Protocol This trial protocol has been provided by the authors to give readers additional information about their work. Protocol for: Hasselqvist-Ax I, Riva G, Herlitz J, et al. Early cardiopulmonary resuscitation in out-of-hospital cardiac arrest. N Engl J Med 2015;372:2307-15. DOI: 10.1056/NEJMoa1405796
This supplement contains the following items: 1. Original protocol, final protocol, summary of changes. 2. Original statistical analysis plan, final statistical analysis plan, summary of changes Page Original protocol 2 Final protocol.4 Summary of changes..6 Original statistical analysis plan.6 Final statistical analysis plan..8 Summary of changes..9 1
Original protocol Protocol for study evaluating the impact of CPR before arrival of ambulance in out of hospital cardiac arrest Ingela Hasselqvist Ax, Gabriel Riva, Johan Herlitz and Leif Svensson Background The survival after out of hospital cardiac arrest is low 1. The chance to survive increases if the patient is found in ventricular fibrillation 2. Other factors such as the witnessed status, place of cardiac arrest and the emergency medical services response time has also been reported to be associated with outcome 3. Since many years back it has been suggested that if cardiopulmonary resuscitation (CPR) is started before arrival of ambulance, the chance to survive will increase 4. In reports from registry databases it has also been found that survival is higher among patients with out of hospital cardiac arrest who received CPR before arrival of emergency medical services 5. However, critical voices have suggested the presence of confounders. Thus, it has been suggested that the higher survival among patients who received bystander CPR might simply be explained by a more rapid activation of the chain of survival i.e. a more rapid call for EMS and ultimately a more rapid defibrillation 6. Furthermore, the association between the estimated delay between collapse and start of CPR and survival has not been clearly described. Based on this background the aim of this project is to a) describe the chance of survival after out of hospital cardiac arrest when CPR is started before arrival of emergency medical services when adjusted for various confounders b) to describe the association between the estimated time from collapse until start of CPR and survival. 2
Methods Patients: Patients will be recruited from The Swedish Cardiac Arrest Registry Time of recruitment: January 1, 1990 December 31, 2011. Inclusion criteria: Out of hospital cardiac arrest with CPR was started and patients were reported to The Swedish Cardiac Arrest Register Exclusion criteria: Crew witnessed out of hospital cardiac arrest and non witnessed out of hospital cardiac arrest. Sample size calculation Between 1990 and 2011 about 60.000 patients have been included in the Swedish Cardiac Register. Of these about 15% have been crew witnessed out of hospital cardiac arrest and about 35% have been non witnessed out of hospital cardiac arrest. Thus we assume that about 50% i.e. 30.000 patients will fulfil the inclusion criteria without exclusion criteria i.e. have a bystander witnessed out of hospital cardiac arrest. Data analyses Hypothesis 1) CPR, when started before arrival of emergency medical services is associated with an increased 30 day survival when adjusted for confounders. 2) There is an association between the estimated delay from collapse until start of CPR i.e. the longer the delay the lower is survival. Confounders that will be included are age, sex, place of out-of-hospital cardiac arrest, initial arrhythmia, etiology, delay from a) collapse to call for ambulance, b) collapse to defibrillation, call for ambulance until arrival of ambulance on scene. Fisher s exact test, Mann-Whitney U test and logistic regression will be used for the statistical analysis. 3
References 1. Atwood C, Eisenberg MS, Herlitz J, Rea TD. Incidence of EMS-treated out-ofhospital cardiac arrest in Europe. Resuscitation 2005;67:75-80. 2. Valenzuela TD, Roe DJ, Nichol G, Clark LL, Spaite DW, Hardman RG. Outcomes of rapid defibrillation by security officers after cardiac arrest in casinos. The New England journal of medicine 2000;343:1206-9. 3. Weaver WD, Cobb LA, Hallstrom AP, Fahrenbruch C, Copass MK, Ray R. Factors influencing survival after out-of-hospital cardiac arrest. Journal of the American College of Cardiology 1986;7:752-7. 4. Cummins RO, Eisenberg MS, Hallstrom AP, Litwin PE. Survival of out-ofhospital cardiac arrest with early initiation of cardiopulmonary resuscitation. The American journal of emergency medicine 1985;3:114-9. 5. Kitamura T, Iwami T, Kawamura T, et al. Nationwide improvements in survival from out-of-hospital cardiac arrest in Japan. Circulation 2012;126:2834-43. 6. Bardy GH. A critic's assessment of our approach to cardiac arrest. The New England journal of medicine 2011;364:374-5. Final protocol Protocol for study evaluating the impact of CPR before arrival of ambulance in out of hospital cardiac arrest Ingela Hasselqvist Ax, Gabriel Riva, Johan Herlitz and Leif Svensson Background The survival after out of hospital cardiac arrest is low 1. The chance to survive increases if the patient is found in ventricular fibrillation 2. Other factors such as the witnessed status, place of cardiac arrest and the emergency medical services response time has also been reported to be associated with outcome 3. Since many years back it has been suggested that if cardiopulmonary resuscitation (CPR) is started before arrival of ambulance, the chance to survive will increase 4. In reports from registry databases it has also been found that survival is higher among patients with out of hospital cardiac arrest who received CPR before arrival of emergency medical services 5. However, critical voices have suggested the presence of confounders. Thus, it has been suggested that the higher survival among patients who received bystander CPR might 4
simply be explained by a more rapid activation of the chain of survival i.e. a more rapid call for EMS and ultimately a more rapid defibrillation 6. Furthermore, the association between the estimated delay between collapse and start of CPR and survival has not been clearly described. Based on this background the aim of this project is to a) describe the chance of survival after out of hospital cardiac arrest when CPR is started before arrival of emergency medical services when adjusted for various confounders b) describe the association between the estimated time from collapse until start of CPR and survival. Methods Patients: Patients will be recruited from The Swedish Cardiac Arrest Registry Time of recruitment: January 1, 1990 December 31, 2011. Inclusion criteria: Out of hospital cardiac arrest with CPR was started and patients was reported to The Swedish Cardiac Arrest Register Exclusion criteria: Crew witnessed out of hospital cardiac arrest and non witnessed out of hospital cardiac arrest. Sample size calculation Between 1990 and 2011 about 60.000 patients have been included in the Swedish Cardiac Register. Of these about 15% have been crew witnessed out of hospital cardiac arrest and about 35% have been non witnessed out of hospital cardiac arrest. Thus we assume that about 50% i.e. 30.000 patients will fulfil the inclusion criteria without exclusion criteria i.e. have a bystander witnessed out of hospital cardiac arrest. Data analyses Hypothesis 1) CPR, when started before arrival of emergency medical services is associated with an increased 30 day survival when adjusted for confounders. 2) There is an association between the estimated delay from collapse until start of CPR i.e. the longer the delay the lower is survival. 5
Confounders that will be included are age, sex, place of out of hospital cardiac arrest, initial arrhythmia, etiology, delay from a) collapse to call for ambulance, b) collapse to defibrillation, call for ambulance until arrival of ambulance on scene. Fisher s exact test, Mann-Whitney U test and logistic regression (including propensity analysis) will be used for the statistical analysis. References 1. Atwood C, Eisenberg MS, Herlitz J, Rea TD. Incidence of EMS-treated out-ofhospital cardiac arrest in Europe. Resuscitation 2005;67:75-80. 2. Valenzuela TD, Roe DJ, Nichol G, Clark LL, Spaite DW, Hardman RG. Outcomes of rapid defibrillation by security officers after cardiac arrest in casinos. The New England journal of medicine 2000;343:1206-9. 3. Weaver WD, Cobb LA, Hallstrom AP, Fahrenbruch C, Copass MK, Ray R. Factors influencing survival after out-of-hospital cardiac arrest. Journal of the American College of Cardiology 1986;7:752-7. 4. Cummins RO, Eisenberg MS, Hallstrom AP, Litwin PE. Survival of out-ofhospital cardiac arrest with early initiation of cardiopulmonary resuscitation. The American journal of emergency medicine 1985;3:114-9. 5. Kitamura T, Iwami T, Kawamura T, et al. Nationwide improvements in survival from out-of-hospital cardiac arrest in Japan. Circulation 2012;126:2834-43. 6. Bardy GH. A critic's assessment of our approach to cardiac arrest. The New England journal of medicine 2011;364:374-5. Summary of changes After advice from the reviewers, we have included a propensity analysis in the final protocol. This will be described more in detail in the statistical analysis plan where also some further changes will be described. Otherwise there were no major changes in the study protocol. Original Statistical Analysis Plan DATA SOURCE The Swedish Cardiac Arrest Register will be used as the data source. ANALYSIS OBJECTIVES There are three primary objectives: 6
1. Whether CPR initiated before arrival of the emergency medical services is associated with improved 30-day survival after out-of-hospital cardiac arrest. 2. Whether delay from collapse to start of CPR is associated with 30- day survival after out- of- hospital cardiac arrest, when adjusting for potential confounders. 3. Whether CPR performed by lay persons and medical educated persons, respectively, initiated before arrival of the emergency medical services is associated with improved 30- day survival after out- of- hospital cardiac arrest, when adjusting for potential confounders. The secondary objective is to assess the association between the estimated delay time from collapse to start of CPR and 30- day survival among patients found in ventricular fibrillation. ANALYSIS SETS All out of hospital cardiac arrests, between January 1, 1990 and December 31, 2011, with CPR started and reported to the Swedish Cardiac Arrest Register will be included in the study. Crew witnessed and non- witnessed cardiac arrests will be excluded. In addition to this final population a subgroup analysis of patients found in ventricular fibrillation will be performed. ENDPOINTS AND COVARIATES The endpoint of 30- day survival is defined as confirmed survival 30 days after the cardiac arrest. Covariates included in the analyses are: For primary objective 2: age (in years), gender (male/ female), etiology (cardiac/non- cardiac), location (at home/other location), initial rhythm (VF or VT/Asystole or PEA), emergency medical services response time (in minutes) and delay from collapse to call for emergency medical service (in minutes). For primary objective 3: age (in years), gender (male/ female), etiology (cardiac/non- cardiac), location (at home/other location), and initial rhythm (VF or VT/Asystole or PEA). For secondary objective: age (in years), gender (male/ female), etiology (cardiac/non- cardiac), location (at home/other location), initial rhythm (VF or VT/Asystole or PEA), delay from collapse to call for emergency medical service (in minutes) and delay from collapse to defibrillation (in minutes). HANDLING OF MISSING DATA No imputation for missing data will be performed, i.e. multivariate analyses will only include patients with complete data. STATISTICAL PROCEDURES Fisher s exact test will be used for univariate group comparisons regarding dichotomous variables. Mann- Whitney U test will be used for univariate group comparisons regarding ordered/continuous variables. 7
Multiple logistic regression in a forward stepwise selection mode will be used to identify independent predictors of 30- day survival. Multiple logistic regression will be used to adjust for potential confounders. Two- sided tests will be used and p- values <0.05 will be considered statistically significant. All statistical analyses will be performed using SAS 9.1 for Windows. Final Statistical Analysis Plan DATA SOURCE The Swedish Cardiac Arrest Register will be used as the data source. ANALYSIS OBJECTIVES The primary objective is to analyze whether CPR initiated before arrival of the emergency medical services is associated with improved 30- day survival after out- of- hospital cardiac arrest, when adjusting for potential confounders. The secondary objective is to assess the association between the estimated delay time from collapse to start of CPR and 30- day survival. ANALYSIS SETS All out of hospital cardiac arrests, between January 1, 1990 and December 31, 2011, with CPR started and reported to the Swedish Cardiac Arrest Register will be included in the study. Crew witnessed and non- witnessed cardiac arrests will be excluded. If information on whether CPR was given before arrival of the EMS is missing the patient will be excluded from the analysis, as will also be the case if information on 30- day survival is missing. In addition to this final population a subgroup analysis of patients found in ventricular fibrillation will be performed. ENDPOINTS AND COVARIATES The endpoint of 30- day survival is defined as confirmed survival 30 days after the cardiac arrest. Covariates included in the analyses are age (in years), gender (male/ female), etiology (cardiac/non- cardiac), location (at home/other location), initial rhythm (VF or VT/Asystole or PEA), year of OHCA, emergency medical services response time (in minutes) and delay from collapse to call for emergency medical service (in minutes). For the subgroup of patients found in ventricular fibrillation also delay from collapse to defibrillation (in minutes) is included as a covariate. 8
HANDLING OF MISSING DATA Main analyses will be performed using only patients with complete data. In addition, for the primary objective, we will also perform multiple imputation using the Markov chain Monte Carlo method. STATISTICAL PROCEDURES Fisher s exact test will be used for univariate group comparisons regarding dichotomous variables. Mann- Whitney U test will be used for univariate group comparisons regarding ordered/continuous variables. Logistic regression will be used for calculation of odds ratios with corresponding confidence intervals and for analyses of interaction. To adjust for potential confounders propensity scoring will be used, where the propensity score will be used as an adjustment factor in a multiple logistic regression model. Two- sided tests will be used and a p- value of <0.05 for the primary objective and <0.01 for all other analyses will be considered statistically significant. All statistical analyses will be performed using SAS 9.3 for Windows. Summary of changes ANALYSIS OBJECTIVES Primary objective was changed to only include analyze of whether CPR initiated before arrival of the emergency medical services is associated with improved 30- day survival after out- of- hospital cardiac arrest and to include adjustment for potential confounders. Secondary objective was changed to assess the association between the estimated delay time from collapse to start of CPR and 30- day survival (formerly the second of three primary endpoints in the original plan). No analyzes of whether CPR performed by lay persons and medical educated persons, respectively, initiated before arrival of the emergency medical services was associated with 30- day survival, was included in the final plan. ANALYSIS SETS Two additional exclusion criteria were introduced, i.e. missing information on whether CPR was given before arrival of the EMS and missing information on 30- day survival. ENDPOINTS AND COVARIATES The year of OHCA was added as an additional covariate. HANDLING OF MISSING DATA An additional analysis for the primary endpoint was introduced by performing multiple imputation using the Markov chain Monte Carlo method. 9
STATISTICAL PROCEDURES Interaction analyses were added and propensity analysis was introduced to adjust for potential confounders. Except for the primary endpoint the significance level was changed from p- values <0.05 to <0.01. 10