Study design and setting
In this study, we analyzed the database of the CRITICAL study, which is a multicenter prospective observational data registry designed to accumulate both pre-and in-hospital information on OHCA treatment. A complete description of the study methodology has been previously reported [8].
Population and settings
The target area of the CRITICAL study is Osaka Prefecture in Japan, which has an area of 1897 km2 and a residential population of 8,839,469 inhabitants as of 2015; 48.1% of the population are males, 25.8% of whom are ≥65 years old [9]. Osaka had 535 hospitals (108,569 beds) in 2013 [10]. A total of 280 hospitals accepted emergency patients from ambulances. Of these, 16 hospitals have critical care medical centers (CCMCs) that can accept severely ill emergency patients [10]. Fifteen CCMCs and one non-CCMC with an emergency care department in Osaka participated in this study. Approximately 30% of patients with OHCA in Osaka are transported to and treated at CCMCs [10]. This CRITICAL study, including a retrospective analysis, was approved by the ethics committee of the Kyoto University (R-1045). The requirement for informed consent was waived.
Data collection and quality control
This registry’s data collection and quality control details have been reported elsewhere [8]. Pre-hospital data on OHCA patients were obtained from the All-Japan Utstein Registry, and data were uniformly collected according to the Utstein-style international guidelines for reporting OHCAs. Each EMS personnel completed a data form in cooperation with the attending physicians in charge of the patient. In this study, a doctor car/helicopter was defined as an ambulance/helicopter in which a physician traveled from the scene of a patient’s collapse to the patient’s arrival at the hospital. The unified protocol on how to dispatch the physician to the location of the occurrence of cardiac arrest and the role of the pre-hospital physician in Japan are not clearly defined. For in-hospital data collection and quality control, the CRITICAL registry has collected substantial data on patients with OHCA after arrival at the hospital, as detailed have been provided in a previous paper. For the current registry, anonymized data were entered into the web sheet either by the physician or medical staff in collaboration with the attending physician in charge of the patient. The pre-hospital and in-hospital data were uploaded to the registry system, logically checked by the computer system, and confirmed by the working group, which consisted of experts in emergency medicine and clinical epidemiology.
Study patients
We enrolled all consecutive patients with OHCA (aged ≥18 years) for whom resuscitation was attempted and then transported to the participating institutions between January 1, 2012, and December 31, 2019. This study excluded patients with OHCA who did not receive cardiopulmonary resuscitation (CPR) from physicians after hospital arrival and those who disagreed with our registry (refusal by the patient or the patient’s family). Additionally, patients with OHCA who were not of medical origin, without witness, transported by a doctor car or helicopter, and whose BT on arrival was not available were excluded. The requirement for obtaining individual informed consent to review patient outcomes was waived.
Outcomes
The primary outcome measure was 1-month survival with favorable neurological outcomes after OHCA. A favorable neurological outcome was defined as a cerebral performance category (CPC) score of 1 or 2 [11]. The secondary outcome measure was the 1-month survival. Outcome data were also prospectively collected and included as follows [8]: 1-month survival and neurological status 1-month after OHCA occurrence, using the CPC scale (category 1, normal cerebral performance; category 2, moderate cerebral disability; category 3, severe cerebral disability; category 4, coma or vegetative state; and category 5, death/brain death). The survivors underwent neurologic assessment by the physician in charge 1-month after the event. The patients in this analysis included those with valid BT data available on hospital arrival. Temperatures were recorded from the ear, rectum, urinary bladder, axilla, and others. The patients were categorized into three groups according to their temperature on arrival at the hospital, based on clinical significance according to previous studies [7, 12]. The low BT group had an initial temperature of 35.9 °C or below; the normothermia group was defined as the group with a temperature from 36.0 °C to 36.9 °C, and the other group was the higher temperature group defined as the temperature of ≥37.0 °C.
Statistical analysis
We described the characteristics of the patients in each BT group. Data are presented as mean ± standard deviation for continuous variables and percentages for categorical variables. Categorical data were compared using the chi-square test. Continuous data were compared using the Kruskal–Wallis test. To assess the association between BT and outcomes, we used a univariable logistic regression model for crude odds ratios (ORs) and performed a multivariable logistic regression analysis to adjust for potential resuscitation factors associated with 1-month survival and 1-month survival with favorable neurological outcomes, and the ORs and 95% confidence intervals (CIs) were calculated. The independent variables considered in this analysis included the following: age (continuous value), sex (male, female), origin of arrest (cardiac, non-cardiac), bystander CPR (no, yes), initial rhythm (shockable [ventricular fibrillation and pulseless ventricular tachycardia] or non-shockable [pulseless electrical activity and asystole], which were defined as the first documented rhythm at the scene), time from emergency medical service (EMS) call to the hospital (continuous variable), return of spontaneous circulation before hospital admission (no, yes), season (spring: March, April, and May; summer; June, July, and August; autumn: September, October, and November; and winter: December, January, and February). This study excluded unwitnessed cases from the analysis. First, because the Utstein template recommends focusing on witnessed cases for resuscitation [11], and second, it is difficult to assess the recorded temperature of unwitnessed cases together with witnessed cases because the time of collapse is not known among unwitnessed cases. For the second analysis, restricted cubic splines (RCS) were used to detect the possible nonlinear dependency of the relationship between BT and outcomes using four knots at prespecified locations according to the percentiles of the distribution of BT (5th, 25th, 75th, and 95% percentiles) [13]. In the analysis, we conducted a post-hoc analysis of the linearity of the association between BT and outcome among OHCA patients, and we found a statistically nonlinear association between neurological outcome and BT among them. Furthermore, we compared the fitness of the linear and nonlinear models for the subjects. As a result, the nonlinear model was better fitted than the linear model by the F test. Moreover, we described the distribution of BT by the first documented rhythm on EMS arrival (ventricular tachycardia [VF]/pulseless ventricular tachycardia [pVT], pulseless electrical activity [PEA]/asystole) and performed subgroup analyses by rhythm using multivariate logistic regression analysis. All p-values were two-sided, and statistical significance was set at p < 0.05. All statistical analyses were performed using STATA version 16.0 SE software (Stata Corp LP) and R studio (Version 1.2.5033). The aforementioned dose-response analyses by RCS were performed using R studio with the package of “rms.”