Termination-of-resuscitation rule in the emergency department for patients with refractory out-of-hospital cardiac arrest: a nationwide, population-based observational study

The 2010 international consensus on cardiopulmonary resuscitation (CPR) and emergency cardiovascular care science with treatment recommendations (CoSTR) recommended validated termination-of-resuscitation (TOR) rules in the field for adult patients with out-of-hospital cardiac arrest (OHCA) [1, 2]. The TOR rules have been implemented to utilise hospital healthcare resources better, reduce hazards to emergency medical service (EMS) providers, and preserve patients’ dignity treatment is futile [3‐5]. The 2020 CoSTR softens the recommendation: the use of TOR rules to assist clinicians in deciding whether to discontinue resuscitation efforts out of the hospital or to transport the patient to a hospital while ongoing CPR, taking into consideration the social acceptability of the potential survivors and the very limited clinical validation of such rules [6]. In some Asian countries, including Japan, no TOR rules in the field can be legally implemented, and it is mandated that all cardiac arrest patients be transported to the hospital [7‐11]. Therefore, in 2013, we developed a TOR rule for emergency department physicians (Goto’s TOR rule [12]) immediately after patient arrival to the hospitals to better utilise hospital healthcare resources. The Goto’s TOR rule for deciding whether to withhold further resuscitation attempts or terminate ongoing resuscitation includes three criteria: no prehospital return of spontaneous circulation (ROSC), initial non-shockable rhythm, and unwitnessed arrest by bystanders. However, the Goto’s TOR rule has a relatively low specificity compared with other TOR rules in the emergency department [13, 14]. The American Heart Association (AHA) and European Resuscitation Council guidelines cautioned that prognostication for outcomes after cardiac arrest should be used very cautiously if the 95% confidence interval (CI) of a diagnostic test is between 90 and 95% because of its imprecision [15, 16]. Moreover, although a longer prehospital EMS-CPR duration is associated with unfavourable outcomes after OHCA [17, 18], there are no TOR rules immediately after hospital arrival, including prehospital EMS-CPR duration as a criterion. In this context, the Goto’s TOR rule should be modified to improve its accuracy by including the EMS-CPR duration in the TOR rule’s criteria.

In this study, using a nationwide population-based registry in Japan, we aimed to develop and validate a modified Goto’s TOR rule that would allow physicians to decide whether to terminate ongoing resuscitation efforts immediately after hospital arrival. Moreover, we validated the other TOR rules (Goto’s TOR [12], Korean Cardiac Arrest Research Consortium [KoCARC] I, [11] and III rules [11]) in the emergency department to compare the performance characteristics of a modified Goto’s TOR rule.

Details of attempted resuscitations performed for 504,561 patients with OHCA between 2016 and 2019 were documented in the database. Ultimately, 465,657 patients (92.2% of registered patients) who experienced OHCA were eligible for analysis. Patients were divided into the development (2016–2017; n = 231,363) and validation groups (2018–2019; n = 234,294). Baseline characteristics of the study participants are presented in Table 1. The overall 1-month survival and CPC 1–2 rates were 6.1% and 3.7%, respectively.

The result of recursive partitioning analysis for the development group in predicting 1-month mortality is depicted in Fig. 2, defined as patients with OHCA meeting all four of the following criteria: (1) initial asystole, (2) unwitnessed arrest (by bystanders or EMS providers), (3) EMS-CPR duration > 20 min, and (4) no prehospital ROSC. The results of the performance of the modified Goto’s rule in predicting 1-month mortality are shown in Table 2. In the development group, 27.5% (95% CI 27.3–27.6%) of patients fulfilled all four criteria and had a survival rate of 0.17% (95% CI 0.15–0.21%). In addition, the modified Goto’s TOR rule had a specificity of 99.2% (95% CI 99.0–99.4%), FPR of 0.8% (95% CI 0.6–1.0%), and PPV of 99.8% (95% CI 99.8–99.9%). In the validation group, 27.8% (95% CI 27.6–28.0%) of the patients met the four criteria and had a survival rate of 0.21% (95% CI 0.18–0.25%), specificity of 99.1% (95% CI 98.9–99.2%), FPR of 0.9% (95% CI 0.8–1.1%), and PPV of 99.8% (95% CI 99.8–99.8%).

The classification accuracy of the modified Goto’s TOR rule in predicting 1-month unfavourable neurological outcomes is shown in Table 3. The rates of CPC categories 1–2 in patients who met all four criteria for the modified Goto’s TOR rule were 0.04% and 0.05% in the development and validation groups, respectively. The specificity of the modified Goto’s TOR rule for the development and validation groups was 99.7%. The PPV of the modified Goto’s TOR rule for the development and validation groups was 99.9%.

Table 4 shows the external validation results of the three TOR rules after hospital arrival in predicting 1-month mortality using the validation group (n = 234,294). The specificity and PPV of the modified Goto’s TOR rule for predicting 1-month mortality were significantly higher than those of the other three TOR rules (all P < 0.001). The FPR of the modified Goto’s TOR rule for predicting 1-month mortality was significantly lower than that of the other three TOR rules (all P < 0.001). The survival rates in patients who met the KoCARC I and III rules (0.46% and 0.43%, respectively) were lower than 1%, but not in those who met the Goto’s TOR rule (1.07%). In addition, the specificity of Goto’s TOR rule was significantly lower than that of the KoCARC I and III rules (89.5% vs 95.5% and 96.3%, all P < 0.001, respectively). Therefore, the FPR of Goto’s TOR rule was higher than that of the KoCARC I and III rules: 10.5% vs 4.5% and 3.7%, respectively. The PPVs of Goto’s TOR rule and the KoCARC I and III rules were 98.9%, 99.5%, and 99.6%, respectively.

Table 5 shows the external validation results of the three TOR rules after hospital arrival in predicting 1-month unfavourable neurological outcomes using the validation group (n = 234,294). The specificity and PPV of the modified Goto’s TOR rule for predicting 1-month CPC 3–5 were significantly higher than those of the other three TOR rules (all P < 0.001). The FPR of the modified Goto’s TOR rule for predicting 1-month CPC 3–5 was significantly lower than that of the other three TOR rules (all P < 0.001). The rates of CPC 1–2 in patients who met the three TOR rules were < 1%. The specificity of Goto’s TOR rule (93.5%) was significantly lower than those of KoCARC I (98.6%) and III (98.8%): all P < 0.001. Therefore, the FPR of Goto’s TOR rule (6.5%) was higher than those of KoCARC I (1.4%) and III (1.2%). The PPVs of all the three TOR rules were > 99%.

Sensitivity analysis was performed after mean value imputation (23.4 min) of the missing data (n = 7,609, 3.0%) for EMS-CPR duration. The sensitivity analysis results are shown in Additional file 1: Results of sensitivity analysis for predicting 1-month mortality and unfavourable neurological outcome, Tables S1 and S2. The specificity and PPV of the modified Goto’s TOR rule for predicting 1-month mortality were 99.1% (95% CI 98.9–99.2%) and 99.8% (95% CI 99.8–99.8%), respectively, which were significantly higher than those of the other three TOR rules (Additional file 1: Table S1: Sensitivity analysis for predicting 1-month mortality (n = 250,517), all P < 0.001). The same was true in predicting 1-month CPC 3–5 (Additional file 1: Table S2; Sensitivity analysis for predicting 1-month unfavourable neurological outcome (n = 250,517)).

In this nationwide, population-based observational study in Japan, we developed and validated a modified Goto’s TOR rule to guide physicians in deciding whether to terminate resuscitation in patients with refractory OHCA immediately after arrival at the emergency department. A modified Goto’s TOR rule was defined to meet the following four criteria: (1) initial asystole, (2) unwitnessed arrest by any layperson, (3) EMS-CPR duration > 20 min, and (4) no prehospital ROSC. Figure 3 shows a flow chart algorithm of how the modified Goto’s TOR rule should be applied. If a patient with OHCA meets all four criteria immediately after arrival at the emergency department, the physician-in-charge should consider terminating resuscitation before performing further resuscitation efforts. Our results demonstrated that the modified Goto’s TOR rule had a specificity of 99.1%, FPR of 0.9%, and PPV of 99.8% for predicting 1-month mortality in the validation group. Moreover, patients who met the modified Goto’s TOR rule had a 1-month survival rate of less than 1% (0.17% and 0.21% in the development and validation groups, respectively), commonly regarded as medical futility [4, 22, 23]. Using the validation data set, we compared the classification accuracy of the three TOR rules (Goto’s TOR [12], KoCARC I [11], and III rules [11]) in the emergency department with that of the modified Goto’s TOR rule in predicting 1-month mortality. The modified Goto’s TOR rule had a higher specificity and PPV than the other three TOR rules in predicting 1-month mortality. These findings suggest that the modified Goto’s TOR rule is preferable to Goto’s TOR and KoCARC I and III rules. When applying the modified Goto’s TOR rule immediately after hospital arrival in the emergency department, CPR efforts could be terminated in approximately 30% of patients without advanced life support in the hospital. Unlike the international TOR rules for EMS personnel [25], the modified Goto’s TOR rule presents no burden to EMS personnel in determining the futility of CPR for patients with OHCA. Since EMS personnel in Japan do not have the authority to decline resuscitation at the scene except death, the modified Goto’s TOR rule for physicians in the emergency department is suitable for its legal authorisation.

In 2013, we analysed data from the All-Japan Utstein Registry during 2005–2009 to develop and validate a TOR rule for emergency physicians immediately after hospital arrival to better utilise hospital healthcare resources [12]. There have been significant changes in the treatment of OHCAs since the original derivation of Goto’s TOR rule. External validation studies for Goto’s TOR rule showed a relatively low specificity of 94.8% (95% CI 92.7–96.4%) [13] or 95.0% (92.8–96.7%) [14] for predicting 1-month mortality compared with other TOR rules in the emergency department (SOS-KANTO 3 [13] and Lee’s rules [14])_. This study also showed a lower specificity of Goto’s TOR rule with 89.5% (95% CI 89.0–90.0%) for predicting 1-month mortality compared with other TOR rules (Table 4). This may partly be explained by the improvement in the 1-month survival rate after OHCA in Japan, from 3.9% (2005–2009) [12] to 6.3% (Table 1. 2016–2019). Therefore, modified Goto’s TOR rule for physicians should be modified periodically with the emergence of new treatments and the evolution of social systems.

The SOS-KANTO 2012 study group [13] and Lee et al. [14] developed TOR rules after hospital arrival in 2017 and 2019, respectively. The SOS-KANTO 3 TOR rule includes three criteria: unwitnessed bystanders, asystole in the field, and emergency department [13]. Lee’s TOR rule was a combination of unwitnessed bystanders, no prehospital ROSC, and asystole in the emergency department [14]. Both rules include unwitnessed arrests by bystanders and asystole in the emergency department as a criterion. The specificities of these TOR rules were 98.6% (97.3–99.4%) [13] and 98.0 (96.4–99.0%) [14] for predicting 1-month mortality. In this study, we could not validate these two TOR rules because of the lack of rhythm data from the All-Japan Utstein Registry in the emergency department. However, the modified Goto’s TOR rule had higher specificity (> 99%) for predicting 1-month mortality in the development and validation groups.

Prehospital EMS-CPR duration is a critical factor associated with survival after OHCA [17, 18]. To date, there have been no TOR rules in the emergency department that include EMS-CPR duration as a criterion. However, the AHA 2010, 2015, and 2020 guidelines support the use of validated TOR rules in the field [25, 27, 28]. The universal TOR Guidelines state that resuscitation can be discontinued in the field by prehospital providers if the following three criteria are met: unwitnessed by EMS providers, no ROSC, and no shocks delivered at any time prior to transport [25]. In North America, it was found that application of the universal TOR Guidelines at 20 min of resuscitation in the field was able to predict futility, identifying 99.3% of survivors and 99.6% with good functional outcomes [29]. In this study, application of the modified Goto’s TOR rule identified 99.1% of survivors (Table 2) and 99.6% of neurologically intact survivors (Table 3). Accordingly, the modified Goto’s TOR rule in the emergency department, including EMS-CPR duration > 20 min, accurately identified potential OHCA survivors, similar to the universal TOR guidelines in the field.

This observational study has several limitations. First, the modified Goto’s TOR rule misclassified 137 survivors in the present validation study, resulting in a misclassification rate of 0.21% (137/65,104). Thirty-one patients (22.6%) were documented to have neurologically intact survival. Unfortunately, we were unable to determine the factors contributing to the outcomes of these patients because we could not access the original patient records. Nevertheless, the modified Goto’s TOR rule had a PPV of 99.8% for predicting 1-month mortality, which is within the acceptable range used by medical ethicists for defining futility [4, 24, 25, 29]. Second, although end-tidal CO₂ < 10 mmHg after 20 min of resuscitation was found to be predictive of futility [28], we did not analyse the results of end-tidal CO₂ monitoring because of the lack of data in the registry. Third, patients who met the modified Goto’s TOR rule after hospital arrival and achieved in-hospital ROSC but did not survive in the emergency department would be candidates for organ donors as an important ancillary benefit of refractory OHCAs. However, we could not analyse the rates of in-hospital ROSC among patients who met the TOR rules owing to a lack of in-hospital data. Fourth, although we used a uniform data collection procedure, a large sample size, and a population-based design, we cannot exclude the possibility of uncontrolled confounders that could have influenced the outcomes, such as pre-existing comorbidities, location of the arrest, quality of bystander CPR or EMS-initiated CPR, and in-hospital treatments, because the study was retrospective and observational. In addition, the extent to which poor outcomes were driven by a self-fulfilling prophecy bias was unknown. Fifth, as with all epidemiological studies, selection bias may have occurred, and the data may have lacked integrity and validity. Finally, the relevance of our results to other communities with different emergency care systems and protocols remains unclear. In particular, in some Asian countries where the TOR rule in the field is not allowed, a validation study for the modified Goto’s TOR rule in the emergency department is required before implementation.

The modified Goto’s TOR rule (which includes the following four criteria: initial asystole, unwitnessed arrest, EMS-CPR duration > 20 min, and no prehospital ROSC) with a > 99% predictor of 1-month mortality is a reliable tool for physicians treating refractory OHCAs immediately after hospital arrival.