Biomechanical comparison of interfragmentary compression in transverse fractures of the olecranon

VOL. 93-B, No. 2, FEBRUARY 2011 245

TRAUMA: RESEARCH

Biomechanical comparison of interfragmentary compression in transverse fractures of the olecranon

J. Wilson, A. Bajwa, V. Kamath, A. Rangan

From James Cook University Hospital, Middlesbrough, United Kingdom

J. Wilson, MRCSEd, MSc, Specialty TraineeNorth West Deanery, 3 Piccadilly Place, Manchester M1 3BN, UK.

A. Bajwa, MPhil, MFSEM, FRCS(Tr&Orth), Orthopaedic Surgeon

A. Rangan, FRCS(Tr&Orth), ProfessorJames Cook University Hospital, Marton Road, Middlesbrough TS4 3BW, UK.

V. Kamath, FRCS(Tr&Orth), Orthopaedic SurgeonVictoria Hospital, Whinney Heys Road, Blackpool, Lancashire FY3 8NR, UK.

Correspondence should be sent to Mr J. Wilson; e-mail: [email protected]

©2011 British Editorial Society of Bone and Joint Surgerydoi:10.1302/0301-620X.93B2. 24613 $2.00

J Bone Joint Surg [Br] 2011;93-B:245-50.Received 6 February 2010; Accepted after revision 26 October 2010

Compression and absolute stability are important in the management of intra-articular fractures. We compared tension band wiring with plate fixation for the treatment of fractures of the olecranon by measuring compression within the fracture. Identical transverse fractures were created in models of the ulna. Tension band wires were applied to ten fractures and ten were fixed with Acumed plates. Compression was measured using a Tekscan force transducer within the fracture gap. Dynamic testing was carried out by reproducing cyclical contraction of the triceps of 20 N and of the brachialis of 10 N. Both methods were tested on each sample. Paired t-tests compared overall compression and compression at the articular side of the fracture.

The mean compression for plating was 819 N (SD 602, 95% confidence interval (CI)) and for tension band wiring was 77 N (SD 19, 95% CI) (p = 0.039). The mean compression on the articular side of the fracture for plating was 343 N (SD 276, 95% CI) and for tension band wiring was 1 N (SD 2, 95% CI) (p = 0.038).

During simulated movements, the mean compression was reduced in both groups, with tension band wiring at -14 N (SD 7) and for plating -173 N (SD 32). No increase in compression on the articular side was detected in the tension band wiring group.

Pre-contoured plates provide significantly greater compression than tension bands in the treatment of transverse fractures of the olecranon, both over the whole fracture and specifically at the articular side of the fracture. In tension band wiring the overall compression was reduced and articular compression remained negligible during simulated contraction of the triceps, challenging the tension band principle.

Transverse fractures of the olecranon arecommon.1 Union with displacement or the for-mation of excessive callus can lead to post-traumatic osteoarthritis of the elbow.2

Anatomical reduction and absolute stability isrecommended for the treatment of articularfractures,3 and in the United Kingdom mostpatients with displaced, transverse fractures ofthe olecranon are treated with tension bandwiring.4 Numerous other methods of fixationhave been described,5 including interfragmen-tary screws, intramedullary screws and Rushpins, with or without figure-of-eight wires, fix-ation with a plate and excision of the proximalfragment with advancement of triceps.6

When applied to fractures of the olecranon,the tension band principle predicts that flexionof the elbow joint compresses the site of thefracture.7 However, no data have been pro-vided to support this theory.6 Previous bio-mechanical studies of techniques of fixation ofthese fractures have concentrated on load tofailure and have measured displacement. We

aimed to identify the most effective method ofcompressing a transverse fracture of the ole-cranon, and to characterise the variation incompression of the fracture during simulatedmovements of the elbow.

Pre-contoured plate fixation was investi-gated as this is a satisfactory way of compress-ing other fractures,8 and there is evidence thatit may have advantages over other methods.9 Itwas compared to tension band wiring, as this isthe most commonly used method in clinicalpractice.6

Materials and MethodsThe experiments were performed on modelulna bones and both methods of fixation wereused on each specimen. This cross-overmethod was used to reduce variability betweenthe samples.10 A disadvantage of this design isthe effect on each model of the previousfixation, which could alter the effectiveness ofthe second technique. In order to mitigate sys-tematic bias that could arise from this, sealed

NAGNAR.A ,HTAMAK.V ,AWJAB.A ,NOSLIW.J642

THE JOURNAL OF BONE AND JOINT SURGERY

envelopes were used to randomly allocate exactly half thesample to tension band wiring first and the other half toplate fixation first. Synthetic material was chosen ratherthan cadaveric tissue, in order to reduce the variabilityencountered when human tissue is used. Even subtle varia-tions in the dimensions of the materials subject to theexperiment would necessitate a much larger sample size tocounteract the problems caused by inexact matching.

The model ulnas (Sawbones, Vashon, Washington) weremanufactured to exactly match the proportions and ana-tomical dimensions of the human ulna. They were also pro-duced from two densities of resin in order to match thestructure of bone.

A film pressure sensor (Tekscan, Boston, Massachusetts)that allowed force measurements at 27 individual pointsper cm2 was selected for this project. The film could also bepierced to allow the passage of the implants used to fix thefracture. The sensor had a pressure range from 0 kPa to13 789 kPa. Prior to each procedure, the sensor was cali-brated using the technique recommended by the supplier.The dynamic testing jig was specifically designed and pro-duced in collaboration with the Durham University Engi-neering Department. It was arranged in the horizontal planeto neutralise the effects of gravity. The force of the tricepspull, generated by the pneumatic piston, was set at 20 N.This was calibrated using a spring force gauge. This valuewas chosen to replicate the forces used in biomechanicalanalysis of movement of the elbow by previous authors.11

This force caused the model ulna to rotate about the pivot,through an arc of between 75° and 125° of flexion. The resis-tance cord, simulating the reciprocal pull of the brachialis,was set at 10 N at its maximum excursion. Again, this wasverified using a spring force gauge and was fixed at half thevalue of the triceps force. Once the piston relaxed the tricepspull, the 10 N force of the simulated brachialis returned themodel ulna to the starting position (Fig. 1).

The two primary outcome measures were the compres-sion force across the whole surface of the fracture and thecompression force adjacent to the articular surface. The

secondary outcome measure was the change in compres-sion of the fracture adjacent to the articular joint surfaceduring simulated contraction of the triceps.

A transverse fracture was created in each model ulnathrough the mid-point of the sigmoid notch. Figure 1 showsthe site of the fracture and the sensor film can be seen in thegap. A cutting jig and a 0.5 mm straight saw blade were usedto standardise the position of the fracture. The pressure sen-sor was then placed between the fragments, which were thenreduced and fixed using tension band wiring or plate fixa-tion. In order to eliminate potential performance bias, all fix-ations were performed by one surgeon (JW). On completionof fixation the sensor was connected to the monitoringequipment and the model ulna placed in the testing jig.

Compression of the fracture was measured during cyclesof simulated elbow movement.Statistical analysis. Data from the experiment wererecorded on a Microsoft Excel database and statisticalanalysis was performed using SPSS v.16 (SPSS Inc., Chi-cago, Illinois).

A power calculation was carried out using Sample Power2.0. A pilot study revealed that compression by tension bandwiring produced a mean compressive force of 78 N over thewhole surface of the fracture. The validity of this value wascorroborated by a previous study by Parent et al,11 whofound compression to be 60 N (SD 13). The pilot data hadan SD of 31 N. A clinically significant difference in compres-sion was considered to be 40 N. This value was chosen as itis half of the average compressive force measured andlarger than the SD. No previous publications have quanti-fied this value. Both methods of fixation were carried outon each specimen, and therefore a paired-samples t-test wasused to compare the mean values. With α set at 0.05, onlyseven model ulnas would be required for a power of 81%.However, to improve the precision of the results and toalign this study with the sample sizes of previous similarbiomechanical experiments,6,11,12 this value was exceededand a sample size of 10 was selected. This gave a power todetect a significant effect of 95%.

The mean values of the outcome measures were calcu-lated and the precision of these values indicated by 95%confidence intervals (CI). The normality of the data wasverified using the Kolmogorov-Smirnov test.13 Mean valueswere compared using a paired-samples two-tailed t-test.13

ResultsIn total, ten model ulnas were used, with tension band wir-ing and plate fixation carried out on each specimen in arandomly allocated order. Using data from the sensor,three-dimensional pressure contour diagrams were gener-ated to illustrate the concentration of the forces. An exam-ple of this for tension band wiring is shown in Figure 2 andthe equivalent diagram for plate fixation, using the samescale, is in Figure 3. These diagrams show the parts of thesensor that are detecting compression. The height andcolour of the peaks represent the magnitude of the force at

Fig. 1

Photograph showing the components of the dynamictesting jig with the model ulna and sensor mounted,following tension band wiring fixation.

-

BIOMECHANICAL COMPARISON OF INTERFRAGMENTARY COMPRESSION IN TRANSVERSE FRACTURES OF THE OLECRANON 247

VOL. 93-B, No. 2, FEBRUARY 2011

each location. Blank areas indicate no compression. Thisvisual comparison shows that tension band wiring onlyproduces compression at the posterior aspect of the frac-ture, away from the articular surface. The technique ofplate fixation produces increased compression over thewhole surface of the fracture.

For statistical comparison, the force through the wholearea of the fracture was calculated for each experiment.The software was also capable of measuring the compres-sive force through specific areas of the sensor (Fig. 4). Thisallowed measurement of compression at the articular(anterior) aspect of the fracture. The individual measure-ments were then combined, and the mean values followingall 20 experiments are shown in Table I. These figures indi-

cate that plate fixation provides a larger force of compres-sion over the whole surface of the fracture than doestension band wiring. Based on these values, the former canproduce more than ten times the compressive force of thelatter. The CIs of the means are shown in Table II.

At the articular side of the fracture, tension band wiringproduces a small compressive force compared to plate fix-ation, with only one of the tension band wiring experi-ments producing a compressive force above zero. Figure 5shows a box-plot which indicates the spread of the mea-surements.

Before parametric tests were performed, the data wereanalysed for normality. The Kolmogorov-Smirnov test wascarried out on the differences between the total

Fig. 2

Three-dimensional pressure contour diagramshowing the distribution of force following ten-sion band wiring fixation.

Fig. 3

Three-dimensional pressure contour diagramshowing the distribution of force followingplate fixation.

Fig. 4

Three-dimensional pressure contour diagramshowing the areas used to measure the totalforce of fracture compression (solid red line)and the compression in the anterior part of thefracture (dashed green line).

Table I. Comparison of the mean compres-sion force produced by the two methods offixation over the whole surface of the frac-ture and at the articular side

TBW* PF†

Mean total compression (N) 77 819Mean articular compression (N) 1 343

* TBW, tension band wiring† PF, plate fixation

Table II. Mean compressive force produced over the whole fracturesurface and at the articular side of the fracture by the two methods offixation

TBW* (95% CI†) PF‡ (95% CI)p-value of paired t-test

Mean total compression (N)

77 (58 to 96) 819 (217 to 1421) 0.039

Mean articular compression (N)

1 (-1 to 3) 343 (67 to 619) 0.038

* TBW, tension band wiring† CI, confidence interval‡ PF, plate fixation

Fig. 5

Box plot showing the mean force of compression pro-duced at the anterior side of the fracture and the totalcompression for the two methods of fixation (notethat the anterior compression produced by tensionband wiring (TBW) is negligible) (PF, plate fixation).

NAGNAR.A ,HTAMAK.V ,AWJAB.A ,NOSLIW.J842

THE JOURNAL OF BONE AND JOINT SURGERY

compression produced by the two methods of fixation (p =0.096), and the same test was repeated for the differences incompression at the anterior part of the fracture (p = 0.141).Both of these p-values indicate that the distribution of thesemeasurements were not significantly different from normal,and hence parametric tests could be applied.

A paired t-test was used to compare the mean compres-sive force over the whole surface of the fracture. The 95%CIs of the mean values were also calculated. The same sta-tistical analysis was applied to the mean value for thecompression in the anterior half of the fracture. Theseresults are shown in Table II. There was a statistically sig-nificant difference between the total compressive forcesproduced by the tension band wiring fixation and thatproduced by plate fixation, as the p-value is < 0.05 and theCIs do not overlap. The result of the t-test relating to theanterior part of the fracture shows that there is also a sta-tistically significant difference between the two methodsof fixation.

With the sensor positioned within the fracture, it was possi-ble to measure the changes in compression that occurred dur-ing simulated movements of the elbow. The compressionthroughout the whole fracture and within its anterior partwere recorded separately. Figure 6 shows an example of thesemeasurements for plate fixation. The model ulna was sub-jected to several cycles of simulated triceps contraction, whichproduced a temporary reduction in the force of compressionat the fracture site. The reduction in compression occurredover the whole surface of the fracture and at the fracture sur-face adjacent to the joint. The downward waves of the tracingseen in Figure 6 correspond to each time the sample under-went a cycle of elbow movement. Figure 7 uses a small sectionof Figure 6 to illustrate how the tracing relates to the differentparts of the cycle of simulated movement of the elbow.

Figure 8 shows an example of the same measurements fortension band wiring. Again, there is a reduction in compres-sion of the fracture during the simulation of triceps contrac-tion. It is important to note no anterior compression wasdetected while the ulna was at rest, and no increase in com-pression was found when contraction of the triceps was sim-ulated. This is contrary to the tension band theory.7

The peak or trough measurements were recorded to com-bine and summarise these data. During simulated move-ments, overall compression was reduced in both groups.For tension band wiring the overall compression wasreduced by a mean of 14 N (7 to 21, 95% CI) and for platefixation by a mean of 173 N (141 to 205, 95% CI).

DiscussionThis study has shown that plate fixation provides signifi-cantly greater compression of transverse fractures of theolecranon than does tension band wiring, both over thewhole surface of the fracture and of the anterior half adja-

Fig. 6

Graph showing variation in fracture compression duringsimulated elbow movements following plate fixation.

600

400

200

00 5

Time (s)

Forc

e (N

)

B

AC

A – Triceps contraction and elbow extension

B – Triceps relaxation and elbow flexion dueto pull of brachialis

C – Elbow at rest

{

{

{

Fig. 7

Section of the graph in Figure 6, showing the variation in compres-sion force during simulated elbow movements.

40

30

20

10

00 5 10 15 20

Time (s)

Forc

e (N

)

BA

C

A – Triceps contraction and elbow extension

B – Triceps relaxation and elbow flexion dueto pull of brachialis

C – Elbow at rest

{Overall force of compression

Fig. 8

Graph showing the variation in fracture compression during sim-ulated elbow movements following tension band wiring. Notethat no anterior compression was detected.

BIOMECHANICAL COMPARISON OF INTERFRAGMENTARY COMPRESSION IN TRANSVERSE FRACTURES OF THE OLECRANON 249

VOL. 93-B, No. 2, FEBRUARY 2011

cent to the articular surface. The CIs of the mean compres-sion over the whole surface of the fracture area do notoverlap, but are quite broad. Plate fixation can produceapproximately two to 24 times more compression over thewhole surface of the fracture than tension band wiring.

The locking olecranon plates (Acumed, Weyhill, UnitedKingdom) used in this study can achieve compression ofthe fracture at two points of the fixation. There are twoslotted holes in the plates. Once the fracture is heldreduced with a clamp, the plate is applied to the ulna andprovisionally held with a 2 mm Kirschner (K)-wirethrough the screw-in K-wire guide and a non-lockingscrew in the first slotted hole in the plate. A second non-locking screw can be inserted in dynamic compressionmode into the distal hole in the plate. Further compressioncan be achieved when inserting the most proximal screwin the plate. This hole accommodates a screw that tra-verses the fracture. A non-locking screw can be selectedwhich can act as a lag screw.14

This study has shown that tension band wiring pro-duces negligible compression of the fracture at the articu-lar side, where primary cortical healing is especiallydesirable. In contrast, plate fixation did produce compres-sion in this anterior area of the fracture, with a meanforce of 343 N.

Fyfe et al12 showed that in transverse fractures of theolecranon, tension band wiring provided more stable fix-ation than intramedullary or plate fixation. Stability wasdefined as resistance to displacement of the fracture frag-ments when a fixed force was applied to the fixation con-struct, but this is not equivalent to measuring thecompression between the fragments. The metal plate usedwas a one-third-tubular plate, which is less rigid than pre-contoured plates.15

A retrospective, non-randomised, single-centre study byRommens et al1 followed 58 patients who had surgicaltreatment for fractures of the olecranon. There was a mix-ture of fracture patterns, only 20 of which were trans-verse. The majority had a tension band wiring fixation,and only five had plate fixation. A further procedure wasnecessary in 15% of cases. No difference between the fix-ation techniques could be concluded owing to the smallnumber in the plate fixation group.

Hume and Wiss9 carried out a prospective randomisedclinical trial to compare the clinical and radiological out-comes of tension band wiring and plate fixation. Detailsof the randomisation process were not described andblinding was not attempted. There were 41 adults withdisplaced olecranon fractures, and the average follow-upwas for only six months. Of the 41 patients, 11 had sus-tained transverse fractures, which were divided equallybetween the two methods of fixation. Overall, plate fixa-tion had a lower complication rate, better clinical out-comes and less displacement of the fracture fragments.The range of movement of the elbow was equivalent in thetwo groups.

Plate fixation achieves greater compression. In clinicaluse it might be expected that the increased bulk of theplate would cause more complications because of itsprominence or through irritation of the skin. However,eight of the 22 patients with tension band wiring com-plained of problems with the metalware, whereas onlyone of 19 patients had a similar problem followingplate fixation.

The measurements taken during simulated movementsof the elbow show that the compressive forces across thefracture alter during contraction of the triceps. However,compression of the fracture reduces with both fixationtechniques, and compression at the anterior part remainsnegligible with tension band wiring. These findings arecontrary to the tension band theory, which states that dueto contraction of the triceps, flexion of the elbow com-presses the fracture site.7 With plate fixation, compressiondoes reduce during simulated elbow movements but neverdiminishes to zero, and remains several times greater thanwith tension band wiring.

The application of the tension band theory to fractureof the olecranon has been questioned by previous authors.Rowland and Burkhart16 presented mathematical calcula-tions, arguing that the standard configuration of tensionband wiring led to the fracture gap opening up on thearticular side of the fracture. They proposed a modifica-tion to the technique but provided no empirical data.Hutchinson et al6 found that in simulated contraction ofthe triceps, the fragments were distracted at both the ante-rior and the posterior sides of the ulna. These observa-tions agree with ours.

These findings indicate that in clinical practice plate fix-ation should be performed in order to effectively compressa transverse fracture of the olecranon. Previous clinicalstudies have not had sufficient power or follow-up toinvestigate the incidence of post-traumatic osteoarthritis.However, the results of this study indicate that plate fixa-tion compresses transverse fractures more effectively thantension band wiring, reducing excess bone formation andsuggesting a reduced risk of osteoarthritis.

During dynamic testing, this investigation revealed thatfracture compression is reduced during contraction of thetriceps. The rehabilitation programme following an ole-cranon fracture aims to avoid stiffness of the elbow byencouraging movement exercises. Clearly, this shouldinvolve passive movements rather than active.

The potential limitations of this study relate largely tothe use of an elbow simulation. The ulna models were spe-cifically designed so that the external dimensions matchedhuman anatomy and the internal structure correspondedto normal bone. The testing jig reproduced contraction ofthe triceps and the reciprocal pull of brachialis, with themagnitudes of forces based on published values.11 How-ever, the effects of blood flow, the healing response ofbone and of the surrounding tissue and skin cannot be eas-ily estimated or modelled, but may be very important.6

250 J. WILSON, A. BAJWA, V. KAMATH, A. RANGAN

THE JOURNAL OF BONE AND JOINT SURGERY

Nevertheless, this study has provided information aboutthe compressive forces within the fracture gap that couldnot be obtained from a study on living human subjects.

Steps were taken in the design of the investigation toensure internal validity. The samples were exactlymatched, the order of the procedures on each specimenwas randomised, a cutting jig was used to produce identi-cal fractures in each model, and each procedure was car-ried out according to the published technique guide.However, blinding of the operator was not possibleand the validity of the findings is therefore reliant on theoperator performing all procedures with the same level oftechnical skill.

It is difficult to standardise tightening of the figure-of-eight wire during tension band wiring. A torque-limitedor torque measuring tool could make this step of the pro-cedure more reproducible. However, the same individualperformed all of the tension band wirings, aiming for thehighest tension achievable without breaking the wires.The mean compression achieved with tension bandwiring was 77 N (95% CI, 58 to 96). This is greater thanthe value of 60 N (95% CI, 47 to 73) found by Parent etal,11 indicating than an effective tensioning techniquewas used.

This study has addressed a specific question regardingcompression in transverse fractures of the olecranon.However, many other fracture configurations and com-binations of injuries occur clinically, and these resultsmay not be applicable in such circumstances. Hence, theexternal validity of these findings for fractures of the ole-cranon in general may be weakened. However, the impli-cations of these results are that, in order to compress atransverse fracture of the olecranon effectively, plate fix-ation should be performed. If tension band wiring isundertaken, contraction of the triceps risks the fractureopening up at the joint surface. Contraction of the tri-ceps should be avoided and during rehabilitation passiveelbow exercises used.

The authors would like to thank A. Bent and the Durham University Engineeringdepartment for their help with the dynamic testing apparatus, the Teeside Cen-tre for Rehabilitation Sciences for accommodating the project and Professor J.Stothard for his support and help. We would also like to thank AcumedTM andSynthesTM who supplied the fixation materials without charge, and TekscanTM

for the supply of the sensor equipment. We also owe thanks to the South TeesResearch and Development Department for funding the project. J. Gray and K.Sanderson from Teeside University were also instrumental in bringing thisproject to completion.

No benefits in any form have been received or will be received from a com-mercial party related directly or indirectly to the subject of this article.

References1. Rommens PM, Küchle R, Schneider RU, Reuter M. Olecranon fractures in adults:

factors influencing outcome. Injury 2004;35:1149-57.

2. Birch R, Singh D. Principles of management of fractures, joint injuries and periph-eral nerve injuries. In: Henry MM, Thompson JN, eds. Clinical surgery. Edinburgh: WBSaunders, 2001:617-32.

3. Schatzker J. Diaphyseal fractures. In: Schatzker J, Tile M. The rationale of operativefracture care. Third ed. Heidelburg: Springer-Verlag, 2005:22.

4. Newman SD, Mauffrey C, Krikler S. Olecranon fractures. Injury 2009;40:575-81.5. Murphy DF, Greene WB, Gilbert JA, Dameron TB Jr. Displaced olecranon frac-

tures in adults: biomechanical analysis of fixation methods. Clin Orthop1987;224:210-14.

6. Hutchinson DT, Horwitz DS, Ha G, Thomas CW, Bachus KN. Cyclic loading ofolecranon fracture fixation constructs. J Bone Joint Surg [Am] 2003;85-A:831-7.

7. Wu CC, Tai CL, Shih CH. Biomechanical comparison for different configurations oftension band wiring techniques in treating an olecranon fracture. J Trauma2000;48:1063-7.

8. Brinker MR, O’Connor DP. Basic sciences. In: Miller MD, ed. Review of orthopae-dics. Fourth ed. Philadelphia: WB Saunders, 2004:18.

9. Hume MC, Wiss DA. Olecranon fractures: a clinical and radiographic comparison oftension band wiring and plate fixation. Clin Orthop 1992;285:229-35.

10. DePoy E, Gitlin LN. Introduction to research: understanding and applying multiplestrategies. Third ed. Missouri: Elsevier Mosby, 2005:101.

11. Parent S, Wedemeyer M, Mahar AT, et al. Displaced olecranon fractures in chil-dren: a biomechanical analysis of fixation methods. J Pediatr Orthop 2008;28:147-51.

12. Fyfe IS, Mossad MM, Holdsworth BJ. Methods of fixation of olecranon fractures:an experimental mechanical study. J Bone Joint Surg [Br] 1985;67-B:367-72.

13. Petrie A, Sabin C. Medical statistics at a glance. Oxford: Blackwell Science,2000:42-82.

14. No authors listed. Acumed congruent elbow plate system: olecranon surgical tech-nique. http://www.acumed/net/sites/default/files/congruent-elbow-plate-system-brochure.pdf (date last accessed 29 October 2010).

15. Hak DJ, Golladay GJ. Olecranon fractures: treatment options. J Am Acad OrthopSurg 2000;8:266-75.

16. Rowland SA, Burkhart SS. Tension band wiring of olecranon fractures: a modifica-tion of the AO technique. Clin Orthop 1992;277:238-42.