- Email: [email protected]
The effect of shoulder arthroplasty on humeral strength: An in vitro biomechanical investigation
Clinical Biomechanics 20 (2005) 1064–1071 www.elsevier.com/locate/clinbiomech
The effect of shoulder arthroplasty on humeral strength: An in vitro biomechanical investigation Anthony M.T. Choo a
a,b
, Robert H. Hawkins a, Brian K. Kwon a, Thomas R. Oxland
a,b,*
Department of Orthopaedics, The University of British Columbia, 910 West 10th Avenue, Vancouver, BC, Canada V5Z 4E3 b Department of Mechanical Engineering, The University of British Columbia, Vancouver, BC, Canada V5Z 4E3 Received 29 May 2004; accepted 24 June 2005
Abstract Background. Periprosthetic humeral fractures are a serious complication of shoulder arthroplasty. While adequate reaming of the canal and insertion of an oversized implant optimizes fit, such maneuvers also weaken the bone and predispose it to fracture. Methods. The impact of the humeral arthroplasty was assessed in vitro on human cadaveric specimens. Strain gauges were attached to the distal diaphyses and the specimens were mounted in a torsion-loading fixture throughout the tests. An initial series examined the effect of reaming of the canal to its clinically appropriate diameter using uniaxial strain gauges. A second series utilized strain rosettes to evaluate the cumulative effects of reaming, broaching, and implant insertion. Findings. Reaming of the canal to its clinically appropriate diameter significantly increased (P = 0.007) uniaxial strain measurements by a mean of 30% with five of eight specimens showing increases of over 49% on at least one of four diaphyseal locations. In the second series, the surface strain was significantly affected by arthroplasty (P < 0.008). Post-hoc analysis showed that the maximum in-plane shear strain following implant insertion was significantly increased relative to strain levels following reaming and broaching (P < 0.009). The direction of the principal strain axes did not significantly change (P > 0.46). Unexpected decreases in some strain measurements were observed as the arthroplasty procedure progressed perhaps reflecting overt mechanical failure within the humeral shaft. Interpretation. The strain increase following reaming suggests a reduction in torsional strength by over 33% which is further reduced following broaching and implant insertion. For the practicing surgeon, post-operative strength can be adversely affected by both canal preparation and implant insertion. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Arthroplasty; Reaming; Strength; Strain
1. Introduction Periprosthetic humeral fractures after shoulder arthroplasty have a reported incidence rate of 1–3% (Cameron and Iannotti, 1999; Schmidt, 1999). Although low, this accounts for roughly 20% of all complications related to shoulder arthroplasty (Cameron and Iannotti, * Corresponding author. Address: Department of Orthopaedics, The University of British Columbia, 910 West 10th Avenue, Vancouver, BC, Canada V5Z 4E3. E-mail address: [email protected] (T.R. Oxland).
0268-0033/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2005.06.013
1999). There are several patient-related factors that predispose to such fractures, with osteopenia related to age or rheumatoid arthritis being undoubtedly the most important (Cameron and Iannotti, 1999; Campbell et al., 1998). While late periprosthetic fractures are seen, the majority occur intraoperatively, and the commonly observed spiral or long oblique fracture pattern suggests that torsional forces dominate the fracture mechanism (Bonutti and Hawkins, 1992; Schmidt, 1999; Wirth and Rockwood, 1994). Although poor bone stock may be inevitable, it is recognized that technical aspects of the surgical procedure such as the reaming, broaching,
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
and implant insertion certainly play an important role in increasing fracture risk. As in hip arthroplasty, aggressive and/or eccentric reaming can result in endosteal notching or cortical perforation, both of which have been reported to be associated with intraoperative periprosthetic fractures (Bonutti and Hawkins, 1992; Campbell et al., 1998). Vigorous impaction of the broach and/ or the forceful insertion of an oversized implant may also subject the bone to high hoop stresses and cause or predispose it to fracture (Wirth and Rockwood, 1994). Several authors have studied the torsional properties of cortical bone when it had been machined to standard geometric cross-sections (Bonfield and Grynpas, 1982; Jepsen and Davy, 1997; Jepsen et al., 1999; Lakes et al., 1990). However, it is difficult to directly extend these results to the geometrically more complex intact bone (Martens et al., 1980, 1981). From a mechanical perspective, endosteal material removal during reaming reduces the cross-sectional area, thereby reducing the specimenÕs polar moment of inertia (J) and hence reduces its strength. Broaching only removes material proximally within the metaphysis and therefore conceptually has less effect on the diaphysis. Subsequent implant insertion introduces hoop strains that further increase fracture risk. Ideally, these changes are modest and have negligible effect on shear strength in torsion where the long bone is typically weakest (Reilly and Burstein, 1975). While technical aspects of the shoulder arthroplasty have been implicated in the generation of periprosthetic fractures, the contributions of reaming, broaching, and implant insertion to the biomechanical integrity of the humerus have not been previously reported. The overall purpose of this study was to determine how these maneuvers affected the torsion strength of the humerus, with the hope that such biomechanical data could assist the surgeon in preventing intraoperative periprosthetic fractures. The objectives of the study were divided into two parts. Part 1 examined whether reaming to a diameter that was deemed appropriate for the particular humeral specimen (i.e. a ‘‘clinically relevant’’ diameter) significantly reduced the torsion strength of the bone. If possible the specimen was reamed beyond the clinical diameter for an assessment of over-reaming. Part 2 determined if any one stage of the arthroplasty procedure—reaming, broaching, or implant insertion—had a particularly detrimental effect.
2. Methods 2.1. Part 1: choice of clinical diameter Eight fresh frozen cadaveric humeral specimens (7 male, 1 female, age 25–84 years) were thawed to room
1065
temperature and osteotomized at the surgical neck to expose the very proximal aspect of the intramedullary canal. While we recognize that the humeral head is typically osteotomized through the anatomic neck for the shoulder arthroplasty procedure, our exposure allowed for precise positioning of the reamer so as to avoid eccentric reaming. A commercial shoulder arthroplasty system (Global Total Shoulder System, DePuy, Warsaw, IN, USA) was used. This system contained rigid hand reamers with diameters increasing from 6 to 16 mm in 2-mm increments. Under the ideal assumption that reaming only affects cross-sectional area, a significant reduction in strength should be accompanied by an increase in diaphyseal strain when the specimen is loaded. Four uniaxial strain gauges (PA06-125AA-350-1 EN, JP Technologies, San Bernardino, CA, USA) were mounted at 45° to the longitudinal axis on the lateral, anterior, medial and posterior (L, A, M, P) aspects of each specimen corresponding to the direction of principal strain for an idealized cylindrical specimen. The gauges were mounted one centimeter proximal to the maximum anticipated reamer depth to account for insertion variability and help ensure measurements were made within the strain field of the advancing reamer tip. All specimens were radiographed in anterior–posterior (AP) and medial–lateral (ML) directions for templating and cortical thickness measurements. The specimens were mounted in a fixture and dead weights were used to apply a continuous torque of 14 N m in external rotation (Fig. 1). Preliminary trials indicated this torque produced approximately 500 microstrain on the diaphysis similar to physiological values reported in the literature (Baggott and Lanyon, 1977; Carter et al., 1980; Cochran, 1972). The overall repeatability of this initial strain level was 5.3% determined from two initial load applications for each specimen. The strain gauge signals were conditioned (Vishay 2120, Measurements Group, Raleigh, NC, USA) and continuously acquired at 25 Hz throughout the entire reaming procedure. Starting with the smallest reamer, specimens were sequentially reamed until the surgeon reached the point at which he would stop intraoperatively, based on the feel and resistance to reaming. The diameter of the reamer at this final reaming was defined as the clinical diameter. The diaphyseal strain response to the second-to-last reaming (2 mm smaller) was used for comparison. If the clinical diameter for a specimen was less than 16 mm (i.e. wider reamers were still available), the specimen was then over-reamed for qualitative data analysis. The uniaxial strain data (e) were normalized to the strain following the application of the 14 N m torque prior to any reaming. The surgeon was blinded to the strain recordings throughout the experiment. Two-way repeated measures ANOVA with Student– Newman–Keuls post-hoc analysis was used to assess
1066
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
Fig. 1. Experimental setup. The humeri were mounted on a torsion fixture and a 14 N m torque was applied with weights. In part 1, uniaxial strain gauges (inset) were mounted on the diaphysis at 45° to the longitudinal axis. In part 2, three-element strain rosettes (inset) were used to measure the in-plane strain.
the reaming decision (second-to-last vs. clinical diameter) and measurement location (L, A, M, P). For two specimens, one of the four strain gauges was damaged during the test and mean substitution was used for the missing data (Fisher and van Belle, 1993). Significance was set at a = 0.05. 2.2. Part 2: comparison of arthroplasty stages Power analysis based on part 1 results used a withinspecimen strain variation of 23% with a root mean square difference of 35% yielding an effect size of d = 1.5 which was considered medium (Murphy and Myors, 1998). For the primary question of interest, the effect of arthroplasty stage on humeral strain, a power of 0.8 at a significance level of 0.05 was achieved with n = 10 (Norman and Streiner, 2000; Zar, 1999). With this number of specimens, the secondary effect of gauge location (found not significant in part 1) had a power of 0.55. Ten fresh frozen cadaveric humeral specimens were tested (8 male, 2 female, age 80–90 years for six specimens with no age data for four). Whereas part 1 measured uniaxial deformations, here strain rosettes (FRA-2-11, Tokyo Sokki Kenkyujo Co., Tokyo, Japan) were used to account for more complex loads introduced by broaching and implant insertion. Each rosette consisted of three uniaxial gauges from which the twodimensional surface strain was calculated. Two rosettes were mounted on the lateral and posterior aspects of the humeral diaphysis in positions that corresponded to the regions where minimum and maximum changes were observed in part 1. The specimen radiographs, preparation, loading, and data acquisition were otherwise the same as in part 1. The humeral arthroplasty procedure was reproduced. The specimens were reamed to the clinical diameter, broached, and a 1 mm oversized implant was inserted
in keeping with the manufacturerÕs protocol. The data from the rosettes were converted to principal strains using standard strain transformation relationships which were derived independent of material properties (Beer and Johnston, 1992; Carter, 1978; Lanyon et al., 1975). The maximum in-plane shearing strain (difference of principal strains), c, was analyzed along with the direction of the principal strain axes (h) relative to the transverse bone axis. Note, strains were analyzed rather than stresses since conversion to the latter requires the additional approximation of material properties (Carter, 1978). As in part 1, strain data were normalized to initial strains for analysis. Two-way repeated measures multivariate ANOVA with Student–Newman–Keuls post-hoc analysis was used to assess the effects of arthroplasty stage (reaming, broaching, implant) and measurement location (L, P) on c and h. One laterally placed rosette was damaged during the test and mean substitution was used for the missing data (Fisher and van Belle, 1993). Significance was set at a = 0.05.
3. Results 3.1. Part 1: clinical diameter decision Clinical diameter reaming showed large transient strain profiles followed by a significant (P = 0.007) elevation of equilibrium strain levels when compared to levels following the second-to-last reaming (Fig. 2). The mean clinical diameter was 14.4 mm (SD 1.9). Two specimens (clinical diameter = 16 mm) exhibited a strain increase of 50% or more while three specimens (clinical diameter = 14 mm) showed an increase greater than 60% on at least one gauge. Averaging all gauges, the overall increase following the second-to-last reaming was 11% and this increased to 30% after clinical
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
1067
3000
2500
P
microstrain
2000
1500
A 1000
L
500
M 0 Over-ream
SL CD -500 50
100
150
200 Time (s)
250
300
350
Fig. 2. Uniaxial strain patterns during reaming. Traces show measurements from gauges mounted on the lateral (L), anterior (A), medial (M) and posterior (P) humeral diaphysis. Vertical boundary lines indicate the start of the second-to-last (SL) reaming, clinical diameter (CD) reaming and over-reaming (qualitative only). Strain levels moderately increased following second-to-last reaming and further increased after clinical diameter reaming. A mixed pattern of increasing and decreasing strains was observed following over-reaming.
diameter reaming. Five specimens were reamed beyond the clinical diameter and three of these fractured. Following over-reaming, a mixed pattern of increasing and decreasing strains was observed on all five specimens. There was no significant difference in strain increase between gauge locations (P = 0.22), however, there was a significant interaction between strain measurement
location and clinical diameter decision (P = 0.032) where larger increases were observed on the AP than ML diaphyses (Fig. 3). The mean change following clinical diameter reaming for combined anterior and posterior data was 46% while the mean change for medial and lateral data was 15%. The mean anterior-to-posterior humeral and canal diameters were 21.5 mm (SD 2.6)
Second to Last
160
Clinical Diameter
Percent change in strain
140 120 100 Mean+SD Mean-SD Mean+SE Mean-SE Mean Outliers
80 60 40 20 0 -20
L
A
M
P
L
A
M
P
Strain gauge Location Fig. 3. Clinical diameter decision. Change in strain levels after the second-to-last reaming compared to levels after reaming at the clinical diameter on the lateral, anterior, medial, and posterior (L, A, M, P) faces of the humeral diaphysis. Clinical diameter reaming caused a significant change in diaphyseal strain (P = 0.007) and there was a significant interaction with gauge location (P = 0.032) with larger increases observed on the narrower AP diaphyses.
1068
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
and 14.1 mm (SD 2.8), respectively. The mean medialto-lateral humeral and canal diameters were larger (23.6 mm SD 2.4 and 16.5 mm SD 2.0). 3.2. Part 2: comparison of arthroplasty stages The equilibrium shear strains following each stage of the arthroplasty procedure were observed to consistently increase in four specimens (Fig. 4a) while a mixed pattern was observed in six specimens (Fig. 4b). The
mean clinical diameter was 12.2 mm (SD 1.8). One specimen failed during reaming at its clinical diameter of 14 mm and hence was excluded from the ANOVA. The three arthroplasty maneuvers had a significant effect on the maximum in-plane shear strain (P = 0.008). The shear strain following implant insertion was significantly higher than after reaming (P = 0.005) and broaching (P = 0.009). The changes in shearing strains after broaching were not significantly different from those observed after reaming (P = 0.450) although
L shear
1500
P shear 1000
microstrain
L max P max 500
0
P min
-500
L min CD=10mm -1000 0 (a)
50
100
Broach 150
200 Time (s)
Implant
250
300
350
400
1500
P shear L shear
1000
microstrain
P max 500
L max
0
P min L min
-500
CD=10mm -1000 0 (b)
50
100
Broach 150 Time (s)
200
Implant 250
300
Fig. 4. Strain patterns during arthroplasty. Traces show in-plane surface strains from rosettes mounted to the lateral (L) and posterior (P) diaphysis. Max and min denote principal strains where positive denotes tension and negative indicates compressive strains. The difference of the principal strains gives the maximum shear strain. Some humeri exhibited progressive increase in strain (a), while others exhibited mixed patterns of increasing and decreasing strain (b).
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071 Posterior
Lateral
100
1069
Percent change in shear strain
80 60 40 20
Mean+SD Mean-SD Mean+SE Mean-SE Mean Outliers
0 -20 -40 -60
Ream
Broach
Implant
Ream
Broach
Implant
Arthroplasty stage Fig. 5. Comparison of arthroplasty stages. Changes in the maximum in-plane shear strain on the lateral and posterior diaphyses were significantly altered by arthroplasty (P = 0.008). Implant insertion significantly increased shear strain compared to reaming (P = 0.005) and broaching (P = 0.009).
the variability was observed to increase on the lateral diaphysis. Again, the effect of gauge location was not significant (P = 0.224). In contrast to part 1, no significant (P = 0.080) interaction between location and arthroplasty stage was observed (Fig. 5) and the AP and ML specimen dimensions were more symmetric. AP mean humeral and canal diameters were measured at 23.0 mm (SD 4.0) and 13.2 mm (SD 4.2). Mean ML diameters were 23.0 mm (SD 2.1) and 13.1 mm (SD 3.9). The presence of mixed strain patterns increased the range of changes observed. Following reaming, the changes in shear strains ranged from 41.1% to 23.6%. Changes ranged from 33% to 65% after broaching and 29% to 51% with the implant inserted. Arthroplasty had no significant effect on the angle of the principal strain axes (P = 0.808). Following the second-to-last reaming, the mean angles of the axes were 46.2° (SD 8.5) and 44.3° (SD 7.3) relative to the transverse humeral axis on the lateral and posterior gauge locations, respectively. After the accumulated effects of the procedure, the mean angles were 41.1° (SD 5.0, lateral) and 44.3° (SD 8.1, posterior) with the implant inserted.
4. Discussion This study aimed to understand the potential reduction in torsional humeral strength during shoulder arthroplasty. We focused on torsion as this appears to be a dominant mechanism for intraoperative fractures (Bonutti and Hawkins, 1992; Schmidt, 1999; Wirth
and Rockwood, 1994). Ideally, strength would be assessed with a torque-to-failure test following a particular stage of the procedure (Jonsson and Stromberg, 1985; Martens et al., 1980; Netz et al., 1980; Pratt et al., 1987). However, a large number of specimens would have been needed to compare the four steps from the second-to-last reaming to implant insertion. Instead, this study utilized continuous monitoring of the changing strain under a static load to indicate strength reductions. Using this repeated measures design, a significant strain increase suggests a significant reduction in the safe margin prior to reaching the fracture threshold. Asymmetric geometry and material anisotropy complicates the determination of strength, however, a coarse estimate may be obtained by considering a two-dimensional cylindrical cross-section with isotropic properties (Appendix A). With this simplification, the mean 30% increase observed in part 1 suggests a strength reduction of approximately 23%. For the narrower anterior-toposterior diameter (21.5 mm), increases in excess of 50% suggest strength reductions of over 33%. Using torque to failure tests, Pratt et al. (1987) found similar reductions (37.5%) in femoral strength following reaming to 12 mm. This simplification likely underestimates the strength reduction since the gauges measure local changes that may not represent the peak strains on the cortical surface. Mathematical models have shown irregular transverse geometry may result in stress variations of ±20% around the cross-section (Kennedy and Carter, 1985). In part 2, the peak changes on the posterior diaphysis following clinical diameter reaming were lower than those observed in part 1 but were comparable to changes
1070
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
measured when humeral diameters were similar. The occurrence of mixed patterns particularly during the latter stages of the procedure was unexpected and was inconsistent with the strength reduction model due strictly to altered cross-sectional area. Instead, the prevalence of these patterns during the over-reaming stage of part 1 suggests it may be an indicator of excessive bone removal. The damage mechanism remains unclear from strain data alone but may be due to interstitial crack formations causing reduced strains in the crackÕs wake (Elias et al., 2000). There were several limitations to this study. First, strain gauges provided only local measurements hence it was difficult to generalize the strain field to regions away from the gauges. Moreover, the part 1 uniaxial gauges were less descriptive than the rosettes used in part 2. Additional measures of strength and damage accumulation such as torsional stiffness (Martens et al., 1981) and crack density (Forwood and Parker, 1989; Jepsen et al., 1999; Netz et al., 1980) combined with finite element analysis (Elias et al., 2000; Huiskes, 1982; Keyak and Rossi, 2000) would provide valuable complementary information. Second, the analysis was based on the equilibrium strain measurements following each arthroplasty step and did not account for the transient strains during each maneuver. These dynamic strains may have been responsible for more severe damage mechanisms than were detectable with the comparison of between—stage strains. Third, the humeri were continuously loaded and this may have increased the damage caused during the experiment. One alternative would have been to load and unload the specimens between surgical steps. However, residual strain offsets following unloading would have impeded strain comparisons before and after each step (Carter et al., 1981; Fondrk et al., 1999). Fourth, the use of a frame that allowed dead-weight vibration introduced a dynamic torsion component particularly during broaching. Clinically, there would also likely be a dynamic torsion component during the broaching stage with the humerus held in external rotation. However, it is uncertain whether the magnitude of the dynamic torque in this study was comparable to that experienced during surgery. In conclusion, reaming the humeral canal until cortical bone contact has the potential to significantly reduce torsional strength, particularly in humeri with narrow outer diaphyses. The risk of intraoperative fracture could potentially be reduced by using finer reamer increments. In addition, cortical strain from implant insertion could be reduced with a reduction in stem over-size. This conceptually simple model of strength reduction due to shoulder arthroplasty quantitatively highlights some of the potential risks when planning and performing the procedure.
Hollow cylinder under torsion
τ
max
σ45o
τ
ri ro T
T
Fig. A.1. Cylinder under torsion. For a cylinder with isotropic material properties, the principal axes lie at 45° to the long axis.
Acknowledgments The technical assistance of Mr. Darrell Goertzen in part 1 and surgical assistance of Dr. David Blanchette in part 2 are gratefully acknowledged. The support of DePuy (Canada) Ltd. is gratefully acknowledged. Thanks also to Dr. D.P. Romilly for the loan of strain measurement equipment.
Appendix A. Approximate strength reduction The maximum shear stress, smax, in a hollow cylinder loaded in torsion is located at the outer circumference and is given by Tro J
smax ¼
where p J ¼ ðr4o r4i Þ 2 is the polar moment of inertia, T is the applied torque, ro is the outer radius, ri is the inner radius of the hollow cylinder (Fig. A.1). T can be increased until smax reaches the failure limit of the material, sfailure. Hence the maximum torsional load that the cylinder can carry is T max ¼
J sfailure ro
For a cylinder under pure torsion, the magnitude of the shear stress (smax) is equal to the stress at 45° (r45°). Since, r45 ð1 þ mÞ E for homogeneous YoungÕs modulus E and PoissonÕs ratio m. Reaming increases ri and hence reduces Tmax: e45 ¼
T max
original
T max
T max original
reamed
J original J reamed J original ereamed eoriginal ¼ ereamed ¼
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
References Baggott, D.G., Lanyon, L.E., 1977. An independent Ôpost-mortemÕ calibration of electrical resistance strain gauges bonded to bone surfaces Ôin vivoÕ. J. Biomech. 10, 615–622. Beer, F.B., Johnston, E.R., 1992. Mechanics of Materials. McGrawHill, New York. Bonfield, W., Grynpas, M.D., 1982. Spiral fracture of cortical bone. J. Biomech. 15, 555–559. Bonutti, P.M., Hawkins, R.J., 1992. Fracture of the humeral shaft associated with total replacement arthroplasty of the shoulder. A case report. J. Bone Joint Surg. Am. 74, 617–618. Cameron, B., Iannotti, J.P., 1999. Periprosthetic fractures of the humerus and scapula: management and prevention. Orthop. Clin. N Am. 30, 305–318. Campbell, J.T., Moore, R.S., Iannotti, J.P., Norris, T.R., Williams, G.R., 1998. Periprosthetic humeral fractures: mechanisms of fracture and treatment options. J. Shoulder Elbow Surg. 7, 406–413. Carter, D.R., 1978. Anisotropic analysis of strain rosette information from cortical bone. J. Biomech. 11, 199–202. Carter, D.R., Smith, D.J., Spengler, D.M., Daly, C.H., Frankel, V.H., 1980. Measurement and analysis of in vivo bone strains on the canine radius and ulna. J. Biomech. 13, 27–38. Carter, D.R., Caler, W.E., Spengler, D.M., Frankel, V.H., 1981. Fatigue behavior of adult cortical bone: the influence of mean strain and strain range. Acta Orthop. Scand. 52, 481–490. Cochran, G.V., 1972. Implantation of strain gages on bone in vivo. J. Biomech. 5, 119–123. Elias, J.J., Frassica, F.J., Chao, E.Y., 2000. The open section effect in a long bone with a longitudinal defect—a theoretical modeling study. J. Biomech. 33, 1517–1522. Fisher, L.D., van Belle, G., 1993. Biostatistics—A Methodology for the Health Sciences. John-Wiley & Sons Inc., New York. Fondrk, M.T., Bahniuk, E.H., Davy, D.T., 1999. Inelastic strain accumulation in cortical bone during rapid transient tensile loading. J. Biomech. Eng. 121, 616–621. Forwood, M.R., Parker, A.W., 1989. Microdamage in response to repetitive torsional loading in the rat tibia. Calcified Tissue Int. 45, 47–53. Huiskes, R., 1982. On the modelling of long bones in structural analyses. J. Biomech. 15, 65–69. Jepsen, K.J., Davy, D.T., 1997. Comparison of damage accumulation measures in human cortical bone. J. Biomech. 30, 891–894.
1071
Jepsen, K.J., Davy, D.T., Krzypow, D.J., 1999. The role of the lamellar interface during torsional yielding of human cortical bone. J. Biomech. 32, 303–310. Jonsson, U., Stromberg, L., 1985. Torsional tests of long bones with computerized equipment. J. Biomed. Eng. 7, 251–255. Kennedy, J.G., Carter, D.R., 1985. Long bone torsion: I. Effects of heterogeneity, anisotropy and geometric irregularity. J. Biomech. Eng. 107, 183–188. Keyak, J.H., Rossi, S.A., 2000. Prediction of femoral fracture load using finite element models: an examination of stress- and strainbased failure theories. J. Biomech. 33, 209–214. Lakes, R.S., Nakamura, S., Behiri, J.C., Bonfield, W., 1990. Fracture mechanics of bone with short cracks. J. Biomech. 23, 967–975. Lanyon, L.E., Hampson, W.G., Goodship, A.E., Shah, J.S., 1975. Bone deformation recorded in vivo from strain gauges attached to the human tibial shaft. Acta Orthop. Scand. 46, 256–268. Martens, M., van Audekercke, R., de Meester, P., Mulier, J.C., 1980. The mechanical characteristics of the long bones of the lower extremity in torsional loading. J. Biomech. 13, 667–676. Martens, M., van Audekercke, R., de Meester, P., Mulier, J.C., 1981. The geometrical properties of human femur and tibia and their importance for the mechanical behaviour of these bone structures. Arch Orthop. Trauma Surg. 98, 113–120. Murphy, K.R., Myors, B., 1998. Statistical Power Analysis—A Simple and General Model for Traditional and Modern Hypothesis Tests. Lawrence Erlbaum Associates, New Jersey. Netz, P., Eriksson, K., Stromberg, L., 1980. Material reaction of diaphyseal bone under torsion. An experimental study on dogs. Acta Orthop. Scand. 51, 223–229. Norman, G.R., Streiner, D.L., 2000. Biostatistics: The Bare Essentials. B.C. Decker Inc., Hamilton. Pratt, D.J., Papagiannopoulos, G., Rees, P.H., Quinnell, R., 1987. The effects of medullary reaming on the torsional strength of the femur. Injury 18, 177–179. Reilly, D.T., Burstein, A.H., 1975. The elastic and ultimate properties of compact bone tissue. J. Biomech. 8, 393–405. Schmidt, A.H., 1999. Periprosthetic fractures of the humerus. Tech. Orthop. 14, 151–156. Wirth, M.A., Rockwood Jr., C.A., 1994. Complications of shoulder arthroplasty. Clin. Orthop., 47–69. Zar, J.H., 1999. Biostatistical Analysis. Prentice Hall, New Jersey.
The effect of shoulder arthroplasty on humeral strength: An in vitro biomechanical investigation Anthony M.T. Choo a
a,b
, Robert H. Hawkins a, Brian K. Kwon a, Thomas R. Oxland
a,b,*
Department of Orthopaedics, The University of British Columbia, 910 West 10th Avenue, Vancouver, BC, Canada V5Z 4E3 b Department of Mechanical Engineering, The University of British Columbia, Vancouver, BC, Canada V5Z 4E3 Received 29 May 2004; accepted 24 June 2005
Abstract Background. Periprosthetic humeral fractures are a serious complication of shoulder arthroplasty. While adequate reaming of the canal and insertion of an oversized implant optimizes fit, such maneuvers also weaken the bone and predispose it to fracture. Methods. The impact of the humeral arthroplasty was assessed in vitro on human cadaveric specimens. Strain gauges were attached to the distal diaphyses and the specimens were mounted in a torsion-loading fixture throughout the tests. An initial series examined the effect of reaming of the canal to its clinically appropriate diameter using uniaxial strain gauges. A second series utilized strain rosettes to evaluate the cumulative effects of reaming, broaching, and implant insertion. Findings. Reaming of the canal to its clinically appropriate diameter significantly increased (P = 0.007) uniaxial strain measurements by a mean of 30% with five of eight specimens showing increases of over 49% on at least one of four diaphyseal locations. In the second series, the surface strain was significantly affected by arthroplasty (P < 0.008). Post-hoc analysis showed that the maximum in-plane shear strain following implant insertion was significantly increased relative to strain levels following reaming and broaching (P < 0.009). The direction of the principal strain axes did not significantly change (P > 0.46). Unexpected decreases in some strain measurements were observed as the arthroplasty procedure progressed perhaps reflecting overt mechanical failure within the humeral shaft. Interpretation. The strain increase following reaming suggests a reduction in torsional strength by over 33% which is further reduced following broaching and implant insertion. For the practicing surgeon, post-operative strength can be adversely affected by both canal preparation and implant insertion. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Arthroplasty; Reaming; Strength; Strain
1. Introduction Periprosthetic humeral fractures after shoulder arthroplasty have a reported incidence rate of 1–3% (Cameron and Iannotti, 1999; Schmidt, 1999). Although low, this accounts for roughly 20% of all complications related to shoulder arthroplasty (Cameron and Iannotti, * Corresponding author. Address: Department of Orthopaedics, The University of British Columbia, 910 West 10th Avenue, Vancouver, BC, Canada V5Z 4E3. E-mail address: [email protected] (T.R. Oxland).
0268-0033/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2005.06.013
1999). There are several patient-related factors that predispose to such fractures, with osteopenia related to age or rheumatoid arthritis being undoubtedly the most important (Cameron and Iannotti, 1999; Campbell et al., 1998). While late periprosthetic fractures are seen, the majority occur intraoperatively, and the commonly observed spiral or long oblique fracture pattern suggests that torsional forces dominate the fracture mechanism (Bonutti and Hawkins, 1992; Schmidt, 1999; Wirth and Rockwood, 1994). Although poor bone stock may be inevitable, it is recognized that technical aspects of the surgical procedure such as the reaming, broaching,
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
and implant insertion certainly play an important role in increasing fracture risk. As in hip arthroplasty, aggressive and/or eccentric reaming can result in endosteal notching or cortical perforation, both of which have been reported to be associated with intraoperative periprosthetic fractures (Bonutti and Hawkins, 1992; Campbell et al., 1998). Vigorous impaction of the broach and/ or the forceful insertion of an oversized implant may also subject the bone to high hoop stresses and cause or predispose it to fracture (Wirth and Rockwood, 1994). Several authors have studied the torsional properties of cortical bone when it had been machined to standard geometric cross-sections (Bonfield and Grynpas, 1982; Jepsen and Davy, 1997; Jepsen et al., 1999; Lakes et al., 1990). However, it is difficult to directly extend these results to the geometrically more complex intact bone (Martens et al., 1980, 1981). From a mechanical perspective, endosteal material removal during reaming reduces the cross-sectional area, thereby reducing the specimenÕs polar moment of inertia (J) and hence reduces its strength. Broaching only removes material proximally within the metaphysis and therefore conceptually has less effect on the diaphysis. Subsequent implant insertion introduces hoop strains that further increase fracture risk. Ideally, these changes are modest and have negligible effect on shear strength in torsion where the long bone is typically weakest (Reilly and Burstein, 1975). While technical aspects of the shoulder arthroplasty have been implicated in the generation of periprosthetic fractures, the contributions of reaming, broaching, and implant insertion to the biomechanical integrity of the humerus have not been previously reported. The overall purpose of this study was to determine how these maneuvers affected the torsion strength of the humerus, with the hope that such biomechanical data could assist the surgeon in preventing intraoperative periprosthetic fractures. The objectives of the study were divided into two parts. Part 1 examined whether reaming to a diameter that was deemed appropriate for the particular humeral specimen (i.e. a ‘‘clinically relevant’’ diameter) significantly reduced the torsion strength of the bone. If possible the specimen was reamed beyond the clinical diameter for an assessment of over-reaming. Part 2 determined if any one stage of the arthroplasty procedure—reaming, broaching, or implant insertion—had a particularly detrimental effect.
2. Methods 2.1. Part 1: choice of clinical diameter Eight fresh frozen cadaveric humeral specimens (7 male, 1 female, age 25–84 years) were thawed to room
1065
temperature and osteotomized at the surgical neck to expose the very proximal aspect of the intramedullary canal. While we recognize that the humeral head is typically osteotomized through the anatomic neck for the shoulder arthroplasty procedure, our exposure allowed for precise positioning of the reamer so as to avoid eccentric reaming. A commercial shoulder arthroplasty system (Global Total Shoulder System, DePuy, Warsaw, IN, USA) was used. This system contained rigid hand reamers with diameters increasing from 6 to 16 mm in 2-mm increments. Under the ideal assumption that reaming only affects cross-sectional area, a significant reduction in strength should be accompanied by an increase in diaphyseal strain when the specimen is loaded. Four uniaxial strain gauges (PA06-125AA-350-1 EN, JP Technologies, San Bernardino, CA, USA) were mounted at 45° to the longitudinal axis on the lateral, anterior, medial and posterior (L, A, M, P) aspects of each specimen corresponding to the direction of principal strain for an idealized cylindrical specimen. The gauges were mounted one centimeter proximal to the maximum anticipated reamer depth to account for insertion variability and help ensure measurements were made within the strain field of the advancing reamer tip. All specimens were radiographed in anterior–posterior (AP) and medial–lateral (ML) directions for templating and cortical thickness measurements. The specimens were mounted in a fixture and dead weights were used to apply a continuous torque of 14 N m in external rotation (Fig. 1). Preliminary trials indicated this torque produced approximately 500 microstrain on the diaphysis similar to physiological values reported in the literature (Baggott and Lanyon, 1977; Carter et al., 1980; Cochran, 1972). The overall repeatability of this initial strain level was 5.3% determined from two initial load applications for each specimen. The strain gauge signals were conditioned (Vishay 2120, Measurements Group, Raleigh, NC, USA) and continuously acquired at 25 Hz throughout the entire reaming procedure. Starting with the smallest reamer, specimens were sequentially reamed until the surgeon reached the point at which he would stop intraoperatively, based on the feel and resistance to reaming. The diameter of the reamer at this final reaming was defined as the clinical diameter. The diaphyseal strain response to the second-to-last reaming (2 mm smaller) was used for comparison. If the clinical diameter for a specimen was less than 16 mm (i.e. wider reamers were still available), the specimen was then over-reamed for qualitative data analysis. The uniaxial strain data (e) were normalized to the strain following the application of the 14 N m torque prior to any reaming. The surgeon was blinded to the strain recordings throughout the experiment. Two-way repeated measures ANOVA with Student– Newman–Keuls post-hoc analysis was used to assess
1066
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
Fig. 1. Experimental setup. The humeri were mounted on a torsion fixture and a 14 N m torque was applied with weights. In part 1, uniaxial strain gauges (inset) were mounted on the diaphysis at 45° to the longitudinal axis. In part 2, three-element strain rosettes (inset) were used to measure the in-plane strain.
the reaming decision (second-to-last vs. clinical diameter) and measurement location (L, A, M, P). For two specimens, one of the four strain gauges was damaged during the test and mean substitution was used for the missing data (Fisher and van Belle, 1993). Significance was set at a = 0.05. 2.2. Part 2: comparison of arthroplasty stages Power analysis based on part 1 results used a withinspecimen strain variation of 23% with a root mean square difference of 35% yielding an effect size of d = 1.5 which was considered medium (Murphy and Myors, 1998). For the primary question of interest, the effect of arthroplasty stage on humeral strain, a power of 0.8 at a significance level of 0.05 was achieved with n = 10 (Norman and Streiner, 2000; Zar, 1999). With this number of specimens, the secondary effect of gauge location (found not significant in part 1) had a power of 0.55. Ten fresh frozen cadaveric humeral specimens were tested (8 male, 2 female, age 80–90 years for six specimens with no age data for four). Whereas part 1 measured uniaxial deformations, here strain rosettes (FRA-2-11, Tokyo Sokki Kenkyujo Co., Tokyo, Japan) were used to account for more complex loads introduced by broaching and implant insertion. Each rosette consisted of three uniaxial gauges from which the twodimensional surface strain was calculated. Two rosettes were mounted on the lateral and posterior aspects of the humeral diaphysis in positions that corresponded to the regions where minimum and maximum changes were observed in part 1. The specimen radiographs, preparation, loading, and data acquisition were otherwise the same as in part 1. The humeral arthroplasty procedure was reproduced. The specimens were reamed to the clinical diameter, broached, and a 1 mm oversized implant was inserted
in keeping with the manufacturerÕs protocol. The data from the rosettes were converted to principal strains using standard strain transformation relationships which were derived independent of material properties (Beer and Johnston, 1992; Carter, 1978; Lanyon et al., 1975). The maximum in-plane shearing strain (difference of principal strains), c, was analyzed along with the direction of the principal strain axes (h) relative to the transverse bone axis. Note, strains were analyzed rather than stresses since conversion to the latter requires the additional approximation of material properties (Carter, 1978). As in part 1, strain data were normalized to initial strains for analysis. Two-way repeated measures multivariate ANOVA with Student–Newman–Keuls post-hoc analysis was used to assess the effects of arthroplasty stage (reaming, broaching, implant) and measurement location (L, P) on c and h. One laterally placed rosette was damaged during the test and mean substitution was used for the missing data (Fisher and van Belle, 1993). Significance was set at a = 0.05.
3. Results 3.1. Part 1: clinical diameter decision Clinical diameter reaming showed large transient strain profiles followed by a significant (P = 0.007) elevation of equilibrium strain levels when compared to levels following the second-to-last reaming (Fig. 2). The mean clinical diameter was 14.4 mm (SD 1.9). Two specimens (clinical diameter = 16 mm) exhibited a strain increase of 50% or more while three specimens (clinical diameter = 14 mm) showed an increase greater than 60% on at least one gauge. Averaging all gauges, the overall increase following the second-to-last reaming was 11% and this increased to 30% after clinical
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
1067
3000
2500
P
microstrain
2000
1500
A 1000
L
500
M 0 Over-ream
SL CD -500 50
100
150
200 Time (s)
250
300
350
Fig. 2. Uniaxial strain patterns during reaming. Traces show measurements from gauges mounted on the lateral (L), anterior (A), medial (M) and posterior (P) humeral diaphysis. Vertical boundary lines indicate the start of the second-to-last (SL) reaming, clinical diameter (CD) reaming and over-reaming (qualitative only). Strain levels moderately increased following second-to-last reaming and further increased after clinical diameter reaming. A mixed pattern of increasing and decreasing strains was observed following over-reaming.
diameter reaming. Five specimens were reamed beyond the clinical diameter and three of these fractured. Following over-reaming, a mixed pattern of increasing and decreasing strains was observed on all five specimens. There was no significant difference in strain increase between gauge locations (P = 0.22), however, there was a significant interaction between strain measurement
location and clinical diameter decision (P = 0.032) where larger increases were observed on the AP than ML diaphyses (Fig. 3). The mean change following clinical diameter reaming for combined anterior and posterior data was 46% while the mean change for medial and lateral data was 15%. The mean anterior-to-posterior humeral and canal diameters were 21.5 mm (SD 2.6)
Second to Last
160
Clinical Diameter
Percent change in strain
140 120 100 Mean+SD Mean-SD Mean+SE Mean-SE Mean Outliers
80 60 40 20 0 -20
L
A
M
P
L
A
M
P
Strain gauge Location Fig. 3. Clinical diameter decision. Change in strain levels after the second-to-last reaming compared to levels after reaming at the clinical diameter on the lateral, anterior, medial, and posterior (L, A, M, P) faces of the humeral diaphysis. Clinical diameter reaming caused a significant change in diaphyseal strain (P = 0.007) and there was a significant interaction with gauge location (P = 0.032) with larger increases observed on the narrower AP diaphyses.
1068
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
and 14.1 mm (SD 2.8), respectively. The mean medialto-lateral humeral and canal diameters were larger (23.6 mm SD 2.4 and 16.5 mm SD 2.0). 3.2. Part 2: comparison of arthroplasty stages The equilibrium shear strains following each stage of the arthroplasty procedure were observed to consistently increase in four specimens (Fig. 4a) while a mixed pattern was observed in six specimens (Fig. 4b). The
mean clinical diameter was 12.2 mm (SD 1.8). One specimen failed during reaming at its clinical diameter of 14 mm and hence was excluded from the ANOVA. The three arthroplasty maneuvers had a significant effect on the maximum in-plane shear strain (P = 0.008). The shear strain following implant insertion was significantly higher than after reaming (P = 0.005) and broaching (P = 0.009). The changes in shearing strains after broaching were not significantly different from those observed after reaming (P = 0.450) although
L shear
1500
P shear 1000
microstrain
L max P max 500
0
P min
-500
L min CD=10mm -1000 0 (a)
50
100
Broach 150
200 Time (s)
Implant
250
300
350
400
1500
P shear L shear
1000
microstrain
P max 500
L max
0
P min L min
-500
CD=10mm -1000 0 (b)
50
100
Broach 150 Time (s)
200
Implant 250
300
Fig. 4. Strain patterns during arthroplasty. Traces show in-plane surface strains from rosettes mounted to the lateral (L) and posterior (P) diaphysis. Max and min denote principal strains where positive denotes tension and negative indicates compressive strains. The difference of the principal strains gives the maximum shear strain. Some humeri exhibited progressive increase in strain (a), while others exhibited mixed patterns of increasing and decreasing strain (b).
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071 Posterior
Lateral
100
1069
Percent change in shear strain
80 60 40 20
Mean+SD Mean-SD Mean+SE Mean-SE Mean Outliers
0 -20 -40 -60
Ream
Broach
Implant
Ream
Broach
Implant
Arthroplasty stage Fig. 5. Comparison of arthroplasty stages. Changes in the maximum in-plane shear strain on the lateral and posterior diaphyses were significantly altered by arthroplasty (P = 0.008). Implant insertion significantly increased shear strain compared to reaming (P = 0.005) and broaching (P = 0.009).
the variability was observed to increase on the lateral diaphysis. Again, the effect of gauge location was not significant (P = 0.224). In contrast to part 1, no significant (P = 0.080) interaction between location and arthroplasty stage was observed (Fig. 5) and the AP and ML specimen dimensions were more symmetric. AP mean humeral and canal diameters were measured at 23.0 mm (SD 4.0) and 13.2 mm (SD 4.2). Mean ML diameters were 23.0 mm (SD 2.1) and 13.1 mm (SD 3.9). The presence of mixed strain patterns increased the range of changes observed. Following reaming, the changes in shear strains ranged from 41.1% to 23.6%. Changes ranged from 33% to 65% after broaching and 29% to 51% with the implant inserted. Arthroplasty had no significant effect on the angle of the principal strain axes (P = 0.808). Following the second-to-last reaming, the mean angles of the axes were 46.2° (SD 8.5) and 44.3° (SD 7.3) relative to the transverse humeral axis on the lateral and posterior gauge locations, respectively. After the accumulated effects of the procedure, the mean angles were 41.1° (SD 5.0, lateral) and 44.3° (SD 8.1, posterior) with the implant inserted.
4. Discussion This study aimed to understand the potential reduction in torsional humeral strength during shoulder arthroplasty. We focused on torsion as this appears to be a dominant mechanism for intraoperative fractures (Bonutti and Hawkins, 1992; Schmidt, 1999; Wirth
and Rockwood, 1994). Ideally, strength would be assessed with a torque-to-failure test following a particular stage of the procedure (Jonsson and Stromberg, 1985; Martens et al., 1980; Netz et al., 1980; Pratt et al., 1987). However, a large number of specimens would have been needed to compare the four steps from the second-to-last reaming to implant insertion. Instead, this study utilized continuous monitoring of the changing strain under a static load to indicate strength reductions. Using this repeated measures design, a significant strain increase suggests a significant reduction in the safe margin prior to reaching the fracture threshold. Asymmetric geometry and material anisotropy complicates the determination of strength, however, a coarse estimate may be obtained by considering a two-dimensional cylindrical cross-section with isotropic properties (Appendix A). With this simplification, the mean 30% increase observed in part 1 suggests a strength reduction of approximately 23%. For the narrower anterior-toposterior diameter (21.5 mm), increases in excess of 50% suggest strength reductions of over 33%. Using torque to failure tests, Pratt et al. (1987) found similar reductions (37.5%) in femoral strength following reaming to 12 mm. This simplification likely underestimates the strength reduction since the gauges measure local changes that may not represent the peak strains on the cortical surface. Mathematical models have shown irregular transverse geometry may result in stress variations of ±20% around the cross-section (Kennedy and Carter, 1985). In part 2, the peak changes on the posterior diaphysis following clinical diameter reaming were lower than those observed in part 1 but were comparable to changes
1070
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
measured when humeral diameters were similar. The occurrence of mixed patterns particularly during the latter stages of the procedure was unexpected and was inconsistent with the strength reduction model due strictly to altered cross-sectional area. Instead, the prevalence of these patterns during the over-reaming stage of part 1 suggests it may be an indicator of excessive bone removal. The damage mechanism remains unclear from strain data alone but may be due to interstitial crack formations causing reduced strains in the crackÕs wake (Elias et al., 2000). There were several limitations to this study. First, strain gauges provided only local measurements hence it was difficult to generalize the strain field to regions away from the gauges. Moreover, the part 1 uniaxial gauges were less descriptive than the rosettes used in part 2. Additional measures of strength and damage accumulation such as torsional stiffness (Martens et al., 1981) and crack density (Forwood and Parker, 1989; Jepsen et al., 1999; Netz et al., 1980) combined with finite element analysis (Elias et al., 2000; Huiskes, 1982; Keyak and Rossi, 2000) would provide valuable complementary information. Second, the analysis was based on the equilibrium strain measurements following each arthroplasty step and did not account for the transient strains during each maneuver. These dynamic strains may have been responsible for more severe damage mechanisms than were detectable with the comparison of between—stage strains. Third, the humeri were continuously loaded and this may have increased the damage caused during the experiment. One alternative would have been to load and unload the specimens between surgical steps. However, residual strain offsets following unloading would have impeded strain comparisons before and after each step (Carter et al., 1981; Fondrk et al., 1999). Fourth, the use of a frame that allowed dead-weight vibration introduced a dynamic torsion component particularly during broaching. Clinically, there would also likely be a dynamic torsion component during the broaching stage with the humerus held in external rotation. However, it is uncertain whether the magnitude of the dynamic torque in this study was comparable to that experienced during surgery. In conclusion, reaming the humeral canal until cortical bone contact has the potential to significantly reduce torsional strength, particularly in humeri with narrow outer diaphyses. The risk of intraoperative fracture could potentially be reduced by using finer reamer increments. In addition, cortical strain from implant insertion could be reduced with a reduction in stem over-size. This conceptually simple model of strength reduction due to shoulder arthroplasty quantitatively highlights some of the potential risks when planning and performing the procedure.
Hollow cylinder under torsion
τ
max
σ45o
τ
ri ro T
T
Fig. A.1. Cylinder under torsion. For a cylinder with isotropic material properties, the principal axes lie at 45° to the long axis.
Acknowledgments The technical assistance of Mr. Darrell Goertzen in part 1 and surgical assistance of Dr. David Blanchette in part 2 are gratefully acknowledged. The support of DePuy (Canada) Ltd. is gratefully acknowledged. Thanks also to Dr. D.P. Romilly for the loan of strain measurement equipment.
Appendix A. Approximate strength reduction The maximum shear stress, smax, in a hollow cylinder loaded in torsion is located at the outer circumference and is given by Tro J
smax ¼
where p J ¼ ðr4o r4i Þ 2 is the polar moment of inertia, T is the applied torque, ro is the outer radius, ri is the inner radius of the hollow cylinder (Fig. A.1). T can be increased until smax reaches the failure limit of the material, sfailure. Hence the maximum torsional load that the cylinder can carry is T max ¼
J sfailure ro
For a cylinder under pure torsion, the magnitude of the shear stress (smax) is equal to the stress at 45° (r45°). Since, r45 ð1 þ mÞ E for homogeneous YoungÕs modulus E and PoissonÕs ratio m. Reaming increases ri and hence reduces Tmax: e45 ¼
T max
original
T max
T max original
reamed
J original J reamed J original ereamed eoriginal ¼ ereamed ¼
A.M.T. Choo et al. / Clinical Biomechanics 20 (2005) 1064–1071
References Baggott, D.G., Lanyon, L.E., 1977. An independent Ôpost-mortemÕ calibration of electrical resistance strain gauges bonded to bone surfaces Ôin vivoÕ. J. Biomech. 10, 615–622. Beer, F.B., Johnston, E.R., 1992. Mechanics of Materials. McGrawHill, New York. Bonfield, W., Grynpas, M.D., 1982. Spiral fracture of cortical bone. J. Biomech. 15, 555–559. Bonutti, P.M., Hawkins, R.J., 1992. Fracture of the humeral shaft associated with total replacement arthroplasty of the shoulder. A case report. J. Bone Joint Surg. Am. 74, 617–618. Cameron, B., Iannotti, J.P., 1999. Periprosthetic fractures of the humerus and scapula: management and prevention. Orthop. Clin. N Am. 30, 305–318. Campbell, J.T., Moore, R.S., Iannotti, J.P., Norris, T.R., Williams, G.R., 1998. Periprosthetic humeral fractures: mechanisms of fracture and treatment options. J. Shoulder Elbow Surg. 7, 406–413. Carter, D.R., 1978. Anisotropic analysis of strain rosette information from cortical bone. J. Biomech. 11, 199–202. Carter, D.R., Smith, D.J., Spengler, D.M., Daly, C.H., Frankel, V.H., 1980. Measurement and analysis of in vivo bone strains on the canine radius and ulna. J. Biomech. 13, 27–38. Carter, D.R., Caler, W.E., Spengler, D.M., Frankel, V.H., 1981. Fatigue behavior of adult cortical bone: the influence of mean strain and strain range. Acta Orthop. Scand. 52, 481–490. Cochran, G.V., 1972. Implantation of strain gages on bone in vivo. J. Biomech. 5, 119–123. Elias, J.J., Frassica, F.J., Chao, E.Y., 2000. The open section effect in a long bone with a longitudinal defect—a theoretical modeling study. J. Biomech. 33, 1517–1522. Fisher, L.D., van Belle, G., 1993. Biostatistics—A Methodology for the Health Sciences. John-Wiley & Sons Inc., New York. Fondrk, M.T., Bahniuk, E.H., Davy, D.T., 1999. Inelastic strain accumulation in cortical bone during rapid transient tensile loading. J. Biomech. Eng. 121, 616–621. Forwood, M.R., Parker, A.W., 1989. Microdamage in response to repetitive torsional loading in the rat tibia. Calcified Tissue Int. 45, 47–53. Huiskes, R., 1982. On the modelling of long bones in structural analyses. J. Biomech. 15, 65–69. Jepsen, K.J., Davy, D.T., 1997. Comparison of damage accumulation measures in human cortical bone. J. Biomech. 30, 891–894.
1071
Jepsen, K.J., Davy, D.T., Krzypow, D.J., 1999. The role of the lamellar interface during torsional yielding of human cortical bone. J. Biomech. 32, 303–310. Jonsson, U., Stromberg, L., 1985. Torsional tests of long bones with computerized equipment. J. Biomed. Eng. 7, 251–255. Kennedy, J.G., Carter, D.R., 1985. Long bone torsion: I. Effects of heterogeneity, anisotropy and geometric irregularity. J. Biomech. Eng. 107, 183–188. Keyak, J.H., Rossi, S.A., 2000. Prediction of femoral fracture load using finite element models: an examination of stress- and strainbased failure theories. J. Biomech. 33, 209–214. Lakes, R.S., Nakamura, S., Behiri, J.C., Bonfield, W., 1990. Fracture mechanics of bone with short cracks. J. Biomech. 23, 967–975. Lanyon, L.E., Hampson, W.G., Goodship, A.E., Shah, J.S., 1975. Bone deformation recorded in vivo from strain gauges attached to the human tibial shaft. Acta Orthop. Scand. 46, 256–268. Martens, M., van Audekercke, R., de Meester, P., Mulier, J.C., 1980. The mechanical characteristics of the long bones of the lower extremity in torsional loading. J. Biomech. 13, 667–676. Martens, M., van Audekercke, R., de Meester, P., Mulier, J.C., 1981. The geometrical properties of human femur and tibia and their importance for the mechanical behaviour of these bone structures. Arch Orthop. Trauma Surg. 98, 113–120. Murphy, K.R., Myors, B., 1998. Statistical Power Analysis—A Simple and General Model for Traditional and Modern Hypothesis Tests. Lawrence Erlbaum Associates, New Jersey. Netz, P., Eriksson, K., Stromberg, L., 1980. Material reaction of diaphyseal bone under torsion. An experimental study on dogs. Acta Orthop. Scand. 51, 223–229. Norman, G.R., Streiner, D.L., 2000. Biostatistics: The Bare Essentials. B.C. Decker Inc., Hamilton. Pratt, D.J., Papagiannopoulos, G., Rees, P.H., Quinnell, R., 1987. The effects of medullary reaming on the torsional strength of the femur. Injury 18, 177–179. Reilly, D.T., Burstein, A.H., 1975. The elastic and ultimate properties of compact bone tissue. J. Biomech. 8, 393–405. Schmidt, A.H., 1999. Periprosthetic fractures of the humerus. Tech. Orthop. 14, 151–156. Wirth, M.A., Rockwood Jr., C.A., 1994. Complications of shoulder arthroplasty. Clin. Orthop., 47–69. Zar, J.H., 1999. Biostatistical Analysis. Prentice Hall, New Jersey.