Mechanics of Dynamic Compression Plate Application in Fracture Fixation

Abstract

Background:

Dynamic compression plating is a fundamental type of bone fracture fixation used to generate interfragmentary compression. The goal of this study was to investigate the mechanics of the surgical application of these plates, specifically how plate prebend, screw location, fracture gap, and applied torque influence the resulting compressive pressures.

Methods:

Synthetic bones with transverse fractures were fixed with locking compression plates. One side of the fracture was fixed with locking screws. On the other side of the fracture, a nonlocking screw was inserted eccentrically to induce interfragmentary compression. A pressure mapping sensor within the fracture gap was used to record the resulting pressure distribution. Plate prebends of 0 mm, 1.5 mm, and 3 mm were tested. Three locations of the eccentric screw, four levels of screw torque, and two initial fracture gap conditions also were tested.

Findings:

With increasing plate prebend, fracture compression pressures shifted significantly toward the far cortex; however, compression force decreased (P<0.05). The 1.5 mm prebend plate resulted in the greatest contact area. Increasing screw torque generally resulted in greater fracture compression force. The introduction of a 1 mm fracture gap at the far cortex prior to dynamic compression resulted in little or no fracture compression.

Interpretation:

The model showed that increasing plate prebend results in an increasing shift of fracture compression pressures toward the far cortex; however, this is accompanied by decreases in compressive force. Initial fracture gaps at the far cortex can result in little or no compression.

Introduction

Dynamic compression with an eccentric screw is a common method used in long bone fracture repair and is available in most plating systems. During this type of fixation, a nonlocking screw is placed eccentrically, and the contact of the curved underside of the screw head with the sloped hole on the plate results in lateral translation of the bone fragment and compression of the fracture site.[1–6] The goal of this compression is to achieve absolute stability, i.e. control shear forces and prevent micromotion and gap formation along the interfragmentary surfaces.[2–6] Compression facilitates primary bone healing without the need for cartilaginous callus synthesis.[2,4,7,8]

A common problem with dynamic compression plating is gapping at the far cortex (away from the plate) following screw insertion.[4,7] As a result, the far cortex may not be stable postoperatively, especially in torsion. To prevent this problem and obtain far cortex compression, surgeons often prebend plates so that they are slightly convex relative to the bone surface. AO recommends a plate prebend of 1–2 mm[4,5,9] however, little research has investigated the influence of prebend on fracture compression pressures.[10]

Another potential detriment is the presence of a fracture gap at the far cortex at the start of dynamic compression. The amount of bone translation is limited based on the geometry of the compression screw head and plate hole. Depending on this maximum translation, an initial fracture gap may not close. Unfortunately, the surgeon often cannot visualize the fracture gap well and thus may not know if adequate compression occurs. In addition to plate prebend and fracture gap, the quality of compression achieved is potentially affected by other surgical variables, including compression screw location (which plate hole is used), and torque applied to the screw.

Although dynamic compression with an eccentric screw is commonly used in fracture treatment, there have been a limited number of studies on how the above surgical factors affect fracture site compression. Several studies have used load cells positioned within the fracture site to measure fracture compression[11–14]; however, a load cell measures an overall force and cannot differentiate between near and far cortex forces and pressures. A recent study of the effect of plate prebend measured cortical strains near the fracture site but not pressures and forces within the fracture site, and it did not investigate effects of other factors[10]. The purpose of the present study was to investigate how plate prebend, screw location, fracture gap, and amount of torque applied to an eccentric compression screw influence the resulting compressive pressure distributions at the site of fracture.

Methods

Materials

Fourth generation composite Sawbones (tube 27 mm outer diameter, 7 mm wall thickness, model 3403–10, Vashon, Washington, USA) were used. Sawbones were cut to a length of 400 mm. A perpendicular, transverse cut was made at the center of each bone model to create the fracture. The fracture gap ends were smoothed with sandpaper to facilitate pressure measurements[11]. Ten-hole 4.5 mm narrow stainless steel Locking Compression Plates (LCPs) with stainless steel 5.0 mm locking screws and 4.5 mm cortex screws (DePuy Synthes, Warsaw, IN, USA) were used. In addition to a flat plate (0 mm prebend), plates with 1.5 mm and 3.0 mm prebends (Fig. 1A) were created using a plate bending press (Depuy Synthes). Plate prebend was confirmed prior to each test using 1.5 mm or 3.0 mm steel rods,[10,12] and plates were reused for multiple samples[10–13].

Figure 1.

Experimental setup and variables. (A) plate prebends; (B) compression screw locations, determined relative to the fracture plane (red line), and locking screw locations (labeled ‘L’) (C) sawbone and plate construct mounted to the fixtures on both ends (the red fixture allowed rotation, and the blue fixture allowed rotation plus free translation in the direction of the bone axis); (D) fracture gap of 1 mm (0 mm not shown); (E) close up of construct showing locations of locking screws, and compression screw—in this example in location 2; (F) example of resulting pressure distribution measured by pressure sensor located in the fracture.

Five bone-implant samples were created for each plate prebend. Due to supply chain limitations at the time, six sawbones were re-used for 2–3 samples each by rotating them such that new screw holes could be created for each sample. Each sample was tested in three compression screw locations (Fig. 1B) and four levels of torque applied to the compression screw as further described below.

Experimental Fixtures and Sensor

Each bone sample was loaded into an experimental fixture which allowed for dynamic compression of the fracture. The bones were constrained at the left and right ends using pin joints, with the compression side also allowing translation using a linear bearing (Figs 1C, Supplemental Digital Content Fig. S-1).

Calibration of the pressure sensor (#4000 sensor, Tekscan, Boston, MA, USA) was performed with a materials test machine (FlexTest 40, MTS, Eden Prairie, MN, USA) according to Tekscan recommendations. The sensor was compressed between sawbones, utilizing incremental loads until pressures near the sensor’s maximum (with sensitivity S-5) were reached.

Construct Preparation

Plates were centered over the fracture and temporarily held by ring clamps (Supplemental Digital Content Fig. S-1). For plates with prebend, the bone fragments were angled so that they were flush with and parallel to the plate. In one fragment, two 5 mm locking screws were placed in the second and fourth locking holes away from the fracture using standard locking drill guides and predrilling with a 4.3 mm drill. The two locking screws were each tightened to a torque of 4.0 Nm confirmed by digital torque wrench (Westward, Lake Forrest, IL, USA). The initial fracture gap between bone fragments, measured at the far cortex, was then set at either 0 or 1 mm (using a spacer; Fig. 1D) with a mock Tekscan sensor in place.

The bone fragments were then held in place by a temporary axial force. The compression screw hole was drilled at either the first, second, or third nonlocking hole (randomly selected) using the 4.5/3.2 mm universal drill guide in the eccentric position and a 3.2 mm drill (DePuy Synthes, Warsaw, IN, USA). A 4.5 mm cortex screw was inserted until it contacted the plate, and an initial torque of 0.5 Nm was achieved and confirmed by digital torque wrench. The temporary supports (and spacer if applicable) were then removed prior to final screw tightening.

Compression Screw Application

The calibrated Tekscan sensor was slid into the fracture gap, such that its top edge was tangential to the top of the outer edge of the near cortex immediately underneath the plate (Fig. 1E). Torque was incrementally applied to the compression screw up to 1.0, 2.0, 3.0, then 4.0 Nm.[12] After each 1.0 Nm torque increase, the torque wrench was removed, and a Tekscan pressure map was recorded (after a timed 30 second relaxation delay to allow for any small transient settling per manufacturer recommendation).

After testing to 4 Nm, the compression screw was removed, whereas the locking side was kept intact. This testing sequence was subsequently repeated two more times with the compression screws placed in the other remaining plate holes (screw locations 1, 2, 3).

Statistics

Using the recorded pressure maps, the fracture compression total force, contact area, and center of force location were determined (I-Scan v 7.60–181, Tekscan, Boston, MA, USA). (“Center of force”, the terminology used by Tekscan, is equivalent here to center of pressure.) The center of force vertical location was expressed as a percentage of the bone diameter, with 0% = near cortex immediately beneath the plate, and 100% = far cortex outer surface. For the initial fracture gap = 0 mm, these three outcomes were analyzed using linear mixed models that accounted for the two repeated factors, i.e. screw location and torque. Only the highest two torque levels were included in the analyses in order to reduce the number of post-hoc comparisons. Residual diagnostics were examined to ensure parametric modeling assumptions were met. For each outcome, there were 18 post-hoc comparisons of interest (among the three plate prebends, three screw locations, and two torque levels) which were corrected for multiple testing using Bonferroni’s method. The familywise error for statistical significance was set at 0.05. All hypotheses tests were two-sided and all analyses were conducted using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA). For comparing total force between an initial fracture gap of 0 mm vs. 1 mm, an unpaired t-test was used.

Results

Samples with a 0 mm prebend (flat) plate resulted in fracture compression pressures concentrated at the near cortex, immediately under the plate (Fig. 2). Conversely, 3 mm prebend resulted in pressures concentrated at the far cortex. The 1.5 mm prebend plate resulted in a much more even distribution of pressures across the fracture. In addition, increasing plate prebend generally led to decreased fracture compression total force (Fig. 3, Supplemental Digital Content Fig. S-2). Increased torque generally resulted in increased compression total force. There was a general trend of increased contact area using screw location 1 (closest to the fracture) compared to the other screw locations.

Figure 2.

Representative samples, having the three different plate prebends, after compression screw tightening. On the right, corresponding pressure results measured in the fracture site are shown, superimposed on an annulus representing the bone end surface. Note the typical results showing contact pressures concentrated at the near cortex (top of annulus) for the 0 mm prebend, near the far cortex for the 3 mm prebend, and a more even distribution for the 1.5 mm prebend. Differences in final angulation of the bones can also be seen.

Figure 3.

Fracture compression results derived from pressure sensor measurements, including total force, contact area, and the center of force (pressure) location. Results are shown for the middle screw location 2 only (results for the other two locations were similar) and initial fracture gap = 0 mm (at far cortex). Error bars represent ± one standard error of the mean.

Differences in fracture compression total force, contact area, and center of force location were statistically significant in many pairwise comparisons (Supplemental Digital Content Table S-1). Focusing on results for the middle screw location 2 and 4.0 Nm torque (Table 2, Fig. 3), the 3 mm prebend plate resulted in significantly lower fracture compression total force (159 ± 62 N) compared to the 0 mm prebend plate (520 ± 73 N, P=0.03). The 0 mm prebend plate also resulted in a center of force location significantly closer to the near cortex (5 ± 3% of bone diameter), compared to both the 1.5 mm prebend (61 ± 32% bone diameter, P=0.005) and 3.0 mm prebend plates (93 ± 3% bone diameter, P<0.001). Additionally, the 1.5 mm prebend generated a significant increase in contact area at the fracture site (84.8 ± 31.8 mm²) compared to both the 0 mm prebend (31.5 ± 9.1 mm², P=0.01) and 3.0 mm prebend (27.4 ± 4.0 mm², P=0.006).

Table 2.

Comparison of fracture compression between plate prebends. Results are shown for the middle (2^nd) screw location, 4.0 Nm torque, and 0 mm initial far cortex fracture gap.

Plate Prebend:	Mean (Standard Deviation)			Pairwise Comparison P-values
	0 mm	1.5 mm	3.0 mm	0 vs. 1.5	1.5 vs. 3.0	0 vs. 3.0
Total Force (N)	520 (73)	239 (229)	159(62)	0.15	1.00	0.03
Contact Area (mm2)	31.5 (9.1)	84.8 (31.8)	27.4 (4.0)	0.01	0.006	1.00
Center of Force Location (% bone diameter)	5 (3)	61 (32)	93 (3)	0.005	0.59	<0.001

For tests with a 1 mm initial fracture gap at the far cortex, in two samples no fracture compression was detected (i.e. no interfragmentary contact), even at the maximum torque. There was a strong trend in total force being less for the 1 mm initial fracture gap, compared to the 0 mm gap (P=0.07, Fig. 4).

Figure 4.

Comparison of total fracture compression force for the 0 mm and 1.0 mm initial fracture gap conditions. In the 1.0 mm group, two samples resulted in no compression force.

Discussion

Surgeons are taught that prebending a dynamic compression plate by 1–2 mm[4,9] on a straight bone prevents gapping at the far cortex during plate application. Our quantitative results are consistent with this teaching. But interestingly this shift in pressures toward the far cortex was accompanied by a significant decrease in fracture compression force (Fig. 3, Table 2, Supplemental Digital Content Fig. S-2). Total force for the 3 mm plate prebend averaged 26–42% of that for the 0 mm prebend, considering all three screw locations and maximum torque. The 1.5 mm prebend resulted in a moderated total force, but the most uniform distribution of pressures at the fracture site and largest contact areas, which are advocated by AO principles[9]. The above results were consistent across the three tested screw locations.

Our experimental model demonstrates that although plate prebend promotes far cortex compression, it may be accompanied by decreased total compression force. Compression is important for primary bone healing because it resists the interfragmentary motion that occurs with external loading, particularly under torsion or a bending load that would tend to separate the far cortex. Increased interfragmentary motion is also generally accompanied by increased plate stresses which, especially combined with poor healing, can lead to hardware failure. The location of compression is more biomechanically advantageous at the far cortex than at the near cortex due to increased distance (moment arm) relative to the plate. However, the accompanying loss in compressive force raises questions around the optimal amount of prebend.

Results from our model of dynamic compression also show that a 1 mm initial far cortex fracture gap results in little or no interfragmentary contact. For this implant system the dynamic compression has a limit of approximately 1.0 mm relative translation of the bone fragment[5,12]. Initial far cortex fracture gaps larger than 1 mm would be expected to result in no interfragmentary contact and compression, unless perhaps an additional compression screw (or articulated tension device) is used.[5] Unfortunately the surgeon may not be able to detect a lack of interfragmentary contact either visually, or by tactile feedback, because the torque being applied to the screw is similar; instead this torque is simply compressing the plate against the bone.

Our fracture compression forces are consistent with previous dynamic compression studies that used load cells.[11–14] Lucas et al.[13] reported mean compressive force of 430 N using the same plate as this study, with no prebend, and torque based on surgeon judgement. Ya’ish et al.[12] reported median force of 365 N using a similar plating system, with a 2 mm prebend and 4 Nm torque. Our findings regarding the effect of plate prebend are also supported by the findings from Ristow et al.[10] Instead of measuring pressures within the fracture site, that study measured strains on the adjacent cortex using digital image correlation, which likely have some relationship with the pressures within the fracture. In their study with synthetic femurs, Ristow et al. noted that the 1 mm prebend plate approximately matched the lateral bow of the femur, and thus were functionally similar to a 0 mm prebend on a straight bone. Their 2 mm prebend plate eliminated the far cortex gapping and resulted in a more evenly distributed strain from near-to-far cortex, which is comparable to our results for 1.5 mm prebend on a straight bone. Unlike our study, the Ristow et al. study did not examine a larger prebend that resulted in primarily far cortex compression, or study effects of screw location, torque, or initial fracture gap.

Dynamic compression with an eccentric screw results in lateral displacement of the bone fragment, but in an asymmetric manner that favors displacement or compression of the near cortex, not far cortex[4,5]. As Fig. 5 demonstrates, the initial near cortex gap size, prior to compression screw insertion, is a driving factor in determining quality of fracture compression (assuming this simplified fracture geometry and an initially closed far cortex). If the initial near cortex gap is much less than the dynamic compression translation, then the near cortex will be highly compressed and the far cortex will open with tilting of the compression screw. If the initial near cortex gap is similar to the dynamic compression translation afforded by the implant, an even pressure distribution across the near and far cortex will result. If the initial near cortex gap is larger than the dynamic compression translation, the near cortex will not close and compression will be focused at the far cortex. Several geometric factors play a role in determining this near cortex gap size before compression. In addition to plate prebend (Fig. 5), this initial near cortex gap size also depends on plate length and bone diameter. The dynamic compression translation depends on the implant system, and drill guide or precise positioning of the screw (and drilled pilot hole) within the plate hole.

Figure 5.

Effects of initial conditions on dynamic compression. (A) For a prebend of 0 mm (and 0 mm initial far cortex fracture gap), the near cortex compresses with large force but cannot be further closed. As the eccentric screw head translates further to the right, it causes the screw to tilt and leads to opening of the far cortex (red circle). (B) For a prebend of 1.5 mm, assuming the bone fragments are brought flush with the plate before compression, a gap of approximately 1 mm at the near cortex is present initially. As the screw head translates relative to the plate 1 mm, that 1 mm near cortex gap is precisely closed and balanced pressures across the fracture are obtained. (C) For a prebend of 3.0 mm, a gap of approximately 1.5 – 2 mm at the near cortex is formed in our model. The screw head translates 1 mm, which is not enough to close the near cortex gap, but induces far cortex pressures. (D) An initial gap at the far cortex 1 mm or greater may not close during dynamic compression.

For the 1.5 mm and 3 mm prebends, the compressive pressures can be attributed to the elastic recoil of the plate that has been flattened relative to its prebend state[4]. In separate preliminary finite element analysis in our lab, a construct with 3 mm prebend provides better torsional stability than one with a smaller prebend, despite the decreased compression. Future biomechanical testing is warranted to compare the postoperative stability of these constructs. In the 0 mm prebend trials only, six of the 60 trials had one Tekscan sensel that exceeded the sensor’s pressure range, meaning that total force for these trials was likely underestimated. However the one sensel represented less than 9% of the contact area in these six trials. Conversely, some local low pressures were likely too low to be detected by the sensors, such as in constructs with the 1.5 mm prebend, resulting in underestimates for contact area and perhaps total force in those situations. The Tekscan pressure range is controlled by choice of sensor type and sensitivity setting; we carefully selected these based on pilot testing with the purpose of minimizing the above competing potential errors.

The present study used composite Sawbones instead of cadaver bones. Sawbones, which were also used in previous related studies[10–15], were used with the rationale that dynamic compression plate mechanics are based primarily on the interaction between the screw, plate, and fracture gap contact, not on bone quality. However Sawbones likely have different screw-bone frictional characteristics compared to cadavers[16], and thus torque and compression force magnitudes could have been affected. Also in osteoporotic bone, higher torques may lead to local failure of bone threads, reducing the achieved fracture compression. In real bone there is also potential loss of compression over time due to viscoelasticity and other properties, but these effects were relatively small in a study with ovine bone[17].

The study used a simplified transverse, planar fracture model. Oblique or more complex fractures could alter results and more study in that area is warranted. The study also focused on the effects of one dynamic compression screw, whereas clinically at least one additional screw would be placed after the dynamic compression screw, on the same side. Additional locking or neutral nonlocking screws can be applied in a variety of configurations but are not intended to alter the fracture compression state. Additional compression screws, placed on the same side or alternating sides of the fracture, can increase fracture compression forces[12–14,18]. Other limitations included the re-use of sawbones and implants. We believe this is acceptable in this application because the variability associated with the manual plate application process is likely dominant compared to any variability in sawbones and plate material properties across samples. (Multiple sawbones often come from the same homogenous lot of material, and geometries are nearly identical.) Although this study did not test additional screw configurations, plate lengths, plate and screw sizes and materials, implant systems, and bone geometries, we believe our main study conclusions would largely hold true for these other conditions, although the specific values may change.

Conclusion

This biomechanical study of dynamic compression plating is the first to examine the effects of plate prebend, screw location, fracture gap, and applied torque on resulting compressive pressures. Our experimental model used a planar transverse fracture and focused on the compression due to a single eccentrically placed screw. Increasing plate prebend was effective in shifting fracture compression pressures toward the far cortex; however, this was accompanied by decreases in compressive force. Initial fracture gaps at the far cortex resulted in little or no compression.

Table 1.

Experimental conditions tested. In each sample and for each screw location, compression screw torques of 1.0, 2.0, 3.0, and 4.0 Nm were tested.

Sample Size	Plate Prebend (mm)	Initial Fracture Gap (Far Cortex) (mm)	Screw Location
5	0	0	1, 2, 3
5	1.5	0	1, 2, 3
5	3.0	0	1, 2, 3
5	1.5	1	2

Highlights:

• This study investigated the effects of surgical variables on fracture compression pressures.

• Plate prebend was effective in shifting compression toward the far cortex.

• However, plate prebend also led to decreases in compression force.

• Initial fracture gaps at the far cortex resulted in little or no compression.