Damage Control Orthopaedics

 THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 87-A · N U M B E R 2 · F E B R U A R Y 2005

D A M A G E C O N T RO L ORTHOPAEDICS

Damage Control Orthopaedics EVOLVING CONCEPTS IN THE TREATMENT OF PATIENTS WHO HAVE SUSTAINED ORTHOPAEDIC TRAUMA BY CRAIG S. ROBERTS, MD, HANS-CHRISTOPH PAPE, MD, ALAN L. JONES, MD, ARTHUR L. MALKANI, MD, JORGE L. RODRIGUEZ, MD, AND PETER V. GIANNOUDIS, MD An Instructional Course Lecture, American Academy of Orthopaedic Surgeons

Many orthopaedic patients who have sustained multiple injuries benefit from the early total care of major bone fractures. However, the strategy is not the best option, and indeed might be harmful, for some multiply injured patients. Since foregoing all early surgery is not the optimal approach for those patients, the concept of damage control orthopaedics has evolved. Damage control orthopaedics emphasizes the stabilization and control of the injury, often with use of spanning external fixation, rather than immediate fracture repair. The concept of damage control orthopaedics is not new; it has evolved out of the rich history of fracture care and abdominal surgery. This article traces the roots of damage control orthopaedics, reviews the physiologic basis for it, describes the subgroups of patients and injury complexes that are best treated with damage control orthopaedics, reports the early clinical results, and provides a rationale for modern fracture care for the multiply injured patient. Definition of Damage Control Orthopaedics Damage control orthopaedics is an approach that contains and stabilizes orthopaedic injuries so that the patient’s

overall physiology can improve. Its purpose is to avoid worsening of the patient’s condition by the “second hit” of a major orthopaedic procedure and to delay definitive fracture repair until a time when the overall condition of the patient is optimized. Minimally invasive surgical techniques such as external fixation are used initially. Damage control focuses on control of hemorrhage, management of soft-tissue injury, and achievement of provisional fracture stability, while avoiding additional insults to the patient. History of Fracture Surgery and Birth of Damage Control Orthopaedics We previously stated that: “Information illustrating the benefits of fracture stabilization after multiple trauma has been gathering for almost a century.”1 We also noted that during this time “fears of the ‘fat embolism syndrome’ also dominated the philosophy in managing polytrauma patients.” Early manipulation of long-bone fractures was considered unsafe2. External fixation, an essential component of damage control orthopaedics, developed slowly and was outpaced by the development of internal

fixation. In Switzerland in 1938, Räoul Hoffmann produced an external fixator frame that allowed the fracture to be mechanically manipulated and reduced3. In 1942, Roger Anderson advocated castless ambulatory treatment of fractures with use of a versatile linkage system, but the device was banned in World War II for being too elaborate3. In 1950, a survey by the Committee on Fractures and Traumatic Surgery of the American Academy of Orthopaedic Surgeons (AAOS) concluded that the complications of external fixation frequently exceed any advantages of the procedure3. Also in 1950, Gavril Abramovich Ilizarov developed the ring system for fractures and deformities, but his device did not reach the West until the late 1970s. On March 15, 1958, Maurice Müller, Hans Willenegger, and Martin Allgöwer convened a group of interested Swiss general and orthopaedic surgeons, including Robert Schneider and Walter Bandi at the Kantonsspital, Chur, Switzerland, to discuss the status of fracture treatment, which usually included traction and prolonged bed rest and led to poor functional results in a high percentage of patients4. On November 6, 1958, these pioneering surgeons established

 THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 87-A · N U M B E R 2 · F E B R U A R Y 2005

the Arbeitsgemeinschaft für Osteosynthesefragen (the Association for the Study of Internal Fixation, or ASIF), or AO, in Biel, Switzerland4. The key objective of the AO was the early restoration of function, whether a patient was being treated for an isolated fracture or for multiple injuries4. Matter noted that this strategy led to “aggressive traumatology involving early total care of the trauma victim, culminating in the statement: This patient is too sick not to be treated surgically.”4 By the 1980s, the accepted care of a major fracture was early or immediate fixation5. Substantiating this approach were eleven studies (ten retrospective and one prospective), with the one by Bone et al.6 being most frequently cited. Bone et al. reported that the incidence of pulmonary complications (adult respiratory distress syndrome, pneumonia, and fat embolism) was higher and the stays in the hospital and the intensive care unit were increased when femoral fixation was delayed. In 1990, Border reported on a comprehensive study of patients with blunt trauma that challenged the accepted practice of immediate definitive fixation7. This changed practice in the early 1990s, and a more selective approach to fracture fixation was used; however, early fixation was still performed in most cases. During the 1990s, more was learned about the parameters associated with adverse outcomes in multiply injured patients and about the systemic inflammatory response to trauma8. It became clear that fracture surgery, especially intramedullary nailing, has systemic physiologic effects. These effects became known as the “second hit” phenomenon. The era of damage control orthopaedics started around 1993. Two reports from one institution9,10 described temporary external fixation of femoral shaft fractures in severely injured patients. From 1989 to 1990, the frequency of using temporary external fixation increased from 10%. The mean duration of external fixation until intramedullary nailing was less than one week. Compared with patients treated with immediate definitive fixa-


tion, those treated initially with external fixation had more severe injuries, with higher injury severity scores and transfusion requirements in the initial twenty-four hours. The term “damage control” began to be used in the orthopaedic literature over the last six to seven years1,9-12. History of Abdominal Damage Control Surgery The concept of damage control surgery was developed first in the field of abdominal surgery. The benefits of controlling hemorrhage and contamination and leaving the abdomen open, in lieu of definite repair of injuries and closure of the abdomen, improved the survival of patients with the lethal triad of hypothermia, acidosis, and coagulopathy. Abdominal damage control surgery was described as the sum total of all maneuvers required to ensure survival of a multiply injured patient who was exsanguinating; its purpose was to control rather than definitely repair injuries13. In the 1940s and 1950s, Arnold Griswold, of Kentucky, used a damage control approach to penetrating injuries of the abdominal cavity14. In 1981, Feliciano et al. reported that nine of ten patients who had undergone hepatic packing for the treatment of exsanguinating hemorrhage survived15. Stone et al., in 1983, described a stepwise approach involving intra-abdominal packing and a laparotomy that was terminated rapidly16. In 1992, Burch et al. reported a 33% survival rate in a group of 200 patients treated with abbreviated laparotomy and a planned reoperation17. Rotondo and Zonies, in 1993, coined the term “damage control” and reported a 58% rate of survival of patients treated with a standardized protocol18. In short, the concept of damage control was first used in abdominal surgery to describe a systematic threephase approach designed to disrupt a lethal cascade of events leading to death by exsanguination13. Phase one involved an immediate laparotomy to control hemorrhage and contamination18. Phase two was resuscitation in the intensive care unit with improvement

of hemodynamics, rewarming, correction of coagulopathy, ventilatory support, and continued identification of injuries. Phase three consisted of a reoperation for removal of intra-abdominal packing, definitive repair of abdominal injuries, and closure and possible repair of extra-abdominal injuries. Damage control surgery in the abdomen has gained widespread acceptance throughout North America and Israel18,19. Physiology of Damage Control Orthopaedics The physiologic basis of damage control orthopaedics is beginning to be understood. Traumatic injury leads to systemic inflammation (systemic inflammatory response syndrome) followed by a period of recovery mediated by a counter-regulatory antiinflammatory response (Fig. 1)20. Severe inflammation may lead to acute organ failure and early death after an injury. A lesser inflammatory response followed by an excessive compensatory anti-inflammatory response syndrome may induce a prolonged immunosuppressed state that can be deleterious to the host. This conceptual framework may explain why multiple organ dysfunction syndrome develops early after trauma in some patients and much later in others. Within this inflammatory process, there is a fine balance between the beneficial effects of inflammation and the potential for the process to cause and aggravate tissue injury leading to adult respiratory distress syndrome and multiple organ dysfunction syndrome. The key players in the host response appear to be the cytokines, the leukocytes, the endothelium, and subsequent leukocyte-endothelial cell interactions21. Reactive oxygen species, eicosanoids, and microcirculatory disturbances also play pivotal roles22. The development of this inflammatory response and its subsequent, often fatal consequences are part of the normal response to injury. When the initial massive injury and shock give rise to an intense systemic inflammatory syndrome with the potential to cause remote organ injury,

 THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 87-A · N U M B E R 2 · F E B R U A R Y 2005


Fig. 1

After trauma, there is a balance between the systemic inflammatory response and the counterregulatory anti-inflammatory response. Severe inflammation can lead to acute organ failure and early death. A lesser inflammatory response coupled with an excessive counter-regulatory antiinflammatory response may also induce a prolonged immunosuppressed state that can be deleterious to the host. SIRS = systemic inflammatory response syndrome, and CARS = counterregulatory anti-inflammatory response syndrome.

this “one hit” can cause an excessive inflammatory response that activates the innate immune system, including macrophages, leukocytes, natural killer cells, and inflammatory cell migration enhanced by interleukin-8 (IL-8) production and complement components (C5a and C3a). When the stimulus is less intense and would normally resolve without consequence, the patient is vulnerable to secondary inflammatory insults that can reactivate the systemic inflammatory response syndrome and precipitate late multiple organ dysfunction syndrome. The second insult may take many forms as a result of a variety of circumstances, such as sepsis and surgical procedures, and is the basis for the decision-making process regarding when and how much to do for a “borderline” multiply injured patient (as will be defined later). Hyperstimulation of the inflammatory system, by either single or multiple hits, is considered by many to be the key element in the pathogenesis of adult respiratory distress syndrome and multiple organ dysfunction syndrome23.

The First and Second-Hit Phenomena Numerous studies have demonstrated that stimulation of a variety of inflammatory mediators takes place in the immediate aftermath of trauma24-27. This response initially corresponds to the first-hit phenomenon25. Hoch et al. reported elevation in plasma concentrations of IL-6 and IL-8 in patients with an injury severity score of ≥25 points28. An immediate increase in expression of neutrophil L-selectin was reported in patients with an injury severity score of ≥16 points29. Similarly, a significant (p < 0.05) increase in the expression of the integrin CD11b was noted in more severely injured patients29. The development of multiple organ dysfunction syndrome has also been associated with a persistent elevation of CD11b expression on both neutrophils and lymphocytes for 120 hours, a finding that is suggestive of neutrophil activation in the early development of leukocytemediated end-organ injury. Several other studies have clearly demonstrated the effect of injury severity on the de-

gree of stimulation of the inflammatory markers8,30. While selective immunostimulation may play a critical role in the development of severe complications after injuries, it is also clear that the governing effect of surgical or accidental trauma on immune function is immunosuppression. Several authors have demonstrated the immunosuppressive effect of trauma31,32. Following trauma, the production of immunoglobulins and interferon decreases and many patients become anergic, as assessed with delayed hypersensitivity skin-testing, and are thus exposed to an increased risk of posttraumatic sepsis33. Defects in neutrophil chemotaxis, phagocytosis, lysosomal enzyme content, and respiratory burst have also been reported. Immunosuppression contributes to the etiology of infection and sepsis after trauma34. The biological profile of the first hit in trauma patients is being defined. Obertacke et al. demonstrated the importance of the first hit by using bronchopulmonary lavage to assess changes in pulmonary microvascular permeability in patients who had sustained multiple trauma35. The permeability of the pulmonary capillaries increased following multiple trauma, and patients in whom adult respiratory distress syndrome later developed had a high correlation (r = 0.81) with increased permeability within just six hours after admission than did those who had had an uneventful recovery. The development of a massive immune reaction in a patient with bilateral femoral fracture who showed a massive inflammatory reaction, which was subsequently hyperstimulated by the surgical procedure itself (bilateral reamed femoral nailing), further supports the importance of the first-hit phenomenon36. Although there was no obvious additional risk factor present (i.e., no chest injury), the patient died from full-blown adult respiratory distress syndrome three days after the injury. This case not only clearly illustrates the existence of biological variation in the inflammatory response to injury, but also confirms the importance of the degree of

 THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 87-A · N U M B E R 2 · F E B R U A R Y 2005


TABLE I Cytokines That Are Important Inflammatory Mediators Group

Examples

Interleukins (IL)

IL-1, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-11, IL-12, IL-13, IL-18

Tumor necrosis factors (TNF)

TNF, lymphotoxin (LT)

Interferons (IFN)

IFN-alpha, IFN-beta, IFN-gamma

Colony stimulating factors (CSF)

G-CSF, M-CSF, GM-CSF

the response to the first hit and the response to the second (surgical) hit that created the final fatal event. The above studies suggest that the degree of the initial injury is important in determining a patient’s susceptibility to posttraumatic complications. The concept that a secondary surgical procedure creates an additional inflammatory insult (a second hit) was specifically addressed in a prospective study of 106 patients with an average injury severity score of 40.6 points37. Forty patients in whom respiratory, renal, or hepatic failure developed, alone or in combination, following a secondary surgical procedure were compared with patients in whom no such complications developed. There was a significant (p < 0.05) elevation of the neutrophil elastase and C-reactive protein levels and a reduction in the platelet counts in the forty patients with systemic complications. Abnormality of those three parameters predicted postoperative organ failure with an accuracy of 79%37. The first and second-hit phenomena in trauma patients were demonstrated in a study in which femoral nailing was considered to be the second hit (Fig. 2)8. That study demonstrated similar responses to reamed and unreamed nailing in terms of neutrophil activation, elastase release, and expression of adhesion molecules. These concepts of biological responses to different stimuli (first and second hits) have now become the basis of our treatment plans and illustrate the impact of the operative procedure on trauma patients at risk for exhaustion of their biological reserve (Fig. 3). Markers of Immune Reactivity Inflammatory markers may hold the key to identifying patients at risk for the

development of posttraumatic complications such as multiple organ dysfunction syndrome (Table I). Common serum markers can be divided into markers of mediator activity such as Creactive protein, tumor necrosis factorα (TNF-α), IL-1, IL-6, IL-8, IL-10, and procalcitonin and markers of cellular activity such as CD11b surface receptor on leukocytes, endothelial adhesion molecules (intercellular adhesion molecule-1 [ICAM-1] and e-selectin), and HLA-DR class-II molecules on peripheral mononuclear cells. C-reactive protein, procalcitonin, TNF-α, IL-1, and IL-8 have not been shown to be reliable markers38-43. However, IL-6 correlates well with the degree of injury, appears to be a reliable index of the magnitude of systemic inflammation, and correlates with the outcome12.

IL-10 inhibits the activity of TNF-α and IL-1, and the levels detectable in the circulation correlate with the initial degree of injury. Persistently high levels of IL10 also correlate with sepsis. However, its role in predicting outcome is still debatable44. Regarding the markers of cellular activity, mixed results have been reported in the literature about the efficacy of endothelial adhesion molecules (ICAM-1 and e-selectin) and the CD11b receptor of leukocytes45. HLADR class-II molecules mediate the processing of antigen to allow for cellular immunity. They are considered to be reliable markers of immune reactivity and a predictor of outcome following trauma46,47. Napolitano et al. reported that the severity of the systemic inflamma-

Fig. 2

Mean plasma elastase concentrations (and 95% confidence intervals) before and after intramedullary nailing of the femur from the time of admission to the emergency room (A&E) to 168 hours after surgery8. The control group is shown by the dotted line. Ind = induction of anesthesia, and Nail Ins. = nail insertion. (Reprinted, with permission, from Giannoudis PV, Smith RM, Bellamy MC, Morrison JF, Dickson RA, Guillou PJ. Stimulation of the inflammatory system by reamed and unreamed nailing of femoral fractures. An analysis of the second hit. J Bone Joint Surg Br. 1999;81:359.)

 THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 87-A · N U M B E R 2 · F E B R U A R Y 2005

tory response syndrome at admission may be an accurate predictor of mortality and the length of stay in the hospital by trauma patients48. In another study, the ratio of IL-6 to IL-10 was found to correlate with injury severity after major trauma, and this ratio was recommended as a useful marker to predict the degree of injury following trauma49. The level of plasma DNA has been found to increase after major trauma and has also been suggested as a potentially valuable prognostic marker for patients at risk50. It appears that, at present, only two markers, IL-6 and HLA-DR class-II molecules, accurately predict the clinical course and outcome after trauma. IL-6 measurement has already been implemented as a routine laboratory test in several trauma centers. Because of the additional laboratory processing required for tests of HLA-DR class-II molecules (antibody staining of cells and flow cytometric analysis), the use of such tests has not found great clinical acceptance. Genetic Predisposition and Adverse Outcomes Biological variation and genetic predisposition are increasingly mentioned as explanations of why serious posttraumatic complications develop in some patients and not in others51. Some individuals may be “preprogrammed” to have a hyperreaction to a given traumatic insult. Genetic polymorphism of the neutrophil receptor for immunoglobulin G, CD16, has been reported and is associated with functional differences in neutrophil phagocytosis52. An inherited predisposition toward high or low levels of HLA-DR expression is further evidence of a genetic component in the immune response to injury46. Additional evidence of genetic predisposition is found in the cytokine genes. The single base pair polymorphism at position –308 in the TNF gene was associated with an increased incidence of sepsis and with a worse outcome after major trauma, postoperative sepsis, and sepsis in a medical intensive care unit53-55. This association depends on the presence of the TNF2 allele. Ho-


Fig. 3

The two-hit theory is shown schematically. The first hit is the initial traumatic event, and the second hit is the definitive orthopaedic procedure, usually femoral nailing. MODS = multiple organ dysfunction syndrome, and ARDS = adult respiratory distress syndrome.

mozygosity for the TNFB2 allele is associated with an increased incidence of severe sepsis and a worse outcome. The risk of posttraumatic sepsis developing is 5.22 times higher in patients who are homozygous for TNFB256. Homozygous patients also have higher circulating TNF-α concentrations and higher multiple organ dysfunction syndrome scores compared with heterozygotes57. IL-6 polymorphisms have been reported and were detected in both the 3′ and the 5′ flanking regions and exon 558,59. The SfaNI polymorphism is located at position 174. A homozygotic constellation of this polymorphism coincided with decreased IL-6 serum levels during inflammation60,61. Polymorphisms in the IL-10 gene have also been demonstrated62. Eskdale et al. reported that stimulation of human blood cultures with bacterial lipopolysaccharide showed large interindividual variation in IL-10 secretion63. They concluded that the ability to secrete IL-10 can vary in humans according to the genetic composition of the IL-10 locus. Recently, isolated case reports of germline defects in the cellular receptor for interferon-gamma (IFN-γ) were described, and the mutations were characterized64,65. Davis et al. conducted

a pilot study of thirty-eight patients who had sustained blunt trauma and found that the microsatellite polymorphism AA correlated strongly with infection66. These findings portend polymorphism in the receptor itself and thus represent a genetic basis for the development of the infection. Early identification of patients at risk for adverse outcomes and complications may allow directed intervention with biological response modifiers in order to improve morbidity and mortality rates. Use of biochemical and genetic markers to identify patients “at risk” after orthopaedic trauma may facilitate clinical decision-making regarding when to switch from early total care to damage control orthopaedics. Patient Selection for Damage Control Orthopaedics Because biomechanical and genetic testing is currently not practical, it remains a clinical decision when to shift from early total care to damage control orthopaedics. Which patient should be treated with damage control orthopaedics instead of early total care after orthopaedic trauma should be decided on the basis of the patient’s overall physiologic status and injury complexes. Many trauma scoring systems

 THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 87-A · N U M B E R 2 · F E B R U A R Y 2005

(e.g., the abbreviated injury scale67, injury severity score68,69, revised trauma score70, anatomic profile71, and Glasgow coma scale72) have been developed in an attempt to describe the overall condition of the trauma patient. However, Bosse et al.73 noted that “there is no score that assists in decision-making during the acute resuscitation phase.” Therefore, it may be that one cannot rely exclusively on a scoring system. Additional data must be synthesized, and the overall status of the patient should be stratified into one of four categories. Patients who have sustained orthopaedic trauma have been divided into four groups: stable, borderline, unstable, and in extremis74. Stable patients, unstable patients, and patients in extremis are fairly easy to define. Stable patients should be treated with the local preferred method for managing their orthopaedic injuries. Unstable patients and patients in extremis should be treated with damage control orthopaedics for their orthopaedic injuries. Borderline patients are more difficult to define. One of us (H.-C.P.) and colleagues defined them as patients with polytrauma and an injury severity score of >40 points in the absence of thoracic injury, or an injury severity score of >20 points with thoracic injury (an abbreviated injury score of >2 points); polytrauma with abdominal trauma (a Moore score75 of >3 points); a chest radiograph showing bilateral lung contusions; an initial mean pulmonary artery pressure of >24 mm Hg; or an increase in pulmonary artery pressure of >6 mm Hg during nailing (Table II)74. Borderline orthopaedic trauma patients are probably best treated with damage control orthopaedics. The term “borderline patient” describes a predisposition for deterioration74. Among other factors, thoracic trauma appears to play a crucial role in this predisposition. However, whether femoral fractures in patients with chest trauma should be treated with definitive stabilization or should be stabilized with a temporary external fixator remains a subject of debate. The clinical situation, including the presence or absence of a criterion


TABLE II Clinical Parameters Used in Hannover, Germany, to Define the “Borderline” Patient for Whom Damage Control Orthopaedics Is Often Preferred Polytrauma + injury severity score of >20 points and additional thoracic trauma (abbreviated injury score >2 points) Polytrauma with abdominal/pelvic trauma (Moore score75 >3 points) and hemorrhagic shock (initial blood pressure 24 mm Hg Increase of >6 mm Hg in pulmonary arterial pressure during intramedullary nailing

indicating borderline status (Table II) and factors associated with a high risk of adverse outcomes (Table III), should determine how the patient is treated. In Louisville, some of the additional clinical criteria that we have used as a basis for shifting to damage control orthopaedics include a pH of 15 points, those

treated with fixation within twentyfour hours after the injury had the highest Glasgow coma scale scores at the time of discharge105. However, since only the mean head abbreviated injury scale score, and not the Glasgow coma scale score on admission, was reported, these results are very difficult to interpret accurately. Hofman and Goris found that the Glasgow coma scale score was better in a group treated with early fixation than it was in a group treated with late fixation, but the difference did not reach significance106. The initial management of a patient with a head injury should be similar to that of other trauma patients, with a focus on the rapid control of hemorrhage and restoration of vital signs and tissue perfusion. A brain injury can be made worse if resuscitation is inadequate or if operative intervention such as long-bone fixation decreases mean arterial pressure or increases intracranial pressure. The

treatment protocol for unstable patients should be based on the individual clinical assessment and treatment requirements rather than on mandatory policies with respect to the timing of fixation of long-bone fractures. In such cases, damage control orthopaedics can provide temporary osseous stability to an injured extremity, functioning as a temporary bridge to staged definitive osteosynthesis, without worsening the patient’s head injury or overall condition. Intracranial pressure monitoring should be utilized in the intensive care unit as well as during surgical procedures in the operating room. Aggressive management of intracranial pressure appears to be related to an improved outcome. Maintenance of cerebral perfusion pressure at >70 mm Hg and intracranial pressure at