Association between neuromuscular blocking agent use and outcomes among out-of-hospital cardiac arrest patients treated with extracorporeal cardiopulmonary resuscitation and target temperature management: A secondary analysis of the SAVE-J II study

Background Neuromuscular blocking agents are used to control shivering in cardiac arrest patients treated with target temperature management. However, their effect on outcomes in patients treated with extracorporeal cardiopulmonary resuscitation is unclear. Methods This study was a secondary analysis of the SAVE-J II study, a retrospective multicenter study of 2175 out-of-hospital cardiac arrest patients treated with extracorporeal cardiopulmonary resuscitation in Japan. We classified patients into those who received neuromuscular blocking agents and those who did not and compared in-hospital mortality and incidence rates of favorable neurological outcome and in-hospital pneumonia between the groups using multivariable regression models and stabilized inverse probability weighting with propensity scores. Results Six hundred sixty patients from the SAVE-J II registry were analyzed. Neuromuscular blocking agents were used in 451 patients (68.3%). After adjusting for potential confounders, neuromuscular blocking agents use was not significantly associated with in-hospital mortality (aHR 0.88; 95% CI, 0.67–1.14), favorable neurological outcome (aOR 0.85; 95% CI, 0.60–1.11), or pneumonia (aOR 1.52; 95% CI, 0.85–2.71). The results for in-hospital mortality (aHR 0.89; 95% CI, 0.64–1.25), favorable neurological outcome (aOR 0.94; 95% CI, 0.59–1.48) and pneumonia (aOR 1.59; 95% CI, 0.74–3.41) were similar after weighting was performed. Conclusions Although data on the rationale for using neuromuscular blocking agents were unavailable, their use was not significantly associated with outcomes in out-of-hospital cardiac arrest patients treated with extracorporeal cardiopulmonary resuscitation and targeted temperature management. Neuromuscular blocking agents should be used based on individual clinical indications.


Introduction
Out-of-hospital cardiac arrest (OHCA) is a significant public health problem, with an estimated global incidence of 55 cases per 100,000 person-years. 13][4] A recent RCT has reported that there was no significant difference in outcome between TTM at 33 °C and early treatment of fever. 5Consequently, current international guidelines recommend actively preventing fever. 6However, TTM is still considered to be a "good practice" for the management of patients with PCAS. 6Shivering is a common complication of TTM which makes it difficult to control temperature, 7 increases oxygen consumption, 8 and reduces brain oxygen supply. 9ontrol of shivering is essential for temperature management to prevent further brain damage. 7xtracorporeal cardiopulmonary resuscitation (ECPR) is cardiopulmonary resuscitation (CPR) combined with extracorporeal membrane oxygenation (ECMO) for refractory cardiac arrest. 10he use of ECPR is increasing worldwide 11,12 and a meta-analysis based on four RCTs [13][14][15][16] reported its benefit for neurological outcome. 17It is inferred that patients who were received ECPR have more severe brain damage and are more vulnerable to further brain injury because of their longer cardiac arrest duration and inflammation by ECMO use 18 which is considered a contributory factor to secondary brain injury. 19Preventing further damage in these patients is essential.
Neuromuscular blocking agents (NMBAs) are frequently used to control shivering during TTM. 7Their use may improve outcomes in OHCA patients by suppressing shivering and optimizing TTM. 20Furthermore, NMBA have been reported to improve outcomes in acute respiratory distress syndrome (ARDS), 21 thus they could potentially be applicable for treating severe hypoxemia or ARDS following OHCA.4][25][26][27][28][29] In addition, few of the studies included patients who underwent ECPR.Therefore, the effects of NMBAs in ECPR patients have not been well studied.This study aimed to investigate the association between NMBAs and outcomes in OHCA patients treated with ECPR.

Study design
This retrospective observational study used data from the Study of Advanced life support for Ventricular fibrillation with Extracorporeal circulation in Japan (SAVE-J II) patient registry to conduct a secondary analysis. 30All patient data were de-identified prior to analysis.Approval was obtained from the institutional review board of Dokkyo Medical University.The requirement for patient consent was waived owing to the retrospective nature of the study.

Participants
The SAVE-J II study enrolled 2157 consecutive patients over 18 years of age who presented to the emergency department with OHCA and underwent ECPR from 36 institutions in Japan between January 1, 2013, and December 31, 2018. 31Details of the ECPR procedure have been previously described. 30Our study included patients from the registry who received ECPR (i.e., ECMO insertion while in a state of cardiac arrest), underwent TTM and received sedatives within 24 hours of admission to the intensive care unit (ICU).These inclusion criteria were selected because all eligible patients in two previous studies that evaluated the association between the use of NMBAs and outcomes received sedatives. 25,28edative agents are recommended in the acute phase of TTM and are considered standard of care for patients with PCAS undergoing TTM. 32atients with non-cardiac conditions such as aortic dissection or aneurysm, hypothermia, primary cerebral disorder, infection, drug intoxication, trauma, suffocation, and drowning were excluded from analysis.We also excluded patients who achieved return of spontaneous circulation at hospital arrival or at ECMO initiation, those who were transferred from another hospital, and patients with incomplete or missing data regarding NMBAs, TTM, sedatives, or outcomes.Furthermore, to exclude patients who died or received a 'do not attempt resuscitation' (DNAR) order within 24 hours of ICU admission, we also excluded those who experienced either event by the second calendar day following their admission.This was because the registry only collected the date of these events, without exact time.

Data collection
Patient demographic and clinical data were collected from the SAVE-J II study registry.Data details are presented in Supplementary Table 1.

Exposure
The primary exposure in this study was any NMBA use, either as a bolus or continuous infusion, at ICU admission or within 24 hours of admission.

Outcomes
The primary outcome was in-hospital mortality.Secondary outcomes were favorable neurological outcome at hospital discharge and pneumonia during hospitalization.Favorable neurological outcome was defined as Cerebral Performance Category score 1 or 2. 33 Pneumonia was defined as the presence of infiltrates on chest radiography in conjunction with detection of bacteria in sputum cultures and at least one of the following: body temperature >38.0 °C, white blood cell count >12,000 cells/mL, and purulent sputum.Pneumonia was chosen as a secondary outcome for multivariable analysis because its definition was pre-established.Other outcomes like ventilatorassociated pneumonia and atelectasis were also collected but were not included in the multivariable analysis, as they were solely subject to physician interpretation in the registry.

Covariates
Adjustments were made for age, sex, witnessed cardiac arrest, bystander CPR, initial rhythm, location of cardiac arrest, and estimated cardiac arrest duration, and target temperature for TTM.The estimated cardiac arrest duration was defined as follows: for patients whose location of cardiac arrest was an ambulance, the time from cardiac arrest to the establishment of ECMO; for patients whose location of cardiac arrest was other than an ambulance, the time from calling an ambulance to the establishment of ECMO.The target temperature for TTM was determined based on the target temperature at the beginning of TTM.Patients with a target temperature 35 °C were classified as 'hypothermia'; those with target temperature !35.5 °C were classified as 'normothermia' according to a previous research. 34If the target temperature at the beginning of TTM was missing, the final target temperature after any adjustment was used.The adjusted covariates were selected based on previous studies [35][36][37][38][39][40][41] and their clinical importance.

Statistical analysis
Patients were grouped according to NMBA use (NMBA and no NMBA groups).Continuous variables are presented as medians with interquartile range (IQR) and were compared using the Mann-Whitney U test.Categorical variables are presented as numbers with percentage and were compared using the chi-square or Fisher's exact test as appropriate.
A multivariable Cox shared frailty model was constructed to investigate the association between NMBA use and hospital mortality while adjusting for covariates as described above.A Cox shared frailty model is a survival analysis model that accounts for unobserved heterogeneity among individuals in the same cluster. 42Multivariable logistic regression was performed with generalized estimating equation (GEE) modeling to investigate the association between NMBA use and secondary outcomes while accounting for patient clustering within hospitals.The adjusted covariates are described above.
Propensity score analysis was also conducted as sensitivity analysis to reduce potential confounders between NMBA use and outcomes.To calculate propensity scores for NMBA use, multivariable logistic regression to adjust for potential confounders (age, sex, witnessed cardiac arrest, bystander CPR, initial rhythm, location of cardiac arrest, estimated cardiac arrest duration, and target temperature for TTM) was performed.The c-statistic was used to assess the model's discriminative ability.Stabilized inverse probability weighting (sIPW) was used to estimate the average treatment effect. 43For patients who received NMBs weighted by the inverse of PS (1/PS) and those who did not receive NMBs weighted by the inverse of 1 minus their PS (1/(1-PS)).To prevent undue influence of extreme weights assigned to patients with a low probability of receiving NMBs, stabilized IPWs were created by multiplying the marginal probability of receiving NMBs by the IPW.Less than 10% of an absolute standardized mean difference for the covariates indicated a good covariate balance. 44Weighted Cox proportional hazards models were used to calculate adjusted hazard ratios (HRs) to evaluate the association between NMBA use and hospital mortality.Weighted logistic regression models were used to calculate adjusted odds ratios (ORs) to evaluate the association between NMBA use and secondary outcomes, both weighted by the stabilized IPW described above.For the weighted analyses, the sandwich variance estimator was used to calculate confidence intervals (CIs) and p-values.
Subgroup analyses were conducted using an interaction term between NMBA use and subgroups to evaluate potential effect modifiers.Subgroups were chosen based on a previous study, 34 the potential for a lower targeted temperature to cause more shivering, and the possibility that patients with longer cardiac arrest duration may have a more severe brain injury and benefit more from NMBA use in preventing secondary brain injury.The subgroup categories were age (<65 vs. !65 years), target temperature for TTM ( 35 °C vs. >35 °C), and estimated cardiac arrest duration ( 50 minutes vs. >50 minutes).The interaction was tested using the Wald test.
All tests were two-tailed.p < 0.05 was considered significant.Statistical analyses were performed using R software version 4.2.0 (The R Foundation for Statistical Computing, Vienna, Austria).

Patient characteristics
Among the 2157 patients in the SAVE-J II registry, 660 met the criteria described above and were included for analysis.NMBAs were administered to 451 patients (the NMBA group).A study flowchart is shown in Fig. 1.Patient characteristics are summarized in Table 1.Patients in the NMBA group were significantly younger (median age, 58 vs. 63 years; p < 0.001), more likely to be male (87.4% vs. 80.4%; p = 0.019), and had a shorter estimated cardiac arrest duration (median time, 50 vs.52 minutes; p = 0.017).Overall, 80.3% of patients were treated with hypothermia.The proportion of patients treated with hypothermia was significantly higher in the NMBA group (88.0% vs. 63.6%;p < 0.001).In the NMBAs group, 60.9% received rocuronium and 39.1% received vecuronium.Among the patients whose infusion methods were available, 91.5% received NMBAs via continuous infusion (Supplementary Table 2).Midazolam was the most used sedatives (68.7%) and propofol usage was significantly lower (31.0%vs. 45.9%;p < 0.001) in the NMBA group.Analgesics were more used in the NMBA group (90.7% vs. 84.2%;p = 0.018).In addition, the incidence rates of pneumonia (31.2% vs. 22.3%; p = 0.019), ventilatorassisted pneumonia (27.3% vs. 14.1%;p < 0.001), and atelectasis (13.8% vs. 6.9%;p = 0.011) were significantly higher in the NMBA group as well.No significant differences were observed in incidence rates of temperature instability events and the various outcomes (Table 2).

Propensity score analysis
The C-statistic for the propensity score model was 0.700.After applying sIPW, the covariates between the NMBA and no NMBA groups were well-balanced, as shown in Table 3.

Pneumonia during hospitalization
Incidence of pneumonia during hospitalization was 31.2% in the NMBA group and 22.3% in the no NMBA group (Table 2).
In the multivariable GEE logistic regression model, NMBA use and pneumonia were not significantly associated before (adjusted OR 1.52; 95% CI, 0.85-2.71;p = 0.16; Table 4), nor  e Patients in whom spontaneous circulation was confirmed at the first emergency medical system team evaluation and then developed cardiac arrest.f Estimated cardiac arrest duration was defined as follows: for patients whose location of cardiac arrest was ambulance, the time from cardiac arrest to the establishment of ECMO; for patients whose location of cardiac arrest was other than ambulance, the time from calling an ambulance to the establishment of ECMO.

Subgroup analyses
In the subgroup analyses, no significant interactions were found between NMBA use and age, target temperature, or estimated cardiac arrest duration (Supplementary Table 3).

Summary of the results
In this retrospective cohort study of OHCA patients who received ECPR and were treated with TTM, we evaluated the association between NMBA use and in-hospital mortality, neurological outcome, e Pneumonia was defined as follows: infiltrates on chest radiography and detection of bacteria in sputum cultures and at least one of the following: body temperature >38.0 °C, white blood cell count >12,000 cells/mL, and purulent sputum.f Ventilator-associated pneumonia was defined based on the treating physician's judgement and medical records.g Atelectasis was defined as new infiltrates on chest radiography during ECMO management.
and incidence of pneumonia during hospitalization.After adjusting for covariates using multivariable regression models and propensity score analysis, no significant association was found between NMBA use and these outcomes.

Relationship to previous studies
4][25][26][27][28][29] Although several observational studies reported an association between NMBA use and better a Patients in whom spontaneous circulation was confirmed at the first emergency medical system team evaluation and then developed cardiac arrest.b Estimated cardiac arrest duration was defined as follows: for patients whose location of cardiac arrest was ambulance, the time from cardiac arrest to the establishment of extracorporeal membrane oxygenation; for patients whose location of cardiac arrest was other than ambulance, the time from calling an ambulance to the establishment of extracorporeal membrane oxygenation.outcomes, 25,26,28,29 our study found no such association.This disparity can be explained by several reasons.
The target population of this study differed from those in previous studies.Previous studies included few patients on ECMO, while ours only included those who underwent ECPR.NMBAs enable strict temperature management by preventing shivering.However, when ECMO is used, blood temperature can be directly regulated; therefore, patient body temperature can be controlled without the use of NMBAs.As a result, there may not be a significant association between NMBA use and patient outcomes.Indeed, the incidence of temperature-instability events in our study did not differ significantly between patients who received NMBAs and those who did not.
Several biases might have been present in previous studies which reported an association between NMBA use and favorable outcomes.First, patients who experience shivering, which is the main reason for administering NMBs, may have a lower degree of cardiac arrest-induced brain injury.The reasons are as follows.
One study has reported that shivering is associated with favorable neurological outcomes in PCAS patients treated with hypothermia, 45 suggesting that shivering is associated with less brain injury.In general, NMBAs are used to treat shivering; consequently, there is an association between the two.Therefore, it is possible that patients who received NMBAs had less brain injury than those who did not and consequently experienced a better outcome.This association is particularly strong when NMBAs are administered in a stepwise fashion according to the severity of shivering.In several previous studies which reported an association between NMBA use and good outcomes, they were administered in a stepwise protocol. 25,28These studies might have been affected by such associations.To address this bias, we adjusted for several factors that influence brain injury related to cardiac arrest.Therefore, the extent of baseline brain injury owing to cardiac arrest seemed to be comparable in the NMBA and no NMBA groups.
Second, there was potential bias caused by patients who died or were subjected to a decision to limit treatment soon after ICU admission.It is common for PCAS patients to experience either of these 46,47 and such patients generally experience a poor outcome. 46,47Furthermore, these patients may not have received an NMBA and were therefore classified in the no NMBA group, which would have resulted in an overestimation of the NMBA effect, as only patients who survived until NMBAs were administered were classified in the NMBA group.This immortality time bias is a common problem of observational studies. 48Several previous studies that reported an association between NMBAs and good outcomes did not mention death or the decision to limit treatment soon after ICU admission. 28,29Therefore, immortality time bias might have been present.To address this bias, we excluded patients who died or received a DNAR order by the second calendar day following their admission.After addressing these biases, our findings are similar to those of previous randomized controlled trials. 23,27Our findings might provide a more accurate estimate of the effects of NMBAs than previous observational studies.
Significant differences were observed in sedatives, with the NMBA group using less propofol.However, the impact of this for patient outcomes appear to be minimal because a study has reported no significant differences in outcomes between propofol and midazolam users among patients treated with ECPR. 49neumonia is a common complication of PCAS 50 and is more likely in patients who receive NMBAs. 28Therefore, we also investigated the association between NMBA use and pneumonia.The incidence of pneumonia was significantly higher in patients treated with NMBAs in the crude analysis; however, after adjusting for covariates, no significant association was found.This is consistent with the findings of a previous study. 28NMBA use may contribute to pneumonia development by suppressing the cough reflex and impeding the elimination of airway secretions.Nevertheless, other factors such as severity of patient illness and differences in ICU management across hospitals may also play a role in pneumonia development.This might explain the lack of a significant association after adjusting for potential confounders.

Significance and implications
Our study found no significant association between NMBA administration and favorable outcomes.While these results may discourage the routine use of NMBAs, they do not necessarily imply that NMBA administration should be avoided.Various physiological observations have suggested the potential adverse effects of shivering, [7][8][9] which is effectively controlled by NMBAs.NMBAs may be a reasonable option for controlling shivering.We believe it is reasonable to evaluate the indications for NMBAs administration on a case-bycase basis.

Strengths and limitations
To the best of our knowledge, this is the first study to evaluate the association between NMBA use and outcomes in OHCA patients who underwent ECPR and TTM.However, it had several limitations.First, data regarding shivering were not available.Therefore, it was not possible to determine whether NMBAs were administered to treat or prevent shivering.If NMBAs were administered to treat shivering, an association between NMBA use and less brain damage would have been established, as mentioned above.Consequently, we adjusted for several factors that influence brain damage, which should have attenuated this bias.Second, NMBA use was evaluated within the first 24 hours of ICU admission.The impact of the timing of NMBA use on outcomes beyond this period is unknown.However, this time frame is reasonable because shivering generally occurs during the TTM induction phase. 32Third, we did not investigate the presence of ARDS, for which NMBAs have been shown to be effective. 21Data regarding the presence of ARDS prior to cardiac arrest were not available; however, we believe that few patients had ARDS because most study patients had a cardiac cause for OHCA.Finally, the study was relatively small sample size and observational in design; therefore, these non-significant results might be due to being under-powered and causality could not be established.A randomized controlled trial with an adequate sample size would be required to confirm these findings.

Conclusions
In

Table 1 -
Patient characteristics overall and according to NMBA use.
a Data are reported as medians [interquartile range] or numbers (percentage).b NMBAs were infused within 24 hours of intensive care unit admission.c NMBAs were not infused within 24 hours of intensive care unit admission.d Pearson's chi-square test, Fisher's exact test, or Mann-Whitney U test as appropriate.
NMBA, neuromuscular blocking agent; TTM, targeted temperature management; ECMO, extracorporeal membrane oxygenation.a Data are reported as medians [interquartile range] or numbers (percentage).b NMBAs were infused within 24 hours of intensive care unit admission.c NMBAs were not infused within 24 hours of intensive care unit admission.d Pearson's chi-square test, Fisher's exact test, or Mann-Whitney U test as appropriate.

Table 3 -
Patient characteristics before and after stabilized inverse probability weighting.

Table 4 -
Multivariate Cox shared frailty model and multivariable logistic regression model with generalized estimating equations: the association between neuromuscular blocking agent use and outcomes., neuromuscular blocking agent; CPC, Cerebral Performance Category; HR, hazard ratio; CI, confidence interval; OR, odds ratio; Ref., reference.Estimated cardiac arrest duration was defined as follows: for patients whose location of cardiac arrest was ambulance, the time from cardiac arrest to the establishment of extracorporeal membrane oxygenation; for patients whose location of cardiac arrest was other than ambulance, the time from calling an ambulance to the establishment of extracorporeal membrane oxygenation.a The multivariate Cox shared frailty model was adjusted for age, sex, witnessed cardiac arrest, bystander cardiopulmonary resuscitation, initial rhythm, location of cardiac arrest, and estimated cardiac arrest duration.b The logistic regression model with generalized estimating equations was adjusted for age, sex, witnessed cardiac arrest, bystander cardiopulmonary resuscitation, initial rhythm, location of cardiac arrest, and estimated cardiac arrest duration. NMBA

Table 5 -
Association between neuromuscular blocking agent use and outcomes after inverse probability weighting., neuromuscular blocking agent; CPC, Cerebral Performance Category; HR, hazard ratio; CI, confidence interval; OR, odds ratio; Ref., reference.Adjusted HR was based on a weighted Cox regression model; adjusted OR was based on a weighted logistic regression model. NMBA this retrospective observational study of ECPR patients who received TTM, NMBA use within 24 hours of ICU admission was not significantly associated with patient outcomes.NMBA use should be based on individual clinical indications.gency and Critical Care Medicine, Nippon Medical School, Tokyo, Japan d Department of Emergency and Critical Care Medicine, St. Luke's International Hospital, Tokyo, Japan e Department of Emergency and Critical Care Medicine, Hyogo Emergency Medical Center, Kobe, Japan f Department of Emergency Medicine, Teikyo University School of Medicine, Tokyo, Japan g Department of Emergency, Disaster and Critical Care Medicine, Kagawa University Hospital, Kagawa, Japan R E F E R E N C E S