Experimental Mechanics (2007) 47:595–607 DOI 10.1007/s11340-007-9056-6

Measurement of Orthopedic Cortical Bone Screw Insertion Performance in Cadaver Bone and Model Materials B. Kincaid & L. Schroder & J. Mason

Received: 12 June 2006 / Accepted: 30 April 2007 / Published online: 2 June 2007 # Society for Experimental Mechanics 2007

Abstract In this study, a method was developed for distinguishing insertion and driving performance between different self-tapping bone screw designs. To measure screw starting load, torque and displacements, a test apparatus was developed utilizing a modified drill press with the capability to measure in-line torque, axial compression load, and axial displacement. Specimens were inserted into cadaver bone to measure a baseline response and a bone analog was developed to mimic the bicortical application of screws in the cadaver model. Recorded data could be used to measure a distinguishable screw starting load and the torque for the insertion of bone screws. The results were similar between the cadaver bone and the bone analog. The average insertion load ranged from 5.4 to 64.5 N in cadaver tests and 9.0–41.0 N in the construct tests. Average first cortex insertion torques ranged from 0.53 to 0.66 N-m in the cadaver tests and 0.29–0.32 N-m in the construct test. Average second cortex insertion torques ranged from 0.70 to 1.03 N-m in the cadaver tests and 0.60–0.63 N-m in the construct tests. This method successfully illuminated differences between several different selftapping screw designs and was also successfully employed to determine the impact of design and manufacturing methods on screw performance. An interesting finding in this study is that axial starting load is very sensitive to screw tip design whereas insertion torque is not. Keywords Bone screw . Self-tap . Load . Insertion torque . Cadaveric analog

B. Kincaid (*) : L. Schroder : J. Mason (SEM member) Corporate Research and Development, Zimmer Inc., PO Box 708, Warsaw, IN 46581, USA e-mail: [email protected]

Introduction Cortical bone screws are used to affix bone plates and other orthopedic devices to bone. Typically in trauma cases, where bone has been broken traumatically, these screws are twice required to penetrate the hard outer portion, or cortex, of long bones such as the femur, tibia, radius, humerus, ulna and fibula. In these cases, screws are inserted with a surgical power drill either through a stab wound in the soft tissue or through a large incision and are required to penetrate both cortical walls of the bone. An example application is shown in Fig. 1. Many screws are self-tapping or both self-drilling and self-tapping. Self-tapping screws require the drilling of a pilot hole before insertion of the screw. Thus, these require two steps for insertion. Self-drilling/self-tapping screws eliminate the need for a pilot hole, and therefore only require one step for insertion, but sometimes they suffer from difficulty penetrating the second cortex. This is due to pitch mismatch in the threads on the drill tip and screw shaft. Most self-drilling/self-tapping screws can not be used bicortically because they strip the threads created in the first cortex when drilling the second cortex. The method described here is applicable to self-tapping screws only, which are more commonly used and do not suffer from this limitation. In general, surgeons prefer a self-tapping screw that takes minimal axial force to self-tap and minimal amount of torque to drive the screw through both cortical walls of the bone yet possess substantial “holding” power. Surgeons frequently describe this attribute as “bite” and have been known to say, “the screw bites well” when they are happy with its performance. This is a very qualitative evaluation that could benefit from a more quantitative, laboratory measure of screw performance. To date the only industry

596

Fig. 1 (a) Example of 3.5 mm bone screw and periarticular femoral plate. (b) an x-ray of application of bone screws to a femur in a compression hip screw

standard for laboratory evaluation of these performance requirements is ASTM F543.02 [1], which is lacking because it does not address either axial self-tap force or the bicortical application of screws. Further, this standard does not thoroughly evaluate the performance of the screw in a test material shown to behave like human cortical bone. When designing such screws it would be useful to be able to measure the total performance in order to improve the usefulness of the screws and improve surgeons’ response. Numerous researchers have reported on the topic of screw insertion into cadaveric bone or bone substitutes. In all of these studies, the authors noted the high degree of variability of cadaveric specimens and cited this as a limitation of their studies or motivation for conducting similar experiments in bone models or analogs [2–17]. Many of these studies were focused on areas other than long bones such as maxillofacial screws and spinal pedicle screws where screw geometries and cortical bone properties differ from cortical bone screws utilized in long bones [2–5, 9, 14, 16]. Several investigators evaluated screw insertion into cadaveric long bones but, unfortunately, the focus of these studies was to evaluate screw failure, via strip, overtorque or pull/push out and thus very little effort is devoted to the measurement and reporting of insertion torque [6, 8]. Furthermore, these studies often used unicortical insertion models, with a few noted exceptions where bicortical fixation was investigated [7, 9, 15, 17]. Four studies focused on insertion torque as well as differences between unicortical and bicortical fixation and the need for uniform bone analogs for experimental purposes [7, 10, 11, 15]. Lawson et al. [7] inserted 4.5 mm self-tapping cortical screws unicortically and bicortically into cadaveric tibia and reported correlations between peak insertion torque and cortical wall thickness. The peak

Exp Mech (2007) 47:595–607

torque was higher for bicortical insertion. These authors also investigated substituting cadaveric specimens with lamb femora but still found scatter in their data. Koistinen et al. [10] inserted 2.7 and 3.5 mm self-tapping cortical screws bicortically into cadaveric femora and noted correlations between peak insertion torque and the degree of bone mineralization (Bone Mineral Density or BMD) as measured by peripheral quantitative computed tomography (pQCT). The authors also noted the relationship of decreasing BMD with increasing age and experimented with surface coatings in an effort to reduce screw insertion forces. In a follow on study, Koistinen et al. [11] investigated inserting the same 2.7 and 3.5 mm self-tapping cortical screws into a variety of bone analogs including bovine bone, wood and Teflon. Unfortunately, no correlations were made between the peak torques obtained in the cadaveric study and the various bone analogs. Furthermore, no effort was made to reproduce cadaveric bicortical insertion characteristics in the bone analogs. Ansell and Scales [15] were among the first to investigate factors which affect screw insertion torques. They examined 4.0 mm self-tapping (ST) and non self-tapping (NST) cortical screws by inserting them into cadaver femora with unicortical and bicortical purchase. Differences in the performance between the ST and NST styles of screws were documented. They also showed that bicortical insertion torques are approximately double unicortical torques. These authors noted a high degree of variability between cadaveric specimens and factors such as age, gender and stature are cited as the main reasons for the data scatter although no correlation between insertion torque and these factors could be definitively established. They also cited the need for a bone analog and experimented with various materials in an attempt to reproduce bicortical insertion behavior. Ansell and Scales [15] were the only investigators who experimented with building a bicortical bone model and directly comparing their results back to values measured in human cadaver long bones. During the course of examining insertion torques of 4.0 mm ST and NST cortical screws, these authors developed bone analogs to reproduce the values measured in cadaveric femora. Their bone analog, made of Delron with 5.5 and 8 mm thick “cortical walls,” showed similar behavior to the cadaver data. Unfortunately, the author’s insertion apparatus and data acquisition were limited by the state of the art equipment available to them at the time. Screw insertions were performed by hand using a hand torque wrench and analog measurement devices that captured peak values only. Koistinen et al. [10, 11] noted that many inaccuracies can be introduced into the measurement without precise, repeatable alignment of the driver, screw and pilot hole as provided by a standardized test apparatus. Furthermore, these same authors noted the improved accuracy of continuous data acquisition and the

Exp Mech (2007) 47:595–607

added friction inherent in the “start/stop” associated with hand insertion, as utilized by Ansell and Scales [15]. This “stick friction” produces an increase in peak torque values as a result. Very little information was found in the literature regarding the self-tapping axial load, the axial force necessary to induce tapping of ST style screws. One group investigated the “pressure,” or stress in the screwdriver shaft, needed to drive ST screws via an instrumented screwdriver but the results did not indicate whether this was a starting load (self-tap force) or a steady state load [18]. To the best of the authors’ knowledge, there has been no reported investigation of the self tapping forces in trauma applications of ST bone screws. In the present study, we attempt to add to the body of knowledge regarding ST screw insertion by measuring the axial force necessary to induce self-tap of ST style screws. Furthermore, our bone model adds a “cancellous” layer to study the behavior of ST screws during transition between materials of dissimilar mechanical properties, further mimicking in vivo conditions. Our study also improves upon the work of Ansell and Scales [15] through the use of the recommendation made by Koistinen et al. [11] and utilizes a standardized apparatus that allows precise, repeatable alignment of the driver, screw and pilot hole as well as continuous rotation and uniform insertion of the screws into the bone model. Furthermore, our apparatus employs a continuous data acquisition system for more precise reporting of measurements. Finally, our study directly compares results obtained in cadavers and to those obtained in the bone model utilizing the same standardize test apparatus.

Experimental Method The experimental effort was divided into two portions: (1) methods to measure the insertion load, advancement torque Fig. 2 Test apparatus for determining self-tapping performance; (a) schematic and (b) actual

597

and screw displacement during screw insertion in both cadaver bone and bone constructs, and (2) the development of a bone model construct that properly mimicked a human long bone during bicortical screwing and tapping. In addition to a description of the screws used, these two portions of the experimental effort are discussed below. Screw Starting Loads To measure screw starting loads, torques and displacements, a test apparatus was developed utilizing a modified drill press with the capability to measure in-line torque, axial compression load, and axial displacement during the process of screw insertion and advancement. For a schematic of the system, see Fig. 2. A drill press was retrofitted with a DC motor and speed control to allow precise control of the rotational speed. The control voltage source allows for rotational speeds of 30 rpm (approximately the speed of manual screw insertion) to 300 rpm (approximately the speed of a surgical power drill). In tests reported here, we use 30 rpm to represent hand insertion. Later we compare our results to surgeon feedback during hand insertion of the screws. To perform the test, the screw is aligned with a pre-drilled pilot hole and lowered onto the bone or bone construct by way of a rack and pinion gear connected to a pulley. See Fig. 2. The pulley has two weights acting in opposition to each other; one is a counter weight that exists to ensure that the initial load on the bone screw is zero, the other is a lead shot receptacle which can be filled to increase the load or emptied to decrease the load as desired by the operator. The load is transmitted through the pulley, the rack and pinion gears, down the drill spindle and into the bone construct or bone where the load is measured on the back face of the bone or construct by a strain gage based, 220 N axial force transducer (Sensotec Model 41, Columbus OH.). This load

598

cell is attached beneath the bone or construct in a threaded hole in the vise fixturing so that compressive or tensile loads are transmitted through the bone or construct to the cell. The spindle is equipped with a linear voltage displacement transducer (LVDT) (Pickering & Co., Model 7310-V2-AO, Plainview, NY.) to measure its linear vertical displacement relative to the drive motor housing during screw insertion. The spindle is also equipped with a 5 N-m rotary torque transducer (Lebow Model 1702-103, Troy, MI.) which measures the torque in the spindle by magnetoelastic strain. This strain measurement is then converted to a torque, based on a calibration performed on an Instron 1125 tension–torsion testing machine with ten points taken on increasing torque and ten points taken on decreasing torque. The load cell was similarly calibrated on the Instron 1125. The LVDT was calibrated using gage blocks inserted directly in the drill press apparatus. All three data, axial force, axial displacement and axial torque, are recorded to a laptop computer throughout the test. A custom software interface was developed in the software program LABTECH (Laboratory Technologies Corp., release 10.0, Wilmington, MA.) to acquire data at 30 Hz and provide real time display of all three measured responses. After the screwdriver has been mounted in the drill press and properly aligned with the pilot hole, the system must be balanced so that there is approximately zero Newton axial force acting on the screw. This is achieved by adjusting the counter weight. Next, the screw is stood upright in the pilot hole and engaged with the screw driver. Then the data acquisition is triggered by the operator before activating the DC motor. At this point, the operator slowly adds lead shot to the axial load weight while monitoring the software display of the axial load, axial torque and displacement. This increases the axial load on the screw in the pilot hole until self-tapping starts. The starting point of selftapping is defined with the example of typical results shown in Fig. 3. Once self-tapping begins no further load is applied and the screw is allowed to penetrate the construct. The test is stopped once the desired level of penetration is achieved. In this investigation, the test was stopped after complete bicortical purchase had been achieved as denoted by a rapidly decreasing torque. The initiation of self-tapping is indicated in the measured parameters. Typical results are shown in Fig. 3. In this test, screws are brought into in contact with the predrilled bone analog at a small axial preload of 1 N and rotated at 30 rpm. Then, the axial load is increased at about 2 N/s, as controlled by the operator watching the load signal in real time, until a steadily increasing torque and increasing axial displacement are observed. In the figure, it can be seen that the axial load initially increases quite rapidly. The torque increases as well as the screw begins to penetrate the specimen, but only when

Exp Mech (2007) 47:595–607

MAX = “starting load”

Axial Load

Unique state indicates self tap has occurred

Torque

Axial Displ.

Fig. 3 Typical result from the starting load evaluation

the self-tapping begins and the screw threads engage the material does the torque increase more rapidly. This corresponded to a rapid drop in axial load and a more rapid increase in the axial displacement due to the screw pulling itself into the bone or bone analog. This load then reduces to zero as the screw “pulls” itself through the bone because the “pulling” results in a force greater than or equal to the applied load but pointed in the opposite direction Therefore, at the onset of self-tapping, each of the three signals measured, axial load, axial displacement and axial torque, changes abruptly giving clear indication of the beginning of this event. The maximum load observed at initiation of self-tap, as indicated in Fig. 3, is reported as the “starting load” in this investigation. An example of the cadaver test utilizing the femur is shown in Fig. 4. Where possible in these cadaver tests, matched pairs of limbs were used. That is, screws were inserted on both right and left matching limbs from the same cadaver in order to limit the effects of the natural variation between cadavers. All cadaver specimens were excised from fresh-frozen cadavers and stripped of all soft tissue. Specimens were then wrapped in saline soaked gauze and frozen until tested. All specimens were allowed to thaw at approximately 4.5°C wrapped in the gauze for a minimum of 24 h prior to initiating any testing. The gauze was removed just before testing began and any remaining soft tissue, including the periosteum, was removed. Before experiments were performed on cadavers, the quality of the bone was assessed. It was expected that the forces and torques measured would correlate with the quality of the bone as previously reported in literature [4, 5, 9, 10, 16]. In these experiments, the term “quality” in relation to the bone refers to the mechanical properties relevant to screw insertion. Among these are bone density, strength, stiffness and toughness. Therefore, prior to testing, the

Exp Mech (2007) 47:595–607

599

Fig. 4 Cadaver experiments are shown. In (a) a typical stripped cadaveric femur is shown with the “insertion zone” denoted. In (b) the hardness measurement is performed on a cadaver femur. In (c) the predrilled femur is tested using the experimental apparatus devised here

degree of mineralization (bone mineral density or BMD) of the cadaveric specimens was assessed through standard Dual Energy X-ray Absorptiometry (DEXA) scans of the femoral neck and head (Lunar Corporation, DPX-Alpha, Madison WI.). As per the machine manufacturer’s recommendation, 150 mm of polymethylmethacrylate was placed beneath the bones to mimic soft tissue (Van Skoyk, personal communication).1 Hardness tests were also performed at each screw insertion site along the length of the cadaver bone using a hand-held Shore D Durometer hardness tester (PTC Instruments, Model # 409, Los Angeles, CA.). The hardness measurement is meant to be a quick and inexpensive way to get a gross approximation of the quality of the bone [19]. Screw insertion sites were limited to the diaphysis (shaft) of the bones. Therefore, prior to testing, markings were made on the specimen approximately 114 mm down from the greater trochanter and 114 mm up from the most distal portion of the condyles to denote the “insertion zone” [see Fig. 4(a)]. The specimen was positioned in the load cell/vise assembly [Figs. 4(b) and 4(c)] to allow lateral to medial screw insertion as is common in clinical practice. Insertions sites were marked from proximal to distal along the length of the bone with a spacing of approximately 8.75 mm between them or two and a half screw diameters and followed the natural curvature of the femur to ensure cortical– cancellous–cortical insertion always occurred. Three hardness readings were taken at each insertion site and averaged to report the Shore D hardness at the insertion site. See Fig. 4(b). Once the hardness tests were complete, a pilot hole was drilled utilizing the appropriate sized drill. See Table 1. The load cell/vise assembly was then immediately fixtured in the test apparatus and the screw test performed. This methodology ensured accurate alignment of the pilot

1

Installing technician, Peter Van Skoyk, 727-409-4403, peterfvs@ mac.com.

hole, screw and driver. The screws were inserted in a staggered order down the length of the bone to account for natural variations in bone hardness and cortical wall thickness that occurs along the length. In this way screws from one type or design were not all inserted in the same area of the bone. The scope of the cadaver work was limited to 3.5 mm cortical screws, the most common size used clinically as determined through sales volume [20]. All the screws examined are described in Table 1. Starting load, torque and vertical displacement were recorded on a laptop computer throughout the test giving results similar to those seen in Fig. 3. Bone Model Constructs Bone models were constructed from solid polymer sheets and polymer foam to represent the cortex and canal of the long bone, respectively. The results from the cadaver tests were used to design the construct such that it gave similar insertion torques and axial insertion loads when compared to cadavers. After some investigation, the following materials were chosen &

Cortex—CELCON® Acetal Co-polymer (Ticona, M25 Grade, Florence, KY) ➢Density (1.41 g/cm3) [21] falls in range of healthy human cortical bone as reported in literature (1.2– 3.0 g/cm3) [22, 23] ➢Shear Strength (66 MPa) [21] falls in range of healthy, human cortical bone as reported in literature: 48–69 Mpa [23] ➢Compressive Strength of 110 Mpa [21] falls in the range of healthy, human cortical bone (measured transversely) as reported in literature: 106–133 Mpa [23] ➢Avg. Hardness (∼85 Shore D) matches human cortical bone average 86.7 (±1.9) Shore D reported in literature [24]

600

Exp Mech (2007) 47:595–607

Table 1 Description of screws examined in the cadaver experiments Specimen ID

n

Catalog number

Major diameter (mm)

Minor diameter (mm)

Thread pitch (mm)

Pilot hole (mm)

Zimmer Periarticular Cortical 2.7/3.5 Zimmer Cortical 3.5 Zimmer Periarticular Locking 3.5 Smith & Nephew Cortical 3.5 Smith & Nephew Locking 3.5 Synthes Cortical 3.5 Synthes Locking 3.5

10 10 10 5 5 5 5

00-2348-040-35 00-4835-040-01 00-2359-040-35 not available not available 204.84 212.117

3.5 3.5 3.5 3.5 3.5 3.5 3.5

2.4 2.4 2.7 2.6 2.7 2.4 2.8

1.2 1.2 1.0 1.2 1.3 1.2 0.8

2.5 2.5 2.7 2.7 2.7 2.5 2.8

&

Cancellous bone or canal—FR3720 Last-o-Foam polyurethane foam (General Plastics, Tacoma, WA) ➢FR-3720 foam density (0.31 g/cm3) [25] falls in range of healthy, human cancellous bone as reported in literature (0.25–0.50 g/cm3) [22, 23] ➢Rigid, closed-cell variety used as industry standard test media for screw testing [1, 26]

Bone model constructs, shown in Fig. 5, consisted of two sheets of 6 mm thick CELCON sandwiched around a 14 mm core of Last-o-Foam FR3720. The construct is bonded together using a thin layer of Gorilla Glue® (The Gorilla Glue Company, Cincinnati OH.) and allowed to cure under a 50 N deadweight for 24 h prior to use. The cortical layer thickness was deemed to be the most crucial to insertion torque performance and therefore the thickness was chosen to mimic the measured torque response of the cadaver experiments as reported by Ansell and Scales [15] as well as independent experiments carried out by the authors. The cancellous layer thickness was also chosen to mimic the response seen in the cadaver experiments. A comparison of the measurements on cadaver bone and bone construct is given in the Results section.

Fig. 5 Bone model construct. The CELCON “corticies” around the foam “canal”

Screw Performance In this investigation, three common sizes (2.7, 3.5 and 4.5 mm) of five different screw designs from two manufacturers (Zimmer Periarticular Locking Screws (ZPLP), Zimmer ZPS Cortical Screws (ZPS), Zimmer Periarticular Cortical Screws (Peri), Zimmer ECT® Cortical Screws (ECT) and Synthes LCP Locking Screws (S-L) were utilized (Table 2). The three sizes chosen are the most common overall as defined by sales [20]. In the first round all these sizes were used; in later tests interest was focused on 3.5 mm screws. The 3.5 mm screw is used in fixation of fractures of the femur, tibia, humerus and sometimes the ulna and radius. The locking screws have slightly larger minor diameters than the cortical screws. The ZPS, Periarticular and ECT screws have design differences in the geometry of the threads as well as manufacturing differences.

Results and Discussion Cadaver Measurements In total, 50 screw tests were performed on the left and right femur from one cadaver and on the left femur of a second cadaver. Shore D hardness evaluations were performed on the lateral side when measuring along the length of the femur from the proximal to distal. The average of three hardness measurements at each insertion location were measured and the values obtained for each insertion site were averaged to report an average hardness value for each cadaveric specimen as well as a composite (pooled) average hardness value for all data. The cadaver specimen information including gender, age, and cause of death (C.O.D.), DEXA BMD results and average Shore D hardness values are presented in Table 3. Scatter plots detailing the hardness evaluations for each cadaver specimen are presented in Fig. 6. In general, no significant differences in hardness values were observed between the osteoporotic specimen and the non-osteoporotic specimens, as first hypothesized. One

Exp Mech (2007) 47:595–607

601

Table 2 Description of the screws examined in the bone model construct experiments Specimen ID

n

Catalog number

Major diameter (mm)

Minor diameter (mm)

Thread pitch (mm)

Pilot hole (mm)

Zimmer Zimmer Zimmer Zimmer Zimmer Zimmer Zimmer Zimmer Zimmer Zimmer Synthes Synthes

10 10 10 10 10 10 10 10 10 10 10 10

00-2359-040-37 00-2359-040-35 00-2359-040-40 00-4827-040-01 00-4835-040-01 00-4840-040-01 00-2347-020-40 00-2347-023-40 00-2351-040-00 00-2319-041-00 212.117 222.594

2.7 3.5 4.5 2.7 3.5 4.5 3.5 4.5 3.5 4.5 3.5 4.0

1.9 2.7 3.4 1.9 2.4 3.0 2.4 3.0 2.4 3.0 2.8 3.0

1.0 1.0 1.0 1.0 1.2 1.75 1.2 1.75 1.2 1.75 0.8 1.75

2.0 2.7 3.7 2.0 2.5 3.2 2.5 3.2 2.5 3.2 2.8 3.0

Periarticular Locking Periarticular Locking Periarticular Locking ZPS Cortical ZPS Cortical ZPS Cortical Periarticular Cortical Periarticular Cortical ECT® Cortical ECT® Cortical Locking Locking

interesting observation regarding the osteoporotic specimen (6485L) was that the hardness values were lower near the extremes of the insertion zone and tended to increase slightly approaching mid-diaphysis. This trend was not as apparent in the non-osteoporotic specimens. In addition, specimen 080306L and 080306R had significantly different hardness values despite being from the same donor and having approximately the same BMD. At present, there is no explanation for this observed difference. The average Shore D hardness values obtained on the three specimens were 75.7 (±3.2), 84.4 (±2.8) and 83.9 (±3.2) with a composite (pooled) average of 81.6 (±4.9). This value was considered low when compared to measurements performed earlier in the lab on younger, healthier cadaver femurs [24] where the average was closer to 86 Shore D. This is consistent with other reports that hardness decreases slightly with age [19]. Correlations between average BMD and average hardness was determined utilizing the non-parametric Spearman rank-order (Rs) calculation. The Spearman correlation was utilized to account for the relatively small sample size (n=3) and likelihood of non-normality of the data. The Spearman correlation coefficient was found to be Rs =0.50. This value may indicate a moderate correlation between hardness and

BMD, however, in the opinion of the authors, no definitive correlation could be established due to the minimum number of available BMD data points (n=3) and the difference in anatomical measurement sites, i.e. the femoral neck (BMD) vs. the diaphysis (hardness). The results for the typical load and torque measurements (ZPS specimen #4) on the cadaver are shown in Fig. 7. For this specimen, the starting load can be seen to be about 10.3 N. The composite average starting load for the 50 tests performed was 23.5±10 N. As can been seen in the figure, the torque exhibits two peaks corresponding to insertion into the first and second cortex respectively. The peak torques ranged from 0.53 to 0.67 N-m for the first cortex and 0.68–1.03 N-m for the second cortex with composite averages of 0.61±0.05 N-m and 0.84±0.16 N-m, respectively, for the 50 tests performed. A complete summary of the cadaveric test data is seen in Table 4 and Fig. 8. The insertion torque results obtained in this portion of the study are consistent with some values previously reported in literature. Ansell and Scales reported averages ranging from 0.10 to 1.0 N-m for 4.0 mm ST screws inserted unicortically and 0.34–2.0 N-m when the screws were inserted bicortically [15]. The values are higher than those reported by Koistinen et al. [10] who reported average, peak insertion torques

Table 3 The cadaver specimen information Specimen ID

Gender and age

C.O.D.

Neck BMD (g/cm2)

T-scorea

Average shore D hardness

080306L 080306R 6485L

Female, 82 years old Female, 82 years old Female, 82 years old

Heart disease Heart disease Cardiac arrest

1.049 1.066 0.565

0.57 0.72 −3.46

75.7 (3.2) 84.4 (2.8) 83.9 (3.2)

Hardness values are presented as the mean (±SD). World Health Organization (WHO) t-scores of −2.5 and lower are considered osteoporotic.

a

602

Exp Mech (2007) 47:595–607

Fig. 6 Hardness measurements along a cadaveric femur. Each point represents the average of three measurements

080306L Shore D Hardness Results

080306R Shore D Hardness Results

92

90

90

88

88

86

86

Shore D Hardness

Shore D Hardness

92

84 82 80 78 76

84 82 80 78 76

74

74

72

72

70

70 68

68 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16

0 1

2 3

4 5

92

6 7

8 9 10 11 12 13 14 15 16 17 18

Screw Insertion Site

Screw Insertion Site

6485L Shore D Hardness Results

90

Shore D Hardness

88 86 84 82 80 78 76 74 72 70 68 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Screw Insertion Site

ranging from 0.36 to 0.45 N-m for 3.5 mm ST screws inserted unicortically into a cadaver femur. In both cases, however, the values are the same order of magnitude. Clearly, differences in starting load performance exist between the various screw designs evaluated. However, due to the large scatter in the data, inherent in cadaveric testing, establishing those differences to a statistically significant level was not practical. Typically, it would be difficult to gather a large enough number of cadaver bones to establish statistical significance in these tests, if significance could be established at all. Specifically, differences in insertion torque are present, particularly in the 2nd cortex, but large 1.0

24

20

Axail Load (N) Axial Displ (mm)

18

Load Displ Torque

2nd Cortex

0.9 0.8

1st Cortex 0.7

16

0.6

14

0.5

12 Cancellous 10

0.4

8

Torque (N-m)

22

0.3

6 0.2

4

0.1

2 0

0.0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46

Time (sec)

Fig. 7 Load, torque and displacement as a function of time for a cadaver specimen

data scatter prevents establishing those differences to a statistically significant level. No correlations could be established between surface hardness and the three measured responses. Correlation coefficients were calculated for each of the measured parameters and surface hardness utilizing a pooled (hole by hole) data set. Pearson’s correlation coefficient’s were found to be R=−0.11, R=0.22 and R=−0.18 for Starting Load, 1st Cortex Torque and 2nd Cortex Torque, respectively, indicating no correlation exists between the three measured parameters and surface hardness. Similarly, no correlation was found between the BMD and any of the three measured parameters. These measurements are useful for estimating the axial load and torque during screw insertion in surgery. However, for testing screw designs, they suffer from some limitations. The first downfall of in vivo data is the variation of bone’s mechanical properties from specimen to specimen. Depending on the age, weight, health and gender of the cadaver, the bone hardness can be quite low and thus the loads and torques measured are likewise low. Additionally, with torque data, cortical wall thickness must be considered. The thicker the bone, the larger the peak torque required to penetrate it. Thickness varies over length of long bones as well as from specimen to specimen. In an attempt to minimize the effect of these variations on the average measured here, specimen insertion location was varied to assure random cortical wall thickness for all tests. However,

Exp Mech (2007) 47:595–607

603

Table 4 Summary of the cadaveric test results Specimen ID

Cadaver specimen

Average starting load (N)

Average 1st cortex torque (N-m)

Average 2nd cortex torque (N-m)

Peri 2.7/3.5×40 mm ZPS 3.5×40 mm ZPLP 3.5×40 mm S&N-C 3.5×40 mm S&N-L 3.5×40 mm S-C 3.5×40 mm S-L 3.5×40 mm

080306L 080306R 6485L 6485L 080306L 080306R 6485L

21.4 (15) 16.9 (10) 9.0 (1.7) 27.1 (4.3) 25.2 (6.2) 23.5 (9.5) 41.4 (13.5)

0.61 0.60 0.64 0.67 0.53 0.55 0.66

1.03 0.68 0.70 0.96 0.78 0.70 1.03

(0.17) (0.09) (0.13) (0.13) (0.06) (0.19) (.022)

(0.14) (0.10) (0.15) (0.16) (0.12) (0.21) (0.33)

Values are presented as the mean (±SD) of the maximum value during screw installation.

wall thickness still contributed to the uncertainty and scatter in the results. Hence, a standardized construct replicating screw behavior in a long bone was developed. Construct Measurements The construct was developed to mimic the cadaver tests, and to provide a uniform test substrate and allow for differentiating the performance of various screw designs. Typical results of measured torques and starting loads in the construct are shown in Fig. 9(a) (torque) and Fig. 9(c) (load). In Fig. 9(a), one can first note that the relative shape of the torque curve agrees with the shape of the torque curve measured in the cadaver in Fig. 9(b). The magnitude, although not in perfect agreement, agrees acceptably with the cadaver measurements. The construct test was very successful in distinguishing between different screw designs. Axial insertion load measurements for twelve different commercially available screw designs are shown in Table 5 and Fig. 10(a). There are several surprising results in this figure. First while some of the designs had an increase in insertion load with diameter, others did not show any variation with diameter and still others showed a decrease in starting load with increasing diameter. Next, the starting loads ranged over an order of magnitude, from 5 to 65 N. Next, while the

Fig. 8 Summary of cadaveric starting load and insertion torque

Zimmer ZPS cortical screw and the Zimmer ECT cortical screw are the same tip design, the results for the 3.5 mm screw from each of these sets are very different. This indicates that some aspect of the manufacturing of the screws, separate from the design, may be affecting their performance. Further, the Zimmer Peri-locking screw had the lowest starting load of all the screws tested at each of the diameters investigated. The results of the various designs were compared to the same size Zimmer ZPLP screw utilizing a 1-tailed, two-sample t-test at the 0.05 level. Prior to conducting the t-test, assumptions regarding the equality of the variances were confirmed utilizing an Ftest. The appropriate form of the t-test was then applied based on the findings of the F-test as denoted in Table 5. Differences were statistically significant (p<0.05, 1tailed t-test (α=0.05)) for all designs with the exception of the 3.5 mm ECT when compared to the same size Zimmer ZPLP screw and are denoted in Fig. 10(a). Peak torque measurements for the two commercially available screw designs tested in the construct in this study, the Zimmer ZPS 3.5 mm (ZPS) and Synthes 3.5 mm cortical (S-C) screws, are shown in Fig. 10(b). Ten specimens (n=10) of each design were tested. The average first cortex peak torque was 0.32±0.04 N-m for the ZPS design and 0.29±0.01 N-m for the S-C design. Second cortex average peak torque values were 0.63±0.06 and

604

Exp Mech (2007) 47:595–607

Fig. 9 Construct (a) torque and (c) load compared to cadaver (b) torque and (d) load

CADAVER 1.0

0.9

0.9

0.8

0.8

0.7

0.7

Torque (N-m)

Torque (N-m)

CONSTRUCT 1.0

0.6 0.5 0.4 0.3 0.2 0.1

0.6 0.5 0.4 0.3 0.2 0.1

0.0

0.0

0

10

20

30

40

50

60

0

10

20

Time (sec)

(a) 12

12

10

10

Axial Load (N)

14

Axial Load (N)

40

50

(b)

14

8 6 4 2

8 6 4 2 0

0 0

1

2

3

4

Time (sec)

(c)

0.60±0.02 N-m, respectively, for the two designs. No statistical differences (p>0.05) were found between the performance of the two designs in either the 1st or 2nd cortex utilizing a 1-tailed, two-sample t-test at the 0.05 level (variances unequal). The composite averages for the first and second cortex were 0.30±0.02 and 0.61±0.02 N-m, respectively. Not surprisingly, given the agreement in torque measurements in Fig. 9, the results are consistent with the cadaver tests. In all cases the torque required to traverse the second cortex is about twice the torque required to traverse the first cortex, as observed by Ansell and Scales [15]. This is logical, the torque is mostly due to friction in the threads and the area of friction roughly doubles when going from one cortex to two. Separate experiments performed by the authors on a one-cortex construct with double the thickness (12 mm) show that the torque scales linearly with construct thickness.

Fig. 10 Summary of construct results: (a) starting load and (b) insertion torque

30

Time (sec)

5

6

7

8

0

1

2

3

4

5

6

7

8

Time (sec)

(d)

The overall starting load performance in the construct tests was compared to the results obtained previously in cadavers. The composite average of the start load for the 120 tests performed in the construct (28.1±6.2 N) agrees well with the composite average obtained previously in the cadavers (23.5±10 N). Differences between the averages are not statistically significant (p=0.29, 1-tailed t-test, equal variance (α=0.05)) indicating the construct is a good predictor of in vivo screw starting loads. A direct comparison of the axial starting load measured in the construct to that measured on the cadavers can be performed for three screw designs; the ZPS 3.5 mm diameter screw, the ZPLP 3.5 mm screw and the S-L 3.5 mm screw. Those results are shown in Fig. 11. The cadaver results are shown in gray, the construct results for the same design are shown in blue. We see that while the values differ significantly for some designs, they do not for others. This is

Exp Mech (2007) 47:595–607

605

Table 5 Summary of the construct starting load test results Specimen ID

n

Average starting load (N)

p-value compared to same size ZPLP

ECT 3.5 mm ECT 4.5 mm ZPS 2.7 mm ZPS 3.5 mm ZPS 4.5 mm Peri 3.5 mm Peri 4.5 mm ZPLP 2.7 mm ZPLP 3.5 mm ZPLP 4.5 mm S-L 3.5 mm S-L 4.0 mm

10 10 10 10 10 10 10 10 10 10 10 10

15.1 (14) 64.5 (24) 15.0 (3.1) 43.3 (10.4) 61.7 (11.8) 35.4 (8.1) 23.2 (9.1) 6.5 (2.5) 9.7 (3.5) 5.4 (3.6) 29.2 (3.8) 28.0 (9.4)

0.13b <0.05b <0.05a <0.05b <0.05b <0.05b <0.05b N/a N/a N/a <0.05a <0.05b

Values are presented as the mean (±SD) of the maximum value during screw installation. a 1-tailed t-test: two-sample assuming equal variances at the 0.05 level. b 1-tailed t-test: two-sample assuming unequal variances at the 0.05 level.

deemed acceptable since the goal of the construct is to generate reproducible measurements in the same range as cadaveric measurements. Further, if one considers the overall averages on all designs examined in either cadaver (23.5±10 N) or construct (28.1±6.2 N), it is concluded that these goals have been adequately achieved. Two interesting results are drawn from the construct tests. First, axial starting loads are very sensitive to design. Similar to twist drill designs, self-tapping flute designs can vary by several means including: (1) the number of cutting “flutes” (2) the flute relief angle and (3) the cutting rake angle. Furthermore, some screws that possessed nominally identical tip geometries, such as ZPS and. ECT 3.5 mm diameter screws, showed differences in starting load performance, while others did not show a difference, for example the ZPS and ECT 4.5 mm screws. The differences can only be attributed to variations in the manufacturing process possibly including: (1) the manufacture methodology, i.e. rolling vs. machining the threads, (2) cutting tool speeds and feeds, (3) deburring operations and (4) post machining processes such as electropolish (EP). Second, torques apparently are not very sensitive to tip design in this study. This is likely due to the fact that all the screws tested here have similar major and minor diameters and similar thread forms. Therefore, it probably cannot be concluded that the construct is sensitive to changes in thread design parameters. That must be proven in another, separate study. To demonstrate the usefulness of this experimental technique and in response to this study, a change was made in the tip design and manufacturing process for the Zimmer

Fig. 11 Summary of comparison of cadaver results to construct results. The cadaver results are shown in gray, the construct results in hatch

ZPS 3.5 mm cortical screw. Results of a series of Design of Experiments (DOE) documenting incremental changes to design and manufacturing practices are shown in Figs. 12 and 13. In both figures, the results from the current design (“controls”) are shown to the left in dark gray while the results of the DOE are tracked through a series of experiments, shown to the right (“experimental”). Those changes resulted in a significant drop in the average axial starting loads and moderate decrease in insertion torque into both corticies. In addition, Fig. 12 also demonstrates the methodology’s ability to detect subtle changes in manufacturing practices as the large variations (“scatter”) in axial starting loads seen in the controls were significantly decreased as reflected in the standard deviations of the final set of DOE results (DOE #8). Since the

Fig. 12 Zimmer ZPS 3.5 mm design and process change axial load results. In the plot, the current design (“control”) is shown in dark gray on the left, successive experiments (DOE) documenting incremental improvements in design and manufacturing practices are shown to the right (“experimental”). It is seen that design and process improvements significantly decreased the axial load and the variability

606

Exp Mech (2007) 47:595–607

significant level (0.003, 1-way ANOVA (α=0.05)). On average, the surgeon’s perceived peak torque was about 85% of the actual peak torque. While surgeon “feel” may seem objective it can be quite accurate and is an important part of orthopedic screw design.

Conclusions

Fig. 13 Zimmer ZPS 3.5 mm design and process change torque results. The insertion torque results in (a) 1st cortex and (b) 2nd cortex. In both plots, the current design (“control”) are shown in dark gray on the left, successive experiments (DOE) documenting incremental improvements in design and manufacturing practices are shown to the right (“experimental”). Only marginal improvement in torque performance is seen

pilot hole size was not changed and neither was the minor diameter of the threads of the screw-in fact, nothing was changed except the self-tapping tip-it is not surprising that torque did not change significantly. (The pilot hole size and minor diameter of the thread form are the key factors in determining insertion torque.) Therefore, it is concluded that simple changes in the tip design and processing can lead to significant changes in self-tapping performance. This was born out when the changed screws were evaluated by a trauma surgeon who inserted the screws into the bone analog developed here and reported them to have better “feel.” i.e. initiated self-tap with less force and required less effort to drive bicortically. Although this feedback is qualitative in nature, it should be noted that Reitman et al. [9] showed experimentally that experienced orthopedic surgeons were routinely able to determine a target maximum insertion torque value by “feel” alone to a statistically

In this study, a new method for the measurement of axial starting load and driving torque through two corticies was developed. Because of the limited availability of human bone samples and the variability from bone to bone, a bone analog construct was also developed and shown to behave like human cadaveric bone during screwing and tapping procedures performed with this apparatus. The method consists of an instrumented drill press that measures axial screw force, screw torque and screw displacement throughout the self-tapping and driving process. Results are compared favorably to cadaver measurements. The results shown here are focused on self-tapping screws only, where a pilot hole is required. If self-drilling/self-tapping screws were to be investigated it would be necessary to validate the construct against cadaver bone again. This study has successfully illuminated differences between several screw designs. Furthermore, this test was able to discern subtle differences in manufacturing techniques for screws of similar tip design. An important part of the tests is the method for application of the axial load on the screw prior to and during the onset of tapping. The axial load profile in the tests was important in identifying the onset of self-tapping; in the cadaver studies or in the construct. Further, in a separate blind, qualitative study a trauma surgeon could “feel” the difference/improvements detected by this test when hand inserting the control and experimental ZPS 3.5 mm screws into the bone analog. Thus, the methodology outlined here shows promise for evaluation of future designs of screws throughout the orthopedic industry. An important part of the results on the screws examined here is that axial starting loads are very sensitive to tip design and manufacturing processes; driving torques apparently are not, except when penetrating the second cortex. Axial loads and insertion torques are sometimes linked. In some designs, when penetrating the second cortex, the cutting efficiency was lower leading the surgeon to apply higher axial loads resulting in higher torques. In other cases, the cutting efficiency was higher, which required less axial load and led to lower torques. However, generally speaking, the lack of sensitivity in the insertion torques observed here is likely due to the fact that the thread designs examined here did not vary significantly from screw to screw, whereas the tip designs did.

Exp Mech (2007) 47:595–607

Overall, this work illuminates the need for good orthopedic screw evaluation methods and demonstrates the usefulness of one such test in evaluating existing or new screw designs. Acknowledgements The authors would like to gratefully acknowledge Dr. Clifford Jones, MD for his professional opinions and assistance in conducting portions of this work.

References 1. ASTM F543 (2002) Standard specification and test method for metallic medical bone screws. American Society for Testing & Materials (ASTM), West Conshohocken, PA. 2. Hsu CC, Chao CK, Wang JL, Hou SM, Tsai YT, Lin J (2005) Increase of pullout strength of spinal pedicle screws with conical core: biomechanical tests and finite element analyses. J Orthop Res 23(4):788–794. 3. Bredbenner TL, Haug RH (2000) Substitutes for human cadaveric bone in maxillofacial rigid fixation research. Oral Surg Oral Med Oral Pathol Oral Radiol Endo 90(5):574–580. 4. Pitzen T, Franta F, Barbier D, Steudel WI (2004) Insertion torque and pullout force of rescue screws for anterior cervical plate fixation in a fatigued initial pilot hole. J Neurosurg Spine 1(2):198–201. 5. Buhler DW, Berlemann U, Oxland TR, Nolte LP (1998) Moments and forces during pedicle screw insertion. In vitro and in vivo measurements. Spine 23(11):1220–1227; discussion 1228. 6. Glauser CR, Oden ZM, Ambrose CG, Willits MB, Coupe KJ (2003) Mechanical testing of small fracture implants for comparison of insertion and failure torques. Arch Orthop Trauma Surg 123(8):388–391. 7. Lawson KJ, Brems J (2001) Effect of insertion torque on bone screw pullout strength. Orthopedics 24(5):451–454. 8. Yerby S, Scott CC, Evans NJ, Messing KL, Carter DR (2001) Effect of cutting flute design on cortical bone screw insertion torque and pullout strength. J Orthop Trauma 15(3):216–221. 9. Reitman CA, Nguyen L, Fogel GR (2004) Biomechanical evaluation of relationship of screw pullout strength, insertional torque, and bone mineral density in the cervical spine. J Spinal Disord Tech 17(4):306–311. 10. Koistinen A, Santavirta SS, Kroger H, Lappalainen R (2005) Effect of bone mineral density and amorphous diamond coatings on insertion torque of bone screws. Biomaterials 26(28):5687–5694.

607 11. Koistinen A, Santavirta S, Lappalainen R (2003) Apparatus to test insertion and removal torque of bone screws. Proc Inst Mech Eng H 217(6):503–508. 12. Hughes AN, Jordan BA (1972) The mechanical properties of surgical bone screws and some aspects of insertion practice. Injury 4(1):25–38. 13. McGuire RA, St John KR, Agnew SG (1995) “Analysis of the torque applied to bone screws by trauma surgeons. Comparisons based on years of experience and material of implant construction.” Am J Orthop 24(3):254–256. 14. You ZH, Bell WH, Schneiderman ED, Ashman RB (1994) Biomechanical properties of small bone screws. J Oral Maxillofac Surg 52(12):1293–1302. 15. Ansell RH, Scales JT (1968) “A study of some factors which affect the strength of screws and their insertion and holding power in bone.” J Biomech 1(4):279–302. 16. Ryken TC, Clausen JD, Traynelis VC, Goel VK (1995) Biomechanical analysis of bone mineral density, insertion technique, screw torque, and holding strength of anterior cervical plate screws. J Neurosurg 83(2):325–329. 17. Trader JE, Johnson RP, Kalbfleisch JH (1979) Bone-mineral content, surface hardness, and mechanical fixation in the human radius: A correlative study. J Bone Joint Surg Am 61(8):1217– 1220. 18. Nunamaker DM, Perren SM (1976) Force measurements in screw fixation. J Biomech 9(11):669–675. 19. Weaver JK (1966) The microscopic hardness of bone. J Bone Joint Surg Am 48(2):273–288. 20. Zimmer, Inc. (2004–2006) Sales Data on file at Zimmer, Inc. 21. CELCON® (M25 Grade) (2004) Material Properties Datasheet, Ticona, Florence, KY. 22. Hobatho MC, Rho JY, Ashman RB (1991) Atlas of mechanical properties of human cortical and cancellous bone. In: Van Der Perre G, Lowet G, Borgwardt-Christensen A (eds) Proceedings COMAC-BME 11.2.6 meeting. Durham, UK, pp 11–13. 23. Cowin S (1989) Bone mechanics, Chapter 6. CRC Press, New York, pp 97–158. 24. Calvert K, Kirkpatrick LA, Blakemore DM, Johnson TS (2005) Poster #366—Femoral cortical wall thickness and hardness evaluation. In: Society for Biomaterials 2005 Annual Meeting, April 27–30. 25. FR 3720 (2004) Material properties datasheet. General Plastics Manufacturing Co., Tacoma, WA. 26. ASTM F1839 (2004) Specification for rigid polyurethane foam for use as a standard material for testing orthopaedic devices and instruments. American Society for Testing & Materials (ASTM), West Conshohocken, PA.