new classification system for long-bone

MAY 2012 | Volume 35 • Number 5

n Feature Article

abstractFull article available online at Healio.com/Orthopedics. Search: 20120426-26

This article describes a novel, clinically oriented classification system for long-bone fractures that is simple, reliable, and useful to predict treatment method, complica-tions, and outcome. The reliability and memorability of the new classification were statistically tested and compared with the AO-Müller/Orthopaedic Trauma Association (AO/OTA) long-bone fracture classification. The proposed classification system was also clinically validated with a targeted pilot study designed for content and clinical outcome retrospectively reviewing 122 closed tibial shaft fractures, which were used as a representative paradigm of long-bone fractures. Statistical evaluation showed that the proposed classification system had improved inter- and intraobserver variation agreement and easier memorability compared with the AO/OTA classification system. The clinical validation study showed its predictive value regarding selection of treat-ment method, complication rate, and injury outcome.

The proposed classification system proved simple, reliable, and memorable. Its clini-cal value appeared strong enough to justify the organization of larger studies for a complete assessment of its clinical usefulness for all long-bone fractures.

Drs Garnavos, Kanakaris, Lasanianos, and Tzortzi are from the Orthopaedic Department, Evangelismos General Hospital, Athens, Greece; Dr Kanakaris is also from the Academic Department of Trauma and Orthopaedics, Leeds Teaching Hospitals, and Dr West is from Leeds Institute of Health Sciences, Centre for Epidemiology & Biostatistics, University of Leeds, Leeds, United Kingdom.

Drs Garnavos, Kanakaris, Lasanianos, Tzortzi, and West have no relevant financial relationships to disclose.

Correspondence should be addressed to: Christos Garnavos, MD, PhD, Orthopaedic Department, Evangelismos General Hospital, 5 Poseidonos Str, Glyfada 16674, Athens, Greece ([email protected]).

doi: 10.3928/01477447-20120426-26

New Classification System for Long-bone Fractures Supplementing the AO/OTA ClassificationChristos Garnavos, MD, PhD; nikolaos k. kanakaris, MD, PhD; nikolaos G. lasanianos, MD, PhD, MsC; Paraskevi tzortzi, MD; robert M. West, PhD

Figure 1: Topography of long-bone segments: D, distal; M, middle; MD, middle to distal; P, proximal; PM, proximal to middle).

1

Figure 2: Morphology of long-bone segments: St, simple-transverse; Ss, simple-spiral; I, intermedi-ate; C, complex.

2

e709

Healio.com The new online home of ORTHOPEDICS | Healio.com/Orthopedics

n Feature Article

Long-bone fractures are common trauma injuries. Compared with periarticular areas, where sev-

eral fracture pattern classifications have been described, long-bone fractures are classified exclusively according to the AO-Müller long-bone classification system and to its 1996 revision by the Orthopaedic Trauma Association (OTA) compendium.1,2

However, the revised version of the AO/OTA classification system has re-ceived criticism for being complicated and for having low inter- and intraob-server variation agreement and reliability. It has never been adequately tested for its usefulness regarding treatment selection and clinical outcomes.3-6 The necessity of a simple classification system for long-bone fractures has been reported4,5,7-10 and is supported by numerous validation studies dealing with the existing AO/OTA system.3,4,6 Moreover, specific guidelines have been described on the optimal way to introduce a novel classification system or revise an existing classification sys-tem.10-12

This article proposes the simple Garnavos classification system for classi-fying long-bone fractures, oriented for ev-eryday clinical use. It was tested and com-pared with the AO/OTA long-bone frac-ture classification system with inter- and intraobserver variation reliability studies. In addition, the current study provides the required evidence that the new classifica-tion system is applicable to a representa-tive sample of long-bone shaft fractures for which it has been designed (face and content validation) and also assesses the correlation of the proposed classification to mechanisms of injury, various treatment options, complication rates, rehabilitation issues, and the main clinical endpoint of long-bone fracture treatment.

Materials and MethodsGarnavos Classification System

Definitions of Long-bone Segments. Femur: The bone segment between 2 paral-

lel lines that cross the long axis of the femur trans-versely. One passes just below the lesser trochanter and the other through the maximum diameter of the femoral condyles.

Tibia: The bone seg-ment between 2 parallel lines that cross the long axis of the tibia transverse-ly. One passes just above the fibular head or the proximal end of the tibial tuberosity and the other just above the base of the medial malleolus or the point where the tibia separates from the fibula.

Humerus: The bone segment between 2 parallel lines that cross the long axis of the humerus transversely. One passes through the surgical neck of the humeral head and the other 1 cm above the proxi-mal tip of the olecranon fossa.

Radius and ulna: The bone segments between 2 parallel lines that cross the long axis of the radius and ulna transversely. One passes through the base of the radial neck and the other through the base of the ulnar head (Figure 1).

Garnavos Classification System. In the new classification system, fractures are first described by their location (topog-raphy) and second by their pattern (mor-phology).

Fracture topography: The long-bone segments are divided into 3 zones of equal length. The most proximal (head-to-foot direction) zone is assigned the capital let-ter P (for proximal), the middle zone is assigned the capital letter M (for middle), and the most distal zone is assigned the capital letter D (for distal). The location of a fracture is described by the appropri-ate capital letter (P, M, or D). The fracture must be completely contained within the relevant zone to be described by only 1 ini-tial. When a fracture extends to .1 zone or occurs in the transition area between 2 zones, its location is described by 2 or 3 initials from proximal to distal. If a fracture

is located in the transition zone between the proximal and middle thirds of a long-bone segment, it is described as PM. If it is located in the transition zone between the middle and distal thirds, it is described as MD. If a fracture extends to the whole bony segment, it is described as PMD.

Classification is based on antero-posterior (AP) and lateral radiographs. However, additional imaging studies, such as oblique radiographs or computed tomography scan, may offer additional information about the extension or com-minution of a fracture. Fractures must be described from the most proximal to the most distal extensions (fissure fracture lines that extend proximally or distally must be included) and not in relation to the center of comminution or to the main location of the fracture. Fractures located mainly within the long-bone segments but extending toward the nearby joint, beyond the line defining the bone segment, should be described with the letters defining the location within the shaft, followed by the small letter j (for joint) (Figure 1; Table 1).

Fracture morphology: Fractures are morphologically described as simple (S), intermediate (I), or complex (C), with these letters following the letter(s) defin-ing the location. Simple fractures are those with no comminution (clear-cut fractures) and are further separated into transverse or slightly oblique (t) and spiral (s). One or 2 minor bony chips should not change the definition of a fracture as simple. Inter-

Figure 1: Topography of long-bone segments: D, distal; M, middle; MD, middle to distal; P, proximal; PM, proximal to middle).

1

e710


Garnavos LonG-bone Fracture cLassiFication system | Garnavos et aL

mediate fractures have 1 or 2 sizable bony fragments, whereas complex fractures have >3 sizable bony fragments or greater comminution (Figure 2; Table 1).

A fracture located between the proxi-mal and middle zones with 1 sizable bony fragment is described as PMI, whereas an oblique fracture with no comminution lo-cated at the distal zone is described as DSt.

In cases where .1 fracture has occurred on the same bone with an intact bony frag-ment with complete canal separating the 2 fractures (segmental fracture), each must be described independently as if the other did not exist. The most proximal fracture is described first, followed by the other fracture(s) in a proximal-to-distal order. If a long bone appears fractured at its proxi-mal zone transversely and a second frac-ture exists at the middle–distal transitional zone with 1 bony fragment, it is described as PSt, MDI.

Intra- and Interobserver Variation Agreement Studies

Radiological Material and Selection of Coders. Anteroposterior and lateral ra-diographs of 120 long-bone shaft fractures (40 femora, 40 tibiae, and 40 humeri) were randomly selected and shown to 6 ortho-pedic physicians of different seniority and experience (2 trainees in the last year of training and 4 qualified orthopedic trauma surgeons). None of these physicians was involved with the preparation or author-ship of the current study. They agreed to participate in the study as independent ob-servers and were blinded to the origin and identity of the creators of the new classifi-cation. All 6 doctors, already familiar with the AO/OTA long-bone fracture classifica-tion system, were introduced to the new classification system by the authors of the current study. They were provided with il-lustrated information for both classification systems and asked to apply either the AO/OTA and the Garnavos classification sys-tem to each long-bone fracture presented as radiographs in a PowerPoint (Microsoft, Redmond, Washington) presentation.

The statistical analysis and evaluation process was based on the comparison of the classification systems by the observ-ers; therefore, no correct answers were set by the authors. The time needed for complete classification of the fractures was recorded for every observer for each anatomical region (femur, tibia, and hu-merus). The same physicians were asked to repeat the procedure (with the same radiographs in the same sequence) 2 to 4 weeks later.

Statistical Methodology. Data were analyzed using k value as a measure of inter- and intraobserver agreement.13,14 Results were defined as poor when k was ,0.20, fair when k was between 0.21 and 0.40, moderate when k was between 0.41 and 0.60, good when k was between 0.61 and 0.80, and very good between 0.81 and 1.00.13,14 All statistical tests were per-formed using SPSS version 10.0 statistical software (SPSS, Inc, Chicago, Illinois). Significance was set at P,.05.

Clinical Validation StudyThe authors retrospectively reviewed

the clinical notes and radiographs of all consecutive patients older than 16 years (N5207) admitted to a Level I trauma center between January 2003 and December 2005 with the diagnosis of a diaphyseal tibial fracture. Inclusion cri-teria were isolated, closed injuries with mild soft tissue trauma (type 0 or 1 ac-cording to Tscherne and Gotzen clas-sification15) and normal preinjury mo-bilization, with full documentation and radiograph series. Exclusion criteria were open fractures, fracture extension to the adjacent joint (knee or ankle), closed fractures with severe soft tissue injury (type 2 or 3 according to Tscherne and Gotzen classification15), and significant comorbidities or coexisting injuries af-fecting postoperative rehabilitation. The fractures were treated individually by experienced trauma surgeons of different seniority, not involved with the prepara-tion or authorship of the current study. At

the time of initial treatment, the surgeons followed no specific protocol, and the chosen treatment was based on personal preference and experience.

After applying the inclusion and exclu-sion criteria, 122 fractures remained and were classified according to the Garnavos classification system by 2 authors (C.G., N.K.K.). Recorded data were entered in a Microsoft Access database. The data included patient demographics, comor-bidities, mechanism of injury, method of treatment, time to operation, length of hospital stay, rehabilitation scheme and mobilization progress, early and late com-plications, secondary interventions, time to union, and time to discharge from out-

Table 1

Garnavos Long-bone FractureClassification System

Topography Morphology

P: Proximal S: Simple (no fragments)t: transverse or

obliques: spiral

M: Middle I: Intermediate (1 or 2 sizable fragments)

D: Distal C: Complex (>3 any size fragments or large comminution)

j: Extension toward the joint

Figure 2: Morphology of long-bone segments: St, simple-transverse; Ss, simple-spiral; I, intermedi-ate; C, complex.

2

e711


n Feature Article

patient clinic. Radiological fracture union was defined as callus formation on AP and lateral radiographs (presence of bridg-ing callus in a minimum 3 of 4 cortices) and clinical fracture union as pain-free weight bearing or stressing at the fracture site according to the existing literature.16 Nonunion was defined as continued pain and motion at the fracture site and the ab-sence of adequate callus formation on AP and lateral radiographs at least 8 months postinjury.16 Delayed union was defined as the absence of radiological progress of callus formation for .4 months postoper-

atively.17 Fracture malunion was defined as AP angulation or rotation of .10°, varus/valgus deformity of .5°, and short-ening of .2 cm.18-20

All operative procedures performed after initial operative fracture fixation were recorded as secondary interven-tions. These included nail dynamization, external fixator pin removal, fascioto-mies, exchange nailing, or change of the fixation method. The clinical suspicion of compartment syndrome was investigated by intracompartmental pressure measure-ments as reported by Frink et al.21

Statistical AnalysisUnivariable and multivariable statis-

tical analyses were performed to verify the original hypothesis that the Garnavos classification system can be related to clinical aspects regarding the manage-ment and outcome of long-bone fractures. For time to union, t test or 1-way analy-sis of variance was applied to compare grouped factors, and test for proportions or chi-square test was applied for com-plications. Univariable analysis was per-formed to identify the variables of inter-est. Any variable with a 10% significance level was taken forward in a multivariable analysis to adjust for other factors.

resultsInterobserver Variation Agreement

Agreement strength improved with the Garnavos classification system compared with the AO/OTA classification system for all fractures assessed. Improvement occurred by 1 level (from fair to moder-ate) for femoral and tibial fractures and by 2 levels (from fair to good) for humeral fractures (Table 2).

Intraobserver Variation Agreement When the same observer classified

the same fractures with both classifica-tion systems 2 to 4 weeks after the first attempt, 1 level of improvement occurred in agreement strength with the Garnavos classification system compared with the AO/OTA classification system (Table 3).

Time ElementComparison of the time needed to clas-

sify fractures of the same long bone during 2 attempts revealed statistically significant differences in favor of the Garnavos clas-sification system (,.01 and ,.05, respec-tively). In addition, statistically significant-ly less time was needed for the classifica-tion with the Garnavos classification sys-tem during the second attempt compared with the time needed for classification of the same fractures with the same clas-sification system during the first attempt

Table 2

Interobserver Agreement

ParameterAO/OTA System

Garnavos System

Femur

Mean k 0.388 0.487

Agreement strength

Fair Moderate

Tibia

Mean k 0.242 0.44

Agreement strength

Fair Moderate

Humerus

Mean k 0.3 0.661

Agreement strength

Fair Good

Abbreviation: AO/OTA, AO-Müller/ Orthopaedic Trauma Association classification.

Table 3

Intraobserver Agreement

ParameterAO-OTA System

Garnavos System

Femur

Mean k 0.38 0.574

Agreement strength

Fair Moderate

Tibia

Mean k 0.38 0.452

Agreement strength

Fair Moderate

Humerus

Mean k 0.38 0.562

Agreement strength

Fair Moderate


Table 4

Time Needed for Fracture Classification at 2 Sessions

Mean (Range)

Session AO/OTA System Garnavos System P

1 44 min 15 s (35 min 8 s to 51 min 50 s)

22 min 44 s (13 min 30 s to 24 min 50 s)

,.01

2 35 min 27 s (28 min 50 s to 64 min 48 s)

15 min 57 s (11 min 25 s to 20 min)

,.05

P .1 ,.05


e712



(P,05). In contrast, no statistically signifi-cant difference existed between the times needed for fracture classification with the AO/OTA classification system between the first and second attempts (P5.1) (Table 4).

Clinical ValidationTables 5 and 6 show demographic

data, mechanism of injury, treatment method, and outcome in correlation with

the Garnavos classification system. Tables 7 and 8 show the distribution of com-plications within the subgroups of the Garnavos classification system.

Univariable Analysis. For statistical methodological reasons, the classification categories were grouped. The fractures of a single zone (P, M, or D) were compared with fractures extending to multiple zones (PM, MD, and PMD). The same simplifi-

cation was made for fracture morphology: simple fractures (Ss and St) were com-pared with multipart fractures (I and C). Univariable analysis identified the follow-ing as significant negative factors related to time to union: age older than 35 years (P5.006), delay to partial or full weight bearing (P,.001), extension of the frac-ture in .2 zones (PM or MD) (P5.006), and presence of complex or intermediate

Table 5

General Parameters by Fracture Topography

Fracture Classification

Variable Total P PM M MD D Segmental

Epidemiology

No. (%) 122 (100) 6 (4.9) 18 (14.7) 16 (13.1) 65 (53.3) 17 (13.9) 3 (2.5)

Median age (range), y 32 (17-74) 33.5 (20-74) 28.5 (17-46) 28.5 (19-49) 34 (18-72) 33 (18-63) 34 (32-35)

No. men:women (ratio) 89:33 (2.7) 5:1 (5) 14:4 (3.5) 14:2 (7) 46:19 (2.4) 10:7 (1.4) 3:0 (N/A)

Mechanism of injury, no. (%)

Fall 61 (50) 3 8 6 35 9 2

Sports 31 (25.4) 1 3 4 18 5 0

Pedestrian accident 16 (13.1) 2 3 2 8 1 1

Motorcycle accident 9 (7.4) 0 2 3 3 1 0

Motor vehicle accident 5 (4.1) 0 2 1 1 1 0

Treatment, no. (%)

Conservativea 18 (14.7) 0 1 0 14 3 0

Operative 104 (85.2) 6 17 16 51 14 3

IMN 82 (67.2) 4 11 16 47 4 3

External fixation 19 (15.6) 2 6 0 4 7 0

ORIF 3 (2.5) 0 0 0 0 3 0

Median time to surgery (range), d 1 (0-29) 0.5 (0-2) 1 (0-9) 2 (0-29) 1 (0-21) 0.5 (0-3) 3 (2-3)

Median hospital stay (range), d 7 (0-28) 5.5 (4-8) 7 (0-24) 9 (4-24) 7 (0-28) 5 (1-17) 6 (4-9)

Rehabilitation and outcome, median (range)

Time to union, wk 22 (16-58) 18 (17-22) 20 (16-58) 20.5 (17-44) 24 (17-56) 21 (18-28) 26 (18-30)

Time of nonweight bearing, wk 4 (0-16) 1.5 (0-8) 3 (0-16) 0 (0-10) 6 (0-12) 6 (0-12) 3 (3-4)

Time to partial weight bearing, wk 6 (0-44) 4 (0-4) 5.5 (0-44) 6 (4-12) 8 (0-44) 4 (1-16) 3 (0-4)

Time to full weight bearing, wk 12 (3-50) 8 (3-12) 12 (4-50) 8 (4-20) 14 (4-48) 12 (7-16) 6 (3-8)

Physiotherapy needed, wk 8 (4-20) N/A 5 (4-12) 6 (4-10) 8 (4-20) 8 (6-12) N/A

Time to discharge, mo 7 (5-33) 6.5 (6-7) 6 (5-26) 6.5 (5-18) 8 (5-33) 7 (5-16) 11 (9-13)

Abbreviations: D, distal; IMN, intramedullary nailing; M, middle; MD, middle to distal; N/A, not applicable; ORIF, open reduction and internal fixation; P, proximal; PM, proximal to middle. aSome fractures initially treated conservatively underwent operative fixation at a later stage.

e713


n Feature Article

fracture morphology (P5.03). Notably, smoking status lay just outside the limit of statistical significance (P,.05), but be-cause smoking delays time to union,22-24 it was included in the parameters of the sub-sequent multivariable analysis (Table 9).

Multivariable Analysis. The multivari-able analysis focused on 2 targets: time to union and complications. To make a suitable adjustment for multiple factors, a multivariable regression was used. The

continuous variables of age and time to full weight bearing were grouped in the univariable analyses for simplicity, but were regarded as continuous for multivari-able analysis. Regarding the correlation between Garnavos classification system categories and the occurrence of compli-cations, multivariable analysis identified that the fracture extension in .1 zone increased the risk of complications (odds ratio53.64). Similarly, a complex mor-

phology doubled the risk of complications (odds ratio52.26), with significance at the 7% level. The time to full weight bear-ing was included as a grouped variable because the effect on the complication rate was nonlinear. Thus, the subgroup of cases that progressed to full weight bear-ing between weeks 8 and 15 had a high risk of complications (odds ratio514.07) when compared with the early mobiliza-tion subgroup (Table 10).

Table 6

General Parameters by Fracture Morphology

Fracture Classification

Variable Total St Ss I C

Epidemiology

No. (%) 122 (100) 24 (19.7) 34 (27.9) 56 (45.9) 8 (6.6)

Median age (range), y 32 (17-74) 28.5 (18-46) 32 (18-74) 32 (17-72) 36.5 (23-49)

No. men:women (ratio) 89:33 (2.7) 19:5 (3.8) 29:5 (5.8) 36/20 (1.8) 5/3 (1.7)

Mechanism of injury, no. (%)

Fall 61 (50) 10 19 26 6

Sports 31 (25.4) 7 11 12 1

Pedestrian accident 16 (13.1) 2 3 10 1

Motorcycle accident 9 (7.4) 3 1 5 0

Motor vehicle accident 5 (4.1) 2 0 3 0

Treatment, no. (%)

Conservativea 18 (14.7) 0 8 10 0

Operative 104 (85.2) 24 26 46 8

IMN 82 (67.2) 18 28 28 8

External fixation 19 (15.6) 3 2 14 0

ORIF 3 (2.5) 3 0 0 0

Median time to surgery (range), d 1 (0-29) 2 (0-29) 1 (0-18) 1 (0-21) 1 (0-3)

Median hospital stay (range), d 7 (0-28) 9 (5-17) 6 (0-24) 6.5 (0-28) 7.5 (4-9)

Rehabilitation and outcome, median (range)

Time to union, wk 22 (16-58) 24 (17-58) 22 (17-54) 22 (16-48) 28 (18-56)

Time of nonweight bearing, wk 4 (0-16) 4 (0-12) 2.5 (0-8) 6 (0-16) 6 (3-12)

Time to partial weight bearing, wk 6 (0-44) 5 (4-44) 4 (0-44) 6 (0-22) 8 (0-36)

Time to full weight bearing, wk 12 (3-50) 13 (4-50) 10 (4-48) 12 (4-24) 20 (3-48)

Physiotherapy needed, wk 8 (4-20) 6 (4-12) 8 (8-8) 8 (4-20) 18 (15-20)

Time to discharge, mo 7 (5-33) 8.5 (5-26) 7 (5-24) 7 (5-25) 11 (6-33)

Abbreviations: C, complex; I, intermediate; IMN, intramedullary nailing; ORIF, open reduction and internal fixation; Ss, simple spiral; St, simple transverse. aSome fractures initially treated conservatively underwent operative fixation at a later stage.

e714



Further Notable Findings. Detailed analysis of the results revealed findings indicating that the Garnavos classifica-tion system can be related to additional clinical parameters than those previously described. However, a larger sample of fractures is needed to prove the statistical strength of these findings.l Mechanism of Injury: The majority of

tibial shaft fractures related to sports, falls, and pedestrian injuries occurred in the MD zone, whereas fractures resulting from motor vehicle injuries were more evenly distributed in 3 zones (PM, M, MD). Six of 8 complex fractures resulted from a fall.

l Treatment Method: Eleven of 14 frac-tures that occurred in the MD zone and were treated conservatively under-went operative fixation at a later stage. Intramedullary nailing was the operative treatment of choice for all fractures that occurred in M zone, for 92.2% of frac-tures in the MD zone, and for all segmen-tal fractures. External fixation was used for half of the fractures that occurred in the D zone and for the majority of frac-tures with an intermediate morphology.

l Rehabilitation Parameters: The median time of nonweight bearing was signifi-cantly less (P,.05) for patients who sus-tained Ss fractures than those who sus-tained I and C fractures (2.5 vs 6 weeks, respectively). The time to full weight bearing was almost double for C fractures compared with St and Ss fractures (20 weeks vs 13 and 10 weeks, respectively). The duration of physiotherapy was sig-nificantly prolonged for C fractures com-pared with St, Ss, and I fractures.

l Complications: Sixteen of 19 delayed unions and all malunions occurred in MD fractures. In addition, delayed union oc-curred more often in St fractures (10/22). Half of the infections (6/12) occurred in the PM zone, whereas all pulmonary em-bolisms (n53) and fat embolisms (n53) were associated with M fractures. All proximal prominences of intramedullary nails occurred in D fractures.

discussionSince its publication, the AO/OTA

long-bone fracture classification system2 has been adopted by the majority of ortho-pedic surgeons and has gained worldwide acceptance for its usefulness in research. However, numerous studies agree that the

AO/OTA long-bone fracture classification system has significant drawbacks, such as complexity and low reliability.3-6,8,9

Newey et al4 reported that the AO clas-sification system is unnecessarily com-plicated and often falls short of playing a useful role in planning and management.

Table 7

Complications by Fracture Topography

Fracture Classification, No.

Complication Total No. (%) P PM M MD D Segmental

Delayed unions 19 (15.6) 0 2 1 16 0 2

Nonunion 3 (2.5) 0 2 0 1 0 0

Malunion 8 (65.6) 0 0 0 8 0 0

Infection 12 (9.8) 1 6 0 1 4 2

Pulmonary embolism 3 (2.5) 0 0 3 0 0 0

Fat embolism 3 (2.5) 0 0 1 2 0 0

Compartment syndrome 12 (9.8) 1 3 3 5 0 1

Loss of reduction, displacement

15 (12.3) 0 1 0 11 3 0

Joint stiffness 3 (2.5) 0 0 1 1 1 0

Anterior knee pain 15 (12.3) 0 5 2 8 0 0

Proximal prominence 3 (2.5) 0 0 0 0 3 0

Abbreviations: D, distal; M, middle; MD, middle to distal; P, proximal; PM, proximal to middle.

Table 8

Complications by Fracture Morphology

Fracture Classification, no.

Complication Total No. (%) St Ss I C

Delayed union 19 (15.6) 8 5 4 2

Nonunion 3 (2.5) 2 1 0 0

Malunion 8 (65.6) 0 0 5 3

Infection 12 (9.8) 3 2 5 2

Pulmonary embolism 3 (2.5) 0 3 0 0

Fat embolism 3 (2.5) 0 0 3 0

Compartment syndrome 12 (9.8) 3 0 6 3

Loss of reduction, displacement 15 (12.3) 0 3 12 0

Joint stiffness 3 (2.5) 0 0 3 0

Anterior knee pain 15 (12.3) 0 2 13 0

Proximal prominence 3 (2.5) 0 0 3 0

Abbreviations: C, complex; I, intermediate; Ss, simple spiral; St, simple transverse.

e715


n Feature Article

Johnstone et al3 reported that a high inci-dence of errors may occur when fractures are coded on an individual basis using this comprehensive system and variabil-

ity may exist when retrieving data from a computer for research purposes and for surgical audit. Martin and Marsh8 reported that the observer agreement for this sys-

tem drops to unacceptable standards at the group and subgroup levels and that further study is warranted to determine how this agreement can be improved. Garbuz et al9

Table 9

Univariable Analysis of Results

Time to Union, wk Complications

Variable Mean6SD Test P Proportion Test P

Sex

Male 24.268.5 32/8950.70

Female 26.7610.4 t51.36 .18 10/3350.64 z50.58 .56

Age group, y

17-34 23.067.5 48/7450.65

35-74 27.6610.5 t52.81 .006 32/4850.67 z50.02 .84

Smoking status

Nonsmoker 24.167.9 59/9350.63

Smoker 27.2611.8 t51.62 .11 21/2950.72 z50.89 .37

Initial treatment

Conservative 22.462.0 12/1850.66

Surgical 25.369.7 t51.25 .21 68/10450.65 z50.11 .92

Treatment type

Conservative 23.162.8 3/750.43

External fixation 26.9612.3 16/2250.73

IMN 24.568.7 59/8850.67

ORIF 23.460.9 F50.53 .66 2/550.40 x2(3)53.63 .30

Time to full weight bearing, wk

0-7 22.464.0 7/2150.33

8-15 22.265.6 52/6350.83

16-50 30.6612.5 F513.63 ,.001 21/3850.55 x2(2)519.49 ,.001

Fracture topography

P 19.062.0 1/650.17

PM 24.6612.1 F52.38 .06 13/1950.68 x2(4)59.18 .06

M 22.466.9 9/1750.53

MD 27.069.4 46/6350.73

D 21.863.7 11/1750.65

P, M, or D 21.665.2 21/4050.525

PM or MD 26.4610.1 t52.82 .006 59/8250.72 z52.12 .03

Fracture morphology

I or C 23.066.7 32/5850.55

Ss or St 26.5610.5 t52.15 .03 48/6450.75 z52.30 .02

Abbreviations: C, complex; D, distal; I, intermediate; IMN, intramedullary nailing; M, middle; MD, middle to distal; ORIF, open reduction and internal fixation; P, proximal; PM, proximal to middle; Ss, simple spiral; St, simple transverse.

e716



identified classification systems in ortho-pedics that have been properly tested for their reliability, but the AO-Müller and AO/OTA classifications for long-bones fractures were not included. They report-ed that many classification systems have been published and widely adopted in orthopedics without enough information available on their reliability.9 Revisions of the AO-Müller fracture classification system introduced by the compendium of the OTA consisted of the reorganiza-tion of the alphanumeric codes to offer a universal system for classifying different anatomical sites.2-7,9,10 Nonetheless, the original concept of the AO-Müller system is maintained, along with its weaknesses.

Audige et al12 described a comprehen-sive staged concept for validating a classi-fication system before its general clinical application. The first phase of the valida-tion pathway included the classification proposal and pilot agreement studies, and the next 2 phases consisted of multicenter agreement studies and clinical studies to assess the clinical relevance and useful-ness of the proposed system.12 The recent compendium of the OTA committee of experts reaffirmed this staged approach of introducing any novel classification

system under the contemporary demand for evidence-based medicine.10 Following these suggestions, the current study in-troduces the new classification system by describing, statistically testing, and clini-cally validating it.

Statistical analysis revealed improved inter- and intraobserver variation agree-ment (constant improvement of at least 1 level), and therefore better reliability, of the Garnavos classification system com-pared with the AO/OTA classification system. Use of the k value for assessment of inter- and intrarater agreement is well established in the medical literature. A difference by 1 level of the k value is con-sidered significant, as well as the consis-tency of this finding in all pairs that were compared.13,14 No statistically significant differences existed between trained and trainee observers regarding the agreement of each classification system.

The time needed for classifying the fractures with the Garnavos classification system during both attempts separated by at least 2 weeks was significantly less compared with the time needed for clas-sifying the same fractures during the same sessions with the AO/OTA classification system, a strong indication that the new

system is less complicated and easier to memorize. This finding is of significant importance in the routine clinical practice of modern trauma. A reliable and easy-to-remember classification system helps communication significantly.

For clinical validation, the Garnavos classification system was applied to a rep-resentative sample of long-bone diaphyseal fractures. The application was easy and swift due to its descriptive nature. This characteristic is well recognized in the lit-erature as an advantage of all descriptive classification systems when compared with hierarchical enumeration coding systems. The univariable and multivariable analy-ses of clinical data revealed significant statistical correlation of the Garnavos clas-sification system with important clinical parameters, such as time to union and com-plication rates. It has been shown that time to union and general complication rates are related to specific groups of the new classification system. Furthermore, strong evidence existed that more clinical param-eters are related to the new classification system, but a larger dataset is required for statistical confirmation. According to this evidence, conservative treatment may not be a good option for fractures occurring in

Table 10

Multivariable Analysis of Results

Variable Coefficient Standard Error t P.(t) 95% CI

Time to union

Age 0.084 0.044 1.92 .057 20.003 to 0.171

Smoking status 2.82 1.36 2.07 .04 0.12 to 5.52

Time to full weight bearing 0.731 0.071 10.35 .000 0.59 to 0.87

PM or MD topography 1.8 1.26 1.43 .156 20.7 to 4.29

I or C morphology 0.14 1.18 0.12 .904 22.2 to 2.48

Complications

Time to full weight bearing 14.07 9.09 4.10 .000 3.97 to 49.88

PM or MD topography 3.64 1.85 2.54 .011 1.35 to 9.84

I or C morphology 2.26 1.02 1.81 .07 0.93 to 5.46

Abbreviations: C, complex; CI, confidence interval; I, intermediate; MD, middle to distal; PM, proximal to middle.

e717


n Feature Article

the MD zone because 11 of 14 fractures in this zone that were treated conservatively underwent operative treatment at a later stage. Intramedullary nailing was the op-erative treatment of choice for all fractures in the M zone and most fractures in the MD zone, a finding that likely reflects the cur-rent treatment policy of the specific trauma center, which is consistent with the con-temporary literature.20

Problems with vascularity and poor soft tissue envelope in the MD zone may be responsible factors for .80% of de-layed unions (16/19) and all malunions (8) occurring in this zone. This finding is consistent with other reports that relate these complications mostly with the mid-dle area of the tibial shaft. Moreover, de-layed union appears to be more associated with the St fracture morphology (10/22), which may be due to the limited contact surface of bony fragments. Half of the in-fections (6/12) occurred in the PM zone, which also could be associated with the specific vascular or anatomic features of this zone, the application of external fixa-tion pins, or the development of compart-ment syndrome and subsequent fascioto-mies. Another interesting finding is that all proximal prominences of an intramed-ullary nail occurred in D fractures. A logi-cal explanation is that in distal fractures, the surgeon uses the longest nail that can go as distal as possible to secure the frac-ture. Therefore, the nail may be prominent at the knee immediately postoperatively or postdynamization. All pulmonary embolisms (n53) and all fat embolisms (n53) were associated with M fractures. Confirmation of this finding with a larger dataset could dictate a change in future anticoagulant strategy. Finally, 6 of 8 complex fractures happened after a fall and not after a motor collision or road traffic accident. However, open fractures or fractures with severe soft tissue damage were excluded from the study.

A major limitation of the current study is the relative small sample size for clinical validation. However, this sample, includ-

ing all consequent tibial shaft fractures admitted in a large Level I trauma center over 3 years, constituted the material for a pilot study according to the criteria set by Audige et al.12 A validation study with an adequate number of cases is warranted.

The current study does not represent an effort to describe and assess all treat-ment options and outcome patterns of closed tibial diaphyseal fractures and has the weaknesses of all retrospective stud-ies. Moreover, fractures extending to adja-cent joints, open fractures, polytrauma pa-tients, and injuries with significant com-promise of the soft tissues or associated injuries that would potentially influence the outcome of the tibial fractures were excluded. Thus, the indications of using the proposed classification system were not exhausted.

The exclusion of open fractures and cases with significant soft tissue injury was dictated by the fact that the new clas-sification system refers strictly to the ele-ment of bony injury. It aims to aid com-munication between clinicians, indicate treatment options, and predict the out-comes of long-bone fractures. For a more comprehensive description of bony and soft tissue injury, the Garnavos classifica-tion system should be combined with the well-established classification systems introduced by Gustilo and Anderson25 and Gustilo et al26 for open fractures or by Tscherne and Gotzen15 for closed frac-tures.

Simple transformation of the investi-gated clinical parameters, such as square root or logarithm, were attempted to im-prove variance stability. However, inter-pretation of the coefficient became more complicated. This led to the use of broad-er topographic and morphologic catego-ries for multivariable analysis. The overall occurrence of complications was consid-erably high due to the fact that even minor complications were included. In addition, some patients experienced .1 complica-tion. With these concessions, the topog-raphy and morphology of the tibial shaft

fractures, as described by the Garnavos classification system, were proven to be significant predictors of complication rates and time to union, together with patient age and smoking status. A larger, multicenter prospective study is justified and may reveal the comprehensive statis-tical significance of the new classification system.

It has been suggested that 2 classifi-cation systems are needed for long-bone fractures: 1 for clinical use and 1 for research purposes.27 This will provide access to all data required for research without burdening the clinician with de-tails he or she cannot use. If the AO/OTA classification of long-bone fractures pro-vides the basis for a general classification for research, the Garnavos classification system may meet the need for a clinically oriented classification system with a high degree of reliability, memorability, and clinical relevance.

conclusionThe Garnavos classification system for

diaphyseal long-bone fractures was found to be easily applicable in a clinical setting. Further research should be conducted in the form of prospective, multicenter, ob-servational agreement studies for the completion of the validation process, something that, to our knowledge, has never been undertaken for any existing classification systems.

references 1. Muller M, Nazarian S, Koch P, Schatzker

J. The Comprehensive Classification of Fractures of Long Bones. Berlin, Germany: Springer-Verlag; 1990.

2. Fracture and dislocation compendium. Orthopaedic Trauma Association Committee for Coding and Classification. J Orthop Trauma. 1996; 10(suppl 1):1-154.

3. Johnstone DJ, Radford WJ, Parnell EJ. Interobserver variation using the AO/ASIF classification of long bone fractures. Injury. 1993; 24(3):163-165.

4. Newey ML, Ricketts D, Roberts L. The AO classification of long bone fractures: an early study of its use in clinical practice. Injury. 1993; 24(5):309-312.

e718



5. Taylor JK. AO fracture classification logos, as evocative signposts. J Orthop Trauma. 1996; 10(2):146.

6. Swiontkowski MF, Agel J, McAndrew MP, Burgess AR, MacKenzie EJ. Outcome vali-dation of the AO/OTA fracture classification system. J Orthop Trauma. 2000; 14(8):534-541.

7. Colton CL. Telling the bones. J Bone Joint Surg Br. 1991; 73(3):362-364.

8. Martin JS, Marsh JL. Current classification of fractures. Rationale and utility. Radiol Clin North Am. 1997; 35(3):491-506.

9. Garbuz DS, Masri BA, Esdaile J, Duncan CP. Classification systems in orthopaedics. J Am Acad Orthop Surg. 2002; 10(4):290-297.

10. Marsh JL, Slongo TF, Agel J, et al. Fracture and dislocation classification compendium–2007: Orthopaedic Trauma Association classification, database and outcomes committee. J Orthop Trauma. 2007; 21(10 suppl):S1-S133.

11. Bland JM, Altman DG. Statistics notes: validating scales and indexes. BMJ. 2002; 324(7337):606-607.

12. Audige L, Bhandari M, Hanson B, Kellam J. A concept for the validation of fracture classifica-tions. J Orthop Trauma. 2005; 19(6):401-406.

13. McKenzie DP, Mackinnon AJ, Peladeau N, et al. Comparing correlated kappas by resam-

pling: is one level of agreement significantly different from another? J Psychiatr Res. 1996; 30(6):483-492.

14. Lui KJ, Kelly C. A note on interval estima-tion of kappa in a series of 2x2 tables. Stat Med. 1999; 18(15):2041-2049.

15. Tscherne H, Gotzen L. Fractures With Soft Tissue Injuries. New York, NY: Springer; 1984.

16. Babhulkar S, Pande K. Nonunion of the di-aphysis of long bones. Clin Orthop Relat Res. 2005; (431):50-56.

17. Phieffer LS, Goulet JA. Delayed unions of the tibia. J Bone Joint Surg Am. 2006; 88(1):206-216.

18. Bridgman SA, Baird K. Audit of closed tibial fractures: what is a satisfactory outcome? Injury. 1993; 24(2):85-89.

19. Feldman DS, Shin SS, Madan S, Koval KJ. Correction of tibial malunion and nonunion with six-axis analysis deformity correction using the Taylor Spatial Frame. J Orthop Trauma. 2003; 17(8):549-554.

20. Schmidt AH, Finkemeier CG, Tornetta P III. Treatment of closed tibial fractures. Instr Course Lect. 2003; 52:607-622.

21. Frink M, Klaus AK, Kuther G, et al. Long term results of compartment syndrome of the lower limb in polytraumatised patients. Injury. 2007; 38(5):607-613.

22. Schmitz MA, Finnegan M, Natarajan R, Champine J. Effect of smoking on tibial shaft fracture healing. Clin Orthop Relat Res. 1999; 36(5):184-200.

23. Harvey EJ, Agel J, Selznick HS, Chapman JR, Henley MB. Deleterious effect of smok-ing on healing of open tibia-shaft frac-tures. Am J Orthop (Belle Mead NJ). 2002; 31(9):518-521.

24. Castillo RC, Bosse MJ, MacKenzie EJ, Patterson BM. Impact of smoking on fracture healing and risk of complications in limb-threatening open tibia fractures. J Orthop Trauma. 2005; 19(3):151-157.

25. Gustilo RB, Anderson JT. Prevention of in-fection in the treatment of one thousand and twenty-five open fractures of long bones: ret-rospective and prospective analyses. J Bone Joint Surg Am. 1976; 58(4):453-458.

26. Gustilo RB, Mendoza RM, Williams DN. Problems in the management of type III (severe) open fractures: a new classification of type III open fractures. J Trauma. 1984; 24(8):742-746.

27. Bernstein J, Monaghan BA, Silber JS, DeLong WG. Taxonomy and treatment—a classification of fracture classifications. J Bone Joint Surg Br. 1997; 79(5):706-707.

e719

new classification system for long-bone

Documents