Transarterial embolisation is associated with improved survival in patients with pelvic fracture: propensity score matching analyses

Despite recent advances in trauma surgery, successful treatment of severe pelvic fractures remains challenging. The overall mortality of pelvic fractures is 5–10%, increasing to 60% in patients with haemodynamic instability [1‐3]. Bleeding is the major cause of death after severe pelvic fracture; the origin of the bleeding can be arterial, venous, or bone-related [4]. Arterial bleeding produces the most severe haemorrhage and frequently results in hypotension [5]. Rapid control of haemorrhage is associated with improved survival [6].

Pelvic angiography and subsequent embolisation is a safe, rapid, and effective technique for patients with pelvic fracture-related arterial haemorrhage in both haemodynamically stable and unstable patients [7‐9]. Transarterial embolisation (TAE) is the mainstay of treatment for arterial bleeding. However, there is no consensus as to the optimal treatment paradigm for patients presenting with severe pelvic fracture [6, 10].

It is difficult to assess the outcome of TAE, as a comparison between published series is hampered by differences in patient populations and because the outcome is often determined by factors other than pelvic haemorrhage alone, such as associated traumatic injuries. No randomised controlled trials have been conducted, and it is unlikely that any will be performed.

Accordingly, in an effort to verify the efficacy of TAE in patients with pelvic injury, we examined outcomes in patients treated with TAE compared to those treated without TAE. We used propensity score matching analyses as the most reliable method to reduce the effects of confounding factors in this retrospective study. We hypothesised that TAE would improve survival at 28 days in trauma victims.

Propensity score matching indicated that TAE is independently associated with improved survival at 28 days in trauma patients with pelvic fracture. To the best of our knowledge, this is the first study to report this relationship using robust statistical methods. Notably, the observed relationship was consistent in the survival at 7 days, and a significantly lower hazard ratio of death from TAE was detected among patients.

Hauschild et al. [12] reported no significant difference in overall mortality rate when comparing the TAE group with the non-embolisation group. However, several studies have demonstrated that control of pelvic arterial haemorrhage using TAE was associated with a 7–35% reduction in mortality [7, 9, 13]. Furthermore, some studies have reported the occurrence of arterial bleeding on angiography in patients with pelvic fracture even in the absence of contrast media extravasation on CT [8, 10, 14]. Non-TAE group patients may have arterial injury even without contrast media extravasation on the CT, and this bleeding, if continuous, may affect the outcome. We therefore believe that the use of TAE should be considered in the management of severely injured patients with pelvic fracture.

Pelvic fractures are often caused by high-energy events. Haemodynamically unstable patients who present with traumatic pelvic fracture are at high risk for mortality and significant morbidity [15, 16]. The mortality rate of pelvic fractures is < 20%, but that of severe pelvic fracture combined with multiple injuries is as high as 30–70%, and the prognosis is poor [12, 17‐20]. In our study, overall mortality was 15% and the mortality of those presenting with haemodynamic instability and severe pelvic injury is similar to the mortality rates in other published series [21, 22].

Most patients with severe pelvic fracture died due to haemorrhage or sepsis/multiple organ failure [21]. The cause of early mortality was often haemorrhage, whereas the cause of late mortality was sepsis/organ failure or traumatic brain injury. Thus, we did not choose survival-to-discharge, but instead survival at 28 days as a primary outcome to distinguish deaths due to bleeding from others. In our propensity matching data, sepsis and organ failure accounted for 3 patients who died after 28 days. We think this explains why overall survival-to-discharge was no different between the TAE and the non-TAE group (p = 0.076). However, survival at 7 and 28 days were significantly higher among patients treated with TAE than among those treated without TAE (p = 0.032; p = 0.039, respectively).

The bony surfaces within a pelvic fracture can cause haemorrhage via arterial and/or venous injury or from the fractured bone itself [18]. Therefore, a multidisciplinary approach using various haemorrhage control modalities is necessary. The overwhelming majority of pelvic bleeding is of venous origin [23]. However, arterial bleeding, while less common than venous bleeding in the pelvis, is more common in patients that are persistently hypotensive. Some have reported that pelvic arterial bleeding is found in > 70% of pelvic fracture patients with transient or no response to fluid resuscitation [8, 14]. TAE is currently accepted in many trauma centres as the preferred method for controlling pelvic arterial haemorrhage [7, 8, 14, 21, 24‐26].

In this study, pelvic TAE was performed in 35% of all patients presenting with pelvic fracture. In other studies, 3.8–9.6% of patients with pelvic fractures required pelvic TAE [18, 22, 27]. Even though there is no contrast media extravasation on CT, patients with haemodynamic instability associated with pelvic fracture without another significant source of bleeding may have arterial bleeding on angiography [8, 10, 14]. Therefore, at our centre, angiography was performed when the expanding hematoma was visible or if there was persistent haemodynamic instability associated with the pelvic fracture even in the absence of extravasation on CT. Our interventional radiologists and the equipment required for TAE are available around the clock, with the attending doctor being responsible for providing the indication for TAE (Fig. 1). We believe this is the reason for the increased proportion of TAE in the present study.

A recent study reported that every hour of delay in pelvic TAE is associated with an increased risk for in-hospital death by 79% [28]. Balogh et al. noted that patients with pelvic fractures and unstable haemodynamics should receive TAE within 90 min after admission, as this reduces blood transfusion volumes and mortality [29]. Any delay exposes patients to unnecessary risks [28‐30]. Therefore, early access to angiography is associated with reduced mortality. Delays to embolisation > 3 h are associated with worse outcomes [21]. In our study, the median time to TAE was 107 min; performing TAE < 2 h after admission led to better survival at 28 days than non-TAE (Fig. 4).

There are two main differences between our study and other studies that suggest that TAE might be useful. First, the number of covariates included in the propensity score calculation was higher than in other studies [14, 22, 25]. We considered that outcome predictors such as the performance of a haemostatic procedure (pelvic stabilisation [external fixation] and/or PPP) or massive transfusion within the initial 24 h after arrival should be entered into the propensity model regardless of their relevance to TAE, because biased distribution of these factors would significantly affect survival. Second, we performed matching through a greedy matching algorithm with a calliper width of less than 0.001 in the logit-transformed propensity score. Accordingly, only 130 (24.2%) patients in the TAE group were matched with those in the non-TAE group. Because the deliberate selection of covariates and the strict matching algorithm could make patient characteristics of the TAE and non-TAE groups more similar, including TRISS-calculated Ps (0.91 in the TAE group vs. 0.92 in the non-TAE group; Table 1), the significant association between TAE and improved survival found in our study suggests that TAE could be beneficial in injured patients with pelvic fracture.

There were several limitations to this study. First, because it was retrospective, the results are not conclusive. Although we demonstrated a higher survival rate at 28 days in patients with pelvic fracture treated with TAE than in those treated without TAE, residual confounding and unmeasured survival predictors could exist as impediments to confirming the efficacy of TAE. Further clinical investigations, including a randomised controlled trial, are needed to validate our results.

Another limitation is the fact that 178 (57.8%) of patients in the TAE group were excluded from the propensity score calculation or matching process, which may limit the generalisability of our findings. The association between TAE and improved survival might not be applicable to patients who were excluded during propensity model development nor in patients outside the database. Although limited in scope, considering that the patient characteristics of the TAE group in this study (in-hospital mortality was 11.2%, median ISS was 27, and median base excess was  − 2.8) were similar to the populations reported in other studies [17, 18, 21], we believe our results might be applicable to severely injured trauma victims around the world.

Finally, this study was confined to patients at a single centre and our study population may not represent all patients with pelvic fracture. For example, our TRISS-calculated Ps are lower than in other studies [23, 31]. This is because patients who were declared dead-on-arrival and/or discharged or transferred from TRU within 24 h were excluded from our study. Therefore, a larger group of patients and further studies (more rigorous research design and external validation) are necessary to draw any definitive conclusions.

In severely injured trauma patients with pelvic injury, TAE is associated with improved survival at 7 and 28 days after injury. The use of TAE should therefore be considered in conjunction with trauma resuscitation during the management of patients with severe pelvic fractures.