Variability in the timeliness of interventional radiology availability for angioembolization of hemodynamically unstable pelvic fractures: a prospective survey among U.S. level I trauma centers

Background Patients with hemodynamically unstable pelvic fractures have high mortality due to delayed hemorrhage control. We hypothesized that the availability of interventional radiology (IR) for angioembolization may vary in spite of the mandated coverage at US Level I trauma centers, and that the priority treatment sequence would depend on IR availability. Methods This survey was designed to investigate IR availability and pelvic fracture management practices. Six email invitations were sent to 158 trauma medical directors at Level I trauma centers. Participants were allowed to skip questions and irrelevant questions were skipped; therefore, not all questions were answered by all participants. The primary outcome was the priority treatment sequence for hemodynamically unstable pelvic fractures. Predictor variables were arrival times for IR when working off-site and intervention preparation times. Kruskal-Wallis and ordinal logistic regression were used; alpha = 0.05. Results Forty of the 158 trauma medical directors responded to the survey (response rate: 25.3%). Roughly half of participants had 24-h on-site IR coverage, 24% (4/17) of participants reported an arrival time ≥ 31 min when IR was on-call. 46% (17/37) of participants reported an IR procedure setup time of 31–120 min. Arrival time when IR was working off-site, and intervention preparation time did not significantly affect the sequence priority of angioembolization for hemodynamically unstable pelvic fractures. Conclusions Trauma medical directors should review literature and guidelines on time to angioembolization, their arrival times for IR, and their procedural setup times for angioembolization to ensure utilization of angioembolization in an optimal sequence for patient survival.


Background
Pelvic fracture management is one of the most complex treatment strategies [1]. Published guidelines offer varying approaches to care for hemodynamically unstable pelvic fractures [2][3][4][5][6]. The World Society of Emergency Surgeons (WSES) and Western Trauma Association (WTA) recommend selective angioembolization after pelvic packing [2,3]. Eastern Association for the Surgery of Trauma (EAST) and Advanced Trauma Life Support (ATLS) suggest angioembolization after circumferential compression device application [5,6]. Trauma Quality Improvement Program (TQIP) [4] utilizes angioembolization after external fixation and pelvic packing, or last when in extremis. There remains a high level of ambiguity on the optimal management of patients with hemodynamic unstable pelvic fractures across guidelines [2][3][4][5][6].
It is known that the time from presentation to angiography affects mortality in cases where angioembolization is needed [7]. Tanizaki et al. found a 4-fold increase in mortality rates for patients who went to angiography 60 min after arrival when compared to those who went within 60 min [7]. This is at least part of the reason that the American College of Surgeons (ACS) requires an interventional radiologist available within 30 min at Level I trauma centers [8].
Although, it has been reported that not all Level I trauma centers have IR on-site, the full extent of IR availability has not been described; therefore it is unclear if angiography within 1 h of arrival is possible [9].

Methods
This anonymous cross-sectional survey of 158 trauma medical directors at United States ACS-verified Level I trauma centers was approved by the Western Institutional Review Board. The contact list was derived from the ACS website, individual trauma center's websites, and via telephone. To view the invitation list, view the Appendix. Coauthors piloted the web-based survey prior to its online dissemination through SurveyMonkey Inc. (San Mateo, California; www.surveymonkey.com). Six invitations, that contained the approved partial waiver of consent, were emailed from March 1, 2018 to June 26, 2018. Participants were called to verify email receipt if they had not responded upon sending the final two invitations. No compensation was provided, and participation was voluntary. Trauma medical directors or an assigned colleague completed the survey and are referred to as "participants".
The study hypotheses were 1) that IR was not on-site and prepared for intervention within 60 min, and 2) arrival times for IR when working off-site and the time for IR to prepare for intervention would be associated with the priority treatment sequence for angioembolization. The survey included 46 questions regarding IR availability and pelvic fracture management practices. To view questions pertaining to this paper, visit: http://bit.ly/SurveyIR. Irrelevant questions were skipped based on prior responses using SurveyMonkey's 'skip logic' , and participants could skip any question; therefore, there are missing responses for individual questions. Analysis was completed on SAS 9.4 (Cary, NC) software. Categorical data were summarized as counts and proportions. The median (interquartile range [IQR]) sequence for angioembolization was compared by both the arrival time for IR, and by the time for IR to prepare for intervention using the Kruskal-Wallis test. Ordinal logistic regression was used to determine if the arrival time for IR, or the time for IR to prepare for intervention was associated with the priority treatment sequence for angioembolization. All hypothesis tests were two-tailed with an alpha of 0.05.

Results
The response rate was 25% (40/158). Of the survey responses, 90% (36/40) completed and 10% (4/40) partially completed the survey; all responses were included. Participating Level I trauma centers' characteristics have been reported [10]. The median (IQR) survey completion time was 11 min (8,21). No pelvic fracture protocol was implemented at 28% (11/40) of participating Level I trauma centers ( Table 1). The most common pelvic fracture guideline followed was the EAST guideline (23% [11/40] Fifty-four percent (20/37) of the represented Level I trauma centers had 24-h on-site IR coverage ( Table 2). The remaining had on-call IR coverage; 13% (2/16) of participants reported IR was on-call for 24 h/day, and 31% (5/ 16) reported IR was on-call for 12 h/day. A majority (71% [12/17]) of participants reported a 21-30-min arrival time for IR when on-call. In addition to arrival times, 46% (17/ 37) of participants reported an IR procedure set-up time of 31-120 min. Most participants provided temporalizing stabilization through circumferential compression devices, pelvic packing, or REBOA while waiting for IR to prepare for intervention (Table 1).
We previously reported the priority treatment sequence for hemodynamically unstable pelvic fractures [10]. The median priority treatment sequence for angioembolization was examined according to the IR arrival time when working off-site and to the time it took IR to prepare for intervention (Table 3). There was no significant relationship between the arrival times, or the intervention preparation time, and median priority sequence of angioembolization. The intervention preparation time, and the arrival time for IR when working off-site, were not significant predictors for the priority treatment sequence of angioembolization, (Table 4). This is evidenced by a lack of significance for these variables as well as a lack of significance in the Hosmer-Lemeshow goodness of fit p-value. No guideline in place 28% (11)  40 Eastern Association for the Surgery of Trauma 23% (9) Hospital developed protocol 18% (7) Western Trauma Association 15% (6) Trauma Quality Improvement Program 8% (3) Advanced Trauma Life Support 5% (2) Agency not specified 5% (2) Does your hospital use both angioembolization and pelvic packing for pelvic fracture management?

Discussion
This study surveyed 25% of ACS-verified Level I trauma centers on angiography practices and IR availability to treat hemodynamically unstable pelvic fractures. We failed to reject the null hypotheses; IR availability was variable across Level I trauma centers and did not significantly affect the priority treatment sequence of angioembolization. A majority of participants utilized angioembolization and pelvic packing, supporting the argument that pelvic packing and angioembolization should be complementary, not competitive, as the treatments target either venous or arterial hemorrhages [11]. Angioembolization primarily treats arterial bleeds, representing 10-20% of hemorrhaging, but cannot treat the majority of hemorrhaging from venous and cancellous sources [2]. Although the priority sequence for angioembolization and pelvic packing continues to be debated, this study observed a reported preference. The majority of participants used angioembolization before pelvic packing. Contrary to this, it has been suggested that pelvic packing may be more efficient when used before angioembolization as it treats the majority of pelvic hemorrhaging [2]. Predicting the need for angioembolization has proven difficult; applying pelvic packing first allows for identification of the bleed source and determination of the need for angioembolization [3,9,[11][12][13]. Additionally, several studies found a shorter time from admission to pelvic packing than angiography [13][14][15][16]. The use of angioembolization before pelvic packing may be due to EAST guideline, being the most commonly followed guideline, recommending angioembolization first [5]. Although Cothren et al. [17] stated preperitoneal pelvic packing can supplant angioembolization needs, this study found that most participants utilized angioembolization and prioritized it earlier than other treatment modalities.
It is our observation that a common reason for pelvic packing application is due to excessive wait times for IR.
Despite the prevalence of angioembolization before pelvic packing, roughly half of the responding Level I trauma centers did not have 24-h on-site IR coverage. Furthermore, many participants reported arrival and IR procedure preparation times in excess of 30 min; some as long as 1-2 h. Ironically, this study revealed a lack of association between the amount of time it took IR to prepare for intervention and the priority treatment sequence of angioembolization for patients with hemodynamically unstable pelvic fractures. Yet, all participants reported utilization of alternative treatments while IR prepared for intervention. Not surprisingly, circumferential compression device was the most common treatment utilized while waiting; which is non-invasive and easily applied [2]. Pelvic packing was also a common treatment modality utilized while IR prepared; a sequence described by Burlew et al. [9] Almost half the participants indicated REBOA was utilized while IR prepared for intervention, suggesting more widespread use than previously reported [18]. The variety of treatment modalities used while waiting is no surprise, given that no guideline provides direction in this situation [2][3][4][5][6]. Therefore, more data is needed to determine the optimal priority treatment when IR is not prepared for intervention.

Limitations
The response rate of 25% was a limitation as the participants responses may not be representative of all Level I trauma centers. The online-only survey format may have negatively impacted the response rate as some trauma medical directors noted a preference towards paper surveys. Some Level I trauma centers had outdated contact information for the trauma medical director which resulted in less email invitations being sent to the participant. Responses may have been subject to self-report and recall biases. Survey anonymity and instructions to have protocols on-hand were precautions to reduce these biases. In addition, mortality data was not collected; therefore we cannot conclude what practices were associated with better outcomes.

Conclusions
The optimal priority treatment sequence for pelvic fractures has not been definitively determined. The reported IR arrival time and time to prepare for intervention did not significantly predict the priority treatment sequence of angioembolization; suggesting the priority treatment sequence was not altered based on these timing metrics. The use of angioembolization first may only be viable to prevent mortality at centers with 24-h on-site IR availability or faster preparation times. Level I trauma centers should review the literature and guidelines on time to angioembolization, their own arrival times for interventional radiology when working off-site, and their intervention preparation times for angioembolization to ensure utilization of the treatment options in an optimal sequence for patient survival. IR interventional radiology, OR odds ratio, CI confidence interval, H-L GOF Hosmer-Lemmeshow goodness of fit, a participants who indicated their interventional radiology department has on-site coverage 24-h a day