Secondary (Indirect) Bone Healing

Whereas primary bone healing proceeds with direct formation of bone across the fracture site, secondary bone healing is characterized by a succession of stages and formation of an intermediate callus prior to bone formation. This process shares similarities with the healing of soft tissues and has arbitrarily been divided into three overlapping phases: inflammation, repair, and remodeling (Fig. 86-1) [1]. Each of these healing phases has unique histologic features, and each can occur in isolation or in concert with the other to achieve bone union. This passage through intermediate tissues of increasing stiffness and strength gradually restores the mechanical stability required for bone formation. Ultimately, any tissue other than bone persisting within a fracture gap represents incomplete healing [2].

Figure 86-1. Phases of secondary bone healing. Inflammation (1) is the most intense and shortest phase of secondary bone healing. The fracture gap then fills with tissues of increasing stiffness and strength, eventually allowing bone formation. Granulation tissue (2) is replaced by a fibrocartilaginous callus (3) that mineralizes into a hard callus (4). Remodeling is the longest phase of secondary healing, resulting in gradual disappearance of the callus (Reproduced with permission from: Griffon DJ: Fracture healing. In Johnson AL, Houlton JEF, Vannini R (eds). AO Principles of Fracture Management in Small Animals. Davos: AO Publishing, 2005, pp. 72-98).

Secondary healing is the most common type of healing encountered in clinical orthopedics, especially since the principles of biologic fracture fixation have gained popularity in man and small animals. The concept of “biologic osteosynthesis” emphasizes the role of soft tissue integrity in bone healing and “less than rigid” fixation of the fracture [3]. This concept contrasts with the “biomechanical approach” to fracture management, aiming for anatomic reduction and rigid fixation, potentially leading to primary bone healing. Instead, the goals of biologic fracture fixation are to restore the overall length and alignment of the bone while limiting surgical approach and manipulation of fragments. Applied to highly comminuted fractures of long bones, biologic fracture management accelerates healing and return of biomechanical strength, thereby lowering complication rates in man and small animals [4,5]. Once considered a sign of failure, callus formation has become a goal in the management of extra articular fractures of long bones.

Staged Tissue Repair in Secondary Bone Healing

Inflammatory Phase

An inflammatory phase begins immediately after traumatic disruption of bone and surrounding soft tissues and persists until formation of cartilage or bone is initiated. This stage is essentially characterized by ischemic bone necrosis, hematoma formation, and formation of a fibrin mesh at the fracture line.

Fractures inevitably disrupt medullary vessels leading to hemorrhage. Although contraction and thrombosis of disrupted vessels minimize blood loss, interruption of blood flow leads to hypoxia and necrosis of bone, characterized histologically by the presence of empty lacunae. Radiographically, this resorption of fragment ends may be recognized as a local loss of radioopacity and widening of the fracture gap 5 to 7 days after injury [6]. Bone resorption is primarily mediated via prostaglandins E1 and E2, inflammatory mediators that may also stimulate angiogenesis and proliferation of osteoprogenitor cells [7]. Mononuclear phagocytes delivered by newly formed vessels assist in removal of necrotic bone and aid in construction of the callus. Macrophages are also believed to orchestrate the orderly sequence of wound healing and play a similar role in fracture repair. They contain several growth factors, including fibroblast growth factor (FGF), initiating fibroplasia in soft tissue as well as bone repair [8-10]. Resorption of fragment ends is particularly obvious in spontaneous fracture healing: the fracture gap widens, thereby lowering interfragmentary strain and minimizing the deformation of local tissues [11].

The lack of mechanical support provided by the fracture hematoma is well established. Its biologic contribution to fracture healing, initially questioned, has gained general acceptance among surgeons. In fact, preservation of the fracture hematoma has become one of the goals of indirect reduction techniques and biologic fracture fixation. The main contribution of the hematoma is the release of growth factors, thereby setting the stage for the repair phase. Transplantation of fracture hematoma has been found to induce endochondral bone formation in ectopic sites, which is consistent with osteoinduction [12]. The hematoma also acts as an osteconductive spacer, providing a scaffold for cells, thereby guiding the size and shape of the callus [11,13]. Local acidity and vascular endothelial growth factor (VEGF) contained in the hematoma encourage vascular ingrowth [14]. The formation of new vessels is also stimulated by mast cells, which are abundant during this phase and release vasoactive substances [8,15]. Within hours, a transient extraosseous blood supply emerges from surrounding soft tissues, revascularizing the hypoxic fracture site [16]. The proliferative vascular response was enhanced by a muscle flap, thereby improving the healing of experimental tibial osteotomies in dogs [17]. This study illustrates the importance of preserving soft tissues around the fracture. Unless infection, excessive motion, or extensive necrosis of the surrounding soft tissues persists at the fracture site, the hematoma is resorbed by the end of the first week [2]. Clinically, the end of the inflammatory stage coincides with a decrease in pain and swelling.

Repair Phase

Granulation Tissue

Within a few days, capillary ingrowth, mononuclear cells, and fibroblasts contribute to the transformation of the hematoma into granulation tissue. This initial stage of the repair phase coincides with a slight gain in mechanical strength because granulation tissue can withstand a tension force up to 0.1 Nm/mm2 [18]. Yet, its tolerance to elongation (up to twice the original length) explains its formation at a stage where interfragmentary deformation remains high.

Connective Tissue

As granulation tissue matures into connective tissue, collagen fibers become more abundant; ultimate tension strength (to 1-60 Nm/mm2) and resistance to elongation (17% maximum) increase. Types I, II, and III collagen are initially deposited, but as the maturation process continues, type I collagen predominates [19]. This interfragmentary fibrous tissue is organized in a diagonal pattern, optimizing its ablity to elongate [11].

"Soft Callus"

Mesenchymal cells within the cambium layer of the periosteum, the endosteum, the bone marrow, and adjacent soft tissues start proliferating during the inflammatory phase and differentiate into chondrocytes during the repair phase (Fig. 86-1). This proliferation and differentiation are triggered by several growth factors, among which bone morphogenetic proteins (BMPS) and transforming growth factor-β (TGF-β) play major roles. Periosteal stripping in immature animals results in production of a callus away from the bone as osteoprogenitor cells are pulled with the periosteum [1]. In mature patients, the periosteum has a tendency to tear rather than strip; osteoprogenitor cells remain attached to the bone and produce a callus in contact with the fracture site. Interfragmentary strain, local blood supply, and tissue oxygenation also affect the elaboration and size of the cartilaginous callus. Local hypoxia encourages mesenchymal cells to differentiate into chondrocytes rather than osteoblasts [20-22]. Whereas the external callus relies entirely on an extraosseous vascularization, the internal or medullary callus developing from the endosteal cell layer receives its blood supply from medullary arterioles [22,23]. The presence of a fibrocartilage layer within the medullary canal temporarily interrupts the medullary blood flow across the fracture gap. The external callus and the internal callus both constitute the "bridging callus" [24]. The full extent of the bridging callus is underestimated on radiographs, because the cartilaginous portion is not visible (Fig. 86-2) [6]. This partly explains the discrepancy between palpation of a large callus and radiographic visualization of a relatively smaller ossified bridge.

This early "soft callus" formed during the first 3 weeks after injury resists compression but its ultimate tensile strength (4-19 Nm/mm2) and elongation at rupture (10-12.8%) are similar to those of fibrous tissue [25]. Production of a prominent external callus is a common finding in well vascularized unstable fractures. The resulting enlargement of the cross-sectional diameter in the injured area greatly increases the resistance to bending: the strength efficiency increases by the third power of the distance to the neutral axis of the bone, and the rigidity increases by the fourth power [26]. Increasing proteoglycan concentrations within the fibrocartilage also contribute to the stiffening of the interfragmentary gap [20].

"Hard Callus"

Callus formation does not become apparent on radiographs until mineralization proceeds. The periosteal component of the callus grows first and appears as a collar around the fracture site. The smaller internal callus forming within the medullary cavity is harder to visualize radiographically owing to superimposition with the external callus [6]. Mineralization of the soft callus proceeds from the fragment ends toward the center of the fracture site and forms a "hard callus", following a process similar to endochondral ossification of growth plates [11]. Mitochondria within chondrocytes first appear to accumulate calcium-containing granules [27]. Under hypoxia and anaerobic metabolism, these intramitochondrial deposits of calcium phosphate are released into the extracellular matrix and become the seeds for growth of apatite microcrystallites. Vascular invasion of fibrocartilage is then coupled with the degradation of nonmineralized matrix compartments by macrophages. Following this resorbing front, blood vessels and osteoprogenitor cells form new trabeculae. Although the mechanical properties of this calcified fibrocartilaginous tissue have not been reported, these structures contribute greatly to the restoration of strength and stiffness within the fracture gap, thus allowing formation of compact bone. The ultimate tensile strength of compact bone is around 130 Nm/mm2, but its modulus of elasticity (resistance to deformation) is high (10,000 Nm/mm2) and ability to elongate is limited to 2% [11]. At the end of the repair phase, bone union is achieved but the structure of the fracture site differs from that of the original bone. Yet, the injured bone has regained enough strength and rigidity to allow low-impact exercise [28]. Maturation of the fibrocartilaginous callus results in radiographic disappearance of the fracture line as the fracture gap gains a radioopacity similar to that of adjacent bone (Fig. 86-2).

Figure 86-2. Radiographic appearance of secondary bone healing. Open lateral radiograph of the femur of a 10-month-old dog with a comminuted diaphyseal fracture (A). Radiograph obtained immediately after closed placement of an interlocking nail (B). A bridging callus is noted on radiographs obtained 6 weeks later (C). Remodeling of the callus can be appreciated 12 weeks after surgery, as the callus becomes more fusiform (D).

Remodeling Phase

This final phase of fracture repair is characterized by a morphologic adaptation of bone to regain optimal function and strength. This slow process may last for 6 to 9 years after initial trauma in man, representing 70% of the total healing time of a fracture (Fig. 86-1) [29]. The balanced action of osteoclastic resorption and osteoblastic deposition is governed by Wolff’s law and modulated by piezoelectricity, a phenomenon by which electrical polarity is created by pressure exerted in a crystalline environment [22,30,31]. Axial loading of long bones creates an electropositive convex surface, onto which osteoclastic activity predominates. On the opposite, concave, surface electronegativity is associated with increased osteoblastic activity. The external callus becomes more fusiform and gradually disappears (Fig. 86-2). Remodeling of the internal callus allows reestablishment of a continuous medullary cavity in the diaphysis of the bone. As medullary blood flow resumes, this extraosseous blood supply subsides.

Influence of the Biomechanical Environment on Bone Healing

If adequate vascularity is a prerequisite for bone healing, the biomechanical environment determines the pattern of repair [1]. Bone formation requires restoration of mechanical stability, which may be achieved by a natural process of healing or by osteosynthesis, with partial or complete stabilization of the fracture fragments. Whereas primary bone formation occurs under extreme conditions of reduction and immobilization, spontaneous healing represents the opposite end of the spectrum and best illustrates the mechanisms involved in secondary bone healing.

Spontaneous healing of complete fractures typically occurs in the presence of highly unstable fragment ends. Bone repair must develop in spite of high interfragmentary strain, defined as the deformation occurring at the fracture site relative to the size of the gap. However, bone formation can occur only in a stable biomechanical environment with an interfragmentary strain lower than 2% [32,33]. Nature’s ways to deal with this unfavorable situation become especially obvious: initial contraction of muscles surrounding the fracture, resorption of fragment ends, orderly repair with tissues suitable for the mechanical environment, and formation of a prominent external callus. Progression from soft to hard callus depends on adequate blood supply and gradual increase in stability at the fracture site. If the fracture gap is well vascularized, uncontrolled interfragmentary motion will stimulate callus formation and the fracture will progress to a cartilaginous callus. However, if this callus is unable to stabilize the fragments, a hypertrophic nonunion and pseudoarthrosis will develop. A stable fracture with an adequate blood supply will allow formation of a mineralized callus; however, initial displacement of bone fragments owing to trauma and muscle contraction frequently results in malunions.

The pattern of healing after external coaptation or semi-rigid internal fixation of fractures is intermediate between the biologic stabilization by a callus formed in spontaneous healing and the callus-free repair obtained after absolute stabilization. Fracture healing after external coaptation resembles spontaneous bone repair except that misalignment of fragments is minimized by closed reduction. Gliding implants such as intramedullary pins and nails typically allow some motion, including axial micromotion during weight bearing and rotational shear between fracture fragments that do not interdigitate. The amount of callus produced after application of external fixators is highly variable, depending on the fracture configuration and the rigidity of the applied frame. Callus formation after plate fixation may occur when the implant is not placed on the tension side of the bone, fracture reduction is not perfect, or when the plate lacks rigidity [34]. These observations led to the concept of biologic fixation, which is especially relevant for comminuted fractures. In these cases, perfect apposition of fragments is unlikely and the surgeon privileges biologic factors over anatomic reduction and mechanical stability [3,35]. The general alignment of the joints is restored but manipulation of fragments and adjacent soft tissues is minimized. A buttress plate or a plate-rod combination is applied across the fracture gap, bridging the entire length of the bone. This "less-than-rigid" surgical fixation is less invasive than traditional plate fixation and results in increased callus production. In one study, the bone density and osteogenesis in comminuted fractures were increased 12 weeks after application of a bridging plate, an intramedullary nail, or an external fixator compared with application of lag screws and compression plate [5]. Similar results have been reported in clinical studies where biologic fixation of comminuted fractures increased callus production and accelerated radiographic union and gain of biomechanical strength, allowing earlier return to function in man and small animals [36,37]. More recently, the concept of "elastic plate osteosynthesis" has been applied to unilateral femoral shaft fractures in 24 immature dogs [38]. Each fracture was treated with a veterinary cuttable plate maintained to the bone with two screws placed as far as possible from the fracture site. Although care should be taken to restore the physiologic degree of femoral torsion, fractures were often bridged with callus by 4 weeks (Fig. 86-3).

Figure 86-3. A femoral fracture in a 4-month-old dog. Elastic plating (A). A veterinary cuttable plate is maintained to the bone with two screws placed as far as possible from the fracture site (B). Note the amount of callus bridging the fracture 3 weeks after repair (C).

Role of Growth Factors and Inflammatory Mediators

Fracture healing is orchestrated by complex interactions between a cascade of growth factors, local cells, and their environment. Although our understanding of the sequence of growth factors involved in fracture healing has drastically improved over the last 50 years, the list of agents affecting bone formation is still growing. Research efforts currently focus on signals triggering the release of growth factors and defining the role of each factor within the sequential tissue differentiation characterizing secondary bone healing. Conflicting results regarding the roles of osteoinductive agents reflect the influence of numerous factors on their effects, including target cells, dose tested, species, and characteristics of the local environment. For example, the efficacy of osteoinductive preparations was initially questioned in primates, until further studies identified the dose-dependent effect of osteoinduction and the lower sensitivity of higher vertebrates compared with rodents [39,40]. In spite of these challenges, a general knowledge of growth factors is required to understand their impact on fracture management and their clinical applications as promoters of bone formation (Fig. 86-4) [9,41].

Figure 86-4. Biologic modulation of bone healing: bone morphogenetic proteins (BMPs) and prostaglandins (PGs). BMP: bone morphogenetic protein; TGF-β: transforming growth factor-β; GDF: growth differentiation factor; CDMP: cartilage-derived morphogenetic protein; FGF: fibroblast growth factor; PDGF: platelet-derived growth factor; IL: interleukin; AA: arachidonic acid; COX: cyclooxygenase; NSAIDs: nonsteroidal anti-inflammatory drugs; PGs: prostaglandins; EP1-4R: prostaglandin receptors 1-4 [41,42,52].

The body of evidence supporting the role of growth factors in bone metabolism has led to the recognition of the inflammatory phase as an essential step in staging the subsequent phases of bone healing. The fracture hematoma, once considered a potential hindrance to fracture healing, is now preserved as an endogenous source of prostaglandins (PGs), kinins, and other noncollagenous proteins [42]. Platelets are the first source of mitogenic factors at the traumatized site [43]. In addition to coagulation factors, they release platelet-derived growth factor (PDGF) and transforming growth factor-β1 (TGF-β1) [9]. Although PDGF appears to stimulate osteoblastic proliferation in vitro, its exact role in fracture repair has not been clearly defined [9]. The fibrin seal forming between bone fragments provides a support for migration of inflammatory cells and their byproducts. Macrophages, neutrophils, and mast cells release growth factors (FGF, PDGF, TGFβ, and FGF), promoting angiogenesis and stimulating fibroblasts. The fibroblast growth factors (FGF) are part of a family of nine structurally related polypeptides, among which acidic FGF (FGF-1 or α-FGF) and basic FGF (FGF-2 or β-FGF) are the most abundant. Both factors encourage growth and differentiation of a variety of cells, including epithelial cells, osteoblasts, and chondrocytes. The mitogenic effects of FGF-1 have been associated with proliferation of chondrocytes, while osteoblasts express FGF-2 receptors [44,45]. Inflammatory cells also produce cytokines (interleukins such as IL-1, IL-6, and tumor necrosis factor or TNF), attracting mesenchymal cells to the fracture site (Fig. 86-4). Pro-inflammatory stimuli and cytokines promote the formation of prostaglandins (PGs) by osteoblasts and osteoclasts. Among these, PG-E2 is the most abundant PG produced by osteoblasts and the most potent PG stimulating both bone formation and resorption [42]. These effects are mediated via interaction with four receptor subtypes (EP1R, EP2R, EP3R, and EP4R), especially EP2R and EP4R [42]. PGs increase the number and activity of osteoclasts [46]. These cells release proteases that dissolve bone mineral matrix and collagen and remove damaged bone, thereby contributing to bone resorption and release of growth factors contained in the matrix. Other studies have documented increased bone formation and turnover after subcutaneous injection of PGE2 in dogs [47,48]. This activity would result from mitogenic properties on osteoblasts and stimulation of undifferentiated cells to engage in osteogenic differentiation [42,46]. In humans, impact loading has been associated with an increased production of PG [49]. These findings support the theory according to which PGs mediate the physiologic response to mechanical loading [42]. Based on their range of activities, PGs are likely to contribute to all phases of fracture healing, including the remodeling phase of secondary bone healing.

In the repair phase, chemotaxis, proliferation, coordination, and differentiation of stem cells into chondrocytes or osteoblasts are orchestrated by numerous growth factors, among which TGF-β and bone morphogenetic proteins (BMPs) play a major role (Fig. 86-4). Both are members of the TGF-β superfamily, a group of dimeric proteins, acting as growth and differentiation factors during embryogenesis and tissue repair in postnatal life. TGF-β1 is released by platelets immediately after a fracture, but the most intense staining occurs during cartilage cell proliferation and endochondral ossification [50]. Although the response to TGF-β varies with dose, species, and biologic environment, this agent stimulates the proliferation of undifferentiated mesenchymal stem cells and induces the expression of BMPs [9,41]. Since Urist discovered the phenomenon of osteoinduction and attributed it to a single protein in 1965, the structure of 16 BMPs has been identified [51,52]. All belong to the TGF-β superfamily, except BMP-1. Among these, BMP-2 through 7 and BMP-9 have been found osteoinductive, meaning that these proteins can provide the primordial signal for mesenchymal stem cells to engage into osteoblastic differentiation [52]. Each BMP exerts its effects by binding to a specific combination of membrane receptors (Type I and II serine threonine sulfate), and activating the intracellular Smad signaling pathway that eventually determines the outcome of the signal [52]. Numerous studies have described the typical sequence of endochondral ossification occurring after ectopic implantation (sites normally devoid of osteoprogenitor cells, such as the subcutaneous tissue) of these agents: recruitment and proliferation of monocytes and mesenchymal stem cells, differentiation into chondrocytes, hypertrophy of chondrocytes, and calcification of the matrix, followed by vascular invasion, osteoblastic differentiation, and bone formation. This sequence ends with remodeling of the new bone, leading to bone marrow formation [53]. This staged differentiation of tissues illustrates the roles of BMPs in natural fracture healing and eventually led to clinical trials of two recombinant proteins, rhBMP-2 and rhBMP-7, in open and nonunion fractures of the tibia in man [54].

Effects of Nonsteroidal Anti-Inflammatory Drugs on Bone Healing

Much interest has been focused on the management of pain in small animals and its impact on bone healing. Nonsteroidal anti-inflammatory drugs (NSAIDs) are now routinely prescribed as part of the perioperative management of fracture patients for their antipyretic, analgesic, and antiphlogistic properties. Preoperative administration of NSAIDs can minimize peripheral and central sensitization to painful stimuli, thereby improving postoperative analgesia and preventing the development of chronic pain syndromes [55]. They do not cause sedation and appear to decrease the doses of opioids required in the multimodal analgesic approach most commonly used to manage trauma cases in the immediate perioperative phase. Combined, these benefits accelerate postoperative return to function, shorten hospitalization, and reduce associated costs for owners [56]. NSAIDs act essentially at the peripheral and regional levels, but are potent enough to justify their use as sole analgesics for patients discharged from the hospital. Their action relies on the inhibition of cyclooxygenase (COX), a catalyzer for the enzymatic conversion of arachidonic acid liberated from cell membranes into prostaglandins, prostacyclin, and thromboxane (Fig. 86-4). This mechanism of action is responsible for the side effects of NSAIDs, including impaired platelet function, renal vasoconstriction, and gastrointestinal ulceration. These limitations have prompted the development of anti-inflammatory agents that inhibit preferentially the cyclooxygenase induced during inflammation (COX-2), while sparing the endogenous isoenzyme (COX-1). COX-2 preferential NSAIDS (coxibs) have a 2- to 100-fold difference in the concentration of drug necessary to inhibit COX-2 versus COX-1 in vitro [57]. These agents have essentially replaced nonselective cyclooxygenase inhibitors in small animal practice. Among these, carprofen has recently been evaluated in dogs undergoing fracture repair [58]. In this prospective study of 26 traumatized dogs, perioperative administration of carprofen did not cause clinically relevant adverse effects on hemostasis or renal function. Nonetheless, the perioperative use of NSAIDs in fracture patients has been tempered by concerns of potential deleterious effects on bone healing. Indeed, NSAIDs may theoretically inhibit bone formation because they interfere with the release of inflammatory mediators, including PGs. COX-2 has been found to be the rate-limiting enzyme in the synthetic pathway of PGs [46]. Any inhibition of COX-2 would be expected to affect the release of PGs and suppress their contribution to bone healing (outlined in the section on growth factors and bone healing). COX-2 expression naturally increases after fracture and suppression of this enzyme results in a relative reduction in osteoblastogenesis [46]. The effects of nonselective NSAIDs, such as indomethacin and aspirin, have been studied extensively, and the majority of experimental in vivo studies report a negative effect on bone healing [59]. Fewer studies have evaluated the effect of COX-1-sparing agents and several report conflicting results. Table 86-1 summarizes the results of selected experimental and all clinical studies published over the last 5 years regarding the effects of COX-2 selective NSAIDs on bone healing [60-68]. Among these, the effects of short-term (10 days) administration of celecoxib (3 and 6 mg/kg) were compared with a negative control in rats with induced femoral fractures [64]. Celecoxib impaired fracture healing and increased the rate of nonunions (26%, 9/34) compared witih the negative control (0/41). The fracture callus at 8 weeks contained more cartilage in the treated group, decreasing its biomechanical properties. Using the same fracture model, other investigators tested the effects of another COX-2 preferential agent, valdecoxib (5 mg/kg/day) administered for 7 or 21 days after fracture [63]. They found no difference in physical, biomechanical, and histologic evaluations of healing after the shorter course of administration. In contrast, 21 days of treatment impaired healing at 21 days compared with the placebo group, but the difference disappeared by 35 days. The temporal effect of COX-2 selective NSAIDs on bone ingrowth was further evaluated using a bone chamber model in rabbits [66]. In this study, rofecoxib decreased bone production if administered continuously for 6 weeks, but had no adverse effect if administered for 2 weeks, whether early or late after implantation of the chamber in the tibia. The overall evidence derived from animal studies is that a short course of COX-1-sparing NSAIDS may alter the callus forming early in the repair phase but has no long-term effect on fracture healing. However, extrapolation to small animal practice warrants caution because these results were obtained on rodents with drug regimens differing from those applied clinically. Few clinical studies have attempted to address the effect of COX-1-sparing NSAIDs in fracture patients (Table 86-1), with a validity limited by their retrospective nature. Better evidence was recently provided in a prospective, randomized, double-blind study of human patients undergoing spinal arthrodesis [68]. In this study, patients receiving celecoxib 1 hour before induction and every 12 hours for 5 days after surgery had lower pain scores and a similar rate of nonunions at 1 year than did patients in the placebo group. Based on the current literature, no firm evidence-based recommendation can be made regarding the use of NSAIDs in fracture treatment. NSAIDs may theoretically delay bone healing, especially if long courses and/or high doses are prescribed in patients with compromised fracture healing. Even in these patients, the potential delay in bone repair must be weighed against the benefits of NSAIDs outlined earlier. In the future, large, randomized clinical trials of agents commercialized for small animals should be developed to assess their effects on objective signs of healing (bone mineral density, size of callus) as well as patient-oriented measures of outcome (removal of implant, return to unrestricted exercise).