 (46%) (n = 81 patients) 15/81 (19%) 3/32 (9%) 7/34 (21%) 5/13 (38%)
(n = 84 cardiac arrests) 45/84 (54%) 10/25 (40%) 25/46 (54%) 10/13 (77%) (n = 81 patients) 18/81 (22%) 2/25 (8%) 9/43 (21%) 7/13 (54%)
P 0.02 0.14 0.44 0.11 0.56 0.86 0.97 0.43
AED: automated external defibrillator. PEA: pulseless electrical activity. VT: pulseless ventricular tachycardia. VF: ventricular fibrillation. Values are number/total number and (percentage). P-Values are for differences between patients in the pre-AED and AED periods assessed by Pearson’s chi-squared. a 8 November 2004–7 November 2007. b 8 November 2007–7 November 2010. c Some patients (4 out of 162) had more than one cardiac arrest during their hospital stay: hospital survival is shown for only the initial rhythm of the first cardiac arrest during the hospital stay.
1540
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
Table 5 Multivariate logistic regression models of survival to hospital discharge when an AED was available and when an AED was applied; N = 160. Variable
Odds ratio
AED available Primary treating unit: surgical Witnessed arrest Shockable initial rhythm Hosmer–Lemeshow P = 0.39. Area under ROC curve = 0.74 AED applied Primary treating unit: surgical Witnessed arrest Shockable initial rhythm Hosmer–Lemeshow P = 0.21. Area under ROC curve = 0.75
1.22 3.26 3.93 5.75
95% confidence interval 0.53–2.84 1.34–7.94 1.51–10.20 2.15–15.39
P 0.64 0.009 0.005 <0.001
1.05 3.29 3.89 5.71
0.45–2.44 1.36–7.99 1.50–10.10 2.14–15.24
0.91 0.008 0.005 0.001
AED: automated external defibrillator. ROC: receiver operating characteristic. Some patients (4 out of 162) had more than one cardiac arrest during their hospital stay: the models were constructed on variables at the time of the first cardiac arrest during the hospital stay. Two cases for which the initial rhythm was not known were excluded.
AED and use of an AED. The multivariate regression models of survival to hospital discharge are shown in Table 5. After controlling for the shock eligibility of the initial rhythm, the patient’s primary treating unit and whether the onset of cardiac arrest was witnessed, the odds ratio for hospital survival when an AED was available was 1.22 (95% CI 0.53–2.84, P = 0.64). After controlling for the shock eligibility of the initial rhythm, the patient’s primary treating unit and whether the onset of cardiac arrest was witnessed, the odds ratio for hospital survival when an AED was applied was 1.05 (95% CI 0.45–2.44, P = 0.91).
3.6. Utilisation of AEDs AEDs were applied in 77 of 84 (92%) cases where an AED was available. Of the 7 cases where an AED was available but not used: in 5 cases advanced life support providers who were capable of performing manual defibrillation were already present at the onset of cardiac arrest; in 1 case ROSC occurred less than one minute after commencing chest compressions, making the AED redundant and; in 1 case it is not known why the AED was not applied.
3.7. Hands-off time with AEDs ECG and event data stored by the AEDs were examined in relation to the initial use of the AEDs for 72 of 77 (94%) cases where AEDs were applied. Of the 5 cases where this data was not examined: in 3 cases it was lost during device maintenance; in 1 case the AED was immediately placed into manual mode so no automated rhythm analysis was performed and; in 1 case the data was uninterpretable. In cases where an automated shock was delivered, the median time from connection of electrode pads to resumption of chest compressions was 23 s (22–24 s). In cases where an automated shock was not given, the median time from connection of electrode pads to resumption of chest compressions was 12 s (12–12 s); for 15% of these patients this interval was 20 s or longer.
3.8. Shock eligibility and automated shock delivery No automated shocks were observed for initial rhythms that were not shockable. An automated shock was delivered in 8 of 13 (62%) cases where the initial rhythm was potentially shockable. Of the 5 cases where no automated shock was given: in 1 case the arrhythmia self-reverted while the device was charging causing the device to ‘dump’ the charge; in 2 cases the arrhythmia was VT and; in 2 cases the rhythm was low amplitude VF. Fig. 1 shows the initial ECG and event data recorded by the AED for one of these cases of VT.
4. Discussion AEDs replaced manual defibrillators in the general patient care areas of a teaching hospital. Most first responders in these areas could not perform manual defibrillation. We analysed prospectively collected cardiac arrest data for the 3 years before and the 3 years after the AED deployment. The analysis included ECG and event data captured by the AEDs. Patient and event characteristics were similar in the periods before and after the AED deployment. ROSC was higher for cases in the AED period but this did not translate into an improvement in survival to hospital discharge. After adjusting for potential confounders, neither the availability of an AED or the use of an AED was associated with survival to hospital discharge. The proportion of patients who survived to hospital discharge was 20% and, while this may appear disappointing, it is similar to rates reported by others.9,10,14 Using a validated technique to determine predicted hospital mortality risk based on hospital admission characteristics, we found the cardiac arrest patients had a median risk of 8.4% and more than a quarter of the patients had a risk greater than 18%. This compares to a representative cohort of acute hospital inpatients in Victoria, Australia, which had a predicted hospital mortality risk under 3%.18 A high level of underlying morbidity is probably an important factor in the poor outcomes observed after in-hospital cardiac arrest. The primary purpose of AEDs is to shorten the time to defibrillation, but the proportion of in-hospital cardiac arrest patients who can potentially benefit from prompt defibrillation is small. Of the cardiac arrest cases where an AED was available, in only 15% was the initial rhythm potentially shockable. Furthermore, the AEDs in this study performed in a manner in keeping with the manufacturer’s specifications and did not shock all potentially shockable rhythms. We did not measure the response times of advanced life support providers to the scenes of the cardiac arrests but estimate it was less than 5 min for most cases and in some cases advanced life support providers were present at the onset of the arrest. Cases certainly occurred where patients received automated defibrillation before the arrival of advanced life support providers, but the extent to which the time to defibrillation was shortened by use of the AED may have only been marginal. We measured the interruption to chest compressions that occurred during automated management of the initial rhythm by examining the ECG and event data captured by the AEDs. In cases where an automated shock was delivered, the median hands-off time was 23 s and, for cases where an automated shock was not given, the median was 12 s. Previous work has found that AEDs are associated with longer pauses in chest compressions than manual defibrillators.12,13 Our study has limitations beyond its single-centre, before-andafter design. First, the relatively small number of patients means
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
1541
Fig. 1. Electrocardiograph and event data recorded by an automated external defibrillator (AED) for a broad QRS complex tachycardia (presumed ventricular tachycardia) in the setting of cardiac arrest: ‘no shock’ was advised by the AED.
a clinically important association between AED use and survival to hospital discharge may be present but we were unable to detect this association. Second, the cardiac arrest events occurred over a 6year period. Although survival to hospital discharge was adjusted for potential confounders, it is difficult to control for the evolution in care over such a period. For example, we have observed an increase in the intensity of medical emergency calls at the hospital as well as a reduction in the hospital’s mortality and cardiac arrest rates.20 Third, for patients in the AED period ECG data was available from the AED to assist with classification of the initial rhythm, where as ECG data of the initial arrest rhythm was rarely available for patients in the pre-AED period. This might have created a bias in the classification of the initial rhythm. If such a bias occurred we believe it is more likely to have affected the categorisation of the non-shockable rhythms into asystole or pulseless electrical activity, rather than the classification of rhythms as shockable or non-shockable. Finally, we only calculated the hands-off time for patients managed with an AED so we cannot comment on the hands-off time when an AED was not used. Further to this, we only estimated the hands-off time attributable to the AEDs and did not investigate human factors that may also have affected the hands-off time.
5. Conclusions The availability of AEDs was not independently associated with survival to hospital discharge after in-hospital cardiac arrest in
the general patient care areas of an Australian teaching hospital. Only a small proportion of patients had an initial cardiac arrest rhythm that was shockable and, in keeping with the manufacturer’s specifications, the AEDs did not shock all potentially shockable rhythms. No inappropriate shocks were observed. The hands-off time associated with automated rhythm management was considerable. Further work is needed to define the role of AEDs in the hospital setting. Conflict of interest statement No conflicts of interest to declare. Acknowledgements We thank the staff of St Vincent’s Hospital Melbourne for their commitment to the pursuit of optimal resuscitation outcomes. We are very grateful to Mr David Reid for his assistance calculating predicted mortality using the HOPE method. References 1. Australian Resuscitation Council. Foreword to guidelines: glossary of terms. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/). 2. Australian Resuscitation Council. Guideline 7: automated external defibrillation in basic life support. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/).
1542
R.J. Smith et al. / Resuscitation 82 (2011) 1537–1542
3. Australian Resuscitation Council. Guideline 8: cardiopulmonary resuscitation. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/). 4. Australian Resuscitation Council. Guideline 11.4: electrical therapy for adult advance life support. Australian Resuscitation Council Guidelines. Melbourne: ARC; December 2010. (Accessed 17 March 2011, at http://www.resus.org.au/). 5. Australian Resuscitation Council. Summary of major changes to ALS guidelines. Australian Resuscitation Council Guidelines. Melbourne: ARC; November 2010. (Accessed 17 March 2011, at http://www.resus.org.au/ files/summary of major changes to als guidelinesmar2011.pdf). 6. Australian Resuscitation Council. Summary of major changes to BLS guidelines. Australian Resuscitation Council Guidelines. Melbourne: ARC; November 2010. Available at http://www.resus.org.au/files/summary-of-majorchangestobls-guidelines-nov10.pdf. 7. Valenzuela TD, Roe DJ, Nichol G, et al. Outcomes of rapid defibrillation by security officers after cardiac arrest in casinos. N Engl J Med 2000;343:1206–9. 8. Hallstrom AP, Ornato JP, Weisfeldt M, et al. Public-access defibrillation and survival after out-of-hospital cardiac arrest. N Engl J Med 2004;351:637–46. 9. Forcina MS, Farhat AY, O’Neil WW, Haines DE. Cardiac arrest survival after implementation of automated external defibrillator technology in the in-hospital setting. Crit Care Med 2009;37:1229–36. 10. Chan PS, Krumholz HM, Spertus JA, et al. Automated external defibrillators and survival after in-hospital cardiac arrest. JAMA 2010;304:2129–36. 11. Haines DE. Automated external defibrillators and the law of unintended consequences. JAMA 2010;304:2178–9. 12. Kramer-Johansen J, Edelson DP, Abella BS, et al. Pauses in chest compression and inappropriate shocks: a comparison of manual and semi-automatic defibrillation attempts. Resuscitation 2007;73:212–20.
13. Pytte M, Pedersen TE, Ottem J, et al. Comparison of the hands-off time during CPR with manual and semi-automatic defibrillation in a manikin model. Resuscitation 2007;73:131–6. 14. Peberdy MA, Kaye W, Ornato JP, et al. Cardiopulmonary resuscitation of adults in the hospital: a report of 14720 cardiac arrests from the National Registry of Cardiopulmonary Resuscitation. Resuscitation 2003;58:297– 308. 15. Smith RJ, Hickey BB, Santamaria JD. Automated external defibrillators and survival after in-hospital cardiac arrest: early experience at an Australian teaching hospital. Crit Care Resusc 2009;11:261–5. 16. Philips Medical Systems. Heartstart FR2+ defibrillator: instructions for use, M3860A, M3861A. 12th ed. Philips Electronics North America Corp.; 2006. 17. Cummins RO, Chamberlain D, Hazinski MF, et al. Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation: the in-hospital ‘Utstein style’. A statement for healthcare professionals from the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa. Circulation 1997;95: 2213–39. 18. Duke GJ, Graco M, Santamaria J, Shann F. Validation of the hospital outcome prediction equation (HOPE) model for monitoring clinical performance. Intern Med J 2009;39:283–9. 19. Kirkwood BR, Sterne JAC. Essential medical statistics. 2nd ed. Melbourne: Blackwell Science Ltd.; 2003. 20. Santamaria J, Tobin A, Holmes J. Changing cardiac arrest and hospital mortality rates through a medical emergency team takes time and constant review. Crit Care Med 2010;38:445–50.