CLINICAL ORTHOPAEDICSAND RELATED RESEARCH Number 339, pp 82-91 0 1997 Lippincolt-Raven Publishers Modified Transverse Locking Nail Fixation of Proximal Femoral Fractures B.H. Ziran, MD; N.A. Sharkey, PhD; T.S. Smith, MS; G. Wang; and M. W. Chapman, MD tween implant constructs. All anatomic specimens failed, with fractures of the proximal fragment involving medial and lateral cortices. Synthetic specimens did not fracture but showed failure with implant deformation at the level of the skeletal defect. The use of high seated transverse locking nails for complex proximal femoral fractures is a viable option and has comparable in vitro mechanical performance with reconstruction nails. Although not shown to be a problem in the present study, clinical evaluation of screws through the medial femoral neck cortex is required. It was hypothesized that transverse locking screws of intramedullary nails, seated above the lesser trochanter, provide equal strength to that of reconstruction nails, and that screws placed through the medial cortex of the femoral neck do not have adverse biomechanical effects during physiologic loading. Synthetic femurs (n = 10) and paired anatomic specimen femurs (n = 14) were tested intact and with an intramedullary device in place. Intact specimens were loaded nondestructively, then a segmental subtrochanteric defect was created and either a high seated transverse locking nail or a reconstruction nail was inserted and statistically locked. Axial and torsional stiffness were determined followed by axial failure testing. Mechanical parameters evaluated were stiffness, displacement, and energy. The implanted specimens did not show any statistically significant difference between transverse or reconstruction screw constructs with any of the measured parameters (stiffness, displacement, and energy). Failure tests in implanted specimens also did not show any statistically significant difference in yield load, yield displacement, or energy to failure be- Fixation of unstable fractures of the proximal femur particularly is challenging because of the large axial forces and bending moments occurring in this regi0n.6JlJ5~14~26 Although standard intramedullary nails offer distinct advantages to plates, in cases where the posteromedial buttress is not intact, nails where the proximal interlocking mechanism uses screws directed up the femoral neck into the femoral head (hereby referred to as reconstruction nails) or plates must be used.24.7. 16,18325 One of the commercially available transverse locking nails (Alta femoral nail, Howmedica, Rutherford, NJ) potentially is useful for such fractures because of the design of the proximal fixation. The proximal holes for crosslocking screws in this nail are closer to the top of the nail than other nail designs, and therefore can be From the Department of Orthopedic Surgery, University of California, Davis, Sacramento. CA. Implants were provided by Howmedica Inc, Rutherford, NJ. Reprint requests to B.H. Ziran, MD, Department of Orthopedic Surgery, University of Pittsburgh, 1010 Kaufman Building, 3471 Fifth Avenue, Pittsburgh, PA 15213-3221. 82 Number 339 June, 1997 placed above the lesser trochanter, achieving fixation in the femoral neck or the femoral head. The purpose of this study was to measure the acute strength of a fixation construct in a simulated unstable proximal femoral fracture, comparing intramedullary nails locked with transverse screws to those locked with reconstruction screws. MATERIALS AND METHODS Experimental Design In the first experiment, synthetic femurs (Sawbones, Pacific Labs, Seattle WA) were tested nondestructively in torsion and axial loading. The specimens were first tested intact, and then after fixation of a simulated unstable proximal femoral fracture. The independent variable of interest was fixation device (nails using transverse screws or reconstruction screws for proximal fixation); the dependent variables measured were stiffness, deformation, energy to peak load, and energy loss in cyclic loading. In the second experiment, the synthetic femurs from the first experiment, and as paired human anatomic specimen femurs with simulated unstable proximal fracture, stabilized with either transverse or reconstruction screw fixation, were loaded to failure in axial compression. The independent variables of interest were specimen type (synthetic, fresh, or embalmed bone) and fixation method (transverse screws or reconstruction screws), and the dependent variables were stiffness, yield displacement, yield load, yield energy, and mode of failure. Modified Transverse Locking Nail Fixation 83 The human anatomic specimen bones used in the second part of the study included two pairs of fresh femurs and five pairs of embalmed femurs. Age of the donors was unknown. All human specimens were radiographed before testing to rule out any obvious pathology. In addition, the specimens underwent computed tomography (CT) scanning at various levels to measure density and describe morphology. Density was recorded from two levels (cephalad and central) in the femoral head, from two intracortical loci on the lateral cortex of the femoral neck, and from one intracortical locus on the medial cortex of the femoral neck. Morphologic measurements included femoral neck diameter at the subcapital level, inner and outer cortical diameters at the subtrochanteric level, offset of the femoral head from the medullary canal, and total cross sectional area at the head, neck, and subtrochanteric levels. Specimen Preparation Distally, femurs were potted in polymethylmethacrylate to the level of the epicondyles, with several embedded screws to prevent rotation. Proximally, the caudad 1/2 of the femoral head was similarly embedded in polymethylmethacrylate in a fashion that allowed loading through the femoral head (Fig 1). The anatomic specimen femurs were stripped of all soft tissues and the distal femoral condyles were removed parallel to the transverse plane. At this point, mounting in polymethylmethacrylate was done identically to the synthetic femurs. After mechanical testing of the intact bones (synthetic and anatomic specimen), an unstable proximal femoral fracture was simulated by reSpecimens moving a 3-cm segment of bone. Transverse saw cuts were made at 1 and 4 cm below the lesser The synthetic bones (n = 14) tested were second trochanter, and the intercalary segment was regeneration composite bones, 42 cm in length, moved. The lesser trochanter also was removed, consisting of a resin covered fiberglass cortex but the lateral cortex at this level was left intact to and a plastic foam filled medullary canal. The allow lateral cortical purchase for both types of stiffness of these bones is close to that of fresh fixation screws (Fig 1). human femurs, and the specimen to specimen In the synthetic specimens, canals were variation is much less (coefficient of variation reamed to 14-mm diameter for implantation of <7.3%). However, the composite bones are cona 13-mm x 380-mm nail. In the human siderably more elastic than fresh human bones, anatomic specimens, canals were reamed 1 with only approximately 1/5 the energy loss per loading cycle in axial compressive loading2L.22 mm larger than the size at which the reamer began cutting the inner endosteal cortex (chat(nonpublished data, Ziran BH, Sharkey NA, ter). With this technique, anatomic specimens Smith TS, Wang G , Chapman M W Comparisons all received 13-mm nails whose lengths were between synthetic and cadaveric bone specimens variable depending on the length of the specimen. in biomechanical testing 1995). 84 Clinical Orthopaedics arid Related Research Ziran et al by the same trauma surgeon (BHZ) who had clinical experience using both fixation systems. Mechanical Testing Fig 1. Osteotomized femur in loading apparatus for axial testing. Note transverse screws placed cephalad to lesser trochanteric level. In specimens fixed with reconstruction screws, the upper fragment was reamed to 17 mm diameter to accommodate the larger proximal segment of the nail. Transverse locking screws were placed so that the most proximal screw exited the medial femoral cortex in the subcapital area and the more distal of the proximal screws exited above the lesser trochanter (Fig 1). The reconstruction screws were placed into the femoral neck and head without penetration of the femoral head. Distally, all nails were statically locked with two screws. Reaming and nailing were done in the laboratory Axial testing was done using an Instron 1122 testing machine (Instron Corp, Canton, MA) equipped with a 5-kN axial load cell. The testing device was interfaced to a personal computer using Asyst scientific software (Asyst Software Technologies, Inc, Rochester, NY) and an analog to digital measurement and control system (Series 500, Keithley Instruments, Inc, Cleveland, OH). Specimens were mounted on a turnstile and loaded along the mechanical axis, with the femoral head centered over the condyles. Simulation of muscle loading was not done.13 For nondestructive testing, specimens were loaded cyclically in compression at 10 mm per minute top peak loads of 1000 N, and loads were recorded at 0.05-mm increments of displacement. Each femur was subjected to five conditioning loads followed by five loading cycles from which data were collected and averaged. Destructive axial testing was done with a ramp load to failure after several low conditioning cycles to 500 N. During testing in torsional loading (synthetic specimens only), the femoral head was mounted by means of an adjustable fixture to a 250-Nm torsional load cell. The base of the femur was mounted to an xy sliding table. The fixture was adjusted so that the axis of rotation was collinear to the axis of the femoral nail (Fig 2). A torsional stepper motor (Model S83-93, Compumotor, Rohnert Park, CA) in series with the load cell was mounted to the Instron crosshead. The specimens were loaded cyclically at a rate of 0.05 rad per second up to 10 Nm of torque. Loading was alternately clockwise and anticlockwise from the neutral position. After five conditioning cycles, data were collected and averaged from five loading cycles (Fig 2). Analysis of Findings The structural stiffness in the nondestructive tests was calculated as the slope of the last six digitized data points before achieving peak load and using elastic portion of the load displacement curve for destructive tests. Energy to peak load was calculated as the area under the load versus displacement curve. Energy loss per cycle was the percentage difference in area under the loading and unloading load versus displacement Number 339 June, 1997 ModifiedTransverse Locking Nail Fixation 85 lyzed using a paired t test. The effects of bone density and morphologic variable measured from the CT scans in relation to mechanical properties were evaluated using a stepwise linear regression. RESULTS Fig 2. Osteotomized specimen mounted for torsional testing. Base mounted on an x y table to allow centering and to reduce extraneous loading. curves. The 95% secant method was used to determine yield load in destructive tests (the intersection of the load versus displacement curve with a line whose slope is 95% of the elastic portion of the load versus displacement curve). A nonpaired test was used for the nondestructive tests using the synthetic femurs. For the destructive axial tests, a two-way analysis of variance (ANOVA) comparing specimen type synthetic, fresh anatomic specimen, embalmed anatomic specimen) and fixation method (transverse screws or reconstruction screws) was done, using a Tukey followup test. Side to side differences in the anatomic specimen groups were ana- In the nondestructive axial loading tests, there were no significant differences in stiffness, displacement, energy to peak load, or energy loss between constructs stabilized with transverse screw or reconstruction screw methods. However, both of these constructs were considerably less stiff and less elastic (greater energy loss) than the intact synthetic femurs (Fig 3). Similarly, in torsional testing the two constructs had very similar mechanical values, but both were 30% to 40% as stiff and less elastic than the intact femurs (Table 1, Fig 4). In the axial tests to failure, ANOVA indicated that there was a significant effect of specimen type (synthetic bone versus anatomic specimen bone) for yield displacement (p < 0.02) and stiffness (p < 0.01). There were no statistically significant differences in any parameter between fixation devices (transverse screw or reconstruction screw) in the anatomic bone specimens or the synthetic bone specimens (Table 2, Figs. 5 6 ) . Gross examination of the failed specimens revealed several differences in failure mechanism between specimen types and fixation devices. In the synthetic specimens, transverse screw specimen constructs failed by bending of the nail at the more distal of the two proximal screw holes (Fig 7); frequently the screw in this hole bent as well. In synthetic specimens with reconstruction screw fixation, failure occurred by bending of the nail at the level where it increases diameter proximally (Fig 8). The most proximal screw often bent, but did not cut out. There were no femoral neck fractures in the synthetic specimens. In human anatomic specimen bone, failure occurred by a combination of bone failure and bending of the fixation device. Typically, bone failure would initiate at a screw hole and progress to a basicervical fracture of the femoral neck. In transverse screws constructs, 86 Clinical Orthopaedics and Related Research Ziran et al TABLE 1. Axial and Torsional NondestructiveTesting Data Test Axial (n = 14) Stiffness (N/mm) Energy to peak load (N-m) Axial displacement (rnm) Energy loss (%) Torsional (n = 14) Stiffness (N/rnm) Energy to peak load (N-rn) Displacement (") Energy loss (%) Intact Reconstruction Screws (n = 7) Transverse Screws (n = 7) 1369 * 100 0.38 .03 0.8 0.1 1.7 533 +. 70 1.04 0.15 2.3 0.4 14.2 * * 532 ~t39 1.05 0.12 2.4 i 0.4 15.1 108 + 10 0.50 & .02 5.4 * 2.0 20.4 126 i 23 0.46 .05 4.4 1.9 20.1 * * 323 0.17 2.2 8.3 * 22 * .01 * .01 * * * There were no statistically significant differences between implant types vide geometric and densitometric inputs for a stepwise linear regression analysis. No associations were found between the geometric or densitometric data and the mechanical behavior of the construct under axial load. Side to side variances and subject to subject variances were not statistically significant. the fracture started in the medial cortex of the femoral neck at the site of the more caudad proximal interlocking screws (Fig 9). In reconstruction screw constructs, the fracture started on the lateral side at the site of the more caudad locking screw, but ultimate failure involved the medial cortex of the neck (Fig 10). In all cases, progression to failure occurred with bending of the implant and fracture of the femoral neck. No differences were observed between fresh and embalmed specimens and no side to side differences in failure were seen. Computed tomographic data of the intact anatomic specimen bones were used to pro- TABLE 2. Anatomic specimen bones (n = 14) Stiffness (N/mm)* Yield load (N) Yield displacement (mm)** Yield energy (N-m) Synthetic bones (n = 14) Stiffness (N/mm)* Yield load (N) Yield displacement (mm)** Yield energy (N-rn) **D < 0.02 The authors found no differences in the mechanical properties of unstable proximal femoral fractures fixed with either transverse or reconstruction locked intramedullary nails. Axial Test to Failure Test *p < 0 01 DISCUSSION Reconstruction Transverse Screws (n = 7) Screws (n = 7) 163 * 44 2705 813 19+7 32 5 21 149 + 57 2763 f 558 23 + 10 36 20 533 * 70 3165 f 930 13 * 4 22 * 11 532 39 2672 796 12+5 17 + 10 * * * * Number 339 June, 1997 ModifiedTransverse Locking Nail Fixation 87 IS- I4 - 12 - 16 10- 86- 4- Fig 3. Axial nondestructive test. No statistically significant differences between implant types. NS = not significant. Reeonsrrvctlon Screw NS 2- 0 Tnnsrcne Screw 0 Reconstruction Screw 0- 3s - 302s- mIS - 10 5- Fig 4. Torsional nondestructive tests. No statistically significant differences between implant types. NS = not significant. 0- NS I Transverse Screw Fig 5. Synthetic axial test to failure. No differences between implant types. NS = not significant. 88 Clinical Orthopaedics and Related Research Ziran et al Reconstruction Screw Transverse Screw YirM Load (N x i 0 2 ) Yield Displacrmenl (mml Sliffnrw (Nlorm I 10) Esrrgy l o Yidd (N-mn) Fig 6. Anatomic specimen axial test to failure. No differences implant types. NS = not significant. To some extent this reflects the fact that in this segmental defect model, much of what is being measured is the mechanical properties of a 13-mm diameter ‘Ti rod, which was the same in all specimen types. The findings further show that the proximal transverse crosslocking screws provide torsional and axial stability similar to that provided by the cephalomedullary locking screws of a reconstruction nail. Although the reconstruction and transverse locked femoral nails were mechanically equivalent in the simulation of an acute, unstable proximal femoral fracture, it is important to note that these constructs provided only 40% of the axial stiffness and 30% of the torsional stiffness of the intact bone.558JOJ3 Therefore, in the actual clinical setting, patients with subtrochanteric fractures stabilized with either of these methods may still require significant weightbearing precautions postoperatively until the fracture has healed Fig 7. Failed transverse locking nail (left) versus intact reconstruction nail (right). Fig 8. Failed reconstruction nail (left) versus intact reconstruction nail (right). Number 339 June, 1997 ModifiedTransverse Locking Nail Fixation 89 Fig 9. Failed anatomic specimen with transverse locked nail. Note basicervical fracture at screw hole (arrow). sufficiently to share the physiologic loads and avoid fatigue failure. The two devices and specimen types also differed in the mechanism of failure. Synthetic specimens (designed as a fiberglass mesh encased in the cortical material) did not show the catastrophicfailure and fracture seen in human bone; failure in synthetic specimens occurred exclusively in the implants. Both implants consistently bent at a predictable site of stress concentration, either at a change of diameter (reconstruction screws) or a screw hole (transverse screws), with occasional bends in the fixation screws.Although anatomic specimens ultimately fractured, there was a variable region of plastic deformation in the load deformation curve. Anatomic specimens fractured at sites of stress concentration in the bone (screw holes) and bent at stress concentrations in the implants (screw holes and diameter changes). Both implant constructs in human anatomic specimen bone ultimately failed through bone in the region of the femoral neck. In transverse locked specimens, the fracture seemed to be related to the medial screw holes, whereas in reconstruction locked specimens, fractures seemed to initiate laterally, al- Fig 10. Failed anatomic specimen with cephalomedullary locked nail. Note fracture of medial cortex and crack at lateral cortex. lowing the medial bone cortex to be forced against the medial wall of the nail (varus deformity) before failure of the femoral neck. Implants in anatomic specimens were not as bent as those in synthetic specimens, indicating a greater (and probably more realistic) contribution to energy absorbed in failure by the human anatomic specimen bone. Few studies have addressed the differences in proximal fixation of intramedullary femoral nails. Kinast et al9JO found that shorter screws in transverse locking systems outperformed (stiffer and stronger) the longer screws in oblique locking systems. The authors did not assess reconstruction systems. Johnson et al* described a loss of mechanical performance with increasing obliquity of proximal screw fixation in standard intmmedullary nail systems. The present study did not have such effects between reconstruction and transverse locking systems, possibly because the effects the reconstruction screws’ more oblique orientation were offset by their increased diameter. 90 Ziran et al The placement of transverse crosslocking screws used in this study requires drilling a hole through the medial cortex of the femoral neck, and there is concern that this defect may constitute a stress riser that increases the risk of femoral neck fracture. Edgarton et a15 found that if the diameter of cortical defects was smaller than 15% of the outer bone diameter, the strength of the bone was not compromised, thus suggesting that such defects may not be clinically important. Even though failure of anatomic specimen bones progressed through these screw holes, it was observed that failure loads of the femurs fixed with the transverse screws were similar to the failure loads of the femurs fixed with the reconstruction screws, which do not penetrate the medial cortex of the femoral neck. Femoral neck diameters measured from CT data were 4 to 4.5 cm, and the transverse screw holes were 4.2 mm, less than the 15% threshold in the study done by Edgarton et al. Furthermore, in the authors’ clinical experience using transverse crosslocked intramedullary nails to fix unstable proximal femoral fractures, femoral neck fractures have not been encountered as a c~mplication.~~ Swiontkowski et a120 found torsional, bending, and side to side density differences of 16.5%, 28.9%, and 12.3%, respectively, in paired anatomic specimen femurs, suggesting that using paired femurs to decrease variance in biologic samples may be only marginally useful.20Minimalside to side and subject to subject differences in mechanical properties of intact femurs were seen in the current study. The attempt to correlate morphologic properties of the proximal femur (geometry and density) with mechanical performance by regression analysis did not show any significant relationships and corroborates findings of previous studies.12.19 One interpretation of this finding would be that subtle differences in surgical technique may be of sufficient mechanical importance to overwhelm morphologic differences. For example, the authors speculate that slight imprecision in placement of the screw holes could result in Clinical Orthopaedics and Related Research preloads in the construct that would affect its mechanical behavior when extrinsic loads are applied. This also may be germane to the observation that the variance in mechanical properties of synthetic bones was much greater when they had been osteotomized and fixed than when they were intact. Analysis of variance indicated that the type of bone (anatomic specimen versus synthetic) was a significant factor in failure mechanics but that the type of implant was not. This observation concurs with other studies that show some of the advantages and limitations of the synthetic bones and the anatomic specimen bones21.22.24 (nonpublished data, Ziran BH, Sharkey NA, Smith TS, Wang G, Chapman MW: Comparisons between synthetic and cadaveric bone specimens in biomechanical testing. J Orthop Trauma [submitted]). Because the synthetic bones are more readily available, more uniform, and less expensive than anatomic specimen bone, their use in this type of comparative testing seems supported. However, differences in stiffness, displacement, and mode of failure between constructs using anatomic specimen versus synthetic bone indicates that the absolute value of measurements made using synthetic bone may not be directly applicable to biologic bone. Moreover, if fatigue testing is contemplated, which clinically would be the most relevant mode of failure testing, the synthetic bones would be inappropriate because their elasticity is greater than that of true bone, and their mechanism of failure is different. Clinically, the transverse locked intramedullary nails offer several advantages that the reconstruction systems do not offer. Use of the reconstruction nail requires imaging of the hip joint, which can be difficult in certain patients and in certain operative positions. Because the indication for using a reconstruction nail occurs relatively infrequently, many surgeons may be less familiar with their use, thus increasing the likelihood of intraoperative difficulty. Moreover, the reconstruction nail is usually twice the cost of Number 339 June, 1997 a standard intramedullary nail. The standard transverse locking femoral nail thus provides a familiar, reliable, and cost effective alternative to the reconstruction nail. The authors have shown that in an in vitro setting, a high seated transverse locking nail shows satisfactory mechanical performance as compared with a reconstruction nail. Furthermore, synthetic bone seems to provide an excellent material for comparative testing. The results of the present study should be interpreted with caution before the completion of clinical studies with long term followup. Acknowledgments The authors thank Michael Madison, PhD, for his editorial assistance. References 1. Asher MA, Tippett JW, Rockwood Jr CA, Zilber S: Compression fixation of subtrochanteric fractures. Clin Orthop 117:202-208,1976. 2. Bergman GD, Winquist RA, Mayo KA, Hansen Jr S T Subtrochanteric fracture of the femur. Fixation using the Zickel nail. J Bone Joint Surg 69A: 1032-1040,1987. 3. Bose WJ, Corces A, Anderson LD: A preliminary experience with the Russell-Taylor reconstruction nail for complex femoral fractures. J Trauma 32: 71-76, 1992. 4. Brien WW, Wiss DA, Becker V, Lehman T: Subtrochanteric femur fractures: A comparison of the Zickel nail, 95 degree blade plate, and interlocking nail. J Orthop Trauma 5:458-464, 1991. 5. Edgarton BC, An KA, Money BF: Torsional strength reduction due to cortical defects in bone. JOR 8:851-855,1990. 6. Fielding JW: Subtrochanteric fractures. Clin Orthop 92:86-99, 1973. 7. Heiple KG, Brooks DB, Samson BL, Burstein AH: A fluted intramedullary rod for subtrochanteric fractures. J Bone Joint Surg 61A:730-737, 1979. 8. Johnson KD, Tencer AF, Blumenthal S, August A, Johnston DWC: Biomechanicalperformance of locked intramedullary nail systems in comminuted femoral shaftfractures. Clin Orthop206:151-161,1986. 9. Kinast C, Bolhofner BR, Mast JW, Ganz R: Subtrochanteric fractures of the femur: Results of treatment with the 95" con&vlar blade-plate. Clin Orthop 238:122-130,1989. 10. Kinast C, Frigg R, Perren S: Biomechanics of the interlocking nail. A study of the proximal interlock. Arch Orthop Trauma Surg 109:197-204, 1990. Modified Transverse Locking Nail Fixation 91 11. Koch JC: The laws of bone architecture. Am J Anat 2 1: 177-298, 1917. 12. Leichter I, Margulies JY,Weinreb A, et al: The relationship between bone density, mineral content, and mechanical strength in the femoral neck. Clin Orthop 163:272-281,1982. 13. McLeish RD, Charnley J: Abduction forces in one legged stance. J Biomech 3:191-209,1970. 14. Oh I, Hams WH: Proximal strain distribution in the loaded femur. An in vitro comparison of the distributions in the intact femur and after insertion of different hip-replacement femoral components. J Bone Joint Surg 60A:75-85, 1978. 15. Rybicki EF, Simonen FA, Weis Jr EB: On the mathematical analysis of stress in the human femur. J Biomech 5:203-215,1972. 16. Sanders R, Regazzoni P: Treatment of subtrochanteric femur fractures using the dynamic condylar screw. J OrthopTrauma 3:206-213, 1989. 17. Sensheimer F Subtrochanteric fractures of the femur. J Bone Joint Surg 60A:300-306, 1978. 18. Smith JT, Goodman SB, Tischenko G: Treatment of comminuted subtrochanteric fractures using the Russell-Taylor reconstruction IM nail. Orthopedics 14:125-129, 1991. 19. Smith MD, Cody DD, Goldstein SA, et al: Proximal femoral bone density and its correlation to fracture load and hip-screw penetration load. Clin Orthop 283:2&25 1, 1992. 20. Swiontkowski MF, Harrington RM, Keller TS, VanPatten PK: Torsion and bending analysis of internal fixation techniques for femoral neck fractures: The role of implant design and bone density. J Orthop Res 5:43344,1987. 21. Szivek JA, Gealer RL: Comparison of the deformation response of synthetic and cadaveric femora during simulated one legged stance. J Appl Biomater 2:277-280, 1991. 22. Szivek JA, Weng M, Karpman R: Variability in the torsional and bending response of a commercially available composite femur. J Appl Biomater 11183-186, 1990. 23. Tencer AF, Johnson KD, Johnston DWC, Gill K: A biomechanical comparison of different methods of stabilization of subtrochanteric fractures of the femur. J Orthop Res 2:297-305,1984. 24. Uta S: Development of synthetic bone models for the evaluation of fracture fixation devices. Nippon Seikeigeka Gakkai Zasshi. J Jpn Orthop Assoc 66: 1156-1 164, 1992. 25. Wiss DA, Brien W W Subtrochanteric fractures of the femur: Results of treatment by interlocking nailing. Clin Orthop 283:231-236,1992. 26. Zickel RE: An intramedullary fixation device for the proximal part of the femur. Nine years' experience. J Bone Joint Surg 58A:866-872, 1976. 27. Ziran BH, Wasan AD, Chapman MW: Complications of intramedullary nails with transverse locking screws through the femoral neck. Trans Orthop Trauma Assoc 1996.