Diagnostic accuracy of MRI for detection of occult instability of type I anterior to posterior pelvic injuries

Type I Young and Burgess anterior posterior compression (APC) pelvic injuries have been classically managed non operatively due to theoretical integrity of sacroiliac joint ligaments (SIJL), though examination under anesthesia (EUA) has been proven occult mechanical instability in up to 50% of these injuries.  We sought to determine the diagnostic accuracy of magnetic resonance (MRI) for detection of occult instability on APC-I injuries when compared to EUA.

METHODS

Diagnostic test study of prospectively recruited patients admitted with APC-I pelvic injuries between 2015 and 2022. All patients consented to participate in this study were subjected to MRI and EUA. The evaluators of each of these tests were blinded.  On MRI evaluation, SIJL were considered compromised when unilateral injury to anterior SIJL was visualized in three or more consecutive images or in bilateral injuries, when injury to the anterior SIJL in two or more consecutive images on each side was observed. Positive EUA was considered a symphyseal diastasis over 25 mm on stress fluoroscopy. Demographic data was collected as recruited and sensitivity, specificity, accuracy, and positive and negative predictive values were calculated. Confidence interval was set at 95%. EUA was considered the gold standard in statistical analysis.

RESULTS

A total of 32 patients mean aged 36 (24-61) years were included. Mean symphyseal diastasis at admission was 17.58 (11 - 25) mm. The median time from injury to EUA was 5 (0-21) days. Positive EUA was observed on 20 patients and 25 patients  presented compromised SIJL. MRI presented a sensitivity of 95% (75.13% - 99.87%), specificity of 50% (21.09% - 78.91%), positive-predictive value of 73% (60.61% to 82.93%), negative-predictive value of 87% (48.66% - 98.08%).

CONCLUSION

Injury to SIJL on MRI presented an accuracy of 77% (58.29% - 89.64%) for the detection of occult pelvic instability on EUA.

The pelvic vascular injury score (P-VIS): a prehospital instrument to detect significant vascular injury in pelvic fractures

PURPOSE

The purpose of this study was to identify predictive factors for peri-pelvic vascular injury in patients with pelvic fractures and to incorporate these factors into a pelvic vascular injury score (P-VIS) to detect severe bleeding during the prehospital trauma management.

METHODS

To identify potential predictive factors, data were taken (1) of a Level I Trauma Centre with 467 patients (ISS ≥ 16 and AISPelvis ≥ 3). Analysis including patient's charts and digital recordings, radiographical diagnostics, mechanism and pattern of injury as well as the vascular bleeding source was performed. Statistical analysis was performed descriptively and through inference statistical calculation. To further analyse the predictive factors and finally develop the score, a 10-year time period (2012-2021) of (2) the TraumaRegister DGU® (TR-DGU) was used in a second step. Relevant peri-pelvic bleeding in patients with AISPelvis ≥ 3 (N = 9227) was defined as a combination of the following entities (target group PVITR-DGU N = 2090; 22.7%): pelvic fracture with significant bleeding (> 20% of blood volume), Injury of the iliac or femoral artery or blood transfusion of ≥ 6 units (pRBC) prior to ICU admission. The multivariate analysis revealed nine items that constitute the pelvic vascular injury score (P-VIS).

RESULTS

In study (1), 467 blunt pelvic trauma patients were included of which 24 (PVI) were presented with significant vascular injury (PVI, N = 24; control (C, N = 443). Patients with pelvic fractures and vascular injury showed a higher ISS, lower haemoglobin at admission and lower blood pressure. Their mortality rate was higher (PVI: 17.4%, C: 10.3%). In the defining and validating process of the score within the TR-DGU, 9227 patients met the inclusion criteria. 2090 patients showed significant peripelvic vascular injury (PVITR-DGU), the remaining 7137 formed the control group (CTR-DGU). Nine predictive parameters for peripelvic vascular injury constituted the peripelvic vascular injury score (P-VIS): age ≥ 70 years, high-energy-trauma, penetrating trauma/open pelvic injury, shock index ≥ 1, cardio-pulmonary-resuscitation (CPR), substitution of > 1 l fluid, intubation, necessity of catecholamine substitution, remaining shock (≤ 90 mmHg) under therapy. The multi-dimensional scoring system leads to an ordinal scaled rating according to the probability of the presence of a vascular injury. A score of ≥ 3 points described the peripelvic vascular injury as probable, a result of ≥ 6 points identified a most likely vascular injury and a score of 9 points identified an apparent peripelvic vascular injury. Reapplying this score to the study population a median score of 5 points (range 3-8) (PVI) and a median score of 2 points (range 0-3) (C) (p < 0.001). The OR for peripelvic vascular injury was 24.3 for the patients who scored > 3 points vs. ≤ 2 points. The TR-DGU data set verified these findings (median of 2 points in CTR-DGU vs. median of 3 points with in PVITR-DGU).

CONCLUSION

The pelvic vascular injury score (P-VIS) allows an initial risk assessment for the presence of a vascular injury in patients with unstable pelvic injury. Thus, the management of these patients can be positively influenced at a very early stage, prehospital resuscitation performed safely targeted and further resources can be activated in the final treating Trauma Centre.

Characterization of Combined Blast- and Fragment-Induced Pelvic Injuries and Hemostatic Resuscitation in Rabbits

INTRODUCTION

To establish a blast- and fragment-induced pelvic injury animal model in rabbits, observe its injury characteristics, and explore the effects of hemostatic resuscitation combined with damage control surgery (DCS) with respect to this injury model.

METHODS

Forty-eight rabbits were randomly allocated to four groups: group A rabbits were subjected to pelvic injury, group B rabbits to pelvic injury + DCS, group C rabbits to pelvic injury + DCS + resuscitation with Hextend, and group D rabbits to pelvic injury + DCS + Hextend + hemostatic resuscitation with tranexamic acid, fibrinogen concentrate, and prothrombin complex concentrate. Simulated blast and fragment-induced pelvic injury was produced by a custom-made machine. We implemented CT scanning and necropsy to assess the injury state and calculated the coefficient of variation (CV) of the cumulative abbreviated injury scale (AIS) to assess the reproducibility of the animal model. Immediately after instrumentation (0 h), and 1 h, 2 h, 4 h, and 8 h after injury, blood samples were taken for laboratory tests.

RESULTS

We found that severe pelvic injury was produced with an AIS CV value of 10.32%, and the rabbits demonstrated severe physiologic impairment and coagulo-fibrinolytic derangements with high mortality. In rabbits of group D, however, physiologic and coagulo-fibrinolytic parameters were significantly enhanced with improved organ function and lowered mortality when compared with the other three groups.

CONCLUSIONS

We herein established in rabbits a blast- and fragment-induced pelvic injury animal model that exhibited high reproducibility, and we demonstrated that hemostatic resuscitation plus DCS was effective in improving the outcome.

Fuel tank injury in a fatal motorcycle accident

Vehicular trauma is the most common cause of pelvic fractures. In motorcycle collisions, the driver strikes or is struck by a fixed object, and is subjected to blunt trauma and deceleration forces. Injuries around the scrotum and perineum of victims in motorcycle accidents are known as a fuel tank injuries (FTI). We report the case of a 26-year-old male motorcyclist involved in a fatal high-speed head-on collision with a car. At autopsy, purple bruising distributed symmetrically over both hips with scrotal lacerations were found. Partial bowel evisceration through an abdominal lacerated wound, extensive abdominal organ injuries and multiple fractures were also found. Considering the dynamics of the accident and the type of motorcycle he had been riding, the pelvic bruising and the scrotal injury were related to violent deceleration following the impact, leading the driver to slide forward against the fuel tank of the motorcycle.FTI is comparatively rare in motorcycle accidents because it only occurs in cases involving a head-on collision, and most drivers try to swerve or correct their direction just prior to the collision. Nevertheless, the most frequent cause of pelvic injuries in motorcyclists is caused by contact with the vehicles fuel tank during the crash. Forensic pathologists should have a better knowledge of FTIs as they are helpful in understand the dynamics of the accident and in distinguishing the driver from the passenger in two-rider motorcycle crashes.

Effect of angioembolization for isolated complex pelvic injury: A post-hoc analysis of a nationwide multicenter trauma database in Japan

BACKGROUND AND IMPORTANCE

Complex pelvic injuries are among the types of trauma with the highest mortality. Treatment strategies should be based on the hemodynamic status, the anatomical type of fracture, and the associated injuries. Combination therapies, including preperitoneal pelvic packing, temporary mechanical stabilization, resuscitative endovascular balloon occlusion of the aorta, and angioembolization, are recommended for pelvic injuries.

OBJECTIVE

To investigate the effect of urgent angioembolization alone on severe pelvic injury-associated mortality.

DESIGN, SETTINGS, AND PARTICIPANTS

We used the Japan Trauma Data Bank database, a multicenter observational study, to retrospectively identify adult patients with isolated blunt pelvic injuries (Abbreviated Injury Scale [AIS] score: 3-5) from 2004 to 2018.

OUTCOME MEASURES AND ANALYSIS

The primary outcome measure was in-hospital mortality. We subdivided patients into two groups, those who underwent urgent angioembolization and non-urgent angioembolization, and compared their mortality rates. We performed multiple imputation and multivariable analyzes to compare the mortality rates between groups after adjusting for known potential confounding factors (age, sex, Glasgow Coma Scale score, systolic blood pressure on hospital arrival, Injury Severity Score, pelvic AIS score, laparotomy, resuscitative endovascular balloon occlusion of the aorta, and external fixation) and for within-hospital clustering using the generalized estimating equation.

MAIN RESULTS

We analyzed 4207 of 345,932 trauma patients, of whom 799 underwent urgent angioembolization. The in-hospital mortality rate was significantly higher in the urgent embolization group than in the non-urgent embolization group (7.4 vs. 4.0%; p < 0.01). However, logistic regression analysis revealed that the mortality rates of patients with urgent angioembolization significantly decreased after adjusting for factors independently associated with mortality (odds ratio: 0.60; 95% confidence interval: 0.37-0.96; p = 0.03).

CONCLUSION

Urgent angioembolization may be an effective treatment for severe pelvic injury regardless of the pelvic AIS score and the systolic blood pressure on hospital arrival.

Preoperative CT simulation of iliosacral screws for treating unstable posterior pelvic ring injury

BACKGROUND

The percutaneous iliosacral screw is a common procedure for treating pelvic posterior ring instability. Traditional X-ray fluoroscopy screw placement has the advantages of decreased bleeding and trauma, but it also has some drawbacks, such as increased radiation exposure and screw dislocation. The purpose of this study was to establish a safe, effective, and quick approach for putting iliosacral screws for the treatment of unstable posterior pelvic ring damage utilizing simulated screws based on preoperative computed tomography (CT) planning.

METHODS

From February 2019 to June 2020, we retrospectively assessed 41 patients with posterior pelvic ring instability who were treated with percutaneous iliosacral screws in our institution, and randomly separated them into two groups: conventional surgery (n = 20) and preoperative planning (n = 21). Pelvic radiographs (anteroposterior, inlet, outlet), as well as normal CT scans of the pelvis, were all taken postoperatively to confirm the screw position. After that, the screw insertion time, the radiation exposure time, and the screw misplacement rate (as assessed by postoperative CT) were all examined. Screw position grading was evaluated by Smith grading.

RESULTS

In the conventional surgery group, 26 screws were inserted in 20 patients, with each screw insertion taking 23.15 ± 4.19 min and 1.02 ± 0.17 min to expose to radiation. Eight of the 26 screws were misplaced (30.8%). In the preoperative planning group, 24 screws were inserted in 21 patients, with each screw taking 19.57 ± 4.05 min to implant and 0.67 ± 0.09 min to expose to radiation. One of 24 screws was misplaced (4.2%). Screw insertion time, radiation exposure time, and screw dislocation rate were all significantly reduced when preoperative planning aided iliosacral screw placement (P < 0.05).

CONCLUSIONS

Preoperative CT simulation of iliosacral screws for placement planning, screw trajectory, and intraoperative screw placement is a safe way for reducing surgical time, radiation exposure, and ensuring accurate screw placement.

[Emergency thoracotomy in a severely injured patient after hemorrhagic shock in traumatic pelvic bleeding : Case report]

A case of in-hospital thoracotomy with subsequent open chest cardiopulmonary resuscitation of a polytraumatized patient is reported. Emergency thoracotomies are rare interventions in challenging situations. Up to now there are only few standards or uniform education and training concepts. The indications are often a borderline decision. The aim of thoracotomy and open resuscitation in combination with a reduction in circulation, for example by cross-clamping the aorta, is to save time to address reversible causes of the hemorrhage, redirect the blood volume into the vital cerebral and coronary circulation and minimize bleeding from subdiaphragmatic bleeding sources. Ultimately, in case of doubt, the thoracotomy can be performed for the patient's benefit with the appropriate indications.

The role of angioembolization and C-clamp fixation: Damaged control orthopaedic in haemodynamically unstable pelvic fracture

INTRODUCTION

Unstable pelvic fracture may emerge to major bleeding complication. Angioembolization is one of method to stop the bleeding effectively. This case series aims to analyze whether the bleeding of unstable pelvic fracture is managed by angioembolization to achieve the better functional outcome.

PRESENTATION OF CASE

Three cases of haemodynamically unstable pelvic fracture were studied retrospectively and prospectively. A staged approach using damage control orthopaedic surgery was performed. Initial resuscitation began from fluid resuscitation, pelvic wrapping using binder. All patients followed with pelvic external fixation, while 2 patients immediately replaced binder to C-Clamp, and 1 patient with anterior frame. Angioembolization was done to all patients. All patients required definitive internal fixation, while only 2 patients reach the definitive surgery. Finally, we measured the functional outcome of all patients using Hannover Pelvic score, Majeed pelvic score, and Iowa Pelvic score.

DISCUSSION

We review some literatures regarding pelvic angioembolization. The previous study suggested to resuscitate patients when the hemodynamic is unstable, the angioembolization procedure is still preferred. The indication and successful definition of this procedure is still unclear, yet it shows decrease of mortality rate of pelvic injury if this procedure starts ahead a schedule.

CONCLUSION

Angioembolization as a part of damaged control orthopaedic has been shown a favorable result in managing unstable pelvic injury.

Management of pelvic injuries in hemodynamically unstable polytrauma patients - Challenges and current updates

Pelvic injuries are notorious for causing rapid exsanguination, and also due to concomitant injuries and complications, they have a relatively higher mortality rate. Management of pelvic fractures in hemodynamically unstable patients is a challenging task and has been variably approached. Over the years, various concepts have evolved, and different guidelines and protocols were established in regional trauma care centers based mainly on their previous experience, outcomes, and availability of resources. More recently, damage control resuscitation, pelvic angioembolization, and acute definitive internal fixation are being employed in the management of these unstable injuries, without clear consensus or guidelines. In this background, we have performed a computerized search using the Cochrane Database of Systematic Reviews, Scopus, Embase, Web of Science, and PubMed databases on studies published over the past 30 years. This comprehensive review aims to consolidate available literature on the current epidemiology, diagnostics, resuscitation, and management options of pelvic fractures in polytraumatized patients with hemodynamic instability with particular focus on damage control resuscitation, pelvic angioembolization, and acute definitive internal fixation.

A systematic review of combined pelvic and acetabular injuries

PURPOSE

Injuries of both pelvic ring and acetabulum as rare very few articles are available in literature. There are no set protocols in defining the injury let alone defining early and definitive management strategies. This article is an attempt to encompass all available data to give us guidelines in managing these injuries.

METHODS

An extensive literature review was carried out on PubMed/Medline, google scholar and Embase databases was done with the eligibility criteria of 1) Case series with a minimum of 20 cases. 2) The patient's outcome reported. 3) Full article available. 4) Article in English. 5) Minimum Jadad score of 3. As per PRISMA guidelines the search was done and gradually filtered down to relevant articles which were 8 in number.

RESULTS

The incidence of these injuries range from 5 to 16%. The transverse acetabular fracture pattern is the commonest followed by associated both column fractures. There is equal propensity of Anteroposterior compression and lateral compression injuries. The injury mechanism appears to transmitted lateral force from the greater trochanter inwards with an implosion injury causing acetabular and pelvic injury as a continuum. The initial management is similar to managing pelvic ring injuries with focus on patient resuscitation, hemodynamic stabilization and temporary stabilization. The injury severity score and the mortality rates are comparable to isolated unstable pelvic ring injuries. Definitive management focuses on fixing the posterior pelvic ring first followed by the acetabular fracture and then the anterior pelvic ring. The displacement rates and outcome is worse than isolated acetabular injuries or pelvic injuries.

CONCLUSION

Combined Pelvic and acetabular injuries are complex injuries which need to be managed initially as we manage pelvic injury and later as we fix as an acetabular fracture meticulously.

Age- and gender-related characteristics of the pubic symphysis and triradiate cartilage in pediatric computed tomography

BACKGROUND

There is little information on the pubic symphysis' normal CT appearance in children.

OBJECTIVE

We sought to generate age-, gender- and maturity-related symphyseal width appearances in CT scans.

MATERIALS AND METHODS

Pelvic CT scans performed for any reason during a 6-year period in patients younger than 18 years were retrospectively analyzed. The symphysis width was measured in the axial plane and the triradiate cartilage was classified as open or closed.

RESULTS

Four hundred twenty-seven CT scans were evaluated and 350 remained for analysis. Age- and gender-related measurements of the symphysis width are illustrated on various centile graphs. When grouping children by age in years 0-6, 7-11, 12-15 and 16-17, mean (standard deviation) symphysis width was 5.4 mm (0.9), 5.3 mm (1.1), 4.1 mm (1.1) and 3.5 mm (1.0), respectively, in girls and 5.9 mm (1.3), 5.4 mm (1.2), 5.2 mm (1.1) and 4.0 mm (1.0), respectively, in boys. Boys and girls were significantly different in the age groups 12-15 years (P<0.001) and 16-17 years (P=0.04). In the mature pelvis, the symphyseal gap is significantly (P<0.001) shorter in both genders, and in girls compared to boys (P=0.04).

CONCLUSION

The pubic symphysis width in children differs according to age, gender and maturity. The reference values published herein may help detect symphyseal injury.

Lumbosacral Plexopathy in Pelvic Injury - A Cause of Hip Instability in Acetabular Fractures: A Report of Two Cases

Lumbosacral plexopathy is a rare clinical entity that results in potentially severe neurological deficit. The clinical presentation of lumbosacral plexopathy includes motor and sensory deficits. However to the best of our knowledge, hip instability secondary to lumbosacral plexopathy has not been reported in current literature. We report two cases of pelvic injury in which recurrent hip subluxation occurred following fixation of acetabular fractures. We attribute this to inadequate hip muscle tension from the associated lumbosacral plexopathy. In patients with acetabular fractures, this may lead to debilitating hip joint instability. In an already traumatized hip joint, this instability has a poor prognosis and can lead to degenerative changes with the eventual need for hip replacement.

Radiological evaluation of the posterior pelvic ring in paediatric patients: Results of a retrospective study developing age- and gender-related non-osseous baseline characteristics in paediatric pelvic computed tomography - References for suspected sacroiliac joint injury

INTRODUCTION

The prevalence of paediatric pelvic injury is low, yet they are often indicative of accompanying injuries, and an instable pelvis at presentation is related to long-term poor outcome. Judging diastasis of the sacroiliac joint in paediatric pelvic computed tomography is challenging, as information on their normal appearance is scarce. We therefore sought to generate age- and gender-related standard width measurements of the sacroiliac joint in children for comparison.

PATIENTS AND METHODS

A total of 427 pelvic computed tomography scans in paediatric patients (<18 years old) were retrospectively evaluated. After applying exclusion criteria, 350 scans remained for measurements. Taking a standard approach we measured the sacroiliac joint width bilaterally in axial and coronal planes.

RESULTS

We illustrate age- and gender-related measurements of the sacroiliac joint width as a designated continuous 3rd, 15th, 50th, 85th and 97th centile graph, respectively. Means and standard deviations in the joint width are reported for four age groups. There are distinct changes in the sacroiliac joint's appearance during growth. In general, male children exhibit broader sacroiliac joints than females at the same age, although this difference is significant only in the 11 to 15-year-old age group.

CONCLUSION

The sacroiliac joint width in children as measured in coronal and axial CT scans differs in association with age and gender. When the sacroiliac joint width is broader than the 97th centile published in our study, we strongly encourage considering a sacroiliac joint injury.

Emergency radiological examination of the externally stabilized pelvis--there is a catch to it: lessons learned from two cases with symphyseal disruption despite initial inconspicuous computed tomography

BACKGROUND

Preclinical and early clinical external pelvic stabilization using commercially available devices has become common in trauma patient care. Thus, in the emergency department an increasing number of patients will undergo radiographic evaluation of the externally stabilized pelvis to exclude injuries. While reports exist where injuries to the pelvis were elusive to radiological examination due to the pelvic immobilization we elaborate on an algorithm to remove an external pelvic stabilizing device, prevent delayed diagnosis of pelvic disruption and thus increase patient safety.

CASE PRESENTATION

We report on two patients with external pelvic stabilization presenting with an inconspicuous pubic symphysis on initial pelvic computed tomography scans. The first patient was an otherwise healthy 51-year old male being run over by his own car. He received external pelvic stabilization in the emergency department. The second patient was a 36-year old male falling from a ladder. In this patient external pelvic stabilization was performed at the scene. In the first patient no pelvic injury was obvious on computed tomography. In the second patient pelvic fractures were diagnosed, yet the presentation of the pubic symphysis appeared normal. Nevertheless, complete symphyseal disruption was diagnosed in both of them upon removal of the external pelvic stabilization and consequently required internal fixation.

CONCLUSION

Based on our experience we propose an algorithm to "clear the initially immobilized pelvis" in an effort to minimize the risk of missing a serious pelvic injury and increase patient safety. This is of significant importance to orthopedic trauma surgeons and emergency physicians taking care of injured patients.

Modeling of unstable pelvic fractures for 20 ° sagittally rotated pelvic displacement

Early and accurate assessment of unstable pelvic fractures decreases mortality and morbidity by improving the outcomes of closed reduction surgery. In some circumstances (such as in developing countries or in acute trauma), accurate computed tomography (CT) 3D reconstructions are difficult to obtain and plain radiography must be relied upon for surgical planning. Therefore, a simulation model of pelvic displacement was evaluated as a basis for improving the assessment of pelvic fracture displacement from plain radiography images. The simulated model was based on a modified anteroposterior (AP) view of the pelvis constructed using anatomical data. Plain radiography imaging data of sagittal 20 ° hemipelvic rotation, similar to that observed in trauma, was obtained from the simulated model and were compared with normal pelvic rings from 10 cadaveric specimens (actual model). For each data set, the anterior superior iliac spine and ipsilateral pubic symphysis were measured on both sides. There was no significant difference between pelvic sides in the simulated and actual models (P = 0.870). Furthermore, no significant difference was found between the rotational degree of each pelvis based on calculated or simulated degrees and the 20 ° rotational angle (P = 0.322). A simulation model based on plain radiography of pelvic displacement could contribute to the optimal surgical management of unstable pelvic fractures without relying on CT.

Emergency primary repair of grade V bladder neck injury complicating pelvic fracture

We report a case of a grade V bladder injury complicating an open-book pelvic fracture following a road traffic accident. The bladder neck injury was primarily repaired in the emergency setting of a poor-resourced area with successful outcome. The dangers of urinary extravasation are still to be considered of importance and we advocate and encourage immediate/emergency open intervention although it remains controversial to say the least in a lesser resourced healthcare set up.

Pelvic Fracture and Risk Factors for Mortality: A Population-Based Study in Taiwan

OBJECTIVE

To study the incidence, demographics, distribution of fracture sites, associated injuries, and risk factors for the outcomes of pelvic fracture on a population basis.

METHODS

An annual claim dataset from 2001 to 2003 was retrieved from the Bureau of National Health Insurance (BNHI) trauma database with any International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) coded as pelvic fracture. Statistical analyses were conducted to discover the association between variables such as age, gender, month of injury, injury mechanisms, associated injuries, and outcome.

RESULTS

Pelvic fracture was diagnosed in 10,109 patients (4,693 males, 46.4%) and 384 patients (3.8%) died. The incidence was 14.97/100,000 person-years. The mean length of stay was 9.3 days (females, 8.8 days; males, 9.9 days; p < 0.01) and the average cost per patient was US$1,475. Both were affected by age, gender, and hospital accreditation level and ownership. There were more female cases in the age groups over 44 years old. Head injury (excluding concussion), associated chest and abdominal injuries, open pelvic fracture, and the transfusion of more than four units of blood significantly increased the risk of mortality.

CONCLUSIONS

The incidence rate of pelvic fracture was higher in females over 44 years of age, but the average medical resource use was higher in males. Associated injuries were stronger positive factors for the risk of mortality than gender, fracture sites, injury mechanisms, and the characteristics of the treating hospitals.