Amplitude spectrum area as an indicator of effective return of spontaneous circulation in prehospital resuscitation—experience from a single regional center in Romania

2.1 Study design

We conducted a retrospective, observational cohort analysis using a database of OHCA patients from 01 July 2014, to 30 April 2015. Data from the register were prospectively collected during this interval primarily to gather post-ROSC biomarker information. The study included all consecutive patients with OHCA, older than 18 years, who experienced VF either as the initial ECG rhythm (iVF) or developed VF during prehospital resuscitation (noted as no-iVF). The eligible population with OHCA comprised patients diagnosed with cardiac arrest by emergency physicians working on the prehospital EMS of Cluj County, specifically the SMURD teams. The SMURD team comprised an emergency physician, a nurse, and two paramedics. The teams had standardized equipment, including a manual Corpuls 3 defibrillator with patches, and they followed the European Resuscitation Council Guidelines protocol for cardiac arrest.

We excluded patients younger than 18 years, those with known end-stage oncological diseases (information obtained from the patient’s family or medical records on site), patients with polytrauma (defined as having at least two injuries, one of which is life-threatening), those with medium or severe head trauma according to the Miller scale, with severe thoracic trauma (defined as rib fractures associated with pneumothorax or hemothorax), and patients with proximal limb amputation. Additionally, we excluded patients with incomplete ECG recordings and ECG strips under 10 seconds from the analysis.

Patients were declared in cardiac arrest if they were found unconscious and without spontaneous breathing, following resuscitation guidelines. The initiation of chest compressions, attachment of defibrillation patches for rapid rhythm assessment, and support of ventilation were immediately and synchronously performed by the members of the SMURD resuscitation team, according to their assigned roles and professional training. Resuscitation medication was provided in accordance with the identified cardiac arrest rhythm, following the recommendations of the resuscitation guidelines.

ROSC was defined as the return of a sustained cardiac rhythm that could allow blood circulation throughout the body after cardiac arrest, according to resuscitation guidelines. Death at the scene was declared when a life-sustaining heart rhythm was not achieved despite following the entire ALS protocol, including asystole lasting more than 20 minutes, and after all potential reversible causes of cardiac arrest had been identified and corrected, in accordance with resuscitation guidelines.

2.2 Source of raw data

The patient’s ECG strip was continuously recorded during CPR and collected on the Corpuls 3 defibrillator card. The defibrillators recorded patient data, which was later extracted, transferred, and analyzed. The ECG strips did not contain any information that could identify individual patients. The prehospital teams collected the following data for each patient: age, sex, urban or rural residence, initial cardiac arrest ECG rhythm (iVF/no-iVF), time to ALS (time from the call being received by the dispatch center to the ambulance arriving at the scene and starting resuscitation), bystander CPR, acute myocardial infarction (AMI), history of cardiovascular disease (CVD), number of defibrillations and dosages of adrenaline during resuscitation.

2.3 Electrocardiogram analysis

The ECG strips recorded by the manual Corpuls 3 defibrillator were set at a speed of 25 cm/s on the second standard bipolar limb lead (DII) and exported to MATLAB (MathWorks, Natick, MA, USA). We analyzed periods of VF during the 10 seconds before each shock delivery, which is the time interval from the detection of a shockable rhythm to the actual delivery of the shock.

Each period was visualized using the memory card of the Corpuls 3 defibrillator, extracted as *.jpg files (Fig. 1), and converted from the time domain to the frequency domain using Fourier Transformation. The AMSA was calculated using the MATLAB platform. AMSA represents the sum of the products of individual amplitudes and their frequencies, according to the formula: AMSA = Σ A_i × F_i, where A_i represents the amplitude and F_i is the frequency. The AMSA values were analyzed individually and divided into two groups: “success”, defined as being admitted to the ED with ROSC after defibrillation (Group 1), and “failure”, defined as prehospital death (Group 2). No filter for artifacts was applied, based on the findings of Young et al. [10], which indicated no significant changes in sensitivity or specificity when using a lower cutoff frequency to filter chest compression artifacts.

Fig. 1.Illustration of 10-second ventricular fibrillation strips used for amplitude spectrum area analysis. DE: Deutschland (screen device inscription).

2.4 Statistical analysis

An independent sample t-test was used to compare normally distributed raw data and summarized as means and standard deviations. The Mann-Whitney test was used to compare groups that did not follow a normal distribution and reported as medians and IQR (Interquartile Range defined as [Q1 to Q3], where Q1 represents the first quartile and Q3 the third quartile). Categorical data are shown as absolute and relative frequencies, and comparisons between groups were performed using the Chi-Squared test or Fisher’s exact test, depending on the theoretical frequencies.

Regression analysis was conducted to identify factors associated with ROSC. Independent variables with a p-value of up to 0.10 were included in the univariable regression analysis using the Wald stepwise methodology. All variables identified in the univariable model with at least a tendency toward statistical significance were included in the multivariable regression analysis.

Sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), and accuracy (Acc) were calculated for different AMSA thresholds. The criteria used to establish the cutoff AMSA values included the minimum AMSA value in Group 1, the arithmetic mean of AMSA in Group 2, the AMSA third quartile in Group 2, and the 90th percentile of AMSA in Group 1.

Statistical analysis was performed using the Statistica program (v8, StatSoft, Tulsa, OK, USA) and JASP (Jeffreys’s Amazing Statistics Program, v.v0.18.30., Amsterdam, Netherlands). A significance level of 5% was applied, with p-values < 0.05 considered statistically significant.

3.1 Characteristics of the patients with OHCA

Of the 73 cases retrieved, 46 met the inclusion criteria and were included for study analysis. The results showed that only 25 (14.20%) defibrillation attempts of 176 “10-second ECG strips” recorded during this study were successful. In 21 cases, defibrillation attempts resulted in ROSC (11.93%), and in 4 cases, a rhythm compatible with life, pulseless electrical activity (PEA), was followed by ROSC (Fig. 2).

Fig. 2.Flow diagram summarizing the out-of-hospital outcome: from subjects with out-of-hospital cardiac arrest (OHCA) identification to the selection of patients included in the study. n: sample size; ED: emergency department; ROSC: return of spontaneous circulation; PEA: pulseless electrical activity; CPR: cardiopulmonary resuscitation.

The patients’ age ranged from 26 to 87 years, with a mean of 58.59 ± 14.19 years. No statistically significant differences were observed between the study groups, except for the incidence of acute myocardial infarction and doses of adrenaline administered during resuscitation (Table 2). Most of the patients were men from urban settings, and half of the patients admitted to the ED had a diagnosis of myocardial infarction. Only 34.78% of patients received bystander assistance for OHCA before the arrival of the ambulance team.

3.2 AMSA and defibrillation efficiency

The number of VF sequences (AMSA values calculated) for each patient recorded during 10 seconds varied from one to eight (median = 3, IQR = [1 to 4]) in Group 1, and from two to thirteen (median = 4, IQR = [2 to 7]) in Group 2 (Mann-Whitney test: p = 0.160). The AMSA values were significantly higher (p = 0.003, Fig. 3) in Group 1, with a mean of 30.77 ± 13.20 mV-Hz (ranging from 13.60 mV-Hz to 72.49 mV-Hz), compared with Group 2, which had a mean of 23.21 ± 10.73 mV-Hz (ranging from 7.43 mV-Hz to 68.85 mV-Hz).

Fig. 3.AMSA values in Group 1 compared with Group 2: success vs. failure in ROSC. The value of median gives the line in the box, the lower and upper box are the values of first and respectively third quartile, while the whiskers are the minimum and maximum values. The ° show raw data. AMSA: amplitude spectrum area; ROSC: return of spontaneous circulation.

Further analysis showed that the values of the last AMSA (mV-Hz) were statistically significantly higher in Group 1 (median = 29.99, IQR= [18.99 to 33.85]) than in Group 2 (median = 15.59, IQR= [10.77 to 25.01]) (Mann-Whitney test: p = 0.007). As expected, although different performances for various AMSA cutoff values were observed, all demonstrated limited abilities in identifying ROSC (Table 3).

3.3 Possible confounders and clinical outcome

Data analysis showed that the time to ALS and the presence of bystanders influenced the evolution of VF and its parameters, including the AMSA value, thereby reducing the chances of survival. A shorter ALS time (≤5 minutes) was more common in Group 1, but the comparison showed only a tendency towards statistical significance when compared to Group 2 (Table 4).

Although the number of initial bystander CPR instances was low (approximately one-third in each group), the absence of bystanders had a statistically significant negative impact on the time to ALS, with longer times observed in Group 2. The probability of success increased when ALS was administered early and bystanders were present (Table 4).

Most patients in our study had a history of cardiovascular disease (CVD) (Table 2). A statistically significant association was observed between CVD and initial VF (iVF), with most patients in Group 1 having both CVD and VF as the initial rhythm, while Group 2 had CVD but no initial VF (Table 4).

The significance obtained by different factors in univariable logistic regression models was not necessarily retained in the multivariable model, except for adrenaline (Table 5). The combination of variables in the multivariable model showed an overall accuracy of 89.1%, with a sensitivity of 88% and a specificity of 90.5%.

4.1 Study findings

Previous studies have reported that myocardial defibrillation produces electrical injury and alters the energy state of the myocardium, potentially creating ectopic sites that could trigger life-threatening arrhythmias [5, 6, 7, 9]. The use of a manual defibrillator and the continuous registration of cardiac rhythm using patches is a particularity of our study, distinguishing it from previous studies using AED defibrillators [12, 13, 14, 15]. We measured AMSA over 10 seconds, in contrast to earlier studies that used shorter periods (2–5 seconds) [13, 15], aligning with more recent studies reporting periods between 8 and 11 seconds [12, 14, 18]. A 10-second period offers a practical “on the field” approach for establishing AMSA values in a prehospital population. However, the specific conditions of our study may impact AMSA values and their effectiveness in predicting ROSC. Severe myocardial ischemia and delayed resuscitation are known to modify AMSA values by decreasing the amplitude of VF waves, thereby reducing the AMSA value. The discrepancies observed in sensitivity and specificity between our present study and current literature could be explained by the difference in the length of analysis (10 seconds vs. 2–5 seconds) and the differences in digitization and calculation methods used to evaluate AMSA values.

When using a longer analysis period of 10 seconds, chest compressions may artifact the rhythm just before defibrillation, potentially leading to an overestimation of AMSA [13, 14, 20]. The literature reports various techniques to reduce or remove chest compression artifacts from ECG analysis [13, 14, 21, 22], and some studies have identified deep-learning strategies for sliding ECG analysis during resuscitation with AEDs [23]. Corpuls 3 can be used to effectively reduce the influence of chest compressions on the VF signal, allowing for analysis without the need for chest compression corrections.

The AMSA value analysis using manual defibrillation revealed higher specificity and accuracy at 27.6 mV-Hz compared to the 15.5 mV-Hz using AEDs in different health system organizations reported in literature [13, 15, 16]. A consistent finding was the decreased number of defibrillations (Table 2) in patients with higher AMSA values and positive short-term outcomes. AMSA values are influenced by factors such as early and high-quality chest compressions, bystander CPR response, and medication [10, 24]. In this context, the regional EMS system and health policies play an essential role in improving short-term outcomes (ROSC and ED admission) by organizing public access defibrillation programs, dispatcher-assisted CPR training, and promoting comprehensive bystander response [25, 26]. Due to the low incidence of bystander CPR (approximately one-third in both groups), the study could not demonstrate the well-documented benefit of bystander CPR on ROSC (Table 2). However, we observed a significantly lower incidence of bystander CPR and an increased time to start ALS (>5 minutes) in the prehospital death group (Group 2), which thereby highlights the role of EMS teams in providing early defibrillation and achieving ROSC, even when bystander CPR is not provided [4, 26]. Insufficient education and awareness of the Romanian population regarding CPR initiation and AED use reduces short-term outcomes, which depend directly on early defibrillation provided by EMS teams (Table 2). Implementing bystander CPR training in Romanian schools, in accordance with the 2021 European Resuscitation Council Guidelines, could improve outcomes for OHCA and VF patients as the training would increase the likelihood of early defibrillation, resulting in higher AMSA values and thereby enhancing the chances for effective defibrillation, ROSC and ED admission.

AMSA reflects the energy state of the myocardium and is dependent on coronary perfusion. Adrenaline administration combined with chest compression could improve AMSA values [8]. In this study, the observed lower dosage of adrenaline in Group 1 compared to Group 2 (Table 2) suggests an association between high AMSA values and a reduced need for vasopressors, which may decrease negative effects after ROSC [4]. Univariable regression analysis for ROSC as an outcome of successful defibrillation identified several significant predictors, including adrenaline (p < 0.001), AMI (p < 0.005), and AMSA value (p < 0.028) (Table 5).

Ristagno et al. [13] reported a higher incidence of successful defibrillation in patients with a history of CVD and an initial VF rhythm. This study confirms that patients with an initial VF rhythm had a higher number of sustained ROSC and were more likely to be admitted to the ED compared to those with a different initial cardiac arrest rhythm (Table 4). Additionally, in cases of CVD, we observed a 32% decrease in success rates when the time to start ALS exceeded 5 minutes. The data confirm that initiating ALS (which includes the first defibrillation attempts in our health system) within the first 5 minutes for patients with VF increases the success of defibrillation, thereby improving the ROSC rate and ED admission (Table 4).

As observed in previous studies, acute myocardial infarction is a significant cause of cardiac arrest [2, 4, 27]. In this study, the results demonstrate a significantly higher ROSC rate and ED admission among VF patients with acute myocardial infarction (52% in Group 1, Table 2), suggesting that effective defibrillation of VF in patients with acute myocardial infarction could be more likely to result in ROSC [27, 28, 29].

4.2 Study limitations

This study had several limitations that should be clarified. First, it was conducted at a single center, using prospectively recorded data over a short period of 10 months; therefore, the number of patients and analyzed ECG tracks was limited. In addition, the reported results reflect only the evaluated cohort and should be validated in larger cohorts. Moreover, the small cohort size might have affected the cutoff AMSA value’s performance, limiting its clinical relevance. However, the results suggest potential utility in establishing a clinically useful cutoff, supporting the need to extend the study to a larger cohort. Second, we did not use any methods to filter ECG artifacts, relying on the defibrillator technology to minimize the influence of chest compressions on the ECG rhythm. Third, due to the limited number of patients, we could not stratify the performances according to factors that could have influenced AMSA changes. Thus, analyzing indicators related to myocardial dysfunction (i.e., the number of defibrillation attempts and adrenaline dosages), individual factors (i.e., presence of CVD), and out-of-hospital conditions (i.e., bystander CPR response, public access defibrillator availability, and time to ALS) could provide key information for defining computer-assisted models to evaluate AMSA efficiency for defibrillation. Further research could develop defibrillator algorithm software to report AMSA values predictive of successful defibrillation.