Sex-Specific Differences in Survival After Out-of-Hospital Cardiac Arrest

A Nationwide, Population-Based Observational Study

Yoshikazu Goto; Akira Funada; Tetsuo Maeda; Hirofumi Okada; Yumiko Goto


Crit Care. 2019;23(263) 

In This Article


Study Design and Setting

This nationwide, population-based observational study included 386,535 adult patients aged ≥ 18 years with OHCA. In all these patients, resuscitation was attempted by EMS personnel in Japan between January 1, 2013, and December 31, 2016. In Japan, nearly 127 million individuals reside in an area of approximately 380,000 km2. Further, approximately two thirds of Japan comprises uninhabited mountainous terrain.[22] The Fire and Disaster Management Agency (FDMA) of Japan supervises a nationwide EMS system, whereas local fire stations operate local EMS systems. In 2017, Japan had 732 fire departments and 5140 ambulance teams.[23] During the study period, all EMS personnel performed CPR following the Japanese CPR guidelines and attempted resuscitation by using automated external defibrillators, inserting airway adjuncts and peripheral intravenous catheters, and administering Ringer's lactate solution.[23–25] Only specially trained emergency life-saving technicians are permitted to insert tracheal tubes and administer intravenous epinephrine after receiving online instructions from a physician.[23] Except in special situations, such as decapitation, incineration, decomposition, rigor mortis, and dependent cyanosis, EMS personnel in Japan are legally prohibited from terminating resuscitation in the field. Most patients with OHCA were given CPR by EMS personnel before transport to a hospital.

Data Collection and Quality Control

In 2005, the FDMA launched an ongoing prospective population-based observational study including all patients with OHCA in Japan who received resuscitation by EMS personnel.[23] EMS personnel and the physician in charge at each center recorded data from the patients using an Utstein-style recommended guideline template.[26,27] The data were transferred to individual local fire stations and subsequently integrated into the registry on the FDMA database server. The database application automatically checked the patient data for consistency, which was again verified by the FDMA. The data were transferred to and stored in a nationwide database that was developed by the FDMA for public use. The FDMA granted us permission to access the anonymized data for this study.

The characteristics included in the dataset were as follows: patient sex and age, etiology of arrest, initially identified cardiac rhythm, presence and relation of bystander witnesses (e.g., family member, a layperson other than family, or EMS personnel), maneuver of bystander CPR, time of collapse, receipt of emergency calls, time of vehicle arrival at the scene and EMS initiation of CPR, 1-month survival, and neurologically intact survival. The etiology of arrest was presumed to be cardiac unless suitable evidence suggested a nonmedical cause (e.g., trauma, accidental hypothermia, hanging, drowning, drug overdose or poisoning, or asphyxia) or another noncardiac cause, such as respiratory or cerebrovascular disease or malignant tumors. The physicians in charge determined the etiology of arrest. Neurological outcomes were defined using the Cerebral Performance Category (CPC) Scale scores (1: good cerebral performance, 2: moderate cerebral disability, 3: severe cerebral disability, 4: coma or vegetative state, 5: death).[26] The CPC scores were determined by the physician in charge.

Study Endpoints

The primary study endpoint was neurologically intact survival (CPC Scale score = 1 or 2 at 1 month). The secondary endpoint was 1-month survival after OHCA.

Statistical Analysis

Continuous variables were either expressed as medians and 25th–75th percentiles or as means and standard deviation. Categorical variables were expressed as numbers and percentages. Effect size and variability were reported as odds ratios (ORs) with 95% confidence intervals (CIs). To determine the differences in 1-month outcomes after OHCA based on sex, the patients were divided into the following eight groups: < 30, 30–39, 40–49, 50–59, 60–69, 70–79, 80–89, and ≥ 90 years. The Kruskal–Wallis and Dunn's post hoc tests were used to compare continuous variables. Chi-square test was used to compare categorical variables, and univariate logistic regression analysis was performed to compare the characteristics and outcomes between men and women. Multivariate logistic regression analyses were performed after adjusting for the differences in patient baseline characteristics for all unmatched patients and each age group for both matched and unmatched patients. Hierarchical propensity score matching was used to adjust for covariates when comparing the outcomes in men and women. Potential prehospital confounders in the analytic model were selected based on biological plausibility and data reported in previous studies. Multivariate logistic regression analysis in both matched and unmatched patients included the following 12 prehospital variables: calendar year (as a categorical variable), Japan geographic regions (rural or urban area), age (as a continuous variable), sex (men or women), presence of a witness (no witness, witnessed by family member, or nonfamily member), initial cardiac rhythm (shockable or nonshockable), cause of arrest (presumed cardiac cause or noncardiac cause), bystander CPR (yes or no), use of advanced airway management (yes or no), epinephrine administration (yes or no), EMS response time (as a continuous variable), and duration between the call to EMS and hospital arrival (as a continuous variable). Hierarchical propensity score matching analyses of the eight groups were performed using a logistic regression model that included the above-mentioned 12 variables (Additional file 1: Table S1–S8). One-to-one nearest neighbor matching was performed between men and women without replacement using a caliper width of 0.20 times of the standard deviation of the logit of the propensity score.[28] Before analyzing the outcomes, the success of the propensity matching procedure was determined by comparing the distributions of the patient characteristics in the matched sample by calculating an absolute standardized difference.[29] An absolute standardized difference of ≥ 0.1 indicated a significant difference between the sexes.[30] The outcomes of the men and women in each age group were compared before propensity matching with either the chi-square test or Fisher's exact test and after propensity matching with the McNemar's test. All data were analyzed using JMP statistical package software version 14-Pro (SAS Institute Inc.; Cary, NC, USA). All the reported tests were two-tailed, and P < 0.005 was considered statistically significant.[31,32]