This cohort study examines data from the Trauma Quality Improvement Program database for patients with pelvic fracture to assess the patterns of hemorrhage control interventions and associated clinical outcomes.
Key Points
Question
What is the current variation in nationwide practices for the management of pelvic fracture–related hemorrhage?
Findings
In this cohort study, more than 1 in 3 patients with pelvic fractures who required early transfusions and invasive hemorrhage control interventions died despite the availability of a variety of advanced hemorrhage control interventions. Only pelvic angioembolization alone was associated with a reduction in mortality.
Meaning
These findings underscore the need for an algorithm to standardize care and achieve improved outcomes in pelvic hemorrhage control.
Abstract
Importance
Management of hemodynamically unstable pelvic fractures remains a challenge. Hemostatic interventions are used alone or in combination. There is a paucity of data on the association between the pattern of hemorrhage control interventions and outcomes after a severe pelvic fracture.
Objective
To characterize clinical outcomes and study the patterns of hemorrhage control interventions in hemodynamically unstable pelvic fractures.
Design, Setting, and Participants
In this cohort study, a retrospective review was performed of data from the 2017 American College of Surgeons Trauma Quality Improvement Program database, a national multi-institutional database of trauma patients in the United States. Adult patients (aged ≥18 years) with pelvic fractures who received early transfusions (≥4 units of packed red blood cells in 4 hours) and underwent intervention for pelvic hemorrhage control were identified. Use and order of preperitoneal pelvic packing (PP), pelvic angioembolization (AE), and resuscitative endovascular balloon occlusion of the aorta (REBOA) in zone 3 were examined and compared against the primary outcome of mortality. The associations between intervention patterns and mortality, complications, and 24-hour transfusions were further examined by backward stepwise regression analyses. Data analyses were performed in September 2021.
Main Outcomes and Measures
Primary outcomes were rates of 24-hour, emergency department, and in-hospital mortality. Secondary outcomes were major in-hospital complications.
Results
A total of 1396 patients were identified. Mean (SD) age was 47 (19) years, 975 (70%) were male, and the mean (SD) lowest systolic blood pressure was 71 (25) mm Hg. The median (IQR) Injury Severity Score was 24 (14-34), with a 24-hour mortality of 217 patients (15.5%), ED mortality of 10 patients (0.7%), in-hospital mortality of 501 patients (36%), and complication rate of 574 patients (41%). Pelvic AE was the most used intervention (774 [55%]), followed by preperitoneal PP (659 [47%]) and REBOA zone 3 (126 [9%]). Among the cohort, 1236 patients (89%) had 1 intervention, 157 (11%) had 2 interventions, and 3 (0.2%) had 3 interventions. On regression analyses, only pelvic AE was associated with a mortality reduction (odds ratio [OR], 0.62; 95% CI, 0.47 to 0.82; P < .001). Preperitoneal PP was associated with increased odds of complications (OR, 1.39; 95% CI, 1.07 to 1.80; P = .01). Increasing number of interventions was associated with increased 24-hour transfusions (β = +5.4; 95% CI, +3.5 to +7.5; P < .001) and mortality (OR, 1.57; 95% CI, 1.05 to 2.37; P = .03), but not with complications.
Conclusions and Relevance
This study found that among patients with pelvic fracture who received early transfusions and at least 1 invasive pelvic hemorrhage control intervention, more than 1 in 3 died, despite the availability of advanced hemorrhage control interventions. Only pelvic AE was associated with a reduction in mortality.
Introduction
Trauma remains the leading cause of death among individuals aged 1 to 46 years,1 and the most common cause of early death is uncontrolled hemorrhage.2,3,4 Nearly 9% of all trauma patients present with severe pelvic fractures associated with mortality rates as high as 32%.5 Approximately 13% of patients with pelvic fracture present with hemodynamic instability from major hemorrhage, which can arise from arterial, venous, and bony sources.6 Pelvic fractures represent a challenging injury pattern, often necessitating major hemorrhage control interventions.7
Damage control resuscitation mandates early definitive hemorrhage control.8 Multidisciplinary efforts are associated with several hemorrhage control interventions in traumatic pelvic fractures.9 These include temporizing measures such as pelvic binders and resuscitative endovascular balloon occlusion of the aorta (REBOA) and operative procedures such as preperitoneal pelvic packing (PP). Pelvic angioembolization (AE) is performed to occlude actively bleeding vessels, specifically arterial bleeding, while external fixation is used to reduce the fractured pelvis and associated hemorrhage. These multidisciplinary interventions may be applied alone or in combination. No standardization of order of applicability has been studied. Indeed, a large multicenter study recently showed that a significant variation existed for the use of hemorrhage control interventions in severely injured patients with pelvic fracture.6
This variation in practice can be explained by the lack of consensus on the optimal treatment paradigm for bleeding pelvic fractures.10 Although the Western Trauma Association devised a critical decision algorithm for the management of severe pelvic trauma,11 it is not clear in which order each intervention must be applied, with multiple treatment options being recommended equivocally. Further, questions remain about the appropriate role and timing of REBOA in pelvic fracture–related hemorrhage. Earlier definitive hemorrhage control for pelvic fracture is associated with improved survival.12,13 We aimed to describe and compare clinical outcomes of patients with pelvic fracture based on the pattern of hemorrhage control interventions used on a nationwide scale. We hypothesized that there is significant variation in practice regarding the management of bleeding pelvic fractures.
Methods
Study Design and Population
This retrospective cohort analysis used data from the 2017 American College of Surgeons (ACS) Trauma Quality Improvement Program database. Data analyses were performed in September 2021. The University of Arizona institutional review board granted this study exemption from approval because the database contains only deidentified data. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Inclusion and Exclusion Criteria
We included all adult patients (aged ≥18 years) who presented with pelvic fractures (pelvic fracture Abbreviated Injury Scale [AIS] score >1), received 4 or more units of packed red blood cells (PRBC) within 4 hours of presentation, and underwent at least 1 pelvic fracture hemorrhage control intervention. Interventions were pelvic AE, preperitoneal PP, and REBOA zone 3. We excluded patients who were transferred from other institutions or dead on arrival.
Data Points
We abstracted the following data points. Demographic data included age, sex, body mass index, and race as captured and classified according to the National Trauma Data Standard, which prioritizes self- or family-reported race identification. We captured emergency department (ED) vital signs that included systolic blood pressure, pulse rate, respiratory rate, Glasgow Coma Scale score, and injury characteristics (mechanism of injury, Injury Severity Score, AIS score for each body region). Comorbidities included diabetes, hypertension, myocardial infarction, chronic obstructive pulmonary disease, chronic kidney disease, cirrhosis, current smoking status, and preinjury anticoagulation. Hospital-related data included ACS trauma center verification level, type and timing of intervention for pelvic fracture hemorrhage control (pelvic AE, preperitoneal PP, REBOA zone 3, pelvic binders, and pelvic fracture external fixation), venous thromboembolism prophylaxis, intensive care unit (ICU) and hospital length of stay, days spent on a ventilator, transfusion parameters (PRBC, platelets, fresh frozen plasma [FFP], and cryoprecipitate), in-hospital complications, and mortality.
Outcomes
Primary outcomes included rates of 24-hour, ED, and in-hospital mortality. Secondary outcomes were major in-hospital complications, including cardiac (cardiac arrest or myocardial infarction), infectious (deep surgical site infections, organ space surgical site infections, osteomyelitis, or sepsis), respiratory (acute respiratory distress syndrome or pneumonia), and venous thromboembolic (deep venous thrombosis or pulmonary embolism) complications. Other complications noted were acute kidney injury, unplanned return to the operating room, unplanned ICU admission, 24-hour PRBC requirements, survivor-only ICU and hospital length of stay, and survivor-only ventilator use.
Stratification and Missing Data Analysis
Patients were stratified first by type and then by number of hemorrhage control interventions they received. We performed multiple imputations using a missing value analysis technique. The original data set was analyzed for random missing data points using the Little test of missing completely at random. The Markov-Chain Monte Carlo method was used for multiple imputations; this refers to a collection of methods for simulating random draws from nonstandard distributions.
Statistical Analysis
Continuous parametric data are reported as mean and SD, continuous nonparametric data as median and IQR, and categorical data as proportions. To compare groups, we used Kruskal-Wallis and Mann-Whitney U tests to compare medians for nonnormally distributed continuous variables. Categorical variables were compared using a χ2 test and means using a t test and analysis of variance.
To control for confounders, we performed multivariate backward stepwise, binary logistic, and linear regression analyses. Potential confounders were entered into the model based on existing literature and univariate analysis demonstrating P < .05. Variables entered included patient demographics, ED vital signs, mechanism of injury, Injury Severity Score, other body region AIS scores, comorbidities (diabetes, hypertension, myocardial infarction, chronic obstructive pulmonary disease), ACS trauma verification level, and type and number of pelvic fracture hemorrhage control interventions. Cutoff P values for entry into and removal from the model were P < .10 and P > .05 respectively; α was set at 5%, and P < .05 was considered statistically significant. All analyses were performed using SPSS version 28 (SPSS).
Results
We included 1396 patients who underwent at least 1 pelvic fracture hemorrhage control intervention and received at least 4 units of PRBC within 4 hours of presentation (eFigure in the Supplement). Overall, mean (SD) age was 47 (19) years, mean lowest systolic blood pressure was 71 (25) mm Hg, median (IQR) pelvis AIS score was 2 (2-4), median (IQR) 9 units (6-16) of PRBC were received within 4 hours of presentation, and 1299 patients (93%) had a blunt injury. Pelvic AE was the most common intervention (774 [55%]). Baseline characteristics of the study cohort are described in Table 1.
Table 1. Baseline Characteristics of the Study Cohort.
| Characteristic | No. (%) | |||
|---|---|---|---|---|
| Overall (N = 1396) | Pelvic AE (n = 774) | Preperitoneal PP (n = 659) | REBOA (n = 126) | |
| Age, mean (SD), y | 47 (19) | 48 (20) | 46 (18) | 44 (18) |
| Sex | ||||
| Male | 975 (69.8) | 544 (70.3) | 458 (69.5) | 84 (66.7) |
| Female | 421 (30.1) | 230 (29.7) | 201 (30.5) | 42 (33.3) |
| Whitea | 918 (65.8) | 519 (67.1) | 429 (65.1) | 76 (60.3) |
| BMI, mean (SD)b | 28 (7) | 28 (7) | 29 (7) | 28 (6) |
| Comorbidities | ||||
| Diabetes | 82 (5.9) | 57 (7.4) | 31 (4.7) | 5 (4.0) |
| Hypertension | 235 (16.8) | 152 (19.6) | 98 (14.9) | 15 (11.9) |
| COPD | 43 (3.1) | 26 (3.4) | 19 (2.9) | 4 (3.2) |
| Myocardial infarction | 8 (0.6) | 6 (0.8) | 2 (0.3) | NA |
| Current smoker | 224 (16.0) | 119 (15.4) | 105 (15.9) | 23 (18.3) |
| CKD | 6 (0.4) | 2 (0.3) | 4 (0.6) | NA |
| Cirrhosis | 31 (2.2) | 18 (2.3) | 16 (2.4) | 1 (0.8) |
| Preinjury anticoagulation | 43 (3.1) | 27 (3.5) | 18 (2.7) | 1 (0.8) |
| Emergency department vital signs, mean (SD) | ||||
| SBP, mm Hg | 101 (35) | 102 (34) | 101 (37) | 101 (35) |
| Lowest SBP, mm Hg | 71 (25) | 71 (23) | 71 (27) | 65 (27) |
| HR /min | 107 (31) | 107 (301) | 107 (32) | 107 (33) |
| RR /min | 21 (8) | 21 (8) | 21 (8) | 21 (9) |
| GCS score, median (IQR) | 15 (9-15) | 13 (3-15) | 4 (3-15) | 7 (3-15) |
| Injury severity | ||||
| ISS, median (IQR) | 24 (14-34) | 27 (17-38) | 17 (12-30) | 33 (21-38) |
| AIS score, median (IQR) | ||||
| Pelvis | 2 (2-4) | 4 (2-5) | 3 (2-5) | 3 (2-4) |
| Head | 2 (1-2) | 2 (0-4) | 2 (0-4) | 2 (0-4) |
| Chest | 2 (2-2) | 2 (1-3) | 2 (1-3) | 2 (1-3) |
| Abdomen | 1 (1-2) | 3 (2-4) | 2 (1-2) | 3 (2-4) |
| Extremities | 2 (2-3) | 2 (2-2) | 2 (1-2) | 2 (2-3) |
| Blunt, No. (%) | 1299 (93.1) | 754 (97.4) | 591 (89.7) | 114 (90.5) |
| Penetrating, No. (%) | 95 (6.8) | 18 (2.3) | 68 (10.3) | 12 (9.5) |
| Blood product transfusion requirements at 4 h, median No. of units (IQR) | ||||
| PRBC | 9 (6-16) | 9 (6-16) | 8 (6-15) | 12 (7-20) |
| Fresh frozen plasma | 6 (3-11) | 6 (4-12) | 6 (4-11) | 9 (4-13) |
| Platelets | 1 (0-2) | 1 (1-3) | 1 (0-2) | 1 (1-3) |
| Cryoprecipitate | 0 (0-1) | 0 (0-2) | 0 (0-1) | 0 (0-1) |
| Hemorrhage control interventions | ||||
| Pelvic AE | 774 (55.4) | NA | 107 (16.2) | 42 (33.3) |
| Time to AE, median (IQR), min | 185 (153-255) | NA | 101 (77-197) | 255 (185-325) |
| Preperitoneal PP | 659 (47.2) | 107 (13.8) | NA | 17 (13.5) |
| Time to PPP, median (IQR), min | 77 (70-82) | 75 (62-87) | NA | 75 (62-87) |
| REBOA zone 3 | 126 (9.0) | 42 (5.4) | 17 (2.6) | NA |
| Time to REBOA, median (IQR), min | 79 (71-83) | 75 (62-87) | 96 (23-169) | NA |
| Pelvic external fixation | 82 (5.9) | 54 (7.0) | 25 (3.8) | 12 (9.5) |
| Time to external fixation, median (IQR), h | 27 (22-62) | 23 (7-36) | 2 (1-3) | 8 (1-40) |
| Pelvic binder | 107 (7.7) | 72 (9.3) | 43 (6.5) | 13 (10.3) |
| Time to binder, median (IQR), h | 11 (11-33) | 11 (3-28) | 11 (1-105) | 11 (1-44) |
| VTE prophylaxis | 934 (67.3) | 549 (70.9) | 428 (64.9) | 74 (58.7) |
| UFH | 284 (20.3) | 166 (21.4) | 127 (19.3) | 23 (18.3) |
| LMWH | 645 (46.2) | 375 (48.4) | 295 (44.8) | 51 (40.5) |
| Time to VTE prophylaxis, median (IQR), d | 3 (2-5) | 2 (2-4) | 3 (2-4) | 3 (2-4) |
| ACS trauma center verification level | ||||
| I | 806 (57.7) | 453 (58.5) | 358 (54.3) | 101 (80.2) |
| II | 259 (18.6) | 150 (19.4) | 130 (19.7) | 3 (2.4) |
| III | 6 (0.4) | 1 (0.1) | 5 (0.8) | NA |
| Unknown | 325 (23.3) | 170 (22.0) | 166 (25.2) | 22 (17.5) |
Abbreviations: ACS, American College of Surgeons; AE, angioembolization; AIS, Abbreviated Injury Scale; BMI, body mass index; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; GCS, Glasgow Coma Scale; HR, heart rate; ISS, Injury Severity Score; LMWH, low-molecular-weight heparin; NA, not applicable; PP, pelvic packing; PPP, preperitoneal pelvic packing; PRBC, packed red blood cells; REBOA, resuscitative endovascular balloon occlusion of the aorta; RR, respiratory rate; SBP, systolic blood pressure; UFH, unfractionated heparin; VTE, venous thromboembolism.
Race was captured and classified according to the National Trauma Data Standard, which prioritizes self- or family-reported race identification. Categories other than White (American Indian, Asian, Black, Hispanic, Pacific Islander, and other race) did not contain large enough numbers to be included in analyses, so they are not reported separately here.
Calculated as weight in kilograms divided by height in meters squared.
The Figure illustrates the proportion and order of interventions and associated mortality rates. The first intervention for hemorrhage control was pelvic AE for 652 patients (47%), preperitoneal PP for 618 patients (44%), and REBOA zone 3 for 126 (9%). The most common pattern of interventions was pelvic AE only (628 [45%]), followed by preperitoneal PP only (538 [39%]), preperitoneal PP followed by pelvic AE (80 [6%]), REBOA zone 3 only (70 [5%]), REBOA zone 3 followed by pelvic AE (39 [3%]), and pelvic AE followed by preperitoneal PP (24 [1.7%]), and 3 patients (0.2%) underwent REBOA zone 3 followed by preperitoneal PP and then pelvic AE. Overall in-hospital mortality was 501 (36%) with the highest mortality rates being in REBOA zone 3/preperitoneal PP (10 [71%]) and REBOA zone 3/preperitoneal PP/pelvic AE (2 [67%]). The lowest mortality rates were seen in pelvic AE only (209 [33%]), preperitoneal PP/pelvic AE (26 [33%]), and preperitoneal PP only (196 [36%]).
Figure. Interventions for Hemorrhage Control and Associated In-Hospital Mortality in Patients With Pelvic Fracture.
AE indicates angioembolization; PP, pelvic packing; REBOA, resuscitative endovascular balloon occlusion of the aorta.
The univariate analysis based on the first pelvic hemorrhage control is provided in Table 2. Patients who underwent REBOA zone 3 first had the highest 24-hour mortality (35 [27.8%]; P < .001), ED mortality (5 [4.0%]; P < .001), and in-hospital mortality rates (61 [48.4]; P = .006). The overall major complication rate was 574 patients (41%). Patients in the REBOA zone 3 first group had higher rates of cardiac complications (26 [20.6%]; P = 04), as well as the highest median (IQR) 24-hour PRBC (16 [9-21]; P < .001), FFP (11 [6-15]; P < .001), and cryoprecipitate (2 [0-3]; P = .007) transfusion requirements.
Table 2. Outcomes of the Study Cohort According to Type of First Pelvic Hemorrhage Control Intervention.
| Outcome measure | No. (%) | P value | |||
|---|---|---|---|---|---|
| Overall (n = 1236) | Pelvic AE (n = 652) | Preperitoneal PP (n = 618) | REBOA (n = 126) | ||
| Mortality | |||||
| 24-Hour | 217 (15.5) | 78 (12.0) | 104 (16.8) | 35 (27.8) | <.001a |
| ED | 10 (0.7) | 4 (0.6) | 1 (0.2) | 5 (4.0) | <.001a |
| In-hospital | 501 (35.9) | 218 (33.4) | 222 (35.9) | 61 (48.4) | .006a |
| Major in-hospital complications | |||||
| Overall | 574 (41.1) | 259 (39.7) | 262 (42.4) | 53 (42.1) | .61 |
| Cardiac complications | 194 (13.9) | 80 (12.3) | 88 (14.2) | 26 (20.6) | .04a |
| Infectious complications | 117 (8.4) | 48 (7.4) | 58 (9.4) | 11 (8.7) | .43 |
| Deep SSI | 37 (2.7) | 15 (2.3) | 18 (2.9) | 4 (3.2) | .74 |
| Organ space SSI | 20 (1.4) | 6 (0.9) | 13 (2.1) | 1 (0.8) | .17 |
| Osteomyelitis | 6 (0.4) | 4 (0.6) | 1 (0.2) | 1 (0.8) | .38 |
| Sepsis | 66 (4.7) | 28 (4.3) | 32 (5.2) | 6 (4.8) | .76 |
| VTE complications | 138 (9.9) | 68 (10.4) | 54 (8.7) | 16 (12.7) | .33 |
| DVT | 96 (6.9) | 46 (7.1) | 39 (6.3) | 11 (8.7) | .60 |
| PE | 50 (3.6) | 24 (3.7) | 20 (3.2) | 6 (4.8) | .69 |
| Respiratory complications | 144 (10.3) | 76 (11.7) | 61 (9.9) | 7 (5.6) | .11 |
| Acute kidney injury | 131 (9.4) | 69 (10.6) | 50 (8.1) | 12 (9.5) | .31 |
| Unplanned return to OR | 92 (6.6) | 42 (6.4) | 40 (6.5) | 10 (7.9) | .82 |
| Unplanned ICU admission | 50 (3.6) | 27 (4.1) | 19 (3.1) | 4 (3.2) | .57 |
| Blood product transfusion requirements at 24 h, median No. of units (IQR) | |||||
| PRBC | 12 (7-20) | 12 (8-21) | 10 (6-17) | 16 (9-21) | <.001a |
| Fresh frozen plasma | 8 (4-15) | 8 (5-17) | 6 (4-13) | 11 (6-15) | <.001a |
| Platelets | 2 (1-4) | 3 (1-4) | 2 (1-3) | 2 (1-5) | <.001a |
| Cryoprecipitate | 1 (0-2) | 1 (0-3) | 0 (0-2) | 2 (0-3) | .007a |
| Hospital course, median (IQR), d | |||||
| Hospital LOS | 16 (5-31) | 15 (4-30) | 17 (7-32) | 14 (1-28) | .005a |
| ICU LOS | 9 (4-17) | 9 (4-17) | 10 (4-16) | 8 (2-16) | .12 |
| Ventilator time | 6 (3-12) | 6 (2-12) | 6 (2-13) | 5 (2-8) | <.001a |
Abbreviations: AE, angioembolization; DVT, deep venous thrombosis; ED, emergency department; ICU, intensive care unit; LOS, length of stay; OR, operating room; PE, pulmonary embolism; PP, pelvic packing; PRBC, packed red blood cells; REBOA, resuscitative endovascular balloon occlusion of the aorta; SSI, surgical site infection; VTE, venous thromboembolism.
Statistically significant.
The univariate analysis based on the number of pelvic hemorrhage control interventions is provided in Table 3. Most patients (n = 1236 [89%]) received only 1 intervention. Patients who received 1 intervention had the lowest rates of unplanned return to the operating room and the lowest 24-hour PRBC, FFP, platelet, and cryoprecipitate transfusion requirements.
Table 3. Outcomes of the Study Cohort According to Number of Pelvic Hemorrhage Control Interventions.
| Outcome measure | No. (%) | P value | ||
|---|---|---|---|---|
| 1 Intervention (n = 1236) | 2 Interventions (n = 157) | 3 Interventions (n = 3) | ||
| Mortality | ||||
| 24-Hour | 195 (15.8) | 22 (14.0) | NA | .64 |
| ED | 9 (0.7) | 1 (0.6) | NA | .98 |
| In-hospital | 437 (35.4) | 62 (39.5) | 2 (66.7) | .32 |
| Major in-hospital complications | ||||
| Overall | 500 (40.5) | 73 (46.5) | 1 (33.5) | .34 |
| Cardiac complications | 169 (13.7) | 25 (15.9) | NA | .58 |
| Infectious complications | 101 (8.2) | 16 (1.2) | NA | .60 |
| Deep SSI | 34 (2.8) | 3 (1.9) | NA | .79 |
| Organ space SSI | 17 (1.4) | 3 (1.9) | NA | .85 |
| Osteomyelitis | 6 (0.5) | NA | NA | .68 |
| Sepsis | 54 (4.4) | 12 (7.6) | NA | .18 |
| VTE complications | 117 (9.5) | 21 (13.4) | NA | .26 |
| DVT | 78 (6.3) | 18 (11.5) | NA | .05 |
| PE | 46 (3.7) | 4 (2.5) | NA | .72 |
| Respiratory complications | 127 (10.3) | 17 (1.8) | NA | .82 |
| Acute kidney injury | 111 (9.0) | 20 (12.7) | NA | .27 |
| Unplanned return to OR | 76 (6.1) | 15 (9.6) | 1 (33.3) | .047a |
| Unplanned ICU admission | 45 (3.6) | 5 (3.2) | NA | .91 |
| Blood product transfusion requirements at 24 h, median No. of units (IQR) | ||||
| PRBC | 11 (6-19) | 16 (9-24) | 15 (14-16) | <.001a |
| Fresh frozen plasma | 7 (4-15) | 11 (6-18) | 9 (6-11) | <.001a |
| Platelets | 2 (1-4) | 3 (1-5) | 4 (3-4) | <.001a |
| Cryoprecipitate | 1 (0-2) | 2 (0-4) | 3 (1-4) | <.001a |
| Hospital course, median (IQR), d | ||||
| Hospital LOS | 16 (6-30) | 14 (2-36) | 9 (1-18) | .16 |
| ICU LOS | 9 (4-16) | 9 (2-16) | 11 (2-19) | .95 |
| Ventilator time | 6 (2-12) | 5 (2-13) | 10 (2-18) | .79 |
Abbreviations: DVT, deep venous thrombosis; ED, emergency department; ICU, intensive care unit; LOS, length of stay; NA, not applicable; OR, operating room; PE, pulmonary embolism; PRBC, packed red blood cells; SSI, surgical site infection; VTE, venous thromboembolism.
Statistically significant.
The results of the multivariate backward stepwise regression analyses assessing for independent associations between divergent interventional patterns for pelvic hemorrhage control and outcome measures are provided in Table 4. Pelvic AE was the only intervention associated with reduced adjusted odds ratio (aOR) of mortality (aOR, 0.62; 95% CI, 0.47 to 0.82; P < .001). Increasing number of interventions was associated with increased adjusted odds of mortality (aOR, 1.57; 95% CI, 1.05 to 2.37; P = .03) and increased 24-hour PRBC transfusion requirements (adjusted β, +5.4; 95% CI, +3.5 to +7.5; P < .001). Preperitoneal PP was the only intervention associated with higher adjusted odds of major complications (aOR, 1.39; 95% CI, 1.07 to 1.80; P = .01). REBOA zone 3 was not independently associated with mortality, complications, or 24-hour PRBC transfusion requirements.
Table 4. Multivariate Backward Stepwise Regression Analyses: Independent Association of Intervention Pattern With Outcomes.
| Variable | Adjusted | P valuea |
|---|---|---|
| In-hospital mortality, OR (95% CI)b | ||
| Pelvic angioembolization | 0.62 (0.47 to 0.82) | <.001 |
| Increasing No. of interventions | 1.57 (1.05 to 2.37) | .03 |
| Major complications, OR (95% CI)c | ||
| Preperitoneal pelvic packing | 1.39 (1.07 to 1.80) | .01 |
| 24-Hour PRBC transfusions, β (95% CI)d | ||
| Increasing No. of interventions | +5.4 (+3.5 to +7.5) | <.001 |
Abbreviations: OR, odds ratio; PRBC, packed red blood cells.
All were statistically significant.
Hosmer-Lemeshow goodness of fit: χ2 = 10.1; P = .26.
Hosmer-Lemeshow goodness of fit: χ2 = 9.9; P = .27.
Unadjusted R2 = 0.112; adjusted R2 = 0.106; Durbin-Watson statistic = 2.035.
Multivariate sensitivity analyses performed separately for each intervention in all patients with pelvic fractures who received 4 or more units of PRBC within 4 hours of presentation demonstrated that pelvic AE was the only intervention significantly associated with reduced mortality (aOR, 0.79; 95% CI, 0.65-0.98; P = .03), while preperitoneal PP (aOR, 0.92; 95% CI, 0.78-1.1; P = .38) and REBOA zone 3 (aOR, 1.45; 95% CI, 0.82-2.56; P = .20) were not. Multivariate sensitivity analyses performed after excluding patients with a chest or head AIS score of 2 or higher in all patients with pelvic fracture who received 4 or more units of PRBC within 4 hours of presentation and received at least 1 intervention demonstrated that pelvic AE was the only intervention significantly associated with reduced mortality (aOR, 0.44; 95% CI, 0.20-0.96; P = .04).
Seven percent of patients (n = 104) had a combination of pelvic AE/preperitoneal PP, of which 80 patients (77%) had preperitoneal PP before pelvic AE. Univariate subanalyses comparing patients who underwent preperitoneal PP first and those who underwent pelvic AE first revealed no significant differences in terms of mortality, complications, and 24-hour PRBC transfusion requirements. eTable 1 in the Supplement provides a univariate subanalysis of outcomes based on the most common intervention patterns. A univariate statistical comparison between those who had pelvic AE as the first intervention vs those who had preperitoneal PP as the first intervention is shown in eTable 2 in the Supplement.
Median (IQR) time to first pelvic hemorrhage control intervention was 122 minutes (62-206) with 1112 of 1396 patients (90%) undergoing the first intervention within 370 minutes. On multivariate backward stepwise and linear regression analyses, time to any first intervention was not associated with mortality (aOR, 0.98; 95% CI, 0.96 to 1.00; P = .10), major complications (aOR, 1.00; 95% CI, 0.98 to 1.01; P = .53), or 24-hour PRBC transfusion requirements (adjusted β, −0.01; 95% CI, −0.06 to +0.04; P = .67).
Discussion
In this retrospective study of interventions for hemodynamically unstable pelvic fractures, more than 1 in 3 patients with pelvic fracture who required early transfusions and invasive hemorrhage control interventions died despite the availability of various advanced interventions. We found significant diverging management regarding the type and order of hemorrhage control interventions. Of the interventions, only pelvic AE was associated with a reduction in mortality. These findings are important when elucidating the ideal algorithm for pelvic fracture management.
Previous studies demonstrated the pelvic fracture–related mortality rate to be about 10%.14,15,16,17 However, these studies assessed all patients with pelvic fracture and not specifically those with concomitant hemodynamic instability or those requiring major hemorrhage control interventions. We found a high mortality rate of 36% among patients with pelvic fracture who were hemodynamically unstable and required early blood product transfusions and hemorrhage control interventions, comparable with that found by Costantini et al.6 Older studies have had similar findings, with mortality rates for a similar subset of patients reported as 21% to 50%.5,18,19,20 Our findings highlight a high nationwide mortality for patients with pelvic fracture who are hemodynamically unstable.
Costantini et al6 found similar variations in pathways of care in a multicenter observational study on the current management of pelvic fracture patients. They found pelvic AE alone was the most common pattern of hemorrhage control intervention, followed by external fixation alone, and preperitoneal PP alone. The combination of interventions was used in a minority of patients. However, Costantini et al6 did not describe the nationwide variation in practice patterns and associated outcomes. Conversely, Brenner et al21 concluded that despite differences in duration of aortic occlusion and rates of ongoing cardiopulmonary resuscitation between their study groups, mortality rates were similar whether hemostasis was achieved after REBOA with PP or with AE. We found a wide variation in outcomes associated with different pathways of pelvic hemorrhage control, with mortality rates ranging from 33% to 71%. We also found that an increasing number of interventions was associated with increased adjusted odds of mortality and 24-hour PRBC transfusion requirements. Perhaps the increasing interventions were not the reason behind worse outcomes; rather, patients were more severely injured and required more interventions. Among patients receiving 2 interventions, there were 4 permutations of the order of intervention identified that were associated with different outcomes. These findings point to the profound practice variation nationwide, highlighting the need for standardized management pathways.
Therapeutic AE is used to diagnose and manage active arterial pelvic fracture bleeding. We found the most common intervention was pelvic AE (55%) in contrast to its use (10%) in the study by Costantini et al.6 The reason behind this higher reported utilization rate may be that we included only patients with pelvic fracture who underwent pelvic hemorrhage control interventions. Pelvic AE was the only intervention significantly associated with reduced mortality, likely because of its definitive nature. The importance of AE in the management of patients with pelvic fracture has been highlighted in previous studies,12,13,22 and earlier time to AE is strongly associated with improved outcomes. Further, Thorson et al22 demonstrated that outcomes were similar regardless of whether pelvic AE was performed before or after patients were taken to the operating room for definitive hemorrhage control. Similarly, we found comparable mortality rates associated with patients managed with pelvic AE followed by PP (37.5%) vs preperitoneal PP followed by AE (32.5%). Nevertheless, assessing the true association of AE with mortality may be beyond the scope of this retrospective study. It becomes difficult to address possible confounders without information on the precise indication for intervention, whether it was considered emergent or not, and the institutional management protocols in place.
We found the second most common intervention was preperitoneal PP (47%). This was much higher than the reported incidence of preperitoneal PP of 7.1% reported by Costantini et al.6 In preperitoneal PP, laparotomy pads serve to tamponade deep pelvic bleeding. Preperitoneal PP was associated with increased adjusted odds of major complications in our study. Our findings show that the precise subset of patients benefiting most from preperitoneal PP use remains to be elucidated, to minimize major complications.
Nine percent of our cohort underwent zone 3 REBOA, considerably higher than the 0.4% incidence reported by Costantini et al.6 All REBOAs placed in their study were performed at a single participating center and assessed all patients with pelvic fracture. Our nationwide study was likely able to detect all the recorded zone 3 REBOAs placed at high-performing US trauma centers and only included patients who received pelvic fracture hemorrhage control interventions. Interestingly, zone 3 REBOA was not independently associated with an improvement in any of our outcome measures. Indeed, on univariate analysis, of the 3 hemorrhage control interventions, REBOA use was associated with the highest mortality and cardiac complication rates and transfusion requirements. These findings cast some light on the possibly limited role that REBOA may play in the acute management of pelvic fracture–related hemorrhage or that the patients undergoing this temporizing maneuver are much sicker at presentation, as noted by the systolic blood pressure in this group compared with others. Asmar et al23 demonstrated that REBOA may be associated with improved outcomes when compared with preperitoneal PP. In stark contrast, and in line with our findings, Mikdad et al24 found that REBOA use was associated with increased mortality compared with the use of PP for patients with pelvic fracture. There is a need for further prospective multicenter studies on the matter of benefits and risks of REBOA.
We aimed to describe the nationwide management of bleeding pelvic fractures. Hence, we believed it was pertinent to mention the ED mortality for patients who survived long enough to undergo interventions to be able to describe the mortality pattern. A total of 10 patients who survived long enough to undergo an intervention died in the ED. However, because hospitalization details are coded after the in-hospital course is complete, we were unable to accurately determine whether these patients truly died in the ED or after being transferred directly to the operating room or interventional radiology suite, before admission orders were placed.
Although the beneficial association of earlier interventions with outcomes for patients with pelvic fracture undergoing AE was previously explored in the literature, we were not able to discern a significant association of time to first intervention with outcomes in our study. This negative finding may be due to the emergently managed nature of the patient cohort. Most patients received the first intervention within 3 hours of admission, and it is likely that further reductions in time to intervention beyond this amount may have a diminished effect on outcomes. Furthermore, assessing the effect of time to intervention may be beyond the scope of this retrospective study, as significant selection bias may lead to more severely injured patients having shorter times to intervention because earlier intervention was warranted. The beneficial effect of earlier interventions may be an intuitive one, and future prospective studies identifying the effects of earlier interventions are warranted.
We propose that the predominant type of bleed (arterial, venous, bony) should be identified before making critical management decisions, and this factor should be incorporated in future algorithms. To that end, faster and higher-quality computed tomographic scans along with increased availability of hybrid operating rooms/trauma bays may allow for earlier imaging before critical decision points even in patients who are markedly unstable.
Limitations
Our study has several limitations. Because of its retrospective nature, we relied on multivariate regression analysis to control for confounders. Concomitant injuries made it difficult to assess the precise contribution of pelvic fracture–related hemorrhage toward outcomes. We were limited by the variables available to us in the database, and we could not provide information regarding pelvic fracture patterns, crystalloid/colloid resuscitation, coagulation test parameters, base deficit/lactate levels, or imaging findings. We were unable to provide sufficient information regarding pelvic external fixation and pelvic binders. We identified an unusually small number of patients who received these commonly used interventions. These discrepancies may be due to erroneous and inaccurate database entries. Further, these findings may not be generalizable to all non-US trauma systems. Future prospective trials on the effect of pelvic fracture hemorrhage control interventions on outcomes are warranted.
Conclusion
This study found that among hemodynamically unstable patients with pelvic fracture who received early transfusions and at least 1 invasive pelvic hemorrhage control intervention, more than 1 in 3 died. Of the variety of diverging intervention patterns, only pelvic AE was associated with a reduction in mortality. Interestingly, preperitoneal PP was associated with increased complication rates. These findings underscore the need for an algorithm to standardize care and achieve improved outcomes in pelvic hemorrhage control.
eFigure. Flow diagram of patients included in the final analysis
eTable 1. Subanalysis: outcomes of the study cohort based on the most common intervention pattern
eTable 2. Comparison of outcomes between those who underwent pelvic AE versus preperitoneal PP
References
- 1.Rhee P, Joseph B, Pandit V, et al. Increasing trauma deaths in the United States. Ann Surg. 2014;260(1):13-21. doi: 10.1097/SLA.0000000000000600 [DOI] [PubMed] [Google Scholar]
- 2.Baker CC, Oppenheimer L, Stephens B, Lewis FR, Trunkey DD. Epidemiology of trauma deaths. Am J Surg. 1980;140(1):144-150. doi: 10.1016/0002-9610(80)90431-6 [DOI] [PubMed] [Google Scholar]
- 3.Sauaia A, Moore FA, Moore EE, et al. Epidemiology of trauma deaths: a reassessment. J Trauma. 1995;38(2):185-193. doi: 10.1097/00005373-199502000-00006 [DOI] [PubMed] [Google Scholar]
- 4.de Knegt C, Meylaerts SA, Leenen LP. Applicability of the trimodal distribution of trauma deaths in a Level I trauma centre in the Netherlands with a population of mainly blunt trauma. Injury. 2008;39(9):993-1000. doi: 10.1016/j.injury.2008.03.033 [DOI] [PubMed] [Google Scholar]
- 5.Demetriades D, Karaiskakis M, Toutouzas K, Alo K, Velmahos G, Chan L. Pelvic fractures: epidemiology and predictors of associated abdominal injuries and outcomes. J Am Coll Surg. 2002;195(1):1-10. doi: 10.1016/S1072-7515(02)01197-3 [DOI] [PubMed] [Google Scholar]
- 6.Costantini TW, Coimbra R, Holcomb JB, et al. ; AAST Pelvic Fracture Study Group . Current management of hemorrhage from severe pelvic fractures: results of an American Association for the Surgery of Trauma multi-institutional trial. J Trauma Acute Care Surg. 2016;80(5):717-723. doi: 10.1097/TA.0000000000001034 [DOI] [PubMed] [Google Scholar]
- 7.Costantini TW, Coimbra R, Holcomb JB, et al. ; AAST Pelvic Fracture Study Group . Pelvic fracture pattern predicts the need for hemorrhage control intervention: results of an AAST multi-institutional study. J Trauma Acute Care Surg. 2017;82(6):1030-1038. doi: 10.1097/TA.0000000000001465 [DOI] [PubMed] [Google Scholar]
- 8.Cannon JW. Hemorrhagic shock. N Engl J Med. 2018;378(4):370-379. doi: 10.1056/NEJMra1705649 [DOI] [PubMed] [Google Scholar]
- 9.Stein DM, O’Toole R, Scalea TM. Multidisciplinary approach for patients with pelvic fractures and hemodynamic instability. Scand J Surg. 2007;96(4):272-280. doi: 10.1177/145749690709600403 [DOI] [PubMed] [Google Scholar]
- 10.Cullinane DC, Schiller HJ, Zielinski MD, et al. Eastern Association for the Surgery of Trauma practice management guidelines for hemorrhage in pelvic fracture: update and systematic review. J Trauma. 2011;71(6):1850-1868. doi: 10.1097/TA.0b013e31823dca9a [DOI] [PubMed] [Google Scholar]
- 11.Tran TLN, Brasel KJ, Karmy-Jones R, et al. Western Trauma Association critical decisions in trauma: management of pelvic fracture with hemodynamic instability, 2016 updates. J Trauma Acute Care Surg. 2016;81(6):1171-1174. doi: 10.1097/TA.0000000000001230 [DOI] [PubMed] [Google Scholar]
- 12.Matsushima K, Piccinini A, Schellenberg M, et al. Effect of door-to-angioembolization time on mortality in pelvic fracture: every hour of delay counts. J Trauma Acute Care Surg. 2018;84(5):685-692. doi: 10.1097/TA.0000000000001803 [DOI] [PubMed] [Google Scholar]
- 13.Schwartz DA, Medina M, Cotton BA, et al. Are we delivering two standards of care for pelvic trauma? availability of angioembolization after hours and on weekends increases time to therapeutic intervention. J Trauma Acute Care Surg. 2014;76(1):134-139. doi: 10.1097/TA.0b013e3182ab0cfc [DOI] [PubMed] [Google Scholar]
- 14.Yoshihara H, Yoneoka D. Demographic epidemiology of unstable pelvic fracture in the United States from 2000 to 2009: trends and in-hospital mortality. J Trauma Acute Care Surg. 2014;76(2):380-385. doi: 10.1097/TA.0b013e3182ab0cde [DOI] [PubMed] [Google Scholar]
- 15.Rothenberger DA, Fischer RP, Strate RG, Velasco R, Perry JF Jr. The mortality associated with pelvic fractures. Surgery. 1978;84(3):356-361. [PubMed] [Google Scholar]
- 16.Chong KH, DeCoster T, Osler T, Robinson B. Pelvic fractures and mortality. Iowa Orthop J. 1997;17:110-114. [PMC free article] [PubMed] [Google Scholar]
- 17.Sathy AK, Starr AJ, Smith WR, et al. The effect of pelvic fracture on mortality after trauma: an analysis of 63,000 trauma patients. J Bone Joint Surg Am. 2009;91(12):2803-2810. doi: 10.2106/JBJS.H.00598 [DOI] [PubMed] [Google Scholar]
- 18.Flint L, Babikian G, Anders M, Rodriguez J, Steinberg S. Definitive control of mortality from severe pelvic fracture. Ann Surg. 1990;211(6):703-706. doi: 10.1097/00000658-199006000-00008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Perkins ZB, Maytham GD, Koers L, Bates P, Brohi K, Tai NR. Impact on outcome of a targeted performance improvement programme in haemodynamically unstable patients with a pelvic fracture. Bone Joint J. 2014;96-B(8):1090-1097. doi: 10.1302/0301-620X.96B8.33383 [DOI] [PubMed] [Google Scholar]
- 20.Cothren CC, Osborn PM, Moore EE, Morgan SJ, Johnson JL, Smith WR. Preperitonal pelvic packing for hemodynamically unstable pelvic fractures: a paradigm shift. J Trauma. 2007;62(4):834-839. doi: 10.1097/TA.0b013e31803c7632 [DOI] [PubMed] [Google Scholar]
- 21.Brenner M, Moore L, Zakhary B, et al. Scalpel or sheath? outcomes comparison between pre-peritoneal pelvic packing and angioembolization for definitive hemorrhage control after REBOA. Endovasc Resusc Trauma Manag. 2020;4(1). doi: 10.26676/jevtm.v4i1.119 [DOI] [Google Scholar]
- 22.Thorson CM, Ryan ML, Otero CA, et al. Operating room or angiography suite for hemodynamically unstable pelvic fractures? J Trauma Acute Care Surg. 2012;72(2):364-370. doi: 10.1097/TA.0b013e318243da10 [DOI] [PubMed] [Google Scholar]
- 23.Asmar S, Bible L, Chehab M, Tang A, Khurrum M, Douglas M, et al. Resuscitative endovascular balloon occlusion of the aorta vs pre-peritoneal packing in patients with pelvic fracture. J Am Coll Surg. 2021;232(1):17-26.e2. doi: 10.1016/j.jamcollsurg.2020.08.763 [DOI] [PubMed] [Google Scholar]
- 24.Mikdad S, van Erp IAM, Moheb ME, et al. Pre-peritoneal pelvic packing for early hemorrhage control reduces mortality compared to resuscitative endovascular balloon occlusion of the aorta in severe blunt pelvic trauma patients: a nationwide analysis. Injury. 2020;51(8):1834-1839. doi: 10.1016/j.injury.2020.06.003 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure. Flow diagram of patients included in the final analysis
eTable 1. Subanalysis: outcomes of the study cohort based on the most common intervention pattern
eTable 2. Comparison of outcomes between those who underwent pelvic AE versus preperitoneal PP