Improving Temporal Trends in Survival and Neurological Outcomes after Out-of-Hospital Cardiac Arrest

Abstract

Background

Considerable effort has gone into improving outcomes from out-of-hospital cardiac arrest (OHCA). Studies suggest that survival is improving; however, prior studies had insufficient data to pursue the relationship between markers of guideline compliance and temporal trends. The objective of the study was to evaluate trends in OHCA survival over an 8-year period that included the implementation of the 2005 and 2010 international CPR guidelines.

Methods and Results

This was a population based cohort study of all consecutive treated OHCA patients of presumed cardiac etiology between 2006 and 2013 in the City of Toronto, Canada and surrounding regions. Temporal changes were measured by chi-square trend test. The association between year of the OHCA and survival were evaluated using logistic regression and joinpoint analysis. A total of 23,619 patients with OHCA met study inclusion criteria. During the study period, survival to hospital discharge doubled (4.8% in 2006 to 9.4% in 2013; p<0.0001), and survival with good neurological outcome increased (6.2% in 2010 to 8.5% in 2013; p =0.005). Improvements occurred in the rates of bystander CPR and AED application, high quality CPR metrics, and in-hospital targeted temperature management. After adjusting for the Utstein variables, survival to hospital discharge (OR: 1.12; 95% CI: 1.09–1.15) and survival with good neurological outcome (OR: 1.13; 95% CI: 1.05–1.22) increased with each year of study.

Conclusions

Survival after OHCA has improved over time. This trend was associated with improved rates of bystander CPR, AED use, high quality CPR metrics, and in-hospital targeted temperature management. The results suggest that multiple factors, each improving over time, may have contributed to the observed increase in survival.

Each year an estimated 50,000 people suffer an out-of-hospital cardiac arrest (OHCA) in Canada.¹ Unfortunately, only 10% of patients survive to hospital discharge.¹ Extensive work has gone into improving OHCA outcomes,^2–4 including increasing community awareness programs, public safety initiatives, and dispatcher assistance at the time of the 911 call. In most regions, bystanders are being encouraged to start cardiopulmonary resuscitation (CPR) and apply an automated external defibrillator (AED) prior to paramedic arrival. Concurrently, advances in science have led to novel approaches aimed at improving outcomes such as performing high-quality CPR and the management of post-cardiac arrest syndrome.

A meta-analysis demonstrated that survival from OHCA has not improved significantly in nearly 30 years (1980–2008), with an aggregate survival between 6.7% and 8.4%.⁵ However, more recent work has cast doubt on this finding. Daya and colleagues showed that survival increased steadily over a 5-year period between 2006 and 2010 among regions participating in the Resuscitation Outcomes Consortium (8.2% to 10.4%).⁶ Similarly, increases in survival have been reported in Denmark,⁷ Sweden,⁸ the Netherlands,⁹ and among cities in the United States participating in the CARES registry.¹⁰

Most experts have attributed these improvements in survival to increased rates of bystander CPR or public AED use.^{7, 9, 10} However, although bystander CPR has been shown to improve survival, there are other important out-of-hospital and in-hospital interventions that are also associated with improved patient outcomes such as the performance of high quality CPR¹¹ and management of post-resuscitation care.¹² Yet few studies have evaluated trends in these potentially influential prehospital and in-hospital variables. Even fewer studies have evaluated temporal changes in neurological outcomes. We sought to evaluate the associations of OHCA treatment interventions by the bystander, the quality of out-of-hospital and in-hospital care with patient outcomes over a seven year period from 2006–2013.

METHODS

Study Design

This was a population based observational cohort study evaluating the temporal trends in survival after OHCA. The study protocol was based on the Resuscitation Outcomes Consortium Registry and Strategies for Post Arrest Care Network protocol which were both approved by the institutional research board. The study was designed and reported according to STROBE standards for cohort studies.¹³ The funders did not have any role in the study. The data, analytical methods and study material will not be made available to other researchers for purposes of reproducing the results or replicating the procedures. The data is governed by data sharing agreements and research board approvals which do not cover the release of the data.

Setting and Study Participants

We included consecutive patients who experienced an OHCA between January 1, 2006 and December 31, 2013. Data was obtained from the Toronto Regional RescuNET cardiac arrest database; Rescu Epistry, which is compliant with the Resuscitation Outcomes Consortium Epistry-Cardiac Arrest and based on the Strategies for Post Arrest Resuscitation Care (SPARC) methodologies that are described elsewhere.^{14, 15} The Rescu Epistry interface is a prospective population-based web registry of consecutive OHCA patients assessed by prehospital care providers. Data is entered into Rescu Epistry from dispatch, prehospital and in-hospital records after each OHCA, and includes patient identifiers, call characteristics, prehospital, and in-hospital interventions and patient outcomes.¹⁵ The catchment area includes 7 urban and rural regions in southern Ontario (Region of Durham, Halton Region, District of Muskoka, Region of Peel, Simcoe County, City of Toronto, York Region). Collectively, paramedics and first responders provide emergency care and transport to a population of 6.6 million people in both urban and rural settings within a geographic area of 17,000 km². Paramedics respond to over 7,000 OHCA per year and transport patients to 42 destination hospitals (both academic and community hospitals with varying size) (Appendix 2). Trained data abstractors manually entered all data with point of entry logic and error checks to minimize errors. CPR process recordings were downloaded directly from the individual defibrillators and uploaded to Rescu Epistry. Data abstractors manually abstracted CPR quality measures from the process recordings. Loss to follow rates for the primary outcome of survival to hospital discharge were <1%. Duplicate data abstraction occurred on a random sample of 10% of abstracted charts for each in-hospital data collector.¹⁴

Regional primary care paramedics can perform CPR and airway management using basic life support techniques or supraglottic airways. Prior to 2012, primary care paramedics performed semi-automated external defibrillation. From 2012 onwards, all were trained to perform manual defibrillation to shorten the pre-shock pause. First responders from local fire departments and advanced care paramedics were dispatched in a tiered fashion to all OHCA patients. Fire services in Ontario were equipped with AEDs and provided basic life support skills, while advanced care paramedics gained intravenous access, performed manual rhythm interpretations, provided advanced airway maneuvers, and administered advanced life support (ALS) medications. Paramedics followed the termination of resuscitation guideline throughout the duration of the study period.³

Eligibility Criteria

We included all adult patients aged ≥ 18 years who sustained a non-traumatic OHCA and were treated by paramedics within the study catchment area. All cardiac arrest patients were included in the model, regardless of any previous event. Each cardiac arrest was classified as one event. We excluded patients who met the criteria for obvious death (i.e., presence of rigor mortis, lividity, decapitation, hemi-section, or decomposition), had a valid “Do Not Resuscitate” (DNR) advanced directive, or had an arrest of obvious etiology (ex. overdose or asphyxiation).

Statistical Methods

We used descriptive statistics to assess the distribution of all variables of interest. Continuous variables were summarized as means and standard deviations, and categorical variables were summarized as counts and percentages. Variables were defined according to the Utstein recomendations.¹⁵ For CPR quality, we used the mean measure (depth, rate, fraction) for the first 5 minutes of the resuscitation, and excluded outliers (as defined by a data point that was greater than or less than one and half the interquartile range). CPR quality was recorded directly from the defibrillators. Only paramedic services using Zoll X series and E series defibrillators (Zoll Medical, Chelmsford, Massachusetts) had available CPR depth, rate and fraction measurements. CPR depth measurements were available from paramedic services using the LifePack 12 or LifePack 15 (PhysioControl, Redmond, Washington). The choice of defibrillator is at the discretion of each individual paramedic service. The yearly reported quality measure was the average value for that year. Adherence was defined as the percentage of total cases where the mean quality measure met the recommended guideline. Guideline adherence was defined in accordance with the 2010 American Heart Association recommendations (chest compression rate ≥ 100 compressions per minute and chest compression depth ≥ 5cm or 2 inches),¹⁶ or the 2015 AHA guidelines when no recommendation was made in 2010 (chest compression fraction ≥ 0.6 and pre-shock pause ≤ 10 seconds).¹¹ Chest compression fraction is defined as the percent of total resuscitation time that compressions were being performed, while pre-shock pause is defined as the time interval between chest compression cessation and shock delivery. Successful targeted temperature management was defined as a temperature ≤ 36°C (96.8°F) within 6 hours of emergency department arrival. Survival with good neurological outcome was collected after 2010, and defined as survival to hospital discharge with a Modified Rankin Score (MRS) ≤ 2.¹⁷

Temporal changes in categorical variables were analyzed using the Cochran-Armitage trend test, while changes in continuous variables were analyzed using an unadjusted linear regression model. For our outcome measures, the 95% confidence intervals around the survival rate were calculated. We used logistic regression analyses to determine the association between predictor variables including episode year and the outcomes of interest (survival to hospital discharge and survival with good neurological outcome). The following Utstein variables were selected a priori and included in the logistic regression: age, sex, initial cardiac rhythm (ventricular fibrillation/pulseless ventricular tachycardia, pulseless electrical activity and asystole), pick-up location, bystander CPR, witnessed status, response time, and episode year (as a continuous variable). Odds ratios (OR) (with 95% confidence intervals) were used to display the effect size of each independent variable. The following additional analyses were performed to evaluate the robustness of our results. We tested for any interaction between episode year and initial rhythm (ventricular fibrillation/pulseless ventricular tachycardia, pulseless electrical activity, or asystole), and if present tested whether temporal trends in survival varied by subgroup defined by initial rhythm. We tested the model to determine if our results were influenced by season. A post-hoc analysis was undertaken to determine the effect of clustering of cardiac arrests among paramedic services.

We used joinpoint analyses to test for any inflection points in temporal trends during the study period. We used the default settings, which allowed up to 3 joinpoints (4 distinct temporal trends), required at least 2 observations between joinpoints, and used a log-linear regression model to assess significant changes in trends over time.

Statistical analyses were carried out using SAS 9.3 (SAS Institute, Cary NC, USA) and joinpoint analyses were performed using Joinpoint Regression program version 4.3.1.0 (Statistical Research and Applications Branch, Surveillance Research Program, National Cancer Institute, Bethesda, Maryland). A two-sided p-value of <0.05 was considered statistically significant for all analyses.

RESULTS

Participants, Descriptive and Outcome Data

From January 1st 2006 to December 31st 2013, there were 44,595 patients who experienced an OHCA, of whom 25,719 (58%) were treated by paramedics. After excluding ineligible cases, 23,619 patients were included in the final primary analysis (Figure 1). There was an absolute increase in the number of OHCAs every year (1826 patients in 2006 to 3610 patients in 2013) as more paramedic services participated in Epistry. However, the percentage of treated patients decreased slightly from 2006 to 2013 (61% to 58%; p=0.03).

Figure 1.

CONSORT Diagram of Eligible Patients

Main Results

During the study, survival to hospital discharge increased from 4.8% in 2006 to 9.4% in 2013 (p < 0.001) (Table 1; Figure 2a). The joinpoint regression identified a linear increase in survival (approximately 0.6% per year) with no inflection points (Appendix 3). A significant interaction was found between year and initial rhythms of VF/VT (p=0.02) and PEA (p = 0.02) for overall survival to hospital discharge; however, there was no significant interaction between initial rhythm and year for survival with good neurological outcome. Increases in survival to hospital discharge were seen in patients whose initial rhythm was VF/VT (17% to 30%; p < 0.001; Figure 2b) and PEA (2.7% to 7.2%; p < 0.001; Figure 2b) as well as bystander witnessed VF/VT patients (16% to 31%; p < 0.001) and paramedic witnessed OHCA (15% to 18%; p = 0.002). There was no change in survival in patients whose initial rhythm was asystole (0.7% to 0.6%; p=0.66; Figure 2b). Survival with good neurological outcome increased from 6.1% in 2010 to 8.5% in 2013 (p<0.001) (Table 1; Figure 2c).

Table 1.

Patient Characteristics and Outcomes

	Overall	2006	2007	2008	2009	2010	2011	2012	2013	Trend Test p-value
Paramedic Assessed OHCA	40002	2999	4132	4755	5057	5501	5518	5842	6198
Paramedic Treated Cases	23619	1826	2463	2831	2955	3241	3245	3448	3610
Treated (% of all OHCA)	59.0	60.9	59.6	59.5	58.4	58.9	58.8	57.3	58.2	0.03
PREHOSPITAL EVENT CHARACTERISTICS
Age (mean +/− SD)	69.9 (16.0)	70.5 (15.5)	69.6 (15.8)	69.4 (16.0)	69.4 (16.2)	69.6 (16.2)	69.8 (15.8)	70.1 (15.8)	70.6 (16.1)	0.0473
Male Sex (%)	63.4	61.1	63.9	64.7	64.4	64.5	63.8	64.2	64.0	0.27
Response Time (mean +/− SD)	6.50 (3.24)	6.53 (3.38)	6.49 (3.56)	6.57 (3.91)	6.49 (3.72)	6.50 (3.00)	6.53 (2.77)	6.45 (2.83)	6.47 (2.77)	0.35
Public Pick Up Location (%)	15.3	16.1	18.0	16.1	15.1	14.8	14.8	13.8	14.7	<0.0001
Bystander (%)CPR*	39.7	35.7	36.5	37.1	38.7	40.3	40.8	41.6	43.7	p<0.001
Bystander AED applied (%)*	2.6	1.6	2.5	2.0	2.4	3.0	2.8	3.0	3.1	0.0002
Bystander AED shock delivered (%)*	42.9	42.3	45.8	35.4	33.9	40.7	45.6	53.4	40.9	0.32
Bystander AED applied (%)†	13.3	7.49	10.85	10.1	13.44	15.22	13.95	16.95	15.47	<0.0001
Bystander AED shock delivered (%)†	66.25	70.0	62.16	71.79	75.0	69.35	65.52	56.72	65.28	0.268
Bystander Witnessed (%)*	45.5	48.8	48.1	45.8	45.1	45.3	45.7	41.7	45.4	<0.0001
Paramedic Witnessed (%)	11.9	9.9	10.9	11.3	11.5	12.5	11.7	12.9	13.0	<0.0001
Initial Shockable Rhythm (%)	22.1	22.5	23.8	23.8	23.5	21.2	21.6	21.3	20.8	0.0005
Transported to Hospital (%)	54.1	51.0	55.2	54.4	55.0	56.5	56.4	52.0	50.6	0.0415
OUTCOME DATA
Survival to Hospital Discharge (%)	7.6	4.8	6.7	6.0	6.6	8.2	8.7	8.5	9.4	<0.0001
Survival to Hospital Discharge (initial VF/VT rhythm) (%)	24.7	17.1	20.1	18.7	20.9	28.6	27.4	28.9	30.1	<0.0001
Survival to Hospital Discharge (initial PEA rhythm) (%)	5.5	2.7	4.4	4.00	4.3	4.7	7.2	6.4	7.3	<0.0001
Survival to Hospital Discharge (initial asystole rhythm) (%)	0.67	0.67	0.66	0.82	0.46	1.01	0.62	0.53	0.63	0.66
Bystander Witnessed VF/VT Survival to Hospital Discharge (%)	23.6	15.5	22.2	19.5	18.8	25.4	26.0	26.0	31.1	<0.0001
Paramedic Witnessed Survival to Hospital Discharge (%)	17.9	15.3	12.9	13.1	17.7	19.6	20.4	21.3	18.4	0.002
Survival (with good outcome) (%)‡	7.6					6.1	7.7	7.8	8.5	0.0005

^*paramedic witnessed arrests excluded

^†among public locations

^‡good outcomes was defined as a Modified Rankin Score ≤ 2

OHCA – Out-of-hospital cardiac arrest; CPR = cardiopulmonary resuscitation; AED = automated external defibrillator; VF = ventricular fibrillation; VT = ventricular tachycardia; PEA = pulseless electrical activity

Missingness for the variables were as follows: age (0%), gender (0%; 1/23619), response time (7.9%; 1861/23619), pick-up location (0.1%; 33/23619), bystander CPR (1.4%; 335/23619), bystander AED (2.6%; 618/23619), witnessed status (0.5%; 170/23619), initial rhythm (3.1%; 742/23619), transport status (0%; 1/23619), survival to hospital discharge (0.3%; 79/23619), survival to hospital discharge by initial rhythm (3.2%; 761/23619); survival with good neurological outcome^‡ (0.6%; 87/13544); survival – VF/VT (0.4%; 15/4027), survival – PEA (0.4%; 20/4811), survival – asystole (0.05%; 5/9428), survival – bystander witnessed VF/VT (0.7%; 22/3134), survival – paramedic witnessed (0.7%; 18/2758).

Figure 2. Survival Graphs.

A. Overall out-of-hospital cardiac arrest survival over time. B. Overall out-of-hospital cardiac arrest survival over time by initial rhythm. C. Functional survival at hospital discharge

Temporal increases were observed in the percentage of public arrests, bystander CPR rates, bystander AED application rates, bystander AED application rates in public settings, mean age, the quality of prehospital CPR metrics in accordance with the 2010 American Heart Association (AHA) guidelines, and successful post arrest targeted temperature management. Decreases were observed in the percentage of bystander witnessed arrests and patients presenting with an initial shockable rhythm (Tables 2 & 3).

Table 2.

Paramedic CPR Quality Metrics

	Overall	2008	2009	2010	2011	2012	2013	Trend Test p-value
Chest Compression Fraction (mean +/− SD)	0.73 (0.12)	0.69 (0.12)	0.68 (0.12)	0.67 (0.12)	0.73 (0.11)	0.77 (0.11)	0.78 (0.10)	<0.0001
Guideline Compliance* (%)	85.3	78.3	75.2	73.4	86.6	92.3	94.7	<0.0001
Chest Compression Rate (mean +/− SD)	107.98 (12.30)	106.15 (14.16)	107.26 (13.84)	110.06 (13.6)	109.4 (12.02)	108.2 (11.4)	106.7 (10.0)	0.74
Guideline Compliance† (%)	74.2	63.3	66.9	76.0	78.6	77.5	76.1	<0.0001
Chest Compression Depth (mean +/− SD)	4.55 (1.13)	4.33 (1.11)	4.36 (1.09)	4.39 (1.10)	4.56 (1.14)	4.66 (1.13)	4.83 (1.09)	<0.0001
Guideline Compliance‡ (%)	33.1	25.6	25.5	28.7	33.3	36.8	42.8	<0.0001
Pre-shock Pause (mean +/− SD)	13.92 (7.69)				14.68 (6.63)	14.24 (7.48)	13.39 (8.14)	0.010
Guideline Compliance§ (%)	33.4				22.7	31.5	38.4	<0.0001

^*Chest Compression Fraction ≥ 0.6

^†Chest Compression Rate ≥ 100 compressions per minute

^‡Chest Compression Depth ≥ 5cm (2 inches)

^§Pre-Shock Pause ≤ 10 seconds

CCF = chest compression fraction

Missingness for the variables were as follows: chest compression fraction (32.6%; 6302/19330), chest compression rate (33.3%; 6453/19330), chest compression depth (42.7%; 8256/19330) and pre-shock pause (51.9%; 1748/3367)

Table 3.

In-hospital Outcomes

	Overall	2008	2009	2010	2011	2012	2013	Trend Test p-value
STEMI Positive (%)*	32.6					30.2	35.1
Angiography < 24hrs (%)†	73.3					73.9	72.8	0.77
Not STEMI & Shockable (%)‡	11.2					11.9	10.5
Angiography < 72hrs (%)‡	21.9					20.6	23.3	0.54
Targeted Temperature Management order given (%)§	52.5	40.5	44.1	49.7	57.6	60.1	58.2	<0.0001
Targeted Temperature Management successful (%)‖	81.4	65.8	71.0	80.1	84.1	88.2	85.6	<0.0001

^*% among patients with ROSC > 20 minutes & 12-lead ECG performed

^†% among patient with STEMI positive ECG

^‡% among non-STEMI patients with initial shockable rhythm

^§% among patients with ROSC > 20 minutes & transported to SPARC hospital

^‖temperature < 36°C within 6hrs of emergency department arrival

STEMI = ST elevation myocardial infarction; ECG = electrocardiogram; ROSC = return of spontaneous circulation; SPARC = Strategies for Post-Arrest Resuscitation Care

Missingness for the variables was as follows: STEMI positive* (3.4%; 63/1832), angiography within 24hrs^† (0%; 0/577), not STEMI & shockable^‡ (4.8%; 57/1192), angiography within 72hrs^‡ (0%; 0/325), targeted temperature management order given (0.02%; 1/4176)^§, targeted temperature management successful among patients with a cooling order (10%; 220/2192)^||

After adjusting for the Utstein variables, overall survival to hospital discharge increased each year over the study period (Table 4). Additionally, having an arrest in a public location, an initial shockable rhythm, and a bystander witnessed arrest with or without bystander CPR were associated with increased odds of survival to hospital discharge, while increasing age and longer 911-responder response time were associated with lower odds of survival. Patient sex and having an unwitnessed arrest with bystander CPR were not associated with survival to hospital discharge. Similar associations were observed for survival with good neurological outcome, with each increasing year associated with improved outcomes (Table 4). Including season (fall, spring, summer, winter) in the model did not change the overall interpretation of the data (Appendix 4). A post-hoc sensitivity analysis reclassifying initial rhythm did not alter the results (Appendix 5), nor did a post-hoc analysis that evaluated clustering of patients within paramedic services (Appendix 6).

Table 4.

Logistic Regression Results for Primary Outcomes

	Survival to Hospital Discharge		Functional Survival to Hospital Discharge
Effect	Adjusted OR (95% CI)	p-value	Adjusted OR (95% CI)	p-value
Age (per year)	0.97 (0.97–0.98)	<0.0001	0.97 (0.97–0.98)	<0.0001
Female Sex	1.00		1.00
Male Sex	0.93 (0.80–1.08)	0.324	0.92 (0.76–1.13)	0.436
Initial Rhythm (Asystole)	1.00		1.00
Initial Rhythm (PEA)	4.70 (3.48–6.35)	<0.0001	8.18 (4.93–13.59)	<0.0001
Initial Rhythm (VF/VT)	29.11 (22.08–38.38)	<0.0001	57.97 (35.79–93.89)	<0.0001
Private Location	1.00		1.00
Public Location	1.61 (1.40–1.85)	<0.0001	1.59 (1.32–1.91)	<0.0001
Unwitnessed with No Bystander CPR	1.00		1.00
Unwitnessed with Bystander CPR	0.96 (0.70–1.30)	0.772	0.94 (0.61–1.43)	0.759
Bystander Witnessed with Bystander CPR	2.38 (1.90–2.98)	<0.0001	2.25 (1.64–3.09)	<0.0001
Bystander Witnessed with no Bystander CPR	1.95 (1.55–2.46)	<0.0001	1.93 (1.39–2.68)	<0.0001
Paramedic Witnessed	6.74 (5.31–8.56)	<0.0001	7.04 (5.05–9.82)	<0.0001
Response Time (per minute)	0.90 (0.87–0.92)	<0.0001	0.87 (0.86–0.93)	<0.0001
Episode Year (per year)	1.12 (1.09–1.15)	<0.0001	1.11 (1.03–1.20)	0.007
Area Under the Curve	0.885		0.901

We included 76.2% (17978/23619) of eligible patients in the survival to hospital discharge logistic regression model and 76.2% (10321/13544) of eligible patients in the functional survival to hospital discharge logistic regression model.

DISCUSSION

Key Results

Using a large, comprehensive, prospective-collected, population based cardiac arrest database, we have shown that survival from OHCA has doubled in this region from 2006 to 2013. Similar increases in survival were seen in subgroups defined by initial cardiac arrest rhythm including VF/VT, PEA and among bystander witnessed VF/VT OHCA. These results align with research from other geographical areas that report increases in OHCA survival to discharge in recent years, as well as survival from VF/VT.^6–10 We also demonstrate temporal improvements in many other factors associated with improved outcomes including the bystander CPR rate, bystander AED application rate, high quality prehospital CPR metrics, and in-hospital use of targeted temperature management.

The observed improvement in survival is unlikely to be attributed to a single cause; rather, it is likely an accumulation of factors that has collectively led to improved patient outcomes. Each of the improvements in care observed in this study was related to a link in the Chain of Survival.^{11, 12, 18} The chain represents important interventions that if effectively implemented, provide patients with the best chance for positive outcomes.^{11, 12, 18} Each link represents a different role that bystanders and clinicians play in cardiac arrest recognition, resuscitation and post-arrest care; each link also relies on the effectiveness of its previous link(s) to optimize patient survival.^{11, 12, 18}

Throughout the study period, treatments in each individual link in the chain of survival have improved. The bystander response to OHCA has improved during the study period; both bystander CPR and AED application rates have increased. Additionally, the quality of prehospital CPR metrics and in-hospital use of targeted temperature management has improved. These improvements reflect advancing science and correlate with recommendations released by the American Heart Association as part of the 2005², the 2010³ and the 2015⁴ CPR guidelines. We believe that it is the collective and progressive strengthening of each link in the Chain of Survival that improved outcomes.

One of the most noticeable changes since 2000 has been the emphasis on high quality CPR. In 2005, the guidelines recommended a universal lay-rescuer compression-to-ventilation ratio of 30:2 for all adult patients and encouraged CPR providers to “push hard and fast”, “release completely” and “minimize interruptions in compressions”.¹⁹ In 2010, the guidelines revised the basic life support sequence from “A–B–C” (airway-breathing-circulation) to “C–A–B” (chest compressions – airway – breathing), thereby further prioritizing chest compressions over ventilations.^{3, 16} The 2015 CPR guidelines continued to recommend the “C–A–B” sequence to minimize time to initiation of chest compressions.¹¹ In this sequence, bystanders were encouraged to provide chest compression-only CPR and emergency dispatchers provided lay rescuers with compression-only CPR instructions to any patient who was not alert and not breathing normally.¹⁶ Bystander CPR is essential for survival, and some have reported better outcomes with compression-only CPR over conventional CPR.¹⁶

In 2005, the guidelines recommended a change to a 1-shock protocol (as opposed to a 3-shock sequence) in order to minimize interruptions in chest compressions. Rapid defibrillation was encouraged for witnessed OHCA, although 5 cycles of up front chest compressions were still recommended for the unwitnessed arrest.²⁰ It wasn’t until the 2010 guidelines that bystanders were encouraged to apply an AED and perform an analysis as quickly as possible regardless of witnessed status. Collectively, when a bystander performs CPR and delivers a shock from an AED, survival to hospital discharge can reach almost 40%.²¹ Throughout our study, the bystander CPR and AED application rates have demonstrated marked improvements, consistent with previous literature.^{7, 9, 10} Unfortunately, despite the clear benefit of AEDs,²¹ they are still underused in OHCA. Although less than 2% of all OHCA patients had a bystander AED applied, their use in public settings doubled (7.5% to 15.5%). In the future, more resources should be allocated to expanding AED coverage and access, especially in high risk areas, while social media strategies and community initiatives should focus on increasing AED awareness and encouraging the public to have the courage to use them.

The 2010 guidelines also emphasize the importance that paramedics and other prehospital care providers perform high quality CPR and rapid defibrillation while minimizing interruptions and delaying advanced airway interventions.¹⁶ Previous research suggests that short preshock pauses and chest compressions of adequate rate and depth are associated with improved outcomes.^22–24 Throughout the study period, paramedic adherence with CPR quality recommendations improved. These improvements in CPR quality could be attributed to monitoring and feedback (both intra arrest and during debriefing), quality improvement initiatives and continued training on performing high quality CPR.^22–24

A new recommendation related to optimizing post arrest care was included in the 2010 guidelines.²⁵ This recommendation followed the publication of two randomized clinical trials^{26, 27} and a more in-depth understanding of the post-arrest syndrome.^{28, 29} The importance of targeted temperature management (TTM) was further re-emphasized in the 2015 guideline update.¹² Additionally, the guideline recommendations were preceded by a local step-wedged randomized trial that evaluated a knowledge translation strategy to increase TTM among receiving hospitals starting in 2008.¹⁷ During our study period, the use and successful implementation of TTM increased. After completion of the RCT in 2010, most receiving hospitals in our catchment area continued to participate in the Strategies for Post-Arrest Resuscitation Care (SPARC) network, where the goal was to standardize, monitor, and improve post-arrest care and encourage best practices. The combination of new guidelines recommendations and participation in the SPARC network helped to optimize post arrest care and may have contributed to the improved outcomes.

Participation in a large clinical trials network may have contributed to the observed improvements in survival.¹⁵ Although numerous randomized controlled trials of cardiac arrest interventions from the Resuscitation Outcomes Consortium have not found a positive association with patient outcomes, participation in these trials may have led to overall improvements in care for a number of reasons. Participation in our research quality network (Rescu Episty) includes monthly conference calls between research staff and the paramedic and fire services, online access to outcome data, assistance with planning and delivering continued medical education sessions, and the option to become involved with publications. Participation also involves quality assurance monitoring and quality improvement programs that encourages re-evaluation and continued adaptation to evidence based practices may have contributed to improved outcomes.¹⁵ Research has shown that the Hawthorne effect exists in prehospital care, and only requires that providers have a perceived demand for good performance, as opposed to any direct observation or feedback.³⁰ In our system, prehospital providers recertify their CPR skills annually and they receive feedback about the patient’s outcome and the team’s performance compared to published guidelines, while destination hospitals get feedback reports on the performance of the institution which are accessible 24–7 off the website.

Limitations

This study had several limitations. First, it was not possible to determine causality of specific interventions to outcomes in a retrospective observational study design. Inferences were based on associations and similar trending outcomes. Second, there were missing data for several variables (such as initial rhythm, TTM and CPR quality metrics) in our statistical modelling. CPR quality data was not collected until 2007 or later and some paramedic services do not measure CPR depth. Good neurological survival as measured by a MRS score was also not collected until 2010. This limited our interpretation of these specific predictor variables on outcomes, particularly in earlier years. Third, individual clinical characteristics (ex. comorbidities) and population health data were not available which may have influenced the temporal changes in outcomes. Fourth, although there were statistically significant differences amongst a number of predictor variables, they may not represent clinically significant differences (ex. CPR depth). In our discussion, we attempted to focus on clinical significance variables.

Generalizability

Although our results may not be generalizable to regions with different health care environments and less resource-rich systems, our study was set in both rural and urban regions (population density range of 15 people /km² to 4150 people/km²) and included both small community hospitals and large academic referral centres (OHCA frequency rate of exposure <15 to >25 cases per month). More importantly, our results suggest that quality measurement and surveillance programs may improve clinical outcomes after OHCA through better compliance with treatment guidelines.

CONCLUSIONS

Survival to hospital discharge and survival with good neurologic function after OHCA have increased over time. In parallel with observed positive trends in survival, several aspects of cardiac arrest treatment known to be associated with improved outcomes, such as bystander interventions, high quality prehospital CPR metrics and in-hospital interventions, have also improved over the same time. Given the observed increase in survival over time, future work should refine our knowledge translation strategies to enable more rapid translation of science into practice.

What is Known

• Recent work cases doubt on a 2010 meta-analysis demonstrating that survival from out-of-hospital cardiac arrest has not improved significantly in nearly 30 years

• No study has comprehensively evaluated out-of-hospital and in-hospital interventions on survival and neurological outcomes over time.

What the Study Adds

• Survival to hospital discharge doubled and survival with good neurological outcome increased by 50%

• Improvements were also seen in the rates of bystander CPR and AED application, prehospital care provider CPR quality, and use of in-hospital targeted temperature management.

Appendix

Rescu Investigators:

• Steven Brooks

• Tim Chan

• Sheldon Cheskes

• Katie Dainty

• Paul Dorian

• Jamie Hutchison

• Dennis Ko

• Steve Lin

• Laurie Morrison

• Barto Nascimiento

• Sandro Rizoli

• Damon Scales

• Rick Swartz

• Richard Verbeek