Vet Comp Orthop Traumatol 2025; 38(03): 119-126
DOI: 10.1055/s-0044-1791701
Original Research

Effect of Plate Screw Configuration on Construct Stiffness and Plate Strain in a Synthetic Short Fragment Small Gap Fracture Model Stabilized with a 12-Hole 3.5-mm Locking Compression Plate

Fabian N. Trefny
1   School of Veterinary Medicine, Murdoch University, Perth, Australia
,
1   School of Veterinary Medicine, Murdoch University, Perth, Australia
,
Giselle L. Hosgood
1   School of Veterinary Medicine, Murdoch University, Perth, Australia
,
Robert E. Day
2   Health Technology Management Unit, East Metropolitan Health Service, Royal Perth Hospital, Perth, Australia
,
Alex Hayes
2   Health Technology Management Unit, East Metropolitan Health Service, Royal Perth Hospital, Perth, Australia
› Author Affiliations

Funding The authors received partial financial support for the implants from DePuy Synthes.
 

Abstract

Objective The aim of the study was to determine the effect of a short and long working length screw configuration on construct stiffness and plate strain in a synthetic, short fragment, small gap fracture model stabilized with a 12-hole 3.5-mm locking compression plate (LCP).

Study Design Six replicates of short and long working length constructs on a short fragment, small gap fracture model underwent four-point bending. Construct stiffness and plate strain were compared across working length and along the plate.

Results With the LCP on the compression surface (compression bending), the short working length had a significantly higher construct stiffness and lower plate strain than the long working length. Conversely, with the LCP on the tension surface (tension bending), transcortical contact between 150 and 155 N induced load sharing at the fracture gap, which significantly increased construct stiffness and decreased plate strain in the long working length. At 100 N (precontact), the short working length had a significantly higher construct stiffness and lower plate strain than the long working length, comparable with our compressing bending results.

Conclusion In compression bending, and before transcortical contact occurred in tension bending, the short working length had a significantly higher construct stiffness and lower plate strain than the long working length. Load sharing due to transcortical contact observed in our model in tension bending will vary with fracture gap, working length, and loading condition. These results must be interpreted with caution when considering clinical relevance or potential in vivo biomechanical advantages.


Introduction

Prior to plate fixation of fractures, a surgeon makes several decisions regarding the number and positioning of screws that will affect biomechanical performance of the construct.[1] [2] [3] [4] In locked constructs with adequate plate–bone standoff distance, plate screw configuration determines the working length, defined as the distance between the innermost screws either side of the fracture.[1] [5]

Previous investigations have shown that short working lengths result in higher construct stiffness and lower plate strain.[2] [4] [6] [7] [8] [9] [10] [11] A long working length is advocated by others to enable stress modulation and relative stability at the fracture gap, on the premise that excessive stiffness may cause stress shielding, impede symmetric callus formation, and may lead to early implant failure.[12] [13] The optimal interfragmentary motion required for bone healing remains unknown, and less stiff implants have been shown to experience higher plate strain and fail after fewer cycles.[3] [4]

Conflicting results on the effect of screw configuration have generated contention in the literature.[2] [3] [7] [8] [9] [14] [15] It was proposed that a long working length will lower stress in the plate bridging the fracture gap, by distributing stress over a longer plate span.[5] [16] [17] [18] [19] [20] [21] Such contradictory views and misinterpretation of the effect of working length may have arisen due to differences in experimental design.[15] This includes how plate strain was measured and whether identical load or deformation was applied to constructs,[2] [17] biomechanical differences between locking versus conventional plates, and confounding by the effect of screw number and configuration.[5] [22] [23]

It has been shown that less stiff constructs with small fracture gaps may endure load sharing due to bone model contact, resulting in increased construct stiffness and decreased plate strain.[2] [4] [17] [24] Some in vitro studies have advocated deliberately increasing working length in these fracture configurations to promote load sharing and transcortical contact stability.[4] [24] Such an approach may not result in a mechanobiological environment suitable for fracture healing.

To further explore the effect of working length and the effect of transcortical contact, this study compared construct stiffness and plate strain of two screw configurations in a synthetic small gap fracture model, replicating the mechanobiological environment of an imperfectly reduced or unstable simple fracture. We hypothesized that in compression and tension bending, the short working length would have higher construct stiffness and lower plate strain than the long working length.


Materials and Methods

Bone Models

A synthetic, short-oblique fracture-gap model was constructed using polyacetal tubes (Delrin acetal polymer, Mulford Plastics, Western Australia, Australia) with an outer diameter of 15.9 mm and inner diameter of 9.5 mm as used in previous studies.[11] [25] Six screw holes and a locating hole at each end of the tube were predrilled on axial midline using a numerically controlled mill with a 2.8-mm drill bit (2.8-mm drill bit, Synthes GmbH, Oberdorf, Switzerland) as per AO recommendations for placement of 3.5-mm locking screws. The tubes were cut at 30 degrees to the long axis, generating a short segment of 45 mm, a long segment of 150 mm, and a fracture gap of 1.75 mm, modeling an imperfectly reduced, noncompressible, short fragment diaphyseal fracture.


Construct Assembly

Six replicates of two differing screw configurations (12 constructs in total) were assembled with 12-hole 3.5-mm locking compression plates (LCP; Synthes GmbH) applied to bone models. All constructs were assembled in random order by a board-certified surgeon (M.G.) using AO technique, maintaining a 1-mm plate standoff distance with a spacer (Tiling spacers, Rubi, Spain). Six 26-mm bicortical locking screws (Veterinary Locking StarDrive, 3.5 mm, Self-Tapping, Synthes GmbH) were placed per construct using an orthopaedic drill (Cordless Driver III, Stryker Instruments, Michigan, United States) coupled with a 1.5-Nm torque limiter (Torque Limiter, 1.5 Nm, Synthes GmbH).


Short Working Length Constructs

Three screws were placed in the short fragment, two screws immediately adjacent to the fracture gap in the long fragment, and a sixth screw at the end of the plate in the long fragment ([Fig. 1A]).

Zoom
Fig. 1 Illustration of an assembled (A) short and (B) long working length 12-hole 3.5-mm locking compression plate (LCP) construct. The short oblique fracture gap is positioned between the third and fourth plate holes (left to right) in both screw configurations. Solid circles depict locked screw positions in the plate.

Long Working Length Constructs

One screw hole was left empty on each side of the fracture gap. Two screws remained in the short fragment, two screws were placed one hole further away from the fracture gap in the long fragment, and a fifth screw at the end of the plate in the long fragment ([Fig. 1B]).


Biomechanical Testing

Constructs were fixed in a custom loading jig with a 3.5-mm bolt through each positioning hole. Four-point bending was applied using a materials testing machine (Instron 5566, Massachusetts, United States) with support and load rollers 336 and 230 mm apart, respectively, using nondestructive, quasi-static ramp loading for three cycles as previously described.[6] [7] [11] [25] Three cycles were chosen based on pilot testing, which found no differences after the initial loading cycle. Load was applied via a 2-kN load cell, parallel to the screw axis under displacement control (10 mm/min) from 5 to 240 N, producing a peak bending moment of 6 Nm. Each construct was alternately tested with the plate on the compression and tension surface of the bone model ([Fig. 2]) to compare compression (gap opening) and tension (gap closing) bending.

Zoom
Fig. 2 Illustration of an assembled construct in loading jigs positioned between load (upper) and support (lower) rollers. The position of load and support rollers is sequentially altered such that each construct endures both (A) compression (gap opening) and (B) tension (gap closing) four-point bending. Under both bending modes, plates remained positioned on the undersurface of the bone model, enabling optical measurement of von Mises' strain from below the construct.

Measurement of Stiffness

Construct stiffness (N/mm) was derived from the slope of the linear elastic part of the load displacement curve between 150 and 240 N based on a preset protocol in Bluehill v2.5.39 (Instron, Massachusetts, United States). In tension bending, R Studio v4.3.2 (R Studio, Massachusetts, United States) was used to fit a segmented linear model within the elastic region of the load displacement curves to detect for expected breakpoints in construct stiffness, coinciding with transcortical contact. When a breakpoint was detected, construct stiffness before (precontact) and after (postcontact) each breakpoint was calculated.


Measurement of Strain

Plates were painted white (Flat White, White Knight, NSW, Australia) followed by a fine black speckle pattern (Flat Black, White Knight) to enable optical strain measurement with three-dimensional (3D) digital image correlation as previously described.[6] [7] [11] [26] [27] The von Mises strain was calculated at six predefined regions of interest (ROI) on the plate surface ([Fig. 3]). Stereo images were obtained with two calibrated high-resolution cameras (Point Grey, Sony ICX625, Richmond, Canada) with 2,448 × 2,048 pixel sensors. Correlation-based displacements of image subsets of the speckle pattern were captured using Vic-Snap (Vic-Snap, Correlated Solutions, South Carolina, United States) and the von Mises strain was calculated with VIC3D (VIC3D, Correlated Solutions). Load (N) was plotted against the von Mises strain and a line of best fit was used to calculate peak (postcontact) strain at a load of 240 N. Under tension bending, precontact strain at a load of 100 N was also reported in both long and short working lengths for comparison.

Zoom
Fig. 3 Illustration depicting six regions of interest (ROI) of an assembled (A) short and (B) long working length 12-hole 3.5-mm locking compression plate (LCP) construct. The short oblique fracture gap is positioned between the third and fourth plate holes (left to right) in both screw configurations. Numbered boxes depict the ROI for strain measurement, positioned axially on the LCP surface. Solid circles depict locked screw positions in the plate.

Statistical Methods

Statistical analysis was performed using SAS v9.8 (SAS Institute, North Carolina, United States) software. Six replicates across two screw configurations were sufficient to detect an effect size as small as 1.75 (power = 0.8; α = 0.05). Response variables of interest were construct stiffness (N/mm) and strain (mm/mm) at each ROI in compression and tension bending. Responses were tested for normality using the Shapiro–Wilk test at p ≤ 0.05. Normally and non-normally distributed data were summarized as mean and 95% confidence interval (CI) or median and interquartile range (IQR), respectively. The effect of working length on construct stiffness was tested with a paired t-test (normal data) or Wilcoxon two-sample rank sum test (non-normal). Two-factor analysis of variance tested the effect of working length and ROI on plate strain. Where there was a significant effect of working length, planned pairwise comparisons were made between working lengths or across ROI against a Bonferroni adjusted p ≤ 0.05. The Davies test[28] was used to detect breakpoints in the load displacement data in tension bending.



Results

Construct Stiffness

In compression bending, construct stiffness was significantly higher (p = 0.0172) for the short working length (median: 74.9 N/mm; IQR: 5.3) than the long working length (median: 57.4 N/mm; IQR: 6.6). In tension bending, construct stiffness was significantly lower (p = 0.0046) for the short working length (mean: 73.4 N/mm; 95% CI: 68.2–78.6) than the long working length (mean: 82.6 N/mm; 95% CI: 78.6–86.5). Construct stiffness was measured near peak load and was therefore recorded after transcortical contact occurred in the bone model in tension bending.


Transcortical Contact

In tension bending, the stiffness response was nonlinear within the elastic region due to transcortical contact occurring within the load range. Visible contact was observed at the fracture gap in both working lengths, but most notably in the long working length. Using a segmented linear model, breakpoints in construct stiffness were identified between 150 and 155 N ([Fig. 4]), coinciding with transcortical contact. This model estimated both pre- and postcontact construct stiffness. Precontact construct stiffness remained significantly higher (p < 0.0001) for the short working length (mean: 62.4 N/mm; 95% CI: 60.3–64.5) than for the long working length (mean: 49.8 N/mm; 95% CI: 49.0–50.6). At each breakpoint, there was a positive slope change where all constructs became stiffer, but this was only significant in the long working length (p < 0.0001). Postcontact construct stiffness was significantly lower (p < 0.0001) for the short working length (mean: 68.2 N/mm; 95% CI: 65.6–70.8) than for the long working length (mean: 81.8 N/mm; 95% CI: 78.6–84.9).

Zoom
Fig. 4 Load displacement curves depicting construct stiffness (N/mm) of six replicates (numbered 1–6) of short (light gray line) and long (dark gray line) working lengths under tension bending. Dashed lines represent an estimate for construct stiffness when fitting a segmented model, showing precontact (segment 1; large dashes) and postcontact (segment 2; small dashes) construct stiffness. Gray circles depict visible breakpoints in construct stiffness in the long working length, coinciding with transcortical contact occurring in the bone model between 150 and 155 N.

Plate Strain

In compression bending, strain was significantly lower for the short working length than for the long working length at ROI 1 (p = 0.0001), ROI 2 (p < 0.0001), ROI 3 (p = 0.0006), ROI 4 (p < 0.0001), and ROI 6 (p = 0.0024; [Table 1]). At no ROI was strain significantly higher in the short working length, including the ROI over the fracture gap. In tension bending, peak strain in the postcontact region (240 N) was significantly higher in the short working length than in the long working length at ROI 1 (p < 0.0001), ROI 2 (p < 0.0001), ROI 3 (p = 0.0457), and ROI 6 (p = 0.0248; [Table 2]). Conversely, strain in the precontact region (100 N) of the load range was significantly lower for the short working length than the long working length at ROI 1 (p < 0.0030), ROI 2 (p < 0.0030), ROI 3 (p = 0.035), ROI 4 (p = 0.0010), and ROI 5 (p < 0.0001; [Table 3]).

Table 1

Mean (95% confidence interval) strain (mm/mm) at peak load (240N) at each numbered region of interest (ROI) on the short and long working length construct under compression bending

Working length

ROI

1

2

3

4

5

6

Short

59.77a (56.00–63.54)

38.53a (33.34–43.72)

39.43a (36.08–42.78)

15.72a (11.98–19.46)

8.18a (7.39–8.97)

67.89a (41.81–82.05)

Long

81.81b (75.66–87.96)

64.20b (59.98–68.42)

59.97b (53.62–66.32)

32.83b (26.78–38.88)

14.69a (12.78–16.60)

90.46b (78.08–102.83)

Note: Within each column, means with the same superscripts (a,b) are not significantly different (Bonferroni adjusted p > 0.05).


Table 2

Mean (95% confidence interval) strain (mm/mm) at 240 N (postcontact) at each numbered region of interest (ROI) on the short and long working length construct under tension bending

Working length

ROI

1

2

3

4

5

6

Short

61.39a (53.73–69.05)

42.99a (33.14–52.83)

49.66a (42.27–57.05)

14.80a (11.46–18.14)

8.78a (7.33–10.23)

74.26a (41.75–106.76)

Long

38.10b (26.94–49.26)

22.38b (18.15–26.62)

33.30b (27.25–39.35)

17.81a (14.92–20.69)

8.10a (7.62–8.57)

12.40b (11.72–13.08)

Note: Within each column, means with the same superscripts (a,b) are not significantly different (Bonferroni adjusted p > 0.05).


Table 3

Mean (95% confidence interval) strain (mm/mm) at 100 N (precontact) at each numbered region of interest (ROI) on the short and long working length construct under tension bending

Working length

ROI

1

2

3

4

5

6

Short

25.58a (22.39–28.77)

17.91a (13.81–22.01)

20.69a (17.61–23.77)

6.17a (4.77–7.56)

3.66a (3.05–4.26)

30.94a (17.40–44.48)

Long

31.94b (29.95–33.93)

25.05 b (23.37–26.73)

25.81b (22.42–29.20)

12.94b (10.39–15.50)

5.92b (5.38–6.46)

38.09a (36.39–39.79)

Note: Within each column, means with the same superscripts (a,b) are not significantly different (Bonferroni adjusted p > 0.05).




Discussion

We accepted our hypothesis in compression bending, that the short working length would have higher construct stiffness and lower plate strain than the long working length. This was also true in tension bending, up to transcortical contact. After contact, the long working length showed higher construct stiffness and lower plate strain than the short working length.

In the long working length, transcortical contact in tension bending resulted in bilinear load displacement curves. Our methodology initially compared construct stiffness and plate strain between constructs at a load range of 150 to 240 N, but a single linear model did not accurately describe the tension bending results. A segmented linear model identified a breakpoint in construct stiffness between 150 and 155 N, which coincided with the point of transcortical contact. Before contact (∼5–150 N), construct stiffness was significantly lower in the long working length than in the short working length, comparable with our results in compression bending. After contact (∼150–155 N), construct stiffness was significantly higher, but only in the long working length. This is likely due to transcortical contact occurring earlier, at lower load than in the short working length. Significant increases in construct stiffness in the short working length would also be expected following transcortical contact. However, a larger bending moment would be necessary before significant increases in construct stiffness become evident in the stiffer, short working length constructs.

In fracture-gap models with the plate on the compression surface, bending progressively widens the gap at the trans-cortex. Increasing working length where transcortical contact will not occur has also been shown by others to decrease construct stiffness, increase plate strain,[4] [6] [7] [8] [9] [11] [20] [24] [25] and reduce fatigue life.[4] Increasing working length increases the lever arm to the fracture gap causing progressively higher bending moments.[5] Higher bending moments acting on the plate cause greater plate deformation and correspondingly higher plate strain.[6] [7] [11] [18] [20] [29]

Conversely, in fracture-gap models with the plate on the tension surface, bending reduces the fracture gap at the trans-cortex. Depending on the fracture-gap width, and other test characteristics, this can result in transcortical fragment contact when the gap is small. When transcortical fragment contact occurs, the model changes from a load-bearing construct to an imperfect load-sharing construct.[2] [4] [24] [29] Four-point bending effectively becomes three-point bending, and the working length is halved, with the point of transcortical contact acting as a fulcrum.

Our results for the long working length in tension bending agree with a previous finite element study.[4] They showed that in axial compression in a gap model where transcortical fracture-gap contact did not occur, increasing working length had a clear negative effect, with decreased construct stiffness and increased plate strain. Where the fracture gap was small enough for contact to occur, they reported significantly increased construct stiffness and decreased plate strain. Their results were in agreement with an anatomically reduced two-part fracture model, where widely spaced screws or those placed away from the fracture also resulted in lower plate strain.[24] Both studies advocated the approach of intentionally leaving vacant plate holes over the fracture gap in simple, small gap fractures to promote transcortical contact stability.

While long working length constructs may promote transcortical contact in small gap fractures under some loading conditions, and, therefore, reduce plate strain, the effect of this process on bone healing has not been evaluated. In a clinical situation, cyclic transcortical contact with weight bearing would very likely create high interfragmentary strain, as when contact occurs interfragmentary strain, by definition, is 100% regardless of the size of the fracture gap, which is not compatible with productive bone healing.[30] [31] [32] In this scenario, bone will be actively resorbed from the fracture ends, widening the fracture gap until strain is low enough to support the formation of soft callus.[32] [33] [34] [35] With ongoing deformation of the plate during each cycle, resorption widens the transcortical defect and correspondingly increases plate strain. Whether fracture healing is likely to progress to union would therefore depend on a patient's capacity to resorb bone and reduce interfragmentary stain to a level compatible with forming fibrous callus within the fatigue life of the implant.

Any potential clinical consequences of load sharing with a long working length should not be overstated given the limitations of our in vitro model without cyclic loading. Nevertheless, we believe the approach of deliberately increasing working length to encourage load sharing in small gap fractures may not result in a mechanobiological environment suitable for fracture healing.

Contrary to our results and those by others,[4] [6] [7] [8] [9] [11] [20] [24] [25] some studies with locked plates did not find an effect of incrementally longer working length on construct stiffness,[5] [18] [36] plate strain,[36] failure load, or fatigue strength.[5] [18] [36] [37] These studies used axial compression in a femoral gap model stabilized with an eccentrically positioned locking plate without a defined plate standoff distance. In a locked construct in compression bending, or one with sufficient plate standoff distance in tension bending, working length is the distance between the screws nearest to the fracture.[3] Conversely, in constructs with insufficient standoff to prevent plate–bone contact in tension bending, working length becomes the length of the fracture gap, regardless of screw position.[3] The lack of significant differences reported by the latter studies is therefore not surprising, given that, despite altering screw positions, working length was determined by the size of the fracture gap, and so was the same between groups.

We used a small fracture gap to replicate the mechanobiological environment of an imperfectly reduced or unstable simple fracture, such as a noncompressible short oblique fracture, or a poorly compressed transverse fracture. Our clinical experience is that many revisions for cyclic plate failure occur with these configurations because interfragmentary strain may be unsustainably high for productive indirect bone healing.[6] [7] [32]

We did not perform fatigue testing, but instead used 3D digital image correlation to measure the von Mises strain at multiple plate ROI.[6] [7] [8] [9] [11] [27] Within the linear elastic region, stress is directly proportional to strain[1] [38] [39] and small decreases in strain can significantly increase implant fatigue life.[3] [39] [40] We consider that surface strain thereby provides useful comparative data, as an implant that experiences higher repetitive strain will have a shorter fatigue life. Nevertheless, measurement of surface strain does not serve as a substitute to cyclic loading.

A limitation of our study was the possible confounding of working length by plate screw number. Our small fragment model prohibited symmetrically increasing working length either side of the fracture without removing one screw from the short fragment. This mimics the clinical scenario of a short fragment long bone fracture, which occurs frequently in companion animals, such as in the canine distal radius.[6] [7] [15] Studies primarily exploring the effect of screw number have also inadvertently altered working length or functional plate length through screw omission adjacent to the fracture gap or at the plate end, respectively.[22] [23] Despite choosing to alter both working length and screw number in the short fragment, in a clinical scenario our preference is to place a third screw whenever possible. Our model still adheres to clinical recommendations to place at least two screws per fragment and leave between zero and three empty plate holes adjacent to the fracture.[4] [17]

The tubes in our model were cylindrical, with tension or compression bending determined by position in the loading jig. In an in vivo scenario, compression or tension surfaces of a native bone vary depending on load direction and bone morphology. Clinical decisions on implant placement may also be based on other features such as bone surface characteristics or ease of surgical access. This study also did not explore potential biomechanical interactions between plate contour and curvature in the transverse plane with respect to bending direction. We did not test our constructs under axial or torsional loads; however, axial compression causes tension bending with an implant eccentric to the mechanical axis of long bones.[3] [9] [20]

In conclusion, in a 12-hole 3.5-mm LCP construct in compression bending, the short working length had a significantly higher construct stiffness and lower plate strain than the long working length. Transcortical contact occurred only in tension bending, causing a bilinear load displacement curve with immediate increases in construct stiffness on transcortical contact. This effect was only significant in the long working length under our experimental conditions. The apparent mechanical advantages of transcortical contact reported in vitro may not be sustainable in vivo given this would inevitably generate high interfragmentary strain, promoting bone resorption at the fracture gap.



Conflict of Interest

None declared.

Acknowledgment

The authors wish to thank DePuy Synthes for partial financial support through provision of implants and to the Royal Perth Hospital for access to their engineering and material testing laboratory.

Note

An abstract of this paper was presented at the annual European College of Veterinary Surgeons Congress, July 6, 2019, Budapest, Hungary. Results from this study were also presented at the AO Vet Symposium, Bridging the Gap: Translating Clinical Research to Clinical Practice, presented November 23, 2020, via Webinar.


Authors' Contribution

All the authors contributed to the conception and design of this study. F.N.T., M.R.G., R.E.D., and A.H. were involved with data acquisition and materials testing. R.E.D. and A.H. provided technical support with material testing. G.L.H. contributed to acquisition of data and data analysis. All the authors contributed to data interpretation and revision of the submitted manuscript.



Address for correspondence

Fabian N. Trefny, BVSc
School of Veterinary Medicine, Murdoch University
Perth
Australia   

Publication History

Received: 01 May 2024

Accepted: 14 September 2024

Article published online:
04 October 2024

© 2024. Thieme. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany


Zoom
Fig. 1 Illustration of an assembled (A) short and (B) long working length 12-hole 3.5-mm locking compression plate (LCP) construct. The short oblique fracture gap is positioned between the third and fourth plate holes (left to right) in both screw configurations. Solid circles depict locked screw positions in the plate.
Zoom
Fig. 2 Illustration of an assembled construct in loading jigs positioned between load (upper) and support (lower) rollers. The position of load and support rollers is sequentially altered such that each construct endures both (A) compression (gap opening) and (B) tension (gap closing) four-point bending. Under both bending modes, plates remained positioned on the undersurface of the bone model, enabling optical measurement of von Mises' strain from below the construct.
Zoom
Fig. 3 Illustration depicting six regions of interest (ROI) of an assembled (A) short and (B) long working length 12-hole 3.5-mm locking compression plate (LCP) construct. The short oblique fracture gap is positioned between the third and fourth plate holes (left to right) in both screw configurations. Numbered boxes depict the ROI for strain measurement, positioned axially on the LCP surface. Solid circles depict locked screw positions in the plate.
Zoom
Fig. 4 Load displacement curves depicting construct stiffness (N/mm) of six replicates (numbered 1–6) of short (light gray line) and long (dark gray line) working lengths under tension bending. Dashed lines represent an estimate for construct stiffness when fitting a segmented model, showing precontact (segment 1; large dashes) and postcontact (segment 2; small dashes) construct stiffness. Gray circles depict visible breakpoints in construct stiffness in the long working length, coinciding with transcortical contact occurring in the bone model between 150 and 155 N.