Study Design
This study was a multicenter, retrospective cohort study conducted using Japan Trauma Data Bank (JTDB) data from 2004 to 2018. JTDB is a nationwide trauma registry established in 2003 by the Japanese Association for Surgery of Trauma and the Japanese Association for Acute Medicine to improve and ensure the quality of trauma care in Japan. During the study period, 291 hospitals, including 95% of all tertiary emergency medical centers in Japan, participated in the JTDB. JTDB collects 92 data elements related to patient and hospital information, such as patient demographics, physiology, abbreviated injury scale (AIS) score, injury severity score (ISS), in-hospital procedures, and survival.
Patient Selection
Patients who were directly transferred to hospital and diagnosed with pelvic fracture (existence of pelvic fracture AIS 2005 codes) were included. In addition, we targeted patients who were aged ≧16 years and who were initially treated with embolization. The following exclusion criteria for patients were defined: 1. AIS grade=6 for any region. 2. Underwent any surgery for hemorrhage control for associated injuries except for external fixation. 3. Lacked information on vital signs of systolic blood pressure (sBP) and heart rate (HR) on hospital arrival. 4. Hemodynamically unstable patients whose sBP <90 mmHg or HR >120 bpm. 5. Lacked information on the time from hospital arrival to embolization. 6. Time from hospital arrival to embolization was over 3 h. 7. Lacked information on outcome. Regarding the time to embolization, we excluded patients who underwent pelvic embolization 3 or more hours after admission as these cases were likely non-emergency cases1.
Study Endpoints
The primary outcome of this study was 30-day mortality, and the secondary outcome was 24-hour mortality.
Statistical Analysis
Study patients were divided into six groups according to 30-min blocks of time to pelvic embolization (0–30, 30–60, 60–90, 90–12, 120–150, and 150–180 min). Univariate analysis was performed comparing patients’ characteristics and trauma severities between the six groups. A Chi-square or Fisher exact test was used for categorical variables, and a Mann–Whitney U test was used for continuous variables. A Cochran–Armitage test for trend was also performed to evaluate outcomes between the six groups. In addition, we adjusted the backgrounds of patients and the trauma severity with regard to clustering by institutions using generalized estimating equation (GEE) models with an independent working correlation matrix. Models were adjusted for age, sex, vital signs at hospital arrival, and ISS, which were selected a priori based on reported findings9,10. We then used marginal standardization based on probability determined from the GEE model to estimate the adjusted 30-day mortality by the six groups. Because the 30-day mortality of the 0–30 min group was zero, we omitted these patients from the GEE model. We showed crude mortality and risk-adjusted 30-day mortality according to groups. As for sensitivity analysis, we estimated a linear relationship between the time to embolization and the 30-day mortality using a GEE model. The model was adjusted for patient demographics, such as age and sex, vital signs and ISS, with regard to clustering by institutions. Statistical significance was defined as a two-sided p-value < 0.05 in all statistical analyses. All analyses were performed using R software (version 3.5.2; R Foundation for Statistical Computing, Vienna, Austria) and Stata software version 15.1 (StataCorp, College Station, TX, USA).