Injury, Int. J. Care Injured 32 (2001) 109–114

Outcome in fracture healing: a review

Roger Wade, James Richardson *Robert Jones and Agnes Hunt Orthopaedic Hospital, Oswestry, Shropshire, SY10 7AG, UK

Accepted 27 June 2000

Abstract

This editorial reviews outcome measures in fracture healing. Individually radiological and clinical methods are inaccurate, buttogether provide a good indicator in clinical practice when a fracture has healed. Other quantitative methods are not practical.Vibrational analysis of fracture healing is complex and limited to research at present. Indirect fracture stiffness measurement ispossible and has been used in clinical trials. Pin loosening reduces accuracy. The measurement of direct fracture stiffness has beenvalidated and using this a decrease in re-fracture rate and a reduction of time to independent weight bearing (on average 3 weeks)was found. The system is accurate and simple to use in the clinical setting and allows the rate of healing to be monitored infractures. Studies using poor outcome measures should be interpreted with care. The study of healing fractures is no exception.© 2001 Elsevier Science Ltd. All rights reserved.

www.elsevier.com/locate/injury

1. Introduction

Measuring outcome accurately is fundamental to thedesign of useful studies. In its strongest form an out-come measure can be given a number; it can be termed‘interval data’. Studies using poor outcome measuresshould be interpreted with care. Conducting an other-wise well designed double blind randomised controlledtrial with an inaccurate outcome measure is pointless.An individual benefits from a precise end-point whenbeing treated, whilst a study group benefits from anaccurate outcome measure for all treatments beingcompared. Studying healing fractures is no exception tothese guidelines.

Arbuthnot Lane in 1914 [1] defined outcome follow-ing a fracture of a tibia treated conservatively, as theearning ability in a group of dockworkers. He showeda 40% loss of earning potential and so argued forinternal fixation of tibial fractures. This is an accurateand precise measure, but only specific to his definedpopulation. The end-point for a professional footballermay be the return to training yet the chairman of alarge organisation may only be required to get in andout of his chauffeur driven car. Each patient has theirown specific criteria; Lane’s approach is sadly not

possible with today’s mobile population. We can nolonger consider the return to work as a final end-point.

Lane’s group of patients’ highlights a further prob-lem; all patients were treated by the same method.Following a tibial fracture the treatment possibilitiesare vast and it is difficult for a group of orthopaedicsurgeons to agree on one treatment method. Differenttreatments cannot always be directly compared and socaution is required when interpreting outcomes. Studiescomparing different methods of treatment must con-sider the associated risks and benefits of each treatmentmodality, in particular the complications. How can wecompare an uncommon but devastating complication,such as infection following internal fixation, with thecommon problem of joint stiffness following plasterremoval?

In 1942 Page [2] called for an objective outcome infracture studies. To define an outcome in fracture heal-ing it is first necessary to consider the end-point. Theobvious answer is a ‘healed’ fracture, but what is ahealed fracture? Classically this is one that allows inde-pendent weight bearing rigid; internal fixation mayallow this from day one. This is not to say the metal-work may be removed and the fracture will remainintact. Failure of the fracture following implant re-moval is a strong outcome measure although it is ratherdramatic; obviously removing splints at a specific timeto see if failure occurs is of no use practically. This is* Corresponding author.

0020-1383/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.PII: S0020-1383(00)00126-1

R. Wade, J. Richardson / Injury, Int. J. Care Injured 32 (2001) 109–114110

highlighted in fractures treated with external splintage;they require timely splint removal to reduce the mor-bidity, but early removal may lead to loss of reduction.This in turn requires more accurate definition of theend-point.

It is also necessary to consider how fractures heal invarious mechanical environments. Rigid internal fixa-tion aims for direct healing, a process which takesmany months with little external callus formation. Withflexible external fixation the aim is for indirect healingwith external callus formation. Methods of defininghealing by the formation of callus are obviously influ-enced by the mechanical environment of the fracture.

There is further confusion due to the use of varyingdefinitions for delayed union and non-union. By defaultit is likely that using non-union as an end-point ingroups of patients with various treatment regimesnegates some of the external influences, because of thetime that has elapsed before non-union is defined.Unfortunately this measure is not what we or ourpatients require, defining the fracture as ununited sev-eral months following the injury is too little too late.Measurement of the rate of healing provides an indica-tor as to whether intervention is required and mayprovide an end-point, although this is not necessarilythe same as the timing of independent weight-bearing.

It can be seen that an end-point can, therefore, beindividualised or valid across a study group but thesemust not be confused. High accuracy is important inlooking at new treatment methods and high precision isimportant in clinical practice. It must be considered

how a fracture is treated and how this will influence thedefinition of the end-point. Ideally the definition wouldcover all these points. We will review various methodsof assessment used.

2. Clinical and radiological methods

Apley and Solomon defined clinical consolidation ascallus which is completely ossified, the fracture is non-tender, no movement can be detected, radiologically thefracture line is almost obliterated and further protectionis unnecessary [3]. This is difficult to measure. Clinicalexamination is unable, reliably, to detect small deflec-tions, less than 3°, at a fracture site [4], a deflection ofthe order of 1° is required to assess a fracture manuallyclose to healing. The probability of correct radiologicalevaluation of the stage of union in tibial fractures hasbeen shown to be only about 50% [5] (Fig. 1). CT scanevaluation [6] and isotope scans [7] have added little tothe overall accuracy of these methods. Ultrasoundimaging in callotasis provides valuable information oncallus quality and dexa-scans of callotasis segments areaccurate using conventional outcome methods and frac-ture stiffness for comparison [8,9]. These rely on thelength of the callotasis segment, somewhat limitingtheir value in healing fractures.

Both radiological methods and clinical assessmentare vital in clinical practice. They allow us to make ajudgement about union, which with many other exter-nal factors allow us to judge when a splint can beremoved. They are precise enough in clinical practice,although it is likely clinicians err mainly on the side ofcaution by keeping splints on far too long. As methodsof comparing similar treatment regimens they havebeen shown to be inaccurate, these inaccuracies areonly increased when differing treatments are compared.Callus is a means to an end but not necessarily themeans to defining an end-point.

3. Fracture stiffness

Fracture stiffness has the potential to provide both ameasure of the rate of healing and an objective defini-tion of union. Since healing is the return of functionand the chief function of bone is the resistance to stress,it is logical to measure the elastic modulus. Fortunatelythe end-point is at a relatively low stiffness comparedwith the intact bone when measurements are relativelyeasy and stiffness correlates with strength [10]. Strengthis defined by failure and as such directly measuringstrength is only possible in the laboratory or by usingfracture failure as an outcome measure.

Stiffness can be measured by vibrational analysis ormechanical testing.

Fig. 1. Which one of these fractures has healed? There is very littledifference in the amount of callus although when the second radio-graph was taken the patient felt the fracture was more stable. The oneon the left had a stiffness of 12 Nm/° and the one on the right astiffness of 18 Nm/°. The fixator was removed on the latter withoutproblems.

R. Wade, J. Richardson / Injury, Int. J. Care Injured 32 (2001) 109–114 111

3.1. Vibrational analysis

The reaction to vibration is a function of stiffness ina healing fracture. As early as 1932 Lippman [11]assessed healing in fractured bone by percussing proxi-mally and auscultating distally. Complex methods ofvibrational analysis have since been developed.

3.1.1. Ultrasound 6elocityUltrasound velocity through the fracture is propor-

tional to the stiffness. However, problems with softtissues have been highlighted [12] and it has been shownthat ultrasound reflects cancellous bone structure inde-pendently of bone density [13].

3.1.2. Impulse responseIn impulse response assessment the bone responds to

a briefly applied force by a free vibration movement.McGaw [14] measured the response to impact using acrystal transducer and found a relationship with theprogression of healing in a fracture. Progress was mon-itored using the ratio of acceleration between the frac-tured limb and uninjured limb [15]. Errors due to softtissues and complex bone shape need to be addressed[16]. This technique has been used to monitor healingtibial fractures in a clinical setting, but remains to bevalidated [17].

3.1.3. Steady state analysisBone responds to forced harmonic vibration by a

steady state if it equals one of the natural frequencies.Steady state analysis relies on this natural frequencywhich is a function of stiffness and mass. A dynamicratio is obtained in comparison to the normal bonewith a good correlation between strain gauge resultsand the ratio. Christensen [18] confirmed that the natu-ral frequency of the healing tibia approaches that of thenormal side during healing. More recently work usingfinite element analysis has shown that torsional stiffnesscan be calculated from vibrational analysis [19].

Using these systems a definition of the end-point‘healing’ is difficult and the practical problems limit theuse of vibration analysis to research at present.

3.2. Mechanical testing

Jorgensen published the first in-depth study of theassessment of the stability of healing tibial fractures bymechanical means. This included measurements on anintact tibia and he was able to demonstrate a pattern ofhealing [20]. The method used involved the applicationof further pins and the consequent risks of an invasiveprocedure. Further studies can be sub-divided into indi-rect and direct stiffness measurements.

3.2.1. Indirect stiffnessAs a fracture heals the load is transferred from the

fixator to maturing callus. This is measured as indirectstiffness and since the external fixator is still in-situthere is no risk of loss of reduction. Burny et al.demonstrated different patterns of healing using straingauges on the external fixator [21]. An objective out-come measure based on stiffness was developed inOxford using a removable strain gauge unit [22]. Pinloosening reduces accuracy [23] but can be monitored[24]. In a prospective randomised trial this techniquedemonstrated that the application of micromovementcorrelated significantly with a faster rate of healing [25].

3.2.2. Direct stiffnessDirect stiffness measurements require the splint to be

removed and are therefore typically measured 6–8weeks after injury in tibial fractures. Angulation at thefracture site is measured alongside the applied load,essentially quantifying clinical assessment of union.

In conservatively treated fractures X-rays have beenused to measure angular change under certain loadingconditions [26] and surface measurements are possibleusing four point bending [27]. Using this system 43closed tibial injuries were studied, the stiffness corre-lated more strongly than callus index with injury sever-ity and functional outcome at 6 months [28]. Once thestiffness reached 7 Nm/° by 20 weeks then no fracturesfailed to heal. Using these stiffness measurements andthe callus index [29], it has been suggested that thecessation of periosteal response before bridging definesdelayed union; whilst non-union is the cessation of bothperiosteal and endosteal healing response withoutbridging.

External fixation allows accurate measurement ofangulation and measurements are not affected greatlyby pin loosening as the measuring device has a lowinherent stiffness. Stiffness measurement of tibial frac-tures in the anterior–posterior plane in patients treatedwith external fixators defines healing as occurring at astiffness of 15 Nm/°. Using this end-point a decreasewas found in the re-fracture rate and a reduction oftime (on average 3 weeks) to independent weight bear-ing [30]. The system is accurate (total error of 3%) [31]and simple to use in the clinical setting. Stiffness hasbeen used to define healing in callotasis patients [32]and more recently a system has been described thatallows torsional stiffness to be measured [33].

Stiffness rises exponentially and therefore allows therate of healing [34] to be monitored. Any delay is seenand intervention can be initiated earlier (Figs. 2 and 3).The measurement takes approximately 10 min andcauses no pain (Fig. 4). A deficiency in the system is theneed to remove the frame to perform the test andtherefore callus cannot be assessed early in healing. Thesystem is designed at present for one-dimensional mea-

R. Wade, J. Richardson / Injury, Int. J. Care Injured 32 (2001) 109–114112

Fig. 2. A plot of increasing stiffness allows the fracture to be monitored throughout healing. (Radiographs 1–6 top left to bottom right). (1) Initialpresentation with Grade IIa open tibial fracture with associated medial malleolar fracture. (2) Treated with debridement, external fixation, andlag screw to malleolus. (3) Five weeks post injury, stiffness 3.0 Nm/°. (4) Eight weeks post injury, stiffness 6.6 Nm/°. (5) Eleven weeks post injury,stiffness 16.1 Nm/°. (6) After removal fixator mobilising free with no problems (advised to avoid heavy use).

surement that may lead to errors due to the fibula orasymmetrical healing. Three-dimensional stiffness mea-surement may address these problems [35].

4. Conclusions

The need for an accurate and precise outcome mea-sure in fracture healing has been highlighted. Radiolog-ical and clinical methods provide the clinician with areasonably precise measure for assessing when a splintcan be removed, but healing times should be interpretedwith care. The comparison of different methods oftreatment using these inaccurate methods is not valid,healing times must be interpreted with caution. Thetime to healing is not always the appropriate outcomemeasure for fractures treated by internal fixation. Di-rect stiffness measurement provides an accurate tool formeasuring healing in fractures treated with externalfixation. This technique has been used to prove thebenefits of micro-movement on fracture healing andmay prove useful in defining the exact role of otherexternal influences.

All factors must be considered when comparing frac-ture treatments. Unfortunately this makes direct com-parison very difficult. It is unlikely that one method ofassessment will define a suitable end-point for all frac-

Fig. 3. The rise in fracture stiffness is exponential up to healing. Theshaded area represents normal healing in tibia. The numbered areasrepresent examples of healing patterns. (1) Normal healing. (2) Dip aspatient failing to weight-bear; improvement when initiated weight-bearing. (3) Delayed union. (4) Non-union.

R. Wade, J. Richardson / Injury, Int. J. Care Injured 32 (2001) 109–114 113

Fig. 4. The measurement of fracture stiffness takes only 10 min in theoutpatient department. This demonstrates that the system can also beused successfully on the femur.

[12] Lowet G, Van der Perre G. Ultrasound velocity measurementin long bones: measurement method and simulation of ultra-sound wave propagation. J Biomech 1996;10(29):1255–61.

[13] Nicholson PHF, Muller R, Lowet G, Cheng XG, HilderbrandT, Ruegsegger P, Van der Perre G, Dequeker J, Boonen S. Doquantitative ultrasound measurements reflect structure indepen-dently of density in human vertebral cancellous bone? Bone1998;23:425–31.

[14] McGaw NH. Osseosonometry. Arch Surg 1942;42:195.[15] Lewis JL, Tarr RR. Fracture healing assessment by impact

response: clinical results. J Bone Joint Surg 1975;57A(SupplI):576.

[16] Thompson GA, Young DR, Orne D. In-vivo determination ofmechanical properties of the human ulna by means of mechan-ical impedance tests: experimental and improved mathematicalmodels. Med Biol Eng 1976;14:253.

[17] Fellinger M, Leitgeb N, Szyszkowitz R, Peicha G, Passler J,Seggl W, Schanner A. Early detection of delayed union inlower leg fractures using computerised analysis of mechanicalvibration reactions of bone for assessing the state of fracturehealing. Acta Orthop Trauma Surg 1994;113:93–6.

[18] Christensen AB. Resonance of the human tibia, method, re-producibility and effect of transection. Acta Orthop Scand1982;53:867.

[19] Lowet G, Dayaun X, Van der Perre G. Study of vibrationalbehaviour of a healing tibia using finite element modelling. JBiomech 1996;29(8):1003–11.

[20] Jorgensen A. Measurement of stability of tibial fractures. Amechanical method. Acta Orthop Scand 1970;35:7–88.

[21] Burny F, Bourgois R, Donkerwolcke M, Moulart F. Utilisa-tion clinique de jauges de contrainte-situation actuelle et per-spective d’avenir. Acta Orthop Belg 1978;44:895–920.

[22] Evans M, Kenwright J, Cunningham JL. Design and perfor-mance of a fracture monitoring transducer. J Biomed Eng1988;10:64–9.

[23] Churches AE, Tanner KE, Harris JD. The oxford externalfixator: fixator stiffness and effects of bone pin loosening. EngMed 1985;14(1):3–11.

[24] Kay PR, Ross ERS, Powell ES. Development and clinical ap-plication of an external fixator monitoring system. J BiomedEng 1989;11:240–4.

[25] Kenwright J, Richardson JB, Goodship AE, Evans M, KellyDJ, Spriggins AJ, Newman JH, Burrough SJ, Harris JD, Row-ley DI. Effect of controlled axial micromovement on healingof tibial fracutres. Lancet 1986;22:1185–7.

[26] Hammer R, Edholm P, Lindholm B. Stability of union aftertibial shaft fracture. J Bone Joint Surg 1984;66B:529–34.

[27] Taktak AFG, Edwards J, Kay P, Laycock DC. Improvementin the monitoring of fracture repair by non-invasive mechani-cal method. In: Micromovement in Orthopaedics, Oxford En-gineering Series 10. UK: Oxford Medical Publications,1993:140–51.

[28] Marsh D. Concepts of fracture union, delayed union andnonunion. Clin Orthop 1998;355S:22–30.

[29] Oni OOA, Dunning J, Mobbs RJ, Gregg PJ. Clinical factorsand the size of external callus in tibial shaft fractures. ClinOrthop 1991;273:278–83.

[30] Richardson JB, Cunningham JL, Goodship AE, O’Connor BT,Kenwright J. Measuring stiffness can define healing of tibialfractures. J Bone Joint Surg 1994;76B(3):389–94.

[31] Hardy JRW, de Jong EJ, Richardson JB. Fracture stiffness. JOrthop Trauma 1994;2(4):177–89.

[31] Hardy JRW, de Jong EJ, Richardson JB. Fracture stiffness. JOrthop Trauma 1994;2(4):177–89.

tures and all fracture treatment methods. The smallrisk of infection when a fracture is internally fixed,the risk of malunion with fractures treated in plasterand the risk of pin sepsis using external fixation needto be specified relative to the definition of the end-point.

References

[1] Lane WA. The Operative Treatment of Fractures, 2nd ed.London: The Medical Publishing Company, 1914.

[2] Page CM. Time to union in the fractured tibia. Br Med J1942;1:305.

[3] Apley A.G., Solomon L. Apleys system of orthopaedics andfractures. 7th edn. Butterworth and Heinemann.

[4] Hammer R, Norrbom H. Evaluation of fracture stability. ActaOrthop Scand 1984;5:330–53.

[5] Hammer R, Hammerby S, Lindholm B. Accuracy of radiologi-cal assessment of tibial shaft fractures in humans. Clin OrthRR 1985;199:233–8.

[6] Schnarkowski P. Tibial shaft fracture healing assessment withC.T. J Comp Ass Tom 1995;19(5):777–81.

[7] Smith M.A., Jones E.A., Strachan R.K., Nicholl J.J., Best J.J.,Tothill P., Hughes S.P. Prediction of fracture healing in thetibia by quantitive radionucleotide imaging, J Bone Joint Surg69B:441–447.

[8] Hughes TH, Maffuli N, Green V, Fixsen JA. Imaging in bonelengthening. Clin Orthop RR 1994;308:50–3.

[9] Tselentakis G, Owen PJ, Richardson JB, Kuiper JH, Had-daway MJ, Dwyer JSM, Evans GA. Dexa-scan can be used todetermine fracture stiffness of callostasis segment. J Bone JointSurg 1996;78B(Suppl 2/3):135.

[10] Cheade MK, Pohl AP, Pearcy MJ, Nawana N. Clinical impli-cations of stiffness and strength in fracture healing. J BoneJoint Surg 1997;79B:9–12.

[11] Lipmann RK. The use of auscultatory percussion for the ex-amination of fractures. J Bone Joint Surg 1932;14:118.

R. Wade, J. Richardson / Injury, Int. J. Care Injured 32 (2001) 109–114114

[32] Dwyer JSM, Owen PJ, Evans GA, Kuiper JH, Richardson JB.Stiffness measurements to assess healing during leg lengthening —A preliminary report. J Bone Joint Surg 1996;78B(2):286–9.

[33] Windhagen et al. A new device to quantify regenerate torsionalstiffness in distraction osteogenesis, J Biomech 1999;Aug32(8):857–860.

[34] Richardson JB, Kenwright J, Cunningham JL. Fracture stiffnessmeasurement in the assessment and management of tibial frac-tures. J Clin Biomech 1992;7:75–9.

[35] Wade RH, Kuiper JH, Richardson JB. Three-dimensional fracturestiffness. Injury 1999;30:562.

.