The association between initial defibrillation dose and outcomes following adult out-of-hospital cardiac arrest resuscitation: a retrospective, multi-agency study

Abstract

Introduction

Effective defibrillation is essential to out-of-hospital cardiac arrest (OHCA) survival. International guidelines recommend initial defibrillation energies between 120 and 360 Joules, which has led to widespread practice variation. Leveraging this natural experiment, we aimed to explore the association between initial defibrillation dose and outcome following OHCA.

Methods

The ESO Data Collaborative (2018-2022) was used for this nationwide, retrospective study of adult (18-80 years of age) non-traumatic OHCA patients who presented with an initially shockable ECG rhythm. We excluded patients if they had ROSC prior to initial defibrillation, a resuscitation-limiting advanced directive, or were residents in a healthcare institution. The primary exposure was initial defibrillation dose, defined as Joules per kilogram of body weight, and the primary outcome was return of spontaneous circulation (ROSC). We included survival to discharge as a secondary outcome. We used multivariable logistic regression modeling to assess the relationship between defibrillation dose and outcome.

Results

We analyzed data from 21,121 patients. Of the 12,160 patients linked to a defibrillator manufacturer, 7,240 (59.5%) were treated using a biphasic truncated exponential (BTE) waveform and 4,920 (40.5%) were treated using a rectilinear biphasic (RLB) waveform. Defibrillation dose (per 1 J/kg increase) was not associated with ROSC (BTE aOR: 0.97 [0.92, 1.01], n=7,240; RLB aOR: 1.00 [0.92, 1.09], n=4,920; all aOR: 1.01 [0.98, 1.04], 21,121) or survival (BTE aOR: 0.98 [0.87, 1.10], n=1,245; RLB aOR: 0.89 [0.70, 1.12], n=775; all aOR: 1.00 [0.92, 1.08], n=2,981).

Conclusions

Initial defibrillation dose was not associated with outcome in this nationwide cohort.

Introduction

Early, effective defibrillation is essential to the survival of victims of out-of-hospital cardiac arrest (OHCA) with an initial cardiac rhythm of ventricular fibrillation or ventricular tachycardia (VF/VT). In 2022, fewer than 1-in-5 patients undergoing resuscitation by emergency medical services (EMS) presented with an initial rhythm of VF/VT. Approximately 45% of these patients achieved prehospital return of spontaneous circulation (ROSC), but only 27% survived to hospital discharge.¹ This low survival rate underscores the need to investigate the optimal delivery of resuscitative interventions to this patient population.

The definitive therapy for victims of VF/VT OHCA is defibrillation, which can terminate these ventricular arrhythmias and allow restoration of the function of the heart. Many factors have been associated with defibrillation success, including the time to defibrillation following cardiac arrest,^{2, 3} pre-shock compression pause duration,^{4, 5} and defibrillator pad position.⁶ While most studies of defibrillation technique to date have focused on defibrillation current waveform^{7, 8} or energy delivery strategy,⁹ one important exposure that has not been extensively studied in the context of contemporary resuscitation is the influence of defibrillation dose (Joules per kilogram of body weight) on outcomes. Resuscitation guidelines permit initial defibrillation energy levels from 120-360 Joules during resuscitation from out-of-hospital cardiac arrest, with the recommendation that the initial defibrillation be delivered using either the manufacturer recommended setting or the maximum available energy setting.^{10, 11} Commercially available cardiac monitor/defibrillators used in the United States have maximum selectable biphasic defibrillation energies ranging from 200 to 360 Joules. This has led to practice variation and a natural experiment that may shed light on the optimal initial defibrillation dose for adult OHCA patients presenting with a shockable rhythm.

Our primary aim was to explore the association between initial defibrillation dose and outcome following OHCA among patients presenting with ventricular fibrillation or ventricular tachycardia. Our primary outcome was ROSC. Secondary outcomes included survival to hospital discharge and survival to discharge to home. We hypothesized that increasing initial defibrillation dose would be associated with greater odds of ROSC.

Methods

Data source

The ESO Data Collaborative 2018-2022 annual datasets were used for this nationwide, retrospective cohort study. This dataset is comprised of electronic health record (EHR) data recorded by EMS clinicians following patient care in the field.¹² The EHR is comprised of data elements that reflect the National EMS Information System (NEMSIS) v3 data standard.¹³ Vital sign and intervention timing data can also be imported into the EHR from cardiac monitor/defibrillators. Participating agencies that use ESO EHR software have their records deidentified and included in annual research datasets without data abstraction on an annual basis. Among transported patients, approximately 20% of EHRs are linked to hospital disposition data from agencies that participate in a bidirectional health data exchange. Use of this deidentified dataset was approved by an institutional review board (protocol #2202524583).

Study design and inclusion criteria

All adult (18-80 years of age) non-traumatic OHCA patients were evaluated for inclusion. Patients were excluded if they had a non-shockable initial electrocardiogram (ECG) rhythm, no documented defibrillation energy, no documented weight, ROSC prior to the first defibrillation attempt, or a resuscitation-limiting advanced directive. Patients were also excluded if they were a resident in a skilled nursing or assisted living facility at the time of their OHCA to minimize the number of included patients with a poor pre-OHCA functional status. The information needed to apply inclusion/exclusion criteria was theoretically available at the time of defibrillation delivery.^{14, 15} Follow-up began at the time of the first defibrillation and ended at hospital arrival, termination of resuscitation, or transfer of care to other EMS clinicians.

Variable definitions and outcome measures

Body weight was derived from the prehospital EHR as estimated by treating EMS clinicians. No measurement of in-hospital weight was captured by this dataset. The defibrillation energy in Joules was defined as the energy setting chosen by EMS clinicians for the first defibrillation attempt. These values were documented following provision of patient care. A subset of defibrillation energy values were directly imported from cardiac monitor/defibrillators into the EHR. The dispatch to defibrillation interval was calculated as the interval from the time of EMS dispatch to the time of the first EMS defibrillation. Initial airway management strategy was derived from the first documented airway attempt, regardless of success. Patients were classified as having an AED used prior to EMS arrival if an AED was applied by firefighters, law enforcement personnel, or lay rescuers. All other variables were derived using pre-exiting data elements from the EHR.

We chose prehospital ROSC as our primary outcome due to the availability of this outcome for our entire cohort and proximity to our exposure of interest. Secondary outcomes included survival to discharge and survival to discharge to home. Survival to discharge home was chosen as a proxy for good functional outcome. Both survival outcomes were derived only for the subset of transported patients with hospital disposition data available.

Statistical analyses

The primary exposure for this analysis was initial defibrillation dose, which was calculated as Joules divided by kilograms of body weight using the initial defibrillation energy setting. We performed multivariable logistic regression to explore the association between defibrillation dose and outcome. A directed acyclic graph (DAG)¹⁶ was generated using a browser-based environment for causal diagram creation (DAGitty v3.1) and used to determine the minimum number of confounding variables necessary to include in our models. [Supplemental Figure 1] The generation of a DAG allows for the illustration of assumptions about the relationship between variables, exposures and outcomes in epidemiologic studies. DAG-guided multivariable adjustment can also be used to prevent an increase in bias via the creation of conditional relationships between covariables that act as ‘colliders’ or mediators.¹⁷ Additional variables were added to our model based on the Utstein template and prior literature examining factors associated with OHCA outcome. Our ‘maximally adjusted’ multivariable logistic regression¹⁸ model included age, sex, witnessed status, bystander CPR, automated external defibrillator (AED) use prior to the arrival of EMS, initial rhythm (ventricular fibrillation, ventricular tachycardia, or unknown AED-shockable rhythm) dispatch to defibrillation interval, OHCA etiology, initial vascular access strategy (intravenous cannula, intraosseous cannula, or no documented vascular access),^19-22 initial airway management strategy (endotracheal intubation, i-gel, King-Laryngeal Tube, laryngeal mask airway, or no advanced airway),²³ and OHCA location (home/residence vs. public) as covariables. We chose to include each style of supraglottic device within the airway factor variable instead of grouping all supraglottic airways together because of a previous study using this dataset²⁴ that suggested a relationship between outcomes and device type. Co-linearity of continuous covariables within our model was assessed using variance inflation factors. [Supplemental Table 1]

Defibrillator manufacturers utilize different biphasic waveforms for current delivery, which may influence defibrillation success. For a subgroup of our cohort, we used the monitor/defibrillator file type uploaded to the EHR to identify the manufacturer of the device (ZOLL vs. Stryker/Philips) used to deliver defibrillation during resuscitation. ZOLL monitor/defibrillators utilize a rectilinear biphasic waveform and Stryker/Philips monitor/defibrillators utilize a biphasic truncated exponential waveform. Our primary and secondary analyses were performed for both the overall cohort and stratified by defibrillation waveform. Stata SE/18 was used for all data management and statistical analysis.

Secondary analyses

Patients at the tails of the defibrillation dose distribution may have been either pathologically underweight (leading to a high calculated dose) or overweight (leading to a low calculated dose). Therefore, patients who received higher doses may have been more likely to have cancer or other diseases that cause cachexia and weight loss, and patients who received lower doses may have been more likely to have obesity and associated comorbidities that have been associated with worse outcomes following OHCA.²⁵ Due to concern for residual confounding, we performed a subgroup analysis among the cohort of patients who weighed 90-110 kilograms. This group of patients was selected from the center of the body weight distribution for our cohort. [Supplemental Figure 2b.]

Because of a possible difference in the defibrillation threshold for ventricular tachycardia and ventricular fibrillation, we examined the association between defibrillation dose and outcomes in subgroups defined by these initial ECG rhythms. Finally, due to the potential influence of defibrillation prior to EMS arrival, we performed an analysis examining the association between defibrillation dose and outcome among patients who did not receive AED defibrillation prior to EMS arrival.

Results

Cohort characteristics

After application of exclusion criteria, 21,121 patients treated by clinicians from 1,461 EMS agencies were eligible for analysis. [Figure 1] Overall, 42.7% of included patients (9,022/21,121) achieved ROSC, 32.8% (977/2,981) survived to hospital discharge, and 22.6% (675/2,981) survived and were discharged to home. The median [IQR] initial defibrillation energy was 200 [200, 274] J, and the median defibrillation dose was 2.3 [1.8, 3.2] J/kg. Of the 12,160 patients (57.6% of total cohort) linked to a monitor manufacturer, 7,240 (59.5%) were treated using Stryker/Philips devices and 4,920 (40.5%) were treated using ZOLL devices. The distribution of initial defibrillation energy, patient body weight, and defibrillation dose for our cohort are displayed in Supplemental Figure 2. The baseline characteristics of included patients for the entire cohort and stratified by defibrillation waveform are displayed in Table 1. The unadjusted relationship between the primary outcome of ROSC and defibrillation dose is displayed in Figure 2.

Figure 1: Flow diagram describing cohort derivation.

A depiction of cohort derivation using described inclusion criteria.

Table 1: Baseline patient characteristics.

	Total Cohort	BTE	RLB
N	21,121	7,240	4,920
	Median (IQR)
Age (years)	62.0 (52.0-70.0)	61.5 (52.0-70.0)	62.0 (52.0-70.0)
Weight (kg)	90.9 (77.3-113.4)	90.7 (77.3-113.4)	90.7 (79.2-113.4)
Dispatch to defib. (min.)	11.1 (8.3-15.8)	11.0 (8.3-15.3)	11.3 (8.2-16.2)
Initial energy (Joules)	200.0 (200.0-274.0)	200.0 (200.0-360.0)	200.0 (144.0-249.0)
Dose (J/kg)	2.3 (1.8-3.2)	2.8 (2.0-3.8)	2.0 (1.5-2.7)
Response time	6.0 (4.3-8.6)	6.0 (4.3-8.4)	6.0 (4.2-8.7)
	n (%)
Sex
Male	15,826 (75.0%)	5,441 (75.2%)	3,639 (74.0%)
Female	5,283 (25.0%)	1,797 (24.8%)	1,278 (26.0%)
Witnessed
No	5,502 (26.0%)	1,923 (26.6%)	1,295 (26.3%)
Yes	13,494 (63.9%)	4,611 (63.7%)	3,139 (63.8%)
Unknown	2,125 (10.1%)	706 (9.8%)	486 (9.9%)
Bystander CPR
No	12,204 (57.8%)	4,206 (58.1%)	2,920 (59.3%)
Yes	7,928 (37.5%)	2,801 (38.7%)	1,897 (38.6%)
Unknown	989 (4.7%)	233 (3.2%)	103 (2.1%)
Rhythm
Ventricular fibrillation	18,611 (88.1%)	6,361 (87.9%)	4,228 (85.9%)
Ventricular tachycardia	1,386 (6.6%)	441 (6.1%)	389 (7.9%)
Unknown AED shockable	1,124 (5.3%)	438 (6.0%)	303 (6.2%)
Etiology
Cardiac	19,235 (91.1%)	6,620 (91.4%)	4,494 (91.3%)
Respiratory/asphyxia	955 (4.5%)	315 (4.4%)	221 (4.5%)
Other	926 (4.4%)	304 (4.2%)	205 (4.2%)
Location
Home/residence	15,705 (74.4%)	5,337 (73.7%)	3,705 (75.3%)
Public	5,416 (25.6%)	1,903 (26.3%)	1,215 (24.7%)
AED use PTA
No	12,134 (57.5%)	4,050 (56.0%)	2,723 (55.4%)
Yes	8,959 (42.5%)	3,180 (44.0%)	2,193 (44.6%)
Vascular access
Intravenous	7,353 (34.8%)	2,355 (32.5%)	1,631 (33.2%)
Intraosseous	12,233 (57.9%)	4,412 (60.9%)	2,979 (60.5%)
None	1,535 (7.3%)	473 (6.5%)	310 (6.3%)
Airway
Endotracheal intubation	9,747 (46.3%)	3,546 (49.0%)	1,946 (39.6%)
i-gel	5,328 (25.3%)	1,805 (25.0%)	1,375 (28.0%)
King-LT	2,766 (13.1%)	862 (11.9%)	766 (15.6%)
LMA	434 (2.1%)	75 (1.0%)	253 (5.2%)
No advanced airway	2,794 (13.3%)	945 (13.1%)	568 (11.6%)
ROSC
No	12,099 (57.3%)	4,067 (56.2%)	2,840 (57.7%)
Yes	9,022 (42.7%)	3,173 (43.8%)	2,080 (42.3%)
Survival to discharge
No	2,004 (67.2%)	829 (66.6%)	515 (66.5%)
Yes	977 (32.8%)	416 (33.4%)	260 (33.5%)
Survival to home
No	2,306 (77.4%)	958 (76.9%)	609 (78.6%)
Yes	675 (22.6%)	287 (23.1%)	166 (21.4%)

Kg = kilograms, J = Joules, CPR = cardiopulmonary resuscitation, AED = automated external defibrillator, PTA = prior to arrival, LT = laryngeal tube, LMA = laryngeal mask airway, ROSC = return of spontaneous circulation.

Figure 2: Unadjusted relationship between defibrillation dose and prehospital return of spontaneous circulation.

Patients were grouped based on decile of defibrillation dose (decile 1 = 0-9^th percentile, decile 2 = 10-19^th percentile, etc.) and stratified by defibrillation waveform. The Y axis describes the percentage of patients who achieved prehospital ROSC.

Multivariable logistic regression modeling

Both overall and stratified by defibrillation waveform, defibrillation dose was not associated with any outcome in this cohort. We failed to detect a significant relationship between defibrillation dose and our outcomes of interest using both DAG-guided and ‘maximally adjusted’ logistic regression models. [Table 2]

Table 2: Multivariable logistic regression analyses.

DAG-guided adjustment (a)	ROSC aOR [95% CI]	Survival to discharge aOR [95% CI]	Survival and discharge home aOR [95% CI]
Initial defibrillation dose (per 1 J/kg, continuous) Truncated exponential	0.97 [0.92, 1.01] n=7,240	0.98 [0.87, 1.10] n=1,245	1.01 [0.89, 1.15] n=1,245
Initial defibrillation dose (per 1 J/kg, continuous) Rectilinear	1.00 [0.92, 1.09] n=4,920	0.89 [0.70, 1.12] n=775	0.77 [0.56, 1.02] n=775
Initial defibrillation dose (per 1 J/kg, continuous) All defibrillation waveforms	1.01 [0.98, 1.04] n=21,121	1.00 [0.92, 1.08] n=2,981	0.99 [0.91, 1.08] n=2,981
Full multivariable model (b)	ROSC aOR [95% CI]	Survival to discharge aOR [95% CI]	Survival and discharge home aOR [95% CI]
Initial defibrillation dose (per 1 J/kg, continuous) Truncated exponential	0.98 [0.93, 1.03]; n=7,212	0.96 [0.85, 1.10]; n=1,243	1.00 [0.86, 1.16]; n=1,243
Initial defibrillation dose (per 1 J/kg, continuous) Rectilinear	1.05 [0.96, 1.15]; n=4,893	0.93 [0.71, 1.22]; n=775	0.75 [0.54, 1.03]; n=775
Initial defibrillation dose (per 1 J/kg, continuous) All defibrillation waveforms *	1.02 [0.99, 1.05] n=20,993	0.99 [0.92, 1.09 n=2,975	0.99 [0.89, 1.09] n=2,975

^*For entire model, see Supplemental Table 2.

DAG = directed acyclic graph, ROSC = return of spontaneous circulation (ROSC), aOR = adjusted odds ratio, J = Joules, kg = kilograms.

Secondary analyses

In the subgroups comprised of patients with a body weight of 90-110 kilograms, patients who did not have an AED applied prior to EMS arrival, and patients with ventricular fibrillation, no relationship was observed between defibrillation dose and outcomes regardless of defibrillation waveform type. Among patients who presented with an initial ECG rhythm of ventricular tachycardia and were treated using a truncated exponential waveform, increasing defibrillation dose was associated with increased odds of ROSC (aOR per J/kg increase: 1.22 [1.02, 1.47]; n=441). [Table 3]

Table 3: Subgroup analyses.

DAG-guided adjustment	ROSC aOR [95% CI] per 1 J/kg increase	Survival to discharge aOR [95% CI] per 1 J/kg increase
90-110 kg Truncated exponential	0.98 [0.88, 1.08] n=2,251	0.87 [0.68, 1.12] n=373
90-110 kg Rectilinear	0.93 [0.77, 1.11] n=1,488	0.79 [0.49, 1.27] n=252
90-110 kg All waveforms	1.02 [0.95, 1.08] n=6,513	0.92 [0.78, 1.08] n=922
No AED use PTA Truncated exponential	0.97 [0.91, 1.04] n=4,050	0.94 [0.80, 1.10] n=662
No AED use PTA Rectilinear	0.98 [0.88, 1.10] n=2,723	0.80 [0.59, 1.09] n=448
No AED use PTA All waveforms	1.01 [0.97, 1.05] n=12,134	0.98 [0.88, 1.09] n=1,664
Ventricular fibrillation Truncated exponential	0.96 [0.92, 1.01] n=6,361	0.97 [0.85, 1.10] n=1,055
Ventricular fibrillation Rectilinear	1.00 [0.91, 1.09] n=4,228	0.83 [0.65, 1.08] n=686
Ventricular tachycardia Truncated exponential	1.22 [1.02, 1.47] n=441	1.39 [0.91, 2.13] n=91
Ventricular tachycardia Rectilinear	1.06 [0.79, 1.42] n=389	1.09 [0.40, 2.97] n=49
Maximally adjusted model plus waveform type covariable * All waveforms	0.99 [0.95, 1.04] n=12,105	0.96 [0.86, 1.07] n=2,018

^*For entire model, see Supplemental Table 3.

DAG = directed acyclic graph, ROSC = return of spontaneous circulation, aOR = adjusted odds ratio, kg = kilograms, AED = automated external defibrillator, PTA = prior to EMS arrival.

In a multivariable logistic regression model that included defibrillator manufacturer as a proxy for defibrillation waveform, defibrillation dose was not associated with any outcome per J/kg increase (aOR for ROSC: 0.99 [0.95, 1.04]; aOR for survival to discharge: 0.96 [0.86, 1.07]; aOR for survival to discharge to home: 0.95 [0.83, 1.08]). Using patients treated with truncated exponential waveform as the reference group, treatment with a rectilinear waveform was not associated with ROSC (aOR: 0.95 [0.87, 1.03]), survival to discharge (aOR: 0.99 [0.78, 1.25]), or survival to discharge to home (aOR: 0.90 [0.68, 1.18]). [Supplemental Table 3]

Discussion

Our multi-agency, nationwide, retrospective cohort study found no statistically significant association between initial defibrillation dose and ROSC or survival following VF/VT OHCA. These results were consistent across multivariable modeling strategies and patient subgroups. There are several strengths to our study, including the large size of the multi-agency cohort and the geographic reach of the dataset we used, which may increase generalizability of our results. We designed our methods to support causal inference and minimize common sources of bias, including the use of a directed acyclic graphs to guide multivariable modeling and adherence to applicable elements of target trial emulation methodology. To establish clear temporality between exposure and outcome, all cases without ROSC and defibrillation timing were excluded.

In the subgroup of patients who presented with ventricular tachycardia and were treated with a defibrillator utilizing a truncated exponential waveform, increasing defibrillation dose was associated with increased odds of ROSC. Defibrillation dose was not associated with increased survival in this small subgroup. It is possible that this finding is due to differences in the physiology of ventricular fibrillation and ventricular tachycardia. However, it is also likely that this finding is due to random chance, as it was the only statistically significant effect detected among the 22 hypothesis tests conducted as subgroup analyses.

Our overall findings are consistent with prior studies that have suggested body weight and body mass index are not associated with defibrillation success.^{26, 27} Increased body weight was strongly associated with worse outcome (per kilogram increase) in our multivariable regression models, but this relationship may have been via a mechanism independent of defibrillation success, including decreased CPR effectiveness in the setting of obesity or the association of body habitus with comorbidity.

The relationships between the chosen covariables for our multivariable regression analyses and our outcomes of interest aligned with previous literature (e.g. increasing dispatch to defibrillation interval was associated with decreased survival^{28, 29}) which supports the validity of our analysis. Our multivariable model suggested that an initial advanced airway attempt using the i-gel or King laryngeal tube supraglottic airway devices was associated with slightly greater odds of survival in comparison to an initial endotracheal intubation attempt, which aligns with previous clinical trials.^{30, 31} Our analysis also found no association between the initial vascular access strategy (IO vs. IV) and survival, which also mirrors the results of previously published randomized controlled trials.^32-34

Limitations

Despite our efforts related to cohort selection and data analysis, our results are subject to the influence of chance, bias, and confounding due to the retrospective, observational nature of this study. Following multivariable adjustment, our results are still subject to residual confounding and the influence of unknown or unmeasured confounders. Confounding by indication may also play a role - although data describing factors influencing the decision of EMS clinicians to select defibrillation energy are sparse in the literature, it is possible that clinician decision making related to energy selection was influenced by factors such as body weight or amplitude of ventricular fibrillation.

One limitation of this work is the inability to determine which defibrillator model was used during resuscitation for the entire study cohort. We used energy level as a proxy for current delivered to the myocardium, which is thought to be the primary determinant of successful defibrillation. The energy level selected by EMS clinicians performing manual defibrillation using a cardiac monitor/defibrillator dictates the amount of voltage charged to the defibrillator’s capacitor. However, different devices can store different amounts of voltage on their capacitor, and the relationship between Joules and voltage may vary. Therefore, the direct comparison of energy settings for different defibrillator models may not be appropriate – delivering a 200 J shock using one defibrillator may not result in the same peak or average current applied to the patient’s heart as delivering a 200 J shock from another defibrillator. In addition, defibrillator manufacturers utilize different biphasic defibrillation waveforms, which may also impact effectiveness. The two most commonly used cardiac monitor/defibrillators in the United States are ZOLL (ZOLL Medical, Chelmsford, MA) and LIFEPAK (Stryker, Redmond, WA) models. ZOLL defibrillators utilize a biphasic rectilinear waveform, whereas LIFEPAK defibrillators utilize a biphasic truncated exponential waveform. Laboratory data from swine models have suggested that use of rectilinear waveforms in comparison to truncated exponential waveforms may lead to increased defibrillation success at lower energies after short durations of VF,³⁵ especially in the setting of high trans-thoracic impedance,³⁶ in addition to earlier successful defibrillation and less post-resuscitation myocardial dysfunction after prolonged cardiac arrest (7 minutes of untreated VF, followed by 5 minutes of BLS prior to the first defibrillation attempt).³⁷ One nationwide prehospital cohort study indirectly evaluated the use of these waveforms to treat OHCA in the real world by comparing the outcomes of patients defibrillated with the ZOLL X-series (rectilinear) to the outcomes of patients treated with the LIFEPAK 15 (truncated exponential), and found no difference in the odds of ROSC at hospital arrival or survival at 30 days.³⁸ Similarly, our analyses using the subgroup of encounters linked to defibrillator manufacturers also did not suggest that defibrillation waveform was a significant predictor of any outcome. These current waveforms have also been compared head-to-head in multiple studies of cardioversion of atrial fibrillation and found to result in similar outcomes.^39-42

Due to variation in the trans-thoracic impedance of OHCA patients, modern biphasic defibrillators employ energy, current, or voltage-based impedance compensation strategies⁴³ based on measurements of impedance obtained via electrotherapy pads prior to defibrillation delivery. The compensation strategy employed has relevance to the relationship between the energy level selected by EMS clinicians and the actual delivered energy. Our subgroup analyses comparing the outcomes of patients treated by different defibrillator types and examining the relationship between defibrillation dose and outcomes within subgroups defined by defibrillator manufacturer may suggest that impedance compensation technique is not a confounder within the context of our analyses. Future study of the influence of impedance compensation techniques on defibrillation success is needed.

We did not have access to biosignal data from defibrillator files for this cohort. Therefore, we were not able to obtain measured transthoracic impedance values or calculate important resuscitation process metrics such as CPR quality. Utilizing defibrillation energy, energy delivery duration, and impedance values to calculate the current delivered may better predict defibrillation success. In addition, because we were partially relying on EMS clinician documentation of Joules, the true number of Joules delivered may have been different than the documented values. Using the number of Joules an EMS clinician intended to deliver to calculate defibrillation dose instead of the Joules delivered following automated impedance adjustment is a pragmatic approach that captures the decision being made by the treating clinician.

The patient weights documented in the EHR by EMS clinicians are likely estimates and may not be accurate. However, previously published data suggests that EMS providers can accurately estimate the weight of OHCA patients. A retrospective analysis by Martin et al. suggested that paramedics are able to estimate the weight of OHCA patients within 10% of their true weight 74% of the time, and within 20% of their true weight 93% of the time (correlation coefficient = 0.93).⁴⁴ Inaccurate estimation of body weight may decrease the internal validity of our investigations via introduction of measurement bias. However, the use of clinician documented body weight does increase the external validity of our study – any decision made in the prehospital setting regarding initial defibrillation dose would have to be made on the basis of EMS clinician estimation of weight.

Independently from the initial defibrillation dose, it is also possible that maximum defibrillation dose influences outcome. Escalation of defibrillation energy after an initial failed shock may lead to successful arrhythmia termination. We have chosen to not examine the association between maximum defibrillation dose and outcome in this observational study due to the inevitable influence of resuscitation time bias⁴⁵ – patients with a higher maximum defibrillation dose will likely have required more shocks with dose escalation, have been in cardiac arrest for a longer interval, and therefore have worse outcomes.

Approximately 40% of the patients in this study had an AED applied prior to EMS arrival, primarily by first responders (firefighters and law enforcement personnel). It is unlikely that the timing and energy of these initial AED defibrillation attempts would have been reliably included in the EMS EHR. Therefore, the first defibrillation documented in the EHR and utilized in this study was likely the initial manual defibrillation delivered by the first arriving EMS clinicians. This may limit the generalizability of our findings.

Our study did not include data regarding several factors that have been shown to influence defibrillation success, including the positioning of electrotherapy pads on the chest wall⁶ or the duration of pre-shock compression pauses.^{4, 5} Future studies incorporating these important variables may be warranted.

Conclusion

Initial defibrillation dose was not associated with patient outcome in this national retrospective cohort. Focusing on the components of defibrillation known to be associated with success - including decreasing the time to defibrillation, shortening pre-shock pauses, and optimizing pad position - should remain the priority.