REVIEWARTICLE

Predictors associated with nonunion and symptomatic malunionfollowing non-operative treatment of displaced midshaftclavicle fractures—a systematic review of the literature

Ann Jørgensen & Anders Troelsen & Ilija Ban

Received: 20 May 2014 /Accepted: 25 June 2014# SICOT aisbl 2014

AbstractPurpose The aim of this study was to survey existing litera-ture in order to identify all reported predictors associated withnonunion or symptomatic malunion in adult patients withdisplaced midshaft clavicle fractures treated non-operatively.Method A systematic literature search in Medline was carriedout in order to identify publications in English, reporting onpredictors for nonunion and malunion in adults with displacedmidshaft clavicle fractures. After applying inclusion and ex-clusion criteria, eight publications were included in this sys-tematic review.Results A total of 2,117 midshaft clavicle fractures wereincluded in the eight publications. All publications reportedon predictors for nonunion but none were found to report onpredictors for malunion. The studies were characterized bydifferent definitions for nonunion and symptomatic malunionif at all present. A total of 13 potential factors associated withnonunion were identified, six of these (displacement, commi-nution, shortening, age, gender and smoking) were reported aspredictors for nonunion. Outcome definitions varied amongthe studies.Conclusion The included publications varied greatly in de-sign, sample size, and quality. Based on the present literaturemost of the predictors were found to be of limited evidence,however displacement seems to be the most likely factor thatcan be used to predict for nonunion. Treating all claviclefractures with displacement surgically would inevitably leadto overtreatment, which is why future studies need to focus onpredictive factors in order to differentiate between patients thatwould benefit from surgery and those who would not.

Keywords Claviclefracture .Midshaftfracture .Non-operativetreatment . Nonunion .Malunion . Risk factors

Introduction

Despite being a common fracture, representing 5–10 % of allorthopaedic fractures [1], evidence of optimal treatment ofdisplaced clavicle fractures is still ambiguous [2]. Allocationof primary treatment of these fractures has gone from system-atic non-operative treatment towards surgical treatment ofmost fractures especially those with shortening and displace-ment of a whole bone width in patients with a high physicaldemand. This shift of treatment strategy is probably due toreports of higher nonunion and symptomatic malunion ratesthan first reported by Neer and Rowe in the 1960s [3, 4].

Recent literature describes primary surgical treatment ofdisplaced midshaft clavicle fractures, with locking plateosteosynthesis as the preferred method, to result in goodfunctionality, high union rates, and few complications [5–7].Non-operative treatment, on the other hand, results in slightlyinferior functional outcome compared to surgical treatment [5,6]. The nonunion rate is reported as high as 20 %, and close to30 % of all patients are reported to have symptomaticmalunion [5]. However, numbers-needed-to-treat (NNT) anal-ysis from several studies indicates that a high number ofsurgical treatments need to take place to avoid one nonunionor malunion compared to non-operative treatment [8–10]. Toavoid overtreatment of displaced midshaft clavicle fracturesfocus should probably be on identification of patients that willbenefit from surgery.

The aim of this study is therefore to survey existing liter-ature in order to identify all reported factors that can be used topredict nonunion or symptomatic malunion in adult patientswith displaced midshaft clavicle fractures treated non-operatively (Fig. 1).

A. Jørgensen (*) :A. Troelsen : I. BanDepartment of Orthopaedic Surgery, Copenhagen UniversityHospital Hvidovre, Kettegård Alle 30, 2650 Hvidovre, Denmarke-mail: annjo@dadlnet.dk

International Orthopaedics (SICOT)DOI 10.1007/s00264-014-2450-7

Method

Publications in English reporting on factors that may influencethe rate of nonunion or symptomatic malunion in adult pa-tients with displaced midshaft clavicle fractures were identi-fied performing a Medline search in September 2013. Allpublications regardless of publication year were included.Clavicle fracture, nonunion, malunion, risk factors, conserva-tive treatment, non-operative treatment, non-surgical treat-ment, middle third fracture, and midshaft fracture were allkeywords used in combination to produce the followingsearch query:

((((((((clavicle fracture malunion) OR clavicle fracturenonunion) AND middle third fracture) OR (((claviclefracture malunion) OR clavicle fracture nonunion) ANDmidshaft fracture)) OR (((clavicle fracture nonunion) ORclavicle fracture malunion) AND conservative treatment))OR (((risk factors) AND clavicle fracture nonunion) ORclavicle fracture malunion)) OR (((clavicle fracturemalunion) OR clavicle fracture nonunion) AND nonsurgical treatment)) OR (((clavicle fracture malunion)OR clavicle fracture nonunion) AND non operativetreatment))

Two of the authors independently surveyed the searchresults by review of title and abstracts to identify poten-tially eligible studies. Primary selection criteria were toinclude only English written publications with a designbased on either randomized clinical trials or prospectiveand retrospective cohort studies including studies with aminimum of 20 patients. The target population was adults.Publications including patients under the age of 18 wereincluded when they were considered adults (adult bonemorphology) by the authors of the publication. The inter-vention of interest was non-operative treatment of patientswith only midshaft clavicle fractures and outcome of inter-est was all possible factors (demographic, lifestyle or frac-ture related factors) influencing the rate of nonunion ormalunion. To ensure eligibility based on the selectioncriteria the same two authors evaluated the manuscript ofthe identified publications. A total of eight publicationsmet the criteria (Fig. 2).

Results

Literature

While all eight publications that were included reported onpredictors for nonunion, none reported on predictors for de-veloping symptomatic malunion. The primary aim of four ofthe eight publications was to identify risk factors associatedwith nonunion ormalunion [8, 11–13] with the remaining fourpublications reporting on predictors as a secondary aim[14–17]. Furthermore, two of the publications had the aim ofreporting onmalunion but published only results for nonunion[13, 14].

The focus of six of the publications was midshaft fractures[8, 12, 14–17] with three including both displaced andundisplaced fractures [12, 14, 16] and three focused oncompletely displaced fractures [8, 15, 17]. The remainingtwo publications included all types of clavicle fractures, butreported outcome and predictors separately for all midshaftfractures [11, 13]. Based on all publications a total of 2,117midshaft clavicle fractures were included with 2,084 of theseinitially treated non-operatively. Of the 2,084 fractures initial-ly treated non-operatively, 1,748 were displaced, 247 were notdisplaced, and 89 fractures had not been accounted for regard-ing displacement. Fifty-five developed symptomaticmalunion. In seven publications 195 developed nonunion [8,13–16]. In the publication by Robinson et al. the cumulatedprobability for nonunion after 24 weeks was 4.5 %. Becauseof the survival analysis method it was not possible to accountfor exact numbers for nonunion [11]. Adolescents were in-cluded in four studies [11, 12, 14, 16] with the remainingstudies reporting on adults aged 18 years or over (Table 1) [8,13, 15, 17].

Outcome definition

Nonunion was defined in four of eight studies reporting onnonunion [8, 11, 13, 17]. Lack of radiological, or clinicalhealing, or lack of both after 24 weeks following injury wasused in two studies [8, 11] with a defining limit of 16 weeks[12] and 52 weeks [17], respectively, in the two other studies.The remaining four studies did not define nonunion [13–16],but in one of the studies they performed surgery six weeksfollowing injury if no sign of callus was found in a symptom-atic patient [13].

In two [16, 17] of four studies [13, 14, 16, 17] aiming atreporting on symptomatic malunion, a definition of symptom-atic malunion was given. Both of these definitions included adefinition of bony malunion in which one study was definedas shortening of more than 20 mm, angulation, or displace-ment, and in the other as displacement more than a bonewidth, or angulation more than 30 degrees. In one of thedefinitions the bony malunion had to be in combination with

Fig. 1 Displaced and comminuted midshaft clavicle fracture

International Orthopaedics (SICOT)

a symptomatic patient [17], whereas this was not defined inthe other study [16].

Predictors for non- and malunion

Statistical estimation of a potential factor as a predictor dif-fered from study to study according to the aim. Thirteendifferent potential factors associated with nonunion were in-vestigated in the eight included studies. Six of these factors(smoking, comminution, displacement, age, gender, fractureshortening) were reported as a predictor or risk factor fordeveloping nonunion in at least one of the publications,whereas the remaining seven factors (final fracture shortening,presence of vertical fragment, presence of associated injury,fracture angulation, fracture translation, occupation, mecha-nism of injury) did not seem to have any predictive effect.

Smoking was described as a predictor of nonunion in onestudy where 73 of 219 patients with a smoking habit devel-oped nonunion (OR 3.76) [8]. It was not found to predictnonunion in two other studies [15, 17]. Fracture comminutionwas found to be a predictor of nonunion in two studies: in one,75 of 352 developed nonunion (OR 1,75) [8] and in the otherthe cumulated probability for nonunion was 4.5 % after24 weeks (RR 1,45) [11]. Two studies found no associationbetween nonunion and fracture comminution [14, 15]. As forfracture displacement four studies found this to be a predictor;in one study, 125 of 941 developed nonunion (OR 1,17) [8]and in another the cumulated probability for nonunion was4.5 % (RR 2,32) [11]. Two other studies found that six of 25and six of 154, respectively, developed nonunion [16, 17].Still one study did not find displacement to be a predictor, andin this study five of 68 developed nonunion [14]. Age and

Total number of publications MEDLINE (Pubmed) Search, n = 242

Potentially relevant publications evaluated to title and abstract, n = 213

Publications for evaluation on design and endpoint, n = 60

Publications excluded, n = 29

Reason- Not in English, n = 29

Publications excluded, n = 153

Reason- Not relevant design, n = 60- Not relevant topic, n = 56- Child or adolescent, n = 10- Case series < 20 patients, n = 26- Duplicate article, n = 1

Publications included, n = 8

Murray 2013Virtanen 2012Postacchini 2009Kirmani2009Robinson 2004Wick 2000Nordqvist 1998Hill 1997

Publications excluded, n = 52

Reason- Not relevant design, n = 8- Not relevant target poplulation, n = 3 - Not relevant end points, n = 36- Case series < 20 patients, n = 3- Article not available, n = 2

Fig. 2 Flowchart of searchstrategy

International Orthopaedics (SICOT)

gender were both found to be predictors of nonunion in onestudy (RR 1.01 and 1.43, respectively) [11] whereas two otherstudies did not find these associations [8, 15]. Clavicle short-ening was found to be a predictor of nonunion in two studies[13, 15]. The first study included only non-united fracturesand reported on shortening as a predictor of nonunion in titleonly without further mentioning in the manuscript. In thisstudy 30 of 33 fractures had shortening of more than 2 cm.There was no mentioning of degree of shortening in theremaining three [13]. The second study reported on initialand not final shortening to be a predictor. Forty-three of 52fractures had initial shortening of 1–25 mm and eight of these

fractures resulted in nonunion. Shortening by at least 2 cmwasfound in six of these eight patients with no reports on short-ening of the remaining two [15]. This is opposed to fourstudies which found no association between shortening andnonunion [8, 11, 14, 17].

Discussion

The search and review of existing literature revealed relativelyfew publications on predictors for nonunion in patients with

Table 1 Characteristics for publications and fractures included in the review

Publication Design Number of midshaftfractures included foranalysis

Number ofnonunions andmalunions

Clavicle fracture typein publication

Age (years)patients withnonunions andmalunions

Aim

Murray et al.2013

Retrospectivecohort

941 125 with nonunionafter displacedmidshaft fractures

Displaced midshaftfractures, nocortical contactbetween mainfragments

18 and older To identify patients with ahigher risk of nonunion

Virtanen et al.2012

RCT 60 Six with nonunionand two withmalunion afterdisplaced midshaftfractures

Displaced midshaftfractures, nocortical contactbetween mainfragments

18–70 Functional outcomefollowing non-operativeor operative treatment

Postacchini et al.2009

Retrospectivecohort

91 Five with nonunionafter displacedmidshaft fractures.Malunions: nonumbers

Midshaft fractures 43–64 (all patientsin publication:13–70)

Long-term outcome afternon-operative treatmentincluding the incidenceof nonunion andmalunion

Kirmani et al.2009

Retrospectivecohort

134 11 with nonunionafter a midshaftfracture (six withvertical fragment,five without)

Midshaft fractures 17–95 Likeliness of undergoingearly or late surgery aftermidshaft fracture withvertical fragment

Robinson et al.2004

Prospectivecohort

581 4.5 % cumulatedprobability after24 weeks, fornonunion after amidshaft fracture

All types of fractures 14 and older Rate of nonunion after non-operative treatment andrisk factors for nonunion

Wick et al. 2000 Retrospectivecase series

33 33 with nonunionafter a midshaftfracture

All types of fractures 18–59 Definition of a certain typeof fracture thatpredisposes to delayedor malunion after non-operative treatment

Nordqvist et al.1998

Retrospectivecase series

225 Seven with nonunionafter a midshaftfracture (one inundisplacedfracture and six indisplaced); 53with malunion

Midshaft fractures 15–70 To analyse long-termoutcome of midshaftfractures in adults andevaluate the clinicalimportance ofdisplacement andfracture comminution

Hill et al. 1997 Retrospectivecase series

52 Eight with nonunionafter displacedmidshaft fractures

Displaced midshaftfractures

18–59 Outcome after non-operative treatment ofdisplaced midshaftfractures

International Orthopaedics (SICOT)

midshaft clavicle fractures treated non-operatively.Interestingly no studies were found reporting on predictorsfor developing symptomatic malunion. Comparing the publi-cations it seems that study aim, outcome parameters, fracturetypes, and fracture characteristics show great variation, mak-ing it difficult to compare results between studies. Also thereis no consensus in defining nonunion as only half of thestudies had a definition, but also because of great variationin duration from injury to announcing a nonunion. Howeverthe review revealed that six of 13 potential factors seemed tobe predictors for nonunion in patients with midshaft claviclefractures. The factors that could influence development ofnonunion included smoking, presence of fracture comminu-tion, complete fracture displacement, increased age, femalesex, and fracture shortening. For each factor found to be apredictor in one or more of the publications, there was at leastone other publication that did not find any association betweenthe factor and development of nonunion (Table 2).

Fracture displacement following a clavicle fracture wasreported to predict pain and dysfunction in two studies bothwith follow-up of more than two years [18, 19]. In this studyfracture displacement was found to be a predictor for non-union in all but one publication reporting on this subject [8,11–13, 15–17]. The studies finding fracture displacement tobe a predictor for nonunion varied in quality, definition ofdisplacement, measurement of displacement and ranged fromsmall to large cohorts and from retrospective designs to RCTdesign [8, 11, 13, 15–17]. Nevertheless, all but one studyfound displacement to be a predictor for nonunion, thuspointing in the direction of this particular predictor to belikely.

Shortening of the clavicle following a midshaft claviclefracture has been found to be associated with inferior clinicaloutcome [20, 21]. In an unpublished survey of Danish hospi-tals treating clavicle fractures, 16 of 20 clinics treat midshaft

clavicle fractures operatively if displacement and shortening ispresent. However, the association between shortening andinferior outcome is controversial [18, 22] and consideringthe risk of nonunion, we found only two studies identifyingshortening as a predictor of developing nonunion [13, 15].Both of these studies were retrospective studies of smallcohorts and a standardized radiological method to evaluatethe shortening was not utilized as recommended [23]. Basedon this it is doubtful that shortening is a risk factor of non-union, especially since four other studies including arandomised controlled trial and two large prospective studiesdid not find any association. Most studies on shortening focuson deficits and symptoms following malunited healing andnot on whether it is a predictor for non- or malunion [20–22].

A publication by Nowak et al. recognized fracture commi-nution as a significant risk factor for pain, strength reduction,and cosmetic defects after a clavicle fracture but not as apredictor for nonunion [24]. We found two cohort studies oflarge sample size reporting comminution to be a predictor fornonunion in midshaft fractures [8, 11] whereas two otherstudies did not find any association between comminutionand nonunion [14, 15]. In general precise definition of com-minution is lacking in all publications and the orientation offragments is not mentioned. Thereby fracture comminution inthe sense used in these publications is covering a wide rangeof heterogenic fractures and soft tissue injuries and compari-son is difficult. There is a need for further studies on this topic.

In the literature age is mentioned as a significant risk factorfor pain, strength reduction, cosmetic defects and paresthesiaand a lower Constant score [22, 24], but also in more recentpublications repeatedly mentioned as a predictor for nonunion[2, 5, 23]. Nowak et al. investigated 222 patients and found noassociation between age and nonunion but in the analysis alltypes of fractures and a mix of undisplaced as well asdisplaced fractures were included which complicates the

Table 2 Predictors for nonunion

Predictors Murrayet al.

Virtanen et al. Postacchiniet al.

Kirmani et al.a Robinsonet al.

Wick et al. Nordqvistet al.

Hill et al. + −

OR No RR or OR No RR or OR RR No RR or OR No RR or OR No RR or OR

Smoking + (3,76) − − 1 2

Fracture comminution + (1,75) − + (1,45) − 2 1

Presence of displacement + (1,17) + − + (2,32) + 4 1

Vertical fragment − 0 1

Age − + (1,01) − 1 2

Gender − + (1,43) − 1 2

Final shortening − 0 1

Shortening − − − − + + 2

+ predictor in the study, − not a predictor

RR risk ratio, OR odds ratioa The relative risk here for a midshaft fracture that displays a vertical fragment to undergo delayed surgery due to symptomatic nonunion is 2.2

International Orthopaedics (SICOT)

interpretation of the results [24]. In this review only Robinsonet al. found age to be a predictor for nonunion but the riskseems only increased by a factor of 0.01 for each year ofincreasing patient age [11]. Nevertheless, the work ofRobinson et al. is constantly used by others to advocate ageas a predictor for nonunion, and in recent publicationsreviewing existing knowledge on clavicle fractures all de-scribe age as a positive predictor by referring to the publica-tion of Robinson et al. [1, 2, 5, 11, 23]. The same seemsapplicable for gender which is found to be a predictor fornonunion only in the publication by Robinson et al. Again thisresult is seen repeated by several others [2, 5, 23]. In thisreview two other studies did not find any relation betweengender and development of nonunion, and in literature genderis not found associated with a worse outcome after a claviclefracture [24] or a lower Constant score after a midshaft frac-ture [22]. Increasing age and gender as a predictor for non-union in midshaft fractures must therefore be viewed asdoubtful.

The effect of smoking on fracture healing in general stillseems controversial as reviews including large number ofpublications finds a negative influence on healing of specificfractures (tibial, spinal, and foot and ankle fractures) but areunable to show a negative effect on bone healing in general[25, 26]. When it comes to clavicle fractures this reviewrevealed that two studies both small in sample size did notfind any association between smoking and nonunion. Thisconcurs with findings of another larger study that concludesthat smoking is not considered as a risk factor for worseoutcome after clavicle fracture [24]. However, in this review,a single study including a large number of adult patients andamongst these 219 with a smoking habit, reported smoking tobe the strongest predictor for nonunion following a midshaftclavicle fracture [8]. A challenge for this publication is the riskof recall bias as there were no records of tobacco use forpatients treated before 1997, and smoking habit was recordedretrospectively for 382 patients.

The question whether smoking is a predictor for nonunionin a midshaft fracture still seems controversial based onexisting literature.

In conclusion six factors were described as predictors fornonunion, however several of these factors have only limitedevidence. Based on the present literature displacement isfound to be the most likely predictor for nonunion and shouldbe taken in account when counselling patients (Table 3). Mostof the studies reporting on potential predictors are small insample size, without standardised definitions and outcomeevaluations and most importantly often not primarily aimedat identifying predicting factors. Second, for each potentialpredictor at least one other study did not find an association tononunion. All existing randomised studies comparing surgicaland non-surgical treatment of displaced midshaft fracturesfavour surgical treatment based on slightly better functional

outcome. With NNT analysis suggesting overtreatment ofpatients if all displaced fractures were to be treated surgically,the future focus and treatment of midshaft fractures needs totake potential predictive factors in account. With inadequateevidence future studies need to focus on this in order todifferentiate between patients that would benefit from surgeryand those who would not.

Conflict of interest The authors declare that they have no conflict ofinterest.

References

1. Robinson CM (1998) Epidemiology and classification. J Bone JointSurg Br 80-B:476–484

2. Khan LA, Bradnock TJ, Scott C, Robinson CM (2009) Fractures ofthe clavicle. J Bone Joint Surg Am 91:447–460. doi:10.2106/JBJS.H.00034

3. Neer CS 2nd (1960) Nonunion of the clavicle. J AMMedAssoc 172:1006–1011

4. Rowe CR (1968) An atlas of anatomy and treatment of mid-clavicular fractures. Clin Orthop Relat Res 58:29–42

5. Ban I, Branner U, Holck K, Krasheninnikoff M, Troelsen A (2012)Clavicle fractures may be conservatively treated with acceptableresults—a systematic review. Dan Med J 59:A4457

6. Liu G, Tong S, Ou S, Zhou L, Fei J, Nan G et al (2013)Operative versus non-operative treatment for clavicle fracture:a meta-analysis. Int Orthop 37:1495–1500. doi:10.1007/s00264-013-1871-z

7. Fridberg M, Ban I, Issa Z, Krashneninnikoff M, Troelsen A (2013)Locking plate osteosynthesis of clavicle fractures: complication andreoperation rates in one hundred and five consecutive cases. IntOrthop 37:689–692. doi:10.1007/s00264-013-1793-9

8. Murray IR, Foster CJ, Eros A, Robinson CM (2013) Risk factors fornonunion after nonoperative treatment of displaced midshaft frac-tures of the clavicle. J Bone Joint Surg Am 95:1153–1158. doi:10.2106/JBJS.K.01275

9. Robinson CM, Goudie EB,Murray IR, Jenkins PJ, AhktarMA, ReadEO et al (2013) Open reduction and plate fixation versus nonopera-tive treatment for displaced midshaft clavicular fractures: a multicen-ter, randomized, controlled trial. J Bone Joint Surg Am 95:1576–1584. doi:10.2106/JBJS.L.00307

Table 3 Events interfering with healing of midshaft clavicle fracturesbased on existing literature

Likely to be a riskfactor for nonunion

Doubtfully a riskfactor for nonunion

Not a risk factorfor nonunion

Presence ofdisplacement

Smoking Vertical fragment

Fracture comminution Final shortening

Age Injury mechanism

Gender Amount of translation

Shortening Fracture angulation

Occupation

Presence of associatedinjuries

International Orthopaedics (SICOT)

10. McKee RC, Whelan DB, Schemitsch EH, McKee MD (2012)Operative versus nonoperative care of displaced midshaft claviclefractures: a meta-analysis of randomized clinical trials. J Bone JointSurg AM 94:675–684. doi:10.2106/JBJS.J.01364

11. Robinson CM, Court-Brown CM, McQueen MM, Wakefield AE(2004) Estimating the risk of nonunion following nonoperative treat-ment of a clavicular fracture. J Bone Joint Surg Am 86-A:1359–1365

12. Kirmani SJ, Pillai SK, Madegowda BR, Shahane SA (2009) Verticalfragment in adult midshaft clavicle fractures: an indicator for surgicalintervention. Orthopedics 32:9–14. doi:10.3928/01477447-20090818-06

13. Wick M, Müller EJ, Kollig E, Muhr G (2001) Midshaft fractures ofthe clavicle with a shortening of more than 2 cm predispose tononunion. Arch Orthop Trauma Surg 121:207–211

14. Postacchini R, Gumina S, Farsetti P, Postacchini F (2010) Long-termresults of conservative management of midshaft clavicle fracture. IntOrthop 34:731–736. doi:10.1007/s00264-009-0850-x

15. Hill JM, McGuire MH, Crosby LA (1997) Closed treatment ofdisplaced middle-third fractures of the clavicle gives poor results. JBone Joint Surg Br 79:537–539

16. Nordqvist A, Petersson C, Redlund-Johnell I (1998) Mid-claviclefractures in adults: end result study after conservative treatment. JOrthop Trauma 12:572–576

17. Virtanen KJ, Remes V, Pajarinen J, Savolainen V, Björkenheim J-M,Paavola M (2012) Sling compared with plate osteosynthesis fortreatment of displaced midshaft clavicular fractures, a randomizedclinical trial. J Bone Joint Surg Am 94:1546–1553. doi: 10.2106/JBJS.J.01999

18. Nowak J, Holgersson M, Larsson S (2004) Can we predict long-termsequelae after fractures of the clavicle based on initial findings? A

prospective study with nine to ten years of follow-up. J ShoulderElbow Surg 13:479–486. doi:10.1016/j.jse.2004.01.026

19. Eskola A, Vainionpää S, Myllynen P, Pätiälä H, Rokkanen P (1986)Outcome of clavicular fracture in 89 patients. Arch Orthop TraumaSurg 105:337–338

20. Lazarides S, Zafiropoulos G, Tydfil M (2006) Conservative treatmentof fractures at the middle third of the clavicle: the relevance ofshortening and clinical outcome. J Shoulder Elbow Surg 15:191–194. doi:10.1016/j.jse.2005.08.007

21. McKee MD, Pedersen EM, Jones C, Stephen DJ, Kreder HJ,Schemitsch EH, Wild LM, Potter J (2006) Deficits following non-operative treatment of displaced midshaft clavicular fractures. J BoneJoint Surg Am 88:35–40. doi:10.2106/JBJS.D.02795

22. Rasmussen JV, Jensen SL, Petersen JB, Falstie-Jensen T, Lausten G,Olsen BS (2011) A retrospective study of the association betweenshortening of the clavicle after fracture and the clinical outcome in136 patients. Injury 42:414–417. doi:10.1016/j.injury.2010.11.061

23. Smekal V, Irenberger A, Struve P, Wambacher M, Krappinger D,Kralinger FS (2009) Elastic stable intramedullary nailing versusnonoperative treatment of displaced midshaft clavicular fractures-arandomized, controlled, clinical trial. J Orthop Trauma 23:106–112.doi:10.1097/BOT.0b013e318190cf88

24. Nowak J, Holgersson M, Larsson S (2005) Sequelae from clavicularfractures are common: a prospective study of 222 patients. ActaOrthop 76:496–502. doi:10.1080/17453670510041475

25. Patel RA, Wilson RF, Patel PA, Palmer RM (2013) The effect ofsmoking on bone healing: a systematic review. Bone Joint Res 2:102–111. doi: 10.1302/2046-3758.26.2000142

26. Al-Hadithy N, Sewell MD, Bhavikatti M, Gikas PD (2012) Theeffect of smoking on fracture healing. Acta Orth Belg 78:285–290

International Orthopaedics (SICOT)