Pelvic Fractures at a New Level 1 Trauma Centre: Who Dies from Pelvic Trauma? The Inkosi Albert Luthuli Central Hospital Experience

Abstract

Objective

To identify the incidence of pelvic trauma, causes of death and factors predicting death with pelvic fractures.

Methods

All pelvic fractures were retrospectively identified from a registry spanning from March 2007 to August 2009. Data was captured on a proforma. Data for survivors, non‐survivors and a subgroup with pelvic injury as the underlying cause of death were compared.

Results

Pelvic fracture incidence was 16% of major trauma cases. Patient with pelvic fractures had 31% mortality and 9% pelvic fracture‐induced mortality. Motor vehicle collisions were the commonest external cause of pelvic fractures (59%); however, the highest mortality was from falls >6 m. The Injury Severity Score (ISS) was 29 in survivors, 36 in non‐survivors, and 54 in the pelvic death subgroup. Type C fracture was a predictor of mortality (P = 0.135). 53% of the cases required transfusion in the first 24 hours. The pelvic death subgroup received a mean of 10.7 units of blood, versus 4 units for survivors and 3.7 units for non‐survivors (P = 0.259).

Conclusion

The overall incidence of pelvic fracture and associated mortality were higher than previously reported. Fracture severity and falls from heights are associated with additional injuries (higher ISS) and mortality. More severe fractures cause deaths directly attributable to the pelvic injury. The requirement for major blood transfusions for pelvic fracture hemorrhage was related to mortality. Female patients appeared to fare worse than males.

Introduction

Patients with pelvic ring disruptions caused by trauma often have multiple associated injuries because such disruptions are commonly caused by high energy trauma1, 2. This patient group also has a high overall mortality rate that has varied between 5% and 30% in previous series3, 4, 5 depending on the severity of the fracture5, 6 as well as on associated injuries1, 3. It is however, less common that the pelvic fracture is the underlying cause of death2, 3. More severe fracture patterns are generally associated with extensive hemorrhage from injured venous plexi and therefore are more likely to necessitate blood transfusions6, 7. Although the mortality of trauma cases with pelvic fractures has declined over the past few decades3, it still represents a serious type of injury and continuous research in this field is important to further decrease the mortality.

In March 2007, the first truly Level 1 equivalent Trauma Centre with an integrated intensive care unit (ICU) was opened in Durban, KwaZulu‐Natal, South Africa, as the pinnacle of care for the development of a regional trauma system. This 10‐bed unit with integrated ICU facilities operates as an exclusive‐type unit, accepting only the most severe major trauma cases that cannot be managed in regional or other teaching hospitals because of the need for either dialysis or specialized surgical/intensive care support. The unit does not accept patients with minor or moderate (single system) trauma, but does receive patients both direct from the trauma scene and as inter‐hospital transfers. The same clinical team manages the resuscitation, imaging, operative care and intensive care of these patients. All personnel in the unit use standard protocols for care and all patients are managed by registered specialist surgeons with certification in critical care, in accordance with the Trauma Surgery sub‐specialty of the Health Professions Council of South Africa. The hospital does not have access to emergency angio‐embolization facilities.

The aim of this study was to further improve the outcome of these injuries by quantifying the incidence of major pelvic trauma presenting to the unit and identifying the mechanisms of, and factors predicting, death in non‐survivors. We hoped to identify predictive factors for pelvic fracture itself being the underlying cause of death.

Materials and Methods

All patients with pelvic fractures admitted to the level 1 Trauma Center with integrated trauma ICU at the Inkosi Albert Luthuli Central Hospital (IALCH) between 1 March 2007 and 31 August 2009 were retrospectively identified and included in this study. Data were extracted from an Ethics Committee‐approved prospective trauma registry (BE207/09) that had been developed in accordance with the Trauma Society of South Africa criteria for Trauma Centers. The study received independent ethical approval from the Biomedical Research Ethics Committee (BE117/09) and the Provincial Health Research Committee of the Province of KwaZulu‐Natal.

Data were extracted from the medical records for analysis and assimilated onto a predetermined proforma. Data collected included sex and age, mechanism of injury, method of admission (direct from trauma scene vs inter‐hospital transfer), pelvic fracture type, additional injuries, blood transfusions and pelvic‐related interventions and complications during admission. The Injury Severity Score (ISS)8 was calculated. Transfusion requirements during the first 24 hours were recorded, as were those at a later stage if related to pelvic surgery. The information in the medical records was cross‐referenced with the independent regional blood bank database to eliminate possible errors in documentation and was found to correlate well.

For the purposes of this study, the only management interventions recorded were the pelvic‐related ones. These were divided into groups according to the procedures that had been performed. These procedures included traction, debridement, external fixation, open reduction with internal fixation, plaster spica (a cast that encompasses the chest, pelvis and femurs that is most commonly used in small children, who cannot be trusted to comply with braces), pelvic wrap‐device, and pelvic packing. Complications during the hospital stay were divided into pelvic‐related and non pelvic‐related complications. Additional injuries on admission were classified as follows: soft tissue injuries; head and face injuries; long bone fractures; chest and lung injuries; injuries in abdominal visceral organs; injuries in abdominal solid organs; and other orthopedic injuries. Prehospital care was not reviewed because some of the patients had been transferred from other hospitals; however, standard prehospital management included intravenous fluids and pelvic‐binding with a sheet or similar device.

Statistical Analysis

Statistical analysis was performed using SPSS for Windows 17.0 (SPSS, Chicago, IL, USA). Pearson X² test was used to compare categorical data and for univariate analysis and one‐way ANOVA to find independent risk factors for mortality. A P‐value <0.05 was considered significant.

Classification of fracture type was based on Tile's classification system9, which comprises type A (stable), type B (rotationally unstable) and type C (vertically and rotationally unstable), and was conducted retrospectively based on clinical notes, reports of plain radiographs and CT scans when available. An experienced trauma surgeon (Timothy C. Hardcastle) supervised the classification. The subclasses of fractures (A1‐2, B1‐3 and C1‐3)9 were not utilized. Cases with isolated acetabular fractures (IAFs) were included in the study but are not classified in the Tile system.

The study patients were divided into survivors, meaning those who survived until discharged from the trauma ICU, and non‐survivors. Among the non‐survivors, a subgroup in which the pelvic fracture was the underlying cause of death was identified; this group is henceforth called the pelvic death subgroup. Causes of death were acquired from autopsy reports from the Department of Forensic Pathology at the hospital, and were cross‐referenced with clinical data. Whether the pelvic fracture was the underlying cause of death was determined by a specialist in forensic pathology (Steven R Naidoo).

Results

Of 548 major trauma cases (ISS > 16) admitted to the trauma unit during the study period, 88 cases with pelvic fractures were identified, giving an overall incidence of pelvic fractures of 16%. Forty‐six cases were male (52%) and the overall mean age was 28 years (range, 2–76 years). Distribution of overall mortality for the different age groups is presented in Fig. 1. Age overall did not meet the requirements for a significant predictor of mortality (P = 0.099). However, age over 45 years was associated with a higher than 50% mortality.

Figure 1.

Comparative age distribution of survivors and non‐survivors with pelvic fractures.

There were 61 survivors and 27 non‐survivors, giving an overall pelvic related mortality of 31%. Among the non‐survivors, the pelvic fracture was determined to be the underlying cause of death in eight cases (30%), the “pelvic death subgroup”. This gave an overall incidence of 9% pelvic fracture‐related death. The pelvic death subgroup had a mean age of 29 years (range, 9–76 years) with a female preponderance (7/8). The most common causes of death among the remaining 19 cases were multiple organ failure (five cases, 26%), hemorrhagic shock (due to other injuries, three cases, 16%), septic shock (three cases, 16%) and other (eight cases, 42%).

Motor vehicle collisions were the commonest external cause of the pelvic fractures (73 cases, 83%), whereas the injury mechanism with the highest overall mortality was falls >6 m (Table 1). The commonest external cause of injury in the pelvic death subgroup, however, was being a pedestrian who was hit by a motor vehicle (five cases, 63%).

Table 1.

ISS and mortality by mechanism of injury

Mechanism of injury	Survivors (cases)	Non‐survivors (cases)	Total number of cases (cases [%])	Mean ISS	Mortality (%)
Pedestrian MVC	28	13	41 (47)	35	32
Passenger MVC	19	6	25 (28)	33	24
Driver MVC	5	2	7 (8)	23	29
Fall >6 m	2	3	5 (6)	42	60
Crush	4	3	7 (8)	22	43
Othera	3	0	3 (3)	36	0
Total	61	27	88 (100)	33	31

MVC, motor vehicle collision.

^aOne train crash, one motor cycle driver crash, one firearm injury.

As to admission type, 32% of cases were admitted directly from the trauma scene (28 cases, mortality 46%) and 68% from referring hospitals (60 cases, mortality 23%). The mean ISS scores for these two groups were 38 and 30, respectively. The overall mean ISS score for all cases was 33 (range, 4–66), with survivors having a mean ISS score of 29 and non‐survivors 36. In the pelvic death subgroup, the mean ISS score was 54. The admission type was a significant predictor of this group as all eight cases (100%) had been brought directly from the trauma scene, implying that paramedics at the scene had accurately identified severe physiological derangement and arranged appropriate prehospital triage to a major trauma centre.

Of the 88 cases, only one had no additional injuries on admission. Thus, the additional injury rate was 99%. The commonest additional injuries were of chest and lung (51 cases, 58%), soft tissue (46 cases, 52%), head and face (35 cases, 40%), and solid abdominal organs (29 cases, 33%). Complications during hospital care occurred among 39 cases (44%), but only 3 (8%) of these had pelvis‐ related complications.

Type C fractures were the least frequent type (Table 2). One patient had a combined type A fracture and an IAF, making the total number of fractures 89. In the pelvic death subgroup, the distribution of fracture types was 50% type B (four cases) and 50% type C (four cases).

Table 2.

ISS and mortality by fracture type9

Fracture type	Survivors (61 cases)	Non‐survivors (27 cases)	Totala	Prevalence (%)	Mean ISS	Mortality, %	P Value
Type A	33	12	45	51	31	27	0.27
Type B	14	7	21	24	35	33	0.48
Type C	4	6	10	11	39	60	0.42
IAF	10	3	13	15	29	23	0.39

^aOne combined Type A and IAF.

Blood transfusion was required in 47 of the 88 cases (53%) during the first 24 hours. The overall mean number of transfused units was 2.3 and among those requiring transfusion 4.3 units. Mortality in the transfused group was slightly higher than for those not requiring blood: 34% and 27%, respectively. In 13 cases who had received transfusion during resuscitation prior to admission to our unit (at referral hospitals), these transfusions were not included in the final data, which may have skewed the data away from statistical significance. Patients in the pelvic death subgroup received a mean of 10.7 transfused units (Table 3), considerably more than the survivor or non‐pelvic death non‐survivor groups. The fracture type with the highest transfusion requirement among the cases requiring transfusion was IAF (Table 4). The units transfused refer to packed red cells or whole blood only. However, freeze‐dried plasma and platelets were transfused as guided by results of blood tests in line with current balanced transfusion guidelines. These were not included in the analysis.

Table 3.

Blood transfusion requirements and ISS by mortality

Patient groups	Transfused patients [cases (%)]	Transfused blood unitsa	Mean ISS
Survivors (61 cases)	31 (51)	4.0	29
Non‐survivors (19 cases)b	13 (68)	3.7	36
Pelvic death (8 cases)	3 (37)	10.7	54

^aAverage number of units among those requiring transfusion.

^bExcluding pelvic death subgroup.

Table 4.

Transfusion requirements by fracture type9

Fracture typea	Cases requiring transfusion (cases [%])	Transfused blood unitsb
Type A (45 cases)	25 (56)	3.8
Type B (21 cases)	9 (43)	4.9
Type C (10 cases)	8 (80)	4.6
IAF (13 cases)	5 (38)	5.4

^aOne combined Type A and IAF.

^bAverage number of units among those requiring transfusion.

Thirty‐four (39%) of the cases underwent interventions regarding their pelvic fractures during their hospital stay. The remainder were managed conservatively (bed rest and mobilization to pain tolerance). Mortality was 27% in the group undergoing interventions and 33% in the group not undergoing interventions (Table 5). Three patients in the pelvic death subgroup had undergone interventions (38%); two had extraperitoneal pelvic packing (for severe ongoing hemorrhage) and one had traction in attempt to reduce a type C fracture.

Table 5.

Outcome by intervention (cases)

Intervention type	Survivors (61 cases)	Non‐survivors (27 cases)	Mortality (%)
Traction	17	7	29
Debridement	4	1	20
External fixation	4	1	20
ORIF	7	1	13
Plaster spica	2	0	0
Sheet	1	0	0
Packing	1	2	67

ORIF, open reduction with internal fixation.

Seventy‐five percent of the patients in the pelvic death subgroup died in the resuscitation room, all of these had been transferred directly from the scene of the injury. With the exception of one patient who died on the operation table during attempted “damage control” surgery, the others all died in the resuscitation bay within the first hour after arrival, despite ongoing appropriate resuscitation. Two patients in the pelvic death subgroup died later, both due to sepsis after open (compound) pelvic fractures, with attempted pelvic packing and multiple other chest and abdominal injuries. The forensic pathologist determined that the other injuries were less severe than the pelvic injuries and that these deaths were related to the open pelvic fractures.

Discussion

By October 2009, the level 1 Trauma Center at the IALCH at which we performed this study had been operational for only three years and had not previously reviewed the care of patients admitted with pelvic fractures. The incidence of pelvic fractures at our center was 16%, which is slightly higher than that previously reported by other major trauma centers, namely 4.5% in children and a range of 9%–15% in adults2, 4, 10. This small difference in incidence compared to other studies may be because IALCH provides the highest level of trauma care in the region and therefore receives only referrals of severely injured cases from other hospitals in the region. Less severe pelvic injuries are managed without transfer to the unit.

Furthermore, the most severely injured cases admitted to the unit arrive directly from the trauma scene by either road ambulance or helicopter. This pre‐hospital triage may have biased our study population towards a higher pelvic fracture incidence, because more severely injured cases have a higher risk of sustaining severe pelvic damage and have a higher associated mortality. The mean ISS was higher in the group admitted directly from the trauma scene (38 compared with 30). This may also explain the higher mortality rate of 31% in our study compared to other studies, as well as the relatively high incidence of death caused directly by the pelvic fracture (9%). The overall mortality in the study patients can be compared to the overall trauma mortality for the whole unit at IALCH during the same period, which was 25%. This figure should be compared with ICU mortality rates, rather than overall trauma mortality rates.

According to Hauschild et al.3, deaths directly attributed to pelvic injury have dropped significantly during the last decades, however this study did not examine incidence figures for different time periods. The present study shows that the most common cause for pelvic fractures was high energy trauma, such as motor vehicle collision, in concurrence with other studies in this field1, 11, 12. However, the mechanism with the highest mortality (60%) as well as the highest mean ISS (42) was fall from height (>6 m), findings which have not been reported in previous studies.

ISS was significantly higher in the pelvic death subgroup than in the non‐survivors and survivors. ISS has been shown to be a better predictor of mortality than fracture classification1. The severity of the fracture types, however, correlated with both mortality and associated injuries (ISS). Type B and type C fractures appeared to portend a higher likelihood of pelvic‐related death. This was not found to be statistically significant and is most likely explained by sample size.

Some commentators have stated that head injuries are a major contributor to death in patients with bleeding pelvic fractures. Kido et al. stated that “Brain injuries should be heavily weighted when evaluating the prognosis of bleeding pelvic fracture cases” 12. In the present study, however, head injuries were not among the three most common causes of death.

Pelvic fractures are known to be associated with significant injuries of the chest, abdomen, diaphragm and other organs. The present study supports the findings of two recent large series that describe similar severe injuries in association with pelvic fractures1, 3.

Magnussen et al. showed that transfusion requirements are linked to the severity of pelvic fractures and that patients with IAFs are as likely as those with other pelvic fractures to require blood transfusions7. In the present study, we did not find a clear correlation between the severity of the fractures and transfusion requirements. However, the subgroup in which pelvic fractures were the underlying cause of death required larger transfusion volumes than did the survivor and overall non‐survivor groups. This difference did not achieve statistical significance. This finding does suggest that patients in the pelvic death subgroup exsanguinated from pelvic bleeding (75% of the causes of death in this subgroup). It is interesting to note that the overwhelming majority of pelvic related deaths occurred in female patients (7:1 ratio). We have no specific explanation for this finding.

The new trauma centre faces a number of challenges that we briefly highlight here to place our study in perspective. Firstly, the center has no angio‐embolization service, despite its physical facilities, because there are no trained interventional radiologists at the hospital.

Secondly, in South Africa there are shortages of blood products and long delivery times because the problem of limited donor numbers has been compounded by the HIV‐AIDS pandemic. This means there are delays in obtaining additional blood products for emergency transfusion beyond the initial two packs of blood available in the emergency refrigerator. Establishment of a written Major Transfusion Policy for the hospital has partially addressed this problem.

Thirdly, many hospitals have limited operation room time and thus a greater proportion of pelvic injuries are managed conservatively compared to the high rate of open reduction and internal fixation in the developed world. This difference in management may have had impacts on outcome, but the numbers are too small to draw definite conclusions regarding this. This limitation has been addressed at this hospital by allocating the Trauma Unit a dedicated general and orthopedic operation room for their trauma cases.

Fourthly and finally, the high rate of motor vehicle collisions (the cause of over 40% of all traumatic injuries in South Africa) is compounded by poor traffic policing and long distances between hospitals. These factors cause delays between prehospital care and reaching definitive care, as do the geography of this very mountainous province of South Africa and the overall shortage of ambulances for the population size. This has been partially addressed by an increased supply of ambulances and prehospital paramedics during and after the FIFA Soccer World cup in 2010.

This study is limited by sample size and the retrospective nature of the data review, despite the data having been prospectively captured during patient admission by using advanced electronic patient records, corroborated by a trauma registry. Larger numbers of patients may have resulted in statistically significant findings. Since this is a new trauma centre, we decided not to compare the results with other local centers as their facilities are very different and the practitioners have much fewer management options available to them. Rather, we decided to compare our results with those in the international literature only.

Conclusions

There were slightly more pelvic fractures and deaths attributed to pelvic injury at this center than previously reported, this discrepancy being explained by the pre‐hospital referral system in effect in the region and the role of this particular hospital. More severe fractures and falls from a height were associated both with additional injuries (higher ISS) and with mortality. More severe fractures were more likely to cause deaths attributed directly to the pelvic injuries. Requirement for major blood transfusions as a result of pelvic fracture hemorrhage was related to mortality.