Trend in survival after out-of-hospital cardiac arrest and its relationship with bystander cardiopulmonary resuscitation: a six-year prospective observational study in Beijing

In this prospective observational study, data of OHCA patients were collected following the “Utstein style” [15] by the Beijing, China, Emergency Medical Service (EMS) from January 2011 (data from February to June in 2011 was not collected) to October 2016. Patients who had a cardiac arrest and for whom an ambulance was dispatched were included in this study. Cardiac arrest was defined as the absence of circulation signs, regardless of whether the assessment was made by the EMS or by a bystander [15]. All cases were followed up to determine hospital discharge or death. This study was approved by the Institutional Review Board of the Capital Medical University (No. 2010SY26). Informed consent from participants was waived by the Institutional Review Board of the Capital Medical University. The study was carried out in accordance with the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines [24].

Beijing has a population of 21.73 million people, with approximately 18.80 million residing in urban areas across 16,410 km² [25]. There are two separate EMS systems in Beijing: one is Beijing 120, a public medical rescue organization of the Beijing Municipal Health Commission; the other is Beijing 999, which is subordinate to the Beijing Red Cross. Although people call 120 or 999 randomly, the reported number of dispatches for patients who called 120 was three times higher than dispatches for patients who called 999. Only cases of cardiac arrest that occurred in the Beijing urban area and that were served by Beijing 120 were included in this study (Fig. S1, Additional file 1).

Rescue ambulances equipped with electrocardiograph machines and defibrillators were dispatched to transfer OHCA patients to hospitals in this study. All ambulances were staffed with one physician, one nurse, and one driver, and they received special training in performing CPR and advanced cardiac life support (ACLS). The EMS physicians and nurses performed CPR and ACLS when indicated (except hypothermia therapy) either on site or on the ambulance following the resuscitation guidelines [26‐28]. The patients were assessed and treated by physicians in the emergency department after transfer to the receiving hospitals prior to hospital admission. It should be noted that automated external defibrillators (AEDs) were not yet available in public areas during the study period except in the Capital International Airport Terminal Three.

Patients’ characteristics were collected prospectively on the ambulances by EMS physicians who were trained to use an Utstein registration sheet for documenting OHCA events. For each OHCA case, the EMS physician completed a registration sheet including the following: (1) general demographic information, such as age and gender; (2) first witness and initial management measures, such as the location of the cardiac arrest, whether the patients received bystander CPR, and whether the witness was trained in CPR; (3) assessment and treatment by the EMS crew, such as patients’ mental status, vital signs, type of initial rhythm when the ambulance arrived, type of rhythm when the patient arrived at the emergency department, defibrillation, tracheal intubation, medication administration, probable cause of the cardiac arrest (cardiac, respiratory, traumatic, drowning, asphyxia, suicide, drug overdose, others), cardiovascular history, achieved ROSC, and survived event; and (4) timing-related information, such as the date and time of the cardiac arrest, time of the emergency call, and time the ambulance arrived. In addition, information was collected by reviewing the medical records regarding whether the patient died in the hospital, survived to admission, and survived to discharge.

Any ROSC was defined “according to a clinical assessment that comprised a palpable pulse or a blood pressure at any point during the resuscitation attempt” [15]. A survived event was defined as “ROSC sustained until the arrival at the emergency department and transfer of care to medical staff at the receiving hospital” [15]. Survival to admission was defined as “patients having ROSC after the treatment in the emergency department and being hospitalized” [15]. Survival to discharge was defined as patients having ROSC at the point of hospital discharge. A resuscitation attempted by EMS personnel was defined as a resuscitation when EMS personnel perform chest compressions or attempt defibrillation [15].

First, we analyzed the characteristics and survival rates among five groups of patients: (1) all OHCA patients, (2) those that EMS attempted to resuscitate, (3) those that EMS did not attempt to resuscitate, (4) those patients who were transported to a hospital with ROSC (that is, those who survived the event), and (5) those patients who were transported to a hospital with ongoing bystander CPR.

Second, we compared the differences in characteristics between patients who survived to discharge and those who did not survive to discharge, among all OHCA patients after excluding 28 patients who were lost-to-follow-up at hospital discharge. Student t or Mann–Whitney U tests were used for continuous data. Chi-square or Fisher’s exact tests were carried out for categorical data. We performed the same analyses among those that EMS attempted to resuscitate, those who survived the event, and those who were transported to a hospital with ongoing bystander CPR.

Third, to evaluate the trends of survival rates and pre-hospital factors over time, a trend chi-square test (linear-by-linear) was conducted for dichotomous data; linear regression was used in case of normally distributed continuous variables. The Jonckheere-Terpstra test was used for non-normally distributed continuous variables. Given that OHCA cases from February to June in 2011 and from November to December in 2016 were not collected, we analyzed only those trends from 2012 to 2015. Analyses were conducted using IBM SPSS Statistics software (version 24, IBM Corp., Armonk, New York, USA).

Finally, to analyze the relationship among bystander CPR, initial shockable rhythm, any ROSC, survived event, and OHCA survival rate, multivariate path analyses with maximum standard likelihood estimation were conducted using IBM SPSS Amos 24.0. The multivariate path analyses were performed among all OHCA patients (model 1) and among those that EMS attempted to resuscitate (model 2). Three steps were carried out to test the mediation effect as follows: (1) The joint significance of the mediation path was used to indicate that full mediation had occurred. (2) The product of the path was calculated to assess the amount of mediation effect. (3) A bootstrapping method with bias-corrected confidence estimates was used to assess the significance of the indirect path.

The acceptable model fitting values for these measures were defined as follows: (1) nonsignificant chi‐square value; (2) comparative fit index > 0.90; (3) goodness of fit index > 0.90; (4) Tucker Lewis index > 0.90; (5) root mean square error of approximation < 0.08; and (6) standardized root mean squared residual < 0.08 [29]. Effect sizes were measured using the standardized estimates and were evaluated at the following levels: 0.1 = small; 0.3 = medium; 0.5 = large. The model was modified until its indices fit well.

A series of sensitivity analyses were conducted to enhance the rigor of the study findings. First, we performed a path model (model 3) assuming that survival to admission was a mediator between survived event and survival to discharge. Second, we conducted a path model (model 4) using survival to admission as a dependent variable. Third, we performed three path models (models 5, 6, and 7) assuming that all patients who were lost-to-follow-up were dead.

As shown in Table 1, overall, the rates of patients who survived to discharge, survived to admission, and survived the event were 0.6% (31 of 5014, excluding 28 patients lost-to-follow-up at hospital discharge), 1.5% (78 of 5042), and 2.4% (121 of 5042), respectively. Among patients that EMS attempted to resuscitate, the rates of patients who survived to discharge, survived to admission, and survived the event were 1.4%, 3.4%, and 5.2%, respectively. Among patients that EMS did not attempt to resuscitate, only 3 patients survived event, and no case survived to admission or discharge. Among patients who were transported to hospital with ongoing bystander CPR, the rates of patients who survived to discharge, survived to admission, and survived the event were 3.1%, 6.1%, and 9.4%, respectively. As shown in Fig. 2, among patients who survived to admission, 39.7% (31 of 78) survived to discharge, 24.4% (19 of 78) died after being hospitalized, and 35.9% (28 of 78) were lost-to-follow-up.

Table 2 displays the trends in survival rates after OHCA, demographic characteristics of enrolled patients (age and gender), bystander CPR, bystander CPR training, and response time from 2012 to 2015. No changes were observed in the rates of survival to discharge (P_{for trend} = 0.569, excluding 28 patients lost-to-follow-up at hospital discharge), survival to admission (P_{for trend} = 0.732), or survived event (P_{for trend} = 0.659) from 2012 to 2015 for all OHCA patients. For patients that EMS attempted to resuscitate, the rates were also stable for survival to discharge (P_{for trend} = 0.846, excluding 28 patients who lost to follow-up at hospital discharge), survival to admission (P_{for trend} = 0.709), or survived event (P_{for trend} = 0.749) from 2012 to 2015. A small increase in the median age was found, from 77 (IQR: 62–84) years in 2012 to 79 (IQR: 64–85) years in 2015 (P_{for trend} = 0.004), but no change was found for gender (P_{for trend} = 0.815). Although no improvement was observed in bystander CPR training from 2012 to 2015 (P_{for trend} = 0.133), the rate of bystander CPR increased from 7.1% to 11.2% (P_{for trend} = 0.005) during this period, with a relative increase rate of 57.7%. A small increase in the response time was detected from 15 (IQR: 11–19) minutes to 16 (IQR: 12–20) minutes during this period (P_{for trend} < 0.001).

As shown in Table 1, among all enrolled OHCA cases, 484 (9.6%) received bystander CPR (271 of 484 of these bystanders had CPR training). The differences in characteristics between patients who survived to discharge and those who did not survive to discharge were summarized in Table 3 and Table S1 (Additional file 1). As shown in Table 3, although significant differences in the survival rates were observed among victims who received or did not receive bystander CPR, the rates of survival to discharge were both very low: only 2.5% (12 of 481) and 0.4% (19 of 4533) among all OHCA patients (P < 0.001); and 2.8% (12 of 429) and 1.0% (19 of 1821) among those that EMS attempted to resuscitate (P < 0.001). As shown in Table S1 (Additional file 1), among patients who survived the event, no difference was found in the rate of survival after OHCA for victims who received or did not receive bystander CPR (P = 0.904). Bystanders with previous CPR training did not change the rate of survival to discharge whether they performed CPR or not (Figure S2, Additional file 1). There was no dispatcher-assisted bystander CPR or bystander AED use during the study period.

Among the enrolled OHCA cases, only 162 (3.2%) had initial shockable rhythm (including ventricular fibrillation or pulseless ventricular tachycardia). Cases with initial non-shockable rhythm (which accounted for 93.0% [4689]) were pulseless, with no electrical rhythm and were in asystole. There were 191 cases (3.8%) with other initial rhythm noted as myocardial infarction, atrial fibrillation, or ventricular escape; however, cardiac arrest occurred during the transfer to the hospital among those cases. Among the 191 cases, only 1 (0.5%) victim survived to discharge. As shown in Table 3, among all OHCA patients, the rate of survival to discharge was 7.1% (11 of 156) in those OHCA victims with initial shockable rhythm, and 0.4% (20 of 4858) in those with initial non-shockable rhythm (including other initial rhythm; P < 0.001). Among those that EMS attempted to resuscitate, the rate of survival to discharge was 7.7% (11 of 143) in those OHCA victims with initial shockable rhythm, and 0.9% (20 of 2,107) in those with initial non-shockable rhythm (including other initial rhythm; P < 0.001; Table 3).

Any ROSC was achieved in 95 OHCA cases (1.9%). Among patients who had any ROSC, 18.9% (18 of 95) survived the event, 14.7% (14 of 95) survived to admission, and 6.3% (6 of 95) survived to discharge. As shown in Table 3, among all OHCA patients, the rate of survival to discharge was 6.7% (6 of 90) in those OHCA victims with any ROSC and 0.5% (25 of 4923) in those without (P < 0.001). Among those that EMS attempted to resuscitate, the rate of survival to discharge was 7.4% (6 of 81) in those OHCA victims with any ROSC and 1.2% (25 of 2,169) in those without ROSC (P < 0.001).

The result of the multivariate path analysis among all OHCA patients is displayed in Fig. 3 (model 1) and Table 4. The tested model had a proper fit. In this model, we excluded 28 patients lost-to-follow-up. As shown in Table 4, no direct relationship was found in bystander CPR and OHCA survival. The indirect effect of bystander CPR on OHCA survival was significant and was mainly mediated by a survived event. Overall, bystander CPR was indirectly associated with an 8.0% (β = 0.080, 95% confidence interval [CI] = 0.064–0.095, P = 0.002) increase in the possibility of survival rate (Table 4, total indirect effect). The total indirect effect of bystander CPR on the survival rate through a survived event was 6.6% (β = 0.066, 95% CI = 0.051–0.081, P = 0.002; Table 4, indirect effect), which accounted for 82.5% (0.066 of 0.080) of the total indirect effect.

A survived event has a significant impact on survival after OHCA. With every 1% increase in the survived event, the possibility of the survival rate will directly increase by 53.5% (β = 0.535, 95% CI = 0.512–0.554, P = 0.003; Table 4). As shown in Fig. 3, although initial shockable rhythm and any ROSC were also positively related to the survival rate, they directly increased the chance of survival by only 6.7% and 3.1%, respectively; their indirect effect on the survival rate through a survived event, however, was 6.4% and 5.6%, respectively. Both the effect of initial shockable rhythm and any ROSC on the survival rate were mediated by a survived event, with every 1% increase in the rates of initial shockable rhythm and any ROSC directly leading to an 11.9% and 10.4% increase in the possibility of a survived event, respectively (Fig. 3).

When analyzing the relationships between bystander CPR, initial shockable rhythm, any ROSC, survived event, and survival rate, we found that bystander CPR affected the survival rate mainly through the “Bystander CPR-Survived Event-Survival to Discharge” path (P2, Table 4). The total indirect effect of bystander CPR on the survival rate through this path was 5.1% (β = 0.051, 95% CI = 0.036–0.066, P = 0.002).

The result of the multivariate path analysis among those patients that EMS attempted to resuscitate is displayed in Fig. 4 (model 2) and Table 5. This tested model also had a proper fit. The pathways and relationships among bystander CPR, initial shockable rhythm, any ROSC, survived event, and survival rate in those patients that EMS attempted to resuscitate were consistent with the results in all OHCA patients.

The sensitivity analyses are displayed in the Figs. S3–S7 (models 3–7). The sensitivity analyses showed that the indirect effect of bystander CPR on survival rate through increasing the survived event rate remained unchanged.

This study has several limitations. First, as only 31 patients survived to discharge, this study may not have the power to explore the effect of the study factors on survival rates. The overall survival rates, however, were still very low (1.2%) even if those patients who were lost-to-follow-up were considered as survivals to discharge. To further verify our findings and to explore the effect of each factor on survival rates, a large-scale study is needed that includes more patients who survived to discharge. Second, this study was not a nationwide OHCA registry prospective study, which may hinder the generalizability of our findings. However, controlling for the inclusion bias, we consecutively screened OHCA patients from the official EMS in Beijing which has the largest number of emergency patients transferred. Third, AED was not available in public areas in Beijing; therefore, bystander defibrillation was not provided during the study period, which might reduce the effect of initial shockable rhythm on OHCA survival. Moreover, in this study, we can not test the effect of early defibrillation on increasing the survival rate. More studies exploring the pathways among early defibrillation, bystander CPR, other predictors of OHCA survival (ROSC, initial shockable rhythm, survived event), and survival rate are warranted. Finally, we did not collect data from February to June in 2011 and from November to December in 2016, which might bias the results of the study.