Kenneth Gundersen

Development of the probability of return of spontaneous circulation in intervals without chest compressions during out-of-hospital cardiac arrest: an observational study

BackgroundOne of the factors that limits survival from out-of-hospital cardiac arrest is the interruption of chest compressions. During ventricular fibrillation and tachycardia the electrocardiogram reflects the probability of return of spontaneous circulation associated with defibrillation. We have used this in the current study to quantify in detail the effects of interrupting chest compressions.MethodsFrom an electrocardiogram database we identified all intervals without chest compressions that followed an interval with compressions, and where the patients had ventricular fibrillation or tachycardia. By calculating the mean-slope (a predictor of the return of spontaneous circulation) of the electrocardiogram for each 2-second window, and using a linear mixed-effects statistical model, we quantified the decline of mean-slope with time. Further, a mapping from mean-slope to probability of return of spontaneous circulation was obtained from a second dataset and using this we were able to estimate the expected development of the probability of return of spontaneous circulation for cases at different levels.ResultsFrom 911 intervals without chest compressions, 5138 analysis windows were identified. The results show that cases with the probability of return of spontaneous circulation values 0.35, 0.1 and 0.05, 3 seconds into an interval in the mean will have probability of return of spontaneous circulation values 0.26 (0.24–0.29), 0.077 (0.070–0.085) and 0.040(0.036–0.045), respectively, 27 seconds into the interval (95% confidence intervals in parenthesis).ConclusionDuring pre-shock pauses in chest compressions mean probability of return of spontaneous circulation decreases in a steady manner for cases at all initial levels. Regardless of initial level there is a relative decrease in the probability of return of spontaneous circulation of about 23% from 3 to 27 seconds into such a pause.

Dynamics and state transitions during resuscitation in out-of-hospital cardiac arrest

BACKGROUND The state or rhythm during resuscitation, i.e. ventricular fibrillation/tachycardia (VF/VT), asystole (ASY), pulseless electrical activity (PEA), or return of spontaneous circulation (ROSC) determines management. The state is unstable and will change either spontaneously (e.g. PEA-->ASY) or by intervention (e.g. VF-->ASY after DC shock); temporary ROSC may also occur. To gain insight into the dynamics of this process, we analyzed the state transitions over time using real-life data. METHODS Detailed recordings from 304 episodes of attempted resuscitation from out-of-hospital cardiac arrests of presumed cardiac etiology were obtained from modified Heartstart 4000 defibrillators. State transitions were visualized and described, and analyzed in terms of a Markov probability model. RESULTS The median number of state transitions was 5 (range 1-39), and more transitions were observed with VF than PEA or asystole as the initial rhythm. Of 105 patients (35%) who regained ROSC at some point during CPR, only 65 (21%) achieved sustained ROSC; suggesting an unrealized survival potential. A 3-min transition probability matrix was estimated: for example, a patient early in VF has a probability of 31% to be in ASY, 32% of still being in VF, 5% to have temporary ROSC, and 2% to have sustained ROSC after 3 min. CONCLUSION The dynamics of resuscitation can be described in terms of state transitions and a Markov probability model. This framework enables prediction of short-term clinical development, supports informed decisions during CPR, and suggests a novel area for research.

Acute ischemic heart disease alters ventricular fibrillation waveform characteristics in out-of hospital cardiac arrest.

BACKGROUND Although ventricular fibrillation waveform characteristics (VFWC) correlate with coronary perfusion pressure and may predict defibrillation outcome, recent animal data indicate that these waveform characteristics are altered in both acute myocardial infarction (AMI) and chronic coronary heart disease (CHD). We wanted to confirm these recent animal data in humans and explore the possibility for such characteristics to identify acute ischemia during cardiac arrest. METHODS Data from all adult patients admitted to hospital after out-of-hospital VF cardiac arrest in Oslo between May 2003 and July 2007 were prospectively collected. Patients were categorized into one of four pre-defined etiologic groups: patients with AMI (AMI only), patients with AMI and CHD (AMI and CHD), patients with previous CHD without evidence for a new AMI (CHD only), and patients with primary arrhythmia (PA). VFWC were analyzed from prehospital ECG tracings, and the different etiologic groups compared using ANOVA. RESULTS One-hundred-and-one patients with ECG recordings usable for VF analysis could confidently be categorized; 16 with AMI only, 34 with AMI and CHD, 41 with CHD only and 10 with PA. The two VFWC median slope (MS) and amplitude spectral area (AMSA) were significantly depressed in patients with AMI only compared to both PA (MS p=0.008, AMSA p=0.035) and CHD only patients (MS p=0.008, AMSA p=0.006). CONCLUSIONS AMI patients have depressed MS and AMSA compared to patients without AMI during VF cardiac arrest. VFWC might be helpful in identifying patients with AMI during cardiac arrest, but prospective clinical studies are warranted to assess its feasibility and clinical benefit.

Which factors influence spontaneous state transitions during resuscitation

BACKGROUND The clinical state (i.e. ventricular fibrillation/tachycardia: VF/VT, asystole: ASY, pulseless electrical activity: PEA, or return of spontaneous circulation, ROSC) during cardiopulmonary resuscitation determines patient management. We investigate how spontaneous transitions (i.e. not forced by DC shock) between these states are influenced by factors like age, gender, bystander CPR, CPR quality, proportion of time spent in a state, or the number of state transitions. METHODS Detailed recordings from CPR attempts in 304 out-of-hospital cardiac arrests in Akershus (Norway), Stockholm (Sweden), and London (UK) were obtained from modified Heartstart 4000 defibrillators. Spontaneous state transitions were studied using a non-parametric intensity regression method that can handle dynamic factors like the state history properly. RESULTS The initial state tended to preserve itself, as did cumulative time in any state. Recent DC shock, bystander CPR, location, response time, gender, compression depth, and ventilation rate were important for some transitions. More ventilation during PEA might possibly avert development to ASY and favour ROSC; otherwise observed variations in CPR quality had little impact. CONCLUSION Using a novel intensity regression approach we studied the influence of various factors on spontaneous (i.e. non-shock) state transitions during CPR. State development was largely determined by the initial state, the proportion of time spent in a state, and the transition frequency; all probably reflecting the underlying aetiology.

Using within-patient correlation to improve the accuracy of shock outcome prediction for cardiac arrest☆

BACKGROUND Analysis of the electrocardiogram (ECG) can to a certain extent predict if a cardiac arrest patient in ventricular fibrillation will get return of spontaneous circulation (ROSC) if defibrillated. The accuracy of such methods determines how useful it is clinically and for retrospective analysis. METHODS AND RESULTS We have tested the accuracy of a new shock outcome prediction algorithm that is the first to use an updating algorithm capable of learning from previous shocks within a resuscitation effort. The algorithm relies on known predictive features from the pre-shock ECG, but for each delivered shock it re-estimates the patient-dependent relationship between predictive feature value and probability of ROSC by incorporating the information from the already performed shocks. The predictive features mean-slope, median-slope, cardioversion-outcome-predictor and amplitude-spectrum-analysis originally had areas under the receiver operating characteristics curve of 0.843, 0.846, 0.837 and 0.819, respectively. The improvements in areas after applying the algorithm were (bootstrap estimate of mean improvement, 95% confidence interval in parentheses): mean-slope, 0.019 (0.00036, 0.042); median-slope, 0.024 (0.0013, 0.048); cardioversion-outcome-predictor, 0.021 (0.0010, 0.051); amplitude-spectrum-analysis, 0.026 (0.0016, 0.051). The predictions for the first shock to each patient were not included when calculating the areas, as for the first shocks the new algorithm has no previous shocks to learn from and give predictions identical to those of the original features. CONCLUSIONS It is possible to improve current shock prediction methods by using an updating algorithm capable of learning from previous shocks within a resuscitation effort.

Chest compression quality variables influencing the temporal development of ROSC-predictors calculated from the ECG during VF

BACKGROUND Predictive measures that reflect the probability of return of spontaneous circulation (ROSC) if the patient is defibrillated can be calculated from the electrocardiogram during ventricular fibrillation (VF) and ventricular tachycardia (VT). It has not been studied how the quality of chest compressions affect the development of such ROSC predictors. MATERIALS AND METHODS We have formulated a model for the effect of chest compressions on the ROSC predictor median-slope (MS). For untreated VF/VT MS is assumed to decay with time and increases in MS are attributed to the effect of chest compressions. The model correlates observed trends in MS with compression quality variables derived from measurements of compression depth and force recorded during out-of-hospital cardiac arrest. Among the quality variables tested were compression rate, depth, duty cycle, leaning depth, force, work and a novel quality indicator termed residual heart force. The model was first developed on an exploration dataset and thereafter validated against independent data. RESULTS When testing the indicators one by one, residual heart force (p<0.0001), force (p<0.0001) and work (p=0.0210) were significantly correlated to MS development. In multivariate analysis, residual heart force (p<0.0001) was the most significant indicator. Adjusting for residual heart force, there was a tendency that increased depth was associated with smaller effect of compressions (p=0.0330). CONCLUSION Using MS as an indicator of the state of the myocardium, force-based compression quality variables are better indicators of efficient CPR than compression depth. A novel indicator termed residual heart force gives the best correlation with observed trends in MS.

Modelling ventricular fibrillation coarseness during cardiopulmonary resuscitation by mixed effects stochastic differential equations

For patients undergoing cardiopulmonary resuscitation (CPR) and being in a shockable rhythm, the coarseness of the electrocardiogram (ECG) signal is an indicator of the state of the patient. In the current work, we show how mixed effects stochastic differential equations (SDE) models, commonly used in pharmacokinetic and pharmacodynamic modelling, can be used to model the relationship between CPR quality measurements and ECG coarseness. This is a novel application of mixed effects SDE models to a setting quite different from previous applications of such models and where using such models nicely solves many of the challenges involved in analysing the available data.

Information Structuring and Symbolic Representation for Analysis of Resuscitation Data

Data from resuscitation episodes offers a multitude of problems to study: shock outcome prediction, automated pulse detection, clinical state transitions,... For each of these problems ECG recordings has to be extracted from large data sets. The criteria for extraction differs from problem to problem. We present a method for information structuring of rhythm, therapy and noise annotations and show how it is possible to define extraction criteria for both the outcome prediction and the pulse detection problem preparing further signal analysis. Furthermore we present a method for making symbolic representations of resuscitation data which serves as a starting point to study the complexity of patient episodes. We demonstrate how this representation technique is applied in Markov and intensity modelling of clinical state transitions during resuscitation