ORIGINAL ARTICLE

The OTA Open Fracture Classification: A Study of Reliability

and Agreement
Julie Agel, MA,* Andrew R. Evans, MD,† John Lawrence Marsh, MD,‡ Thomas A. DeCoster, MD,§
Douglas W. Lundy, MD,¶ James F. Kellam, MD,k Clifford B. Jones, MD,# and Gregory L. DeSilva, MD**

Key Words: interrater reliability, OTA open fracture classification

Objectives: To determine the reliability of the Orthopaedic Trauma
Association (OTA) Open Fracture Classification. (J Orthop Trauma 2013;27:379–385)

Design: Video-based reliability study.

Setting: Orthopedic meetings and grand rounds. INTRODUCTION
Fracture classification systems are important for research
Patients/participants: Orthopedic surgeons.
and clinical care. They provide a means to communicate, to
Intervention: None. combine similar injuries and distinguish different patterns, and
to stratify injury severity. To be scientifically sound and justify
Main Outcome Measurements: Interobserver reliability assess- wide spread utilization, fracture classifications need to be
ment classification. reliable, reproducible, responsive, clinically relevant, and valid.
Results: The results demonstrate the system to have high reliability Effectively classifying open fractures presents chal-
and much improvement compared with published Gustilo–Anderson lenges because there are multiple different tissues involved in
classification reliability studies. Overall interrater reliability (k) values the injury, each of which contributes in an important way to
were highest for arterial injury, with near perfect agreement across all stratifying injury severity, determining optimal treatment, and
raters and within each value. Skin injury, bone loss, and contamination to the eventual patient outcome and risk of complication,
demonstrated moderate to substantial levels of agreement. Muscle particularly infection. Despite these challenges, systemati-
injury had the most disagreement between raters but still demonstrat- cally classifying open fractures is an important step in
ing a fair level of interrater agreement, which is a level of agreement directing treatment, predicting complications, and assessing
superior to the literature related to the Gustilo–Anderson classification. prognosis. Less severe open fractures may have a negligible
Levels of agreement were similar between attending surgeons and infection rate, whereas severe open fractures have very high
residents for all categories. infection rates.1 Some less severe fractures can be treated
without antibiotics or surgical debridement, whereas more
Conclusions: This study, which included a diverse multicenter severe injuries may require complex antibiotic regimens, mul-
multinational cohort of orthopaedic surgeons and residents, of the tiple surgical debridements, and staged reconstruction for
OTA Open Fracture Classification demonstrated moderate to optimal results. However, defining severity for the different
excellent interobserver reliability. types of injured tissues involved and incorporating it into
a standardized classification remains a challenge.
The Gustilo–Anderson classification of open fractures,
which has been used successfully used for more than 35 years,
Accepted for publication December 7, 2012. has significant shortcomings.2–4 The classification was initially
From the *Department of Orthopaedic Surgery and Sports Medicine, Harbor- based on open tibial fractures.2 It was an important advance-
view Medical Center, Seattle, WA; †Department of Orthopaedics, UPMC ment of evidence-based classification of open fractures when it
Mercy, Pittsburg, PA; ‡Department of Orthopaedic Surgery, University of
Iowa Hospitals and Clinics, Iowa City, IA; §Department of Orthopaedics, was introduced in 1976.3 There have been significant advances
University of New Mexico, Albuquerque, NM; ¶Resurgens Orthopaedics, in open fracture care since 1976, and new techniques have been
Marietta, GA; kDepartment of Orthopaedic Surgery, Carolinas Medical introduced to deal with these fractures. Significant problems and
Center, Charlotte, NC; #Orthopaedic Associates of Michigan, Grand Rapids, shortcomings have been identified with the Gustilo–Anderson
MI; and **Department of Orthopaedic Surgery, University of Arizona, open fracture classification. In particular, the reliability and
Tucson, AZ.
The authors report no funding or conflict of interest. reproducibility of this classification has been demonstrated to
Supplemental digital content is available for this article. Direct URL citations be suboptimal.5 Using Gustilo–Anderson criteria, there is very
appear in the printed text and are provided in the HTML and PDF poor agreement between 2 different surgeons as to which grade
versions this article on the journal’s Web site (www.jorthotrauma.com). is appropriate for a given injury and even the same surgeon does
Reprints: Julie Agel, MA, Department of Orthopaedic Surgery and Sports
Medicine, Harborview Medical Center, 325 Ninth Avenue, Box 359798,
not assign the same grade when shown the same injury at 2
Seattle, WA 98104-2499. E-mail: [email protected]. different times. Other problems with the classification have
Copyright © 2013 by Lippincott Williams & Wilkins become evident. Severe open fractures cannot be subclassified

J Orthop Trauma  Volume 27, Number 7, July 2013 www.jorthotrauma.com | 379

Agel et al J Orthop Trauma  Volume 27, Number 7, July 2013

based on potentially important and varied characteristics, PATIENTS AND METHODS

treatment, and outcomes because there are only 2 categories Institutional Review Board approval was obtained for
of severe fractures (types 3B and 3C). The Gustilo–Anderson this study at all centers where video materials were recorded.
classification has undergone unofficial modification in com- All video materials were deidentified. Twelve videos of
mon usage over time creating more variability in the assess- separate open fracture cases, representing a convenience
ment. An important category of the Gustilo–Anderson sample, were obtained and reviewed by members of the OTA
classification (3B) is determined by how the soft tissue Classification and Outcomes Committee. Six videos were
wound is treated (that is, the need for flap coverage signifies selected for the study based upon evaluation of image and
a 3B category), which is problematic since indications for video quality, diversity of injury severity, and anatomic
flap coverage have changed over the last 30 years and vary distribution. The fractures presented in these videos included
between clinicians. The introduction of wound vacuum tech- 3 tibial shaft fractures; 1 distal tibial pilon fracture, 1 distal
nology has dramatically reduced the frequency of free flap humeral fracture, and 1 malleolar ankle (distal fibular and
procedures in many situations. This problem, based on tibia mallelolar) fracture dislocation. All videos were
change in practice, vividly demonstrates why classification recorded at the time of initial operative debridement, which
should direct treatment rather than treatment (flap coverage) occurred 6 hours postinjury at the primary receiving
directing classification. Overall, there are enough problems institution. They were edited and arranged following a stan-
with the Gustilo–Anderson open fracture classification (poor dardized format designed to illustrate key portions of the
reliability and reproducibility) and enough changes in open assessment and debridement procedure pertinent to classifying
fracture care techniques over the past 30 years to warrant injury severity.
considering an updated classification system for open The videos were presented to each group of raters in
fractures. a PowerPoint presentation format (Microsoft). Each case
The Classification and Outcomes Committee of the
presentation contained a brief clinical history describing the
Orthopaedic Trauma Association (OTA) has developed a new
mechanism of injury and description of injuries sustained in
classification for open fractures designed to build on the
addition to the open fracture. Orthogonal plain preoperative
foundation of the Gustilo–Anderson classification and over-
radiographs of the fracture were included, followed by
come some of its shortcomings.6 This new system classifies
sequential videotaped examination of the skin injury, muscle
open fractures of the upper extremity, lower extremity, and
injury, vascular status, bone injury, and inspection for
pelvis allowing for the following: (1) a greater stratification
of injury severity, (2) better assessment of the factors that must contamination. Fracture treatment information, postoperative
be considered for treatment, (3) the various injured tissues to radiographs, and final wound closure or coverage were not
be accounted for separately, and (4) optimal communication included in any presentation to avoid introduction of surgeon
for clinical care and research. With experience and widespread bias into the assessment of injury severity by independent
acceptance, it may be possible to use this system to direct raters. Presentation of the vascular examination was accom-
treatment and predict outcome in a manner with greater panied by a verbal statement of physical findings given the
reliability and reproducibility. challenges associated with demonstration of pulse palpability,
The OTA Open Fracture Classification (See Appendix 1, capillary refill, skin color, and temperature in video format
Supplement Digital Content 1, http://links.lww.com/BOT/A77) alone. Contamination was easier than vascular examination to
is composed of 5 categories as follows: skin injury (S), demonstrate on video, although a statement about injury
muscle injury (M), arterial injury (A), bone loss (B), and environment or depth and nature of contamination accompa-
contamination (C). Each category is stratified into 3 nied the video documentation.
subcategories ordered according to severity, mild, moderate, Members of the Classification and Outcomes Commit-
and severe. These subcategories are designed to be parallel tee presented the PowerPoint presentations with the 6 cases to
across the 5 categories. The classification nomenclature the Southeastern Fracture Symposium 2011, Southwest
focuses on the pathoanatomic characteristics of injury. The Orthopedic Trauma Association annual meeting January
development process and face validity of the classification 2010, two local institutional Grand Rounds attended by
have been previously published.6 orthopaedic faculty and/or residents, and at a meeting of the
The purpose of this study was to evaluate the interob- Arbeitsgemeinschaft für Osteosynthesefragen in Davos,
server reliability of the OTA Open Fracture Classification Switzerland. Participants at these events formed the pool of
system when applied by practicing orthopaedic surgeons and raters for this study. All video presentations were projected
orthopaedic residents to open fractures of the upper and lower for raters on a central screen, and each rater independently
extremity.7 The raters classified fractures based on viewing marked the classifications for each category of injury on
videos of open fractures in a PowerPoint presentation prepared a standardized data sheet displaying the OTA Open Fracture
for the purpose of the study. The study tested the hypothesis Classification (see Appendix 1, Supplement Digital Content 1,
that the OTA open fracture classification has good to excellent http://links.lww.com/BOT/A77). All videos were presented in
interrater reliability and percent agreement amongst orthopae- the same order and time frame at each site. The surgeons who
dic surgeons and residents. The effects of rater experience on treated the patients presented in the videos did not participate
the interrater reliability of the classification and the degree to in any of the rating sessions. The classification scheme and its
which the quality of videos affected the raters’ confidence in nomenclature were briefly presented to the raters immediately
their ratings of the open fractures were also assessed. before presentation of the open fracture videos.

380 | www.jorthotrauma.com Ó 2013 Lippincott Williams & Wilkins

J Orthop Trauma  Volume 27, Number 7, July 2013 The OTA Open Fracture Classification

One hundred thirty-six independent raters, comprised of moderate to substantial levels of agreement. Muscle injury had
91 attending orthopaedic surgeons and 45 orthopaedic the most disagreement across raters but still demonstrated a fair
residents at variable levels of training, completed open level of interrater agreement, which is a level of agreement
fracture classification forms after viewing the video presenta- superior to the Gustilo–Anderson classification.
tions of the open fracture (Table 1). In some cases, not all Levels of agreement were similar between attending
raters completed classification forms for all 6 video presenta- surgeons and residents for all categories, although agreement
tions. Incomplete fracture classification forms were not was slightly higher for the resident cohort in every category
included in this analysis. except arterial injury (Table 3) where near perfect agreement
The completed raw data forms were mailed to one of was demonstrated for all reviewers.
the investigators (J.A.) who compiled the data. Data for each The percent agreement across all raters within each
classification category (skin injury, muscle injury, etc.), and classification category and severity subcategory for each
for each severity subcategory (1, 2, 3), was reviewed and video case is presented in Table 4. This represents the degree
analyzed for interrater reliability and percent agreement. All to which identical ratings were given for the categories for
analysis was done using SPSS v 17 (Chicago, IL) or SASv9 each video. The severity subcategory within the arterial injury
(Cary, NC). Interrater agreement for multiple raters was done category representing no arterial injury was the only category
using the Inter_Rater Macro for SAS8 which allows for to reach perfect agreement amongst all raters and was
Kappa scores for multiple raters on a scale with multiple achieved for 3 of 6 videos. Classification categories of muscle
values. Kappa scores were interpreted similar to other studies injury and contamination consistently demonstrated the low-
as follows: ,0.20 = poor agreement, 0.20–0.40 = fair agree- est percent agreement across almost all videos. There was
ment, 0.40–0.60 = moderate, 0.60–0.80 = good agreement, a trend for greater levels of agreement amongst raters for
and 0.80–1.0 = excellent agreement.9 Interrater reliability (k) classification subcategories representing lesser severity inju-
values were also calculated separately for the cohort of ries (mild and moderate) and somewhat less agreement for
attending orthopaedic surgeons and the cohort of orthopaedic more severe injuries (moderate and severe).
residents. The measures of reliability were used to determine Twenty attending orthopaedic surgeons assessed and rated
the ability of the classification to differentiate between videos. the quality of the videos and the overall presentations of the
The measure of percent agreement was used to determine the cases. The results are presented in Supplement Digital Content 2
degree to which the different raters gave each variable within (Appendix 2, http://links.lww.com/BOT/A78). Overall, video
each video the same value. As an assessment of quality of the quality and arrangement within the presentations was felt to be
video presentations, participants were asked to grade the ade- adequate by most participants. There was a lack of participant
quacy of the videos in 5 categories during one of the presenta- agreement on the appropriateness of the amount of material
tion sessions. The 5 categories were based upon image quality, presented and whether the experience of classifying open
the arrangement of videos within the presentation, the length of fracture severity outside the operating theater was contrived.
video shown an impression of whether the video material rep- This indicates that at least some raters perceived that it was
resented the setting of the initial fracture debridement or seemed difficult to assess and classify open fractures in this video
contrived, and whether adequate time was allowed for video presentation format.
viewing, interpretation, and fracture classification.

DISCUSSION
RESULTS This study demonstrated that the new OTA Open
Each category of the classification was assessed sepa- Fracture Classification has good reliability of classification
rately. The results shown in Table 2 demonstrate the system to amongst a wide variety of orthopaedic surgeons. These data
have high reliability and much improvement compared with support continued development and utilization of the OTA
that found for the Gustilo–Anderson classification. Overall Open Fracture Classification.
interrater reliability (k) values were highest for arterial injury, The strength of this study is the broad, multicentered,
with near perfect agreement across all raters and within each international cohort of orthopaedic surgeons and residents
value. Skin injury, bone loss, and contamination demonstrated who participated as raters. This was accomplished by using
videos as the mechanism to most closely replicate the
experience of actually performing or observing the initial
TABLE 1. Sites of Open Fracture Video Presentations and open fracture debridement. Reliability was assessed by kappa
Experience Levels of Raters scores to determine the degree to which observers using the
Site Attending Surgeons Residents open fracture classification agreed on how to differentiate
SEFS 22 3 amongst various injuries. Percent agreement was assessed to
Emory University 15 determine the degree to which the scores in each category
SWOTA 15 given by the raters were identical. These 2 analyses combined
University of Pittsburgh 27 provide information on the overall quality of the measurement
AO meeting 54 tool being evaluated; its ability to discriminate or identify
different levels of severity, and the ability for different users
SEFS, Southeastern Fracture Symposium; SWOTA, Southwest Orthopedic Trauma
Association. of the instrument to choose the same value when assessing the
same injury.

Ó 2013 Lippincott Williams & Wilkins www.jorthotrauma.com | 381

Agel et al J Orthop Trauma  Volume 27, Number 7, July 2013

TABLE 2. Interrater Reliability for the 5 Categories and 3 Levels of Severity (Kappa Values)
Kappa (k)
Score Definition Values
Skin Injury
1 Can be approximated 0.79
2 Cannot be approximated 0.01
3 Extensive degloving 0.79
Overall 0.69
Muscle Injury
1 No muscle in area, no appreciable muscle necrosis, some muscle injury with intact muscle function 0.51
2 Loss of muscle but the muscle remains functional, some localized necrosis in the zone of injury that 0.21
requires excision, intact muscle–tendon unit
3 Dead muscle, loss of muscle function, partial or complete compartment excision, complete disruption of 0.52
a muscle–tendon unit, muscle defect does not approximate
Overall 0.40
Arterial Injury
1 No injury 0.9
2 Artery injury without ischemia 0.9
3 Artery injury with distal ischemia 0.0
Overall 0.9
Bone Loss
1 None 0.70
2 Bone missing or devascularized but still some contact between proximal and distal fragments 0.65
3 Segmental bone loss 0.00
Overall 0.65
Contamination
1 None or minimal contamination 0.52
2 Surface contamination (easily removed not embedded in bone or deep soft tissues 0.16
3 Contaminant embedded in bone or deep soft tissues or high risk environmental conditions (barnyard, 0.69
fecal, dirty water, etc.)
Overall 0.48

When the 6 open fracture cases were presented, the OTA nomenclature contained within the muscle injury and contam-
Open Fracture Classification demonstrated moderate to excel- ination categories will be re-evaluated. However, limits to the
lent agreement between the raters. This level of agreement was degree of agreement possible for some of these categories
better than found by Brumback et al5 for the Gustilo–Anderson regardless of category wording may be present. This is partic-
classification using a similar video format. The average interob- ularly the case when employing the current study design of
server agreement seen for the Gustilo–Anderson classification video assessment of injury. Assessment of muscle injury and
was 60%, whereas all categories of the OTA Open Fracture contamination might be expected to be better and more reliable
Classification in the current study demonstrated an average of with direct observation of the wound.
86% overall agreement for all categories (between 52% and The level of agreement between residents was similar to
100% agreement for individual categories). the agreement between attending surgeons. This finding was
This study also separately assessed the reliability of each unexpected because more experience should lead to greater
of the 5 categories. The arterial injury category had excellent interrater reliability for a classification scheme. This supports
agreement among the raters for all 6 video presentations the wording appropriateness of each category. However, this
(k = 0.9), which would be expected as the information to classification was new to all of the observers minimizing the
assign the correct category was given in the video. The only effect of previous experience. In addition, more experienced
disagreement seemed to be error in recording results on the part surgeons may have had preconceived ideas that made it
of the rater. The raters also had a good level of agreement for difficult for them to reliably use these new definitions. They
skin injury and bone loss (k = 0.65 and 0.69, respectively). The may have substituted their own opinion of importance of
most disagreement was identified for muscle injury and con- various open fracture characteristics for the ones identified by
tamination, which had moderate agreement (k = 0.40 and 0.48, the classification itself when assigning cases to categories.
respectively). Muscle injury and contamination are areas This level of agreement between residents and attending
where category definitions are the most subjective and vari- surgeons show this system may be readily applied in teaching
able. The committee members have reassessed category defi- hospitals and trauma centers.
nitions after each stage of validating the classification to make In vivo reliability testing for this type of classification
them as clear as possible. Based on the results of this study, the system is probably the best scenario. However having

382 | www.jorthotrauma.com Ó 2013 Lippincott Williams & Wilkins

J Orthop Trauma  Volume 27, Number 7, July 2013 The OTA Open Fracture Classification

for purposes of comparison or accuracy measurement. This

TABLE 3. Overall Kappa (k) Values Within Attending Physician
would have been a useful additional measure to have for
and Resident Physician Cohorts
judging the clarity of the videos and setting a benchmark for
Arterial Bone Loss Contamination Muscle Skin the raters and a limitation of this project.
Attending 0.91 0.65 0.43 0.36 0.66 Additional limitations of this study design include the
Resident 0.89 0.77 0.58 0.39 0.76 limited number of cases with the lower extremity cases
limited to the tibial shaft and distal tibia and only one upper
extremity case. Because 5 ratings were required for each
multiple observers classify the same injury in-vivo is case, and each was based on assessment of video material, 6
logistically difficult and limited to those present in the cases were felt to be the maximum number that was feasible
operating room. The use of video presentations allowed for volunteer raters in a conference format. We chose
each case to be classified by multiple raters. The data a variety of anatomic locations but did not include all
demonstrate that some of the videos created more disagree- anatomic locations. We included a majority of tibia fractures
ment than others. Brumback and Jones5 cited an absence of because open fractures of the tibial shaft remain the pro-
complaints about the quality of video or video experience in totypical injury for such studies for many reasons including
a blank “comments” box as potential data to support the val- their relative frequency and the resultant collective experi-
idity of their video presentation methodology. In this study, ence, research, and teachings of orthopaedic surgeons
the data in Supplement Digital Content 2 (Appendix 2, derived from that relative experience. Open fractures involv-
http://links.lww.com/BOT/A78) indicate that the raters did ing other anatomic locations contain a multitude of nuances
not think the videos were perfect representations of the actual that could make classification reliability in these areas
debridement of an open fracture. Some of the respondents felt different from what we have reported here. The Gustilo-
there might be too little material presented in the videos for Anderson classification was not provided by the surgeons
adequate classification. Although video documentation of an whose videos were used nor was it requested from the raters.
initial debridement rather than first-hand observation of the In future trials, this would be a useful addition to the
findings remains a weakness of this study, this experimental methodology.
model is one of the only ways that interrater reliability can be At this time, no summative scoring method for this
assessed amongst a large sample size. The treating surgeon did classification exists. Therefore, there are 125 different
not provide a “gold standard” intraoperative injury classification combinations of the 5 categories and 3 severities. This is

TABLE 4. Percent Agreement of Raters for Each Video Case

Bone
Variable Arterial Loss Contamination Muscle Skin
Video 1: Diaphyseal tibial fracture—AO/OTA 42C2
1 100% 87% 89% 78% 97%
2 0% 12% 10% 22% 3%
3 0% 1% 1% 0% 0%
Video 2: Diaphyseal tibial fracture—AO/OTA 42B2
1 4% 2% 33% 1% 2%
2 96% 94% 57% 16% 6%
3 1% 5% 11% 83% 92%
Video 3: Ankle fracture-dislocation—AO/OTA 44B1
1 100% 99% 72% 52% 91%
2 0% 1% 27% 43% 8%
3 0% 0% 2% 5% 1%
Video 4: Intra-articular distal humerus fracture—AO/OTA 13C3
1 99% 20% 3% 34% 91%
2 1% 76% 20% 58% 5%
3 0% 4% 77% 8% 4%
Video 5: Intra-articular distal tibial pilon fracture—AO/OTA 43C2
1 100% 6% 82% 95% 96%
2 0% 93% 14% 5% 5%
3 0% 2% 4% 0% 0%
Video 6: Diaphyseal tibial fracture—AO/OTA 42C2
1 98% 6% 2% 1% 4%
2 1% 93% 5% 65% 14%
3 1% 1% 93% 34% 83%

Ó 2013 Lippincott Williams & Wilkins www.jorthotrauma.com | 383

Agel et al J Orthop Trauma  Volume 27, Number 7, July 2013

too many to gain widespread clinical use. However, any nomenclature, which injury based and not treatment based,
scoring method that summarizes all 5 categories of injury requires further scrutiny particularly in the muscle injury
into a single number or grouping will potentially lose the and contamination categories in which interobserver
specificity of injury severity inherent in independent assess- agreement was the weakest. In addition to reliability
ment of each of the 5 categories. It is hoped that further testing, the degree to which patient outcomes are stratified
research using the OTA Open Fracture Classification will between the categories of the classification is a current
allow recognition of common important patterns for the research interest.
classification of severe injuries (subclassification of Gustilo–
Anderson 3B) in a clinically relevant manner. Furthermore, REFERENCES
grouping may lessen the affect of one category over another. 1. Dellinger EP, Miller SD, Wertz MJ, et al. Risk of infection after open
However broad appeal of the classification for routine non- fracture of the arm or leg. Arch Surg. 1988;123:1320–1327.
research use may be facilitated by a summary score or by 2. Gustilo RB. Management of open fractures. An analysis of 673 cases.
Minn Med. 1971;54:185–189.
grouping of categories into a 3 (or other small number of 3. Gustilo RB, Anderson JT. Prevention of infection in the treatment of one
categories) category summary. One potential summary score thousand and twenty-five open fractures of long bones: retrospective and
is to use the most severe of any category to describe the prospective analyses. J Bone Joint Surg Am. 1976;58:453–458.
overall injury. Therefore, an injury with skin severity 2, 4. Gustilo RB, Simpson L, Nixon R, et al. Analysis of 511 open fractures.
Clin Orthop Relat Res. 1969;66:148–154.
muscle 2, bone loss 1, contamination 1, and arterial injury 5. Brumback RJ, Jones AL. Interobserver agreement in the classification of
1 would be classified as a 2 or “moderate” because the high- open fractures of the tibia. The results of a survey of two hundred and forty-
est severity of any category is 2. The OTA classification and five orthopaedic surgeons. J Bone Joint Surg Am. 1994;76:1162–1166.
outcomes committee continue to evaluate the potential of 6. Evans AR, Agel J, DeSilva GL, et al; Orthopaedic Trauma Association:
these and other options. Open fracture study group. A new classification scheme for open fractures.
J Orthop Trauma. 2010;24:457–465.
In conclusion, in this study, which included a diverse 7. Kottner J, Audige L, Brorson S, et al. Guidelines for reporting reliability
multicenter multinational cohort of orthopaedic surgeons and agreement studies (GRRAS) were proposed. J Clin Epidemiol. 2011;
and residents, of the OTA Open Fracture Classification 64:96–106.
demonstrated moderate to excellent inter-observer reli- 8. Gwet K. Inter-rater reliability: dependency on trait prevalence and
marginal homogeneity. Statistical Methods for Inter-rater Reliability
ability. This is despite the fact that the new classification Assessment. 2002;2:1–9.
provides more detailed injury severity discrimination 9. Altman DG. Practical Statistics for Medical Research. London, United
particularly in the more severe injuries. The classification Kingdom: Chapman and Hall/CRC; 1991.

Invited Commentary

I read this manuscript with interest, as I believe that advancement in the classification of open fractures is necessary to be able to
compare the results of open fracture care. One unifying problem with the scientific study of open fractures remains the difficulty
in creating cohort groups, with so many variables of injury inherent in open fractures. Without differing trauma centers being able
to reproducibly group like fractures together, we are at a loss to discover which open fractures need certain treatments, and what
others are better served by alternative methods. And, while certainly advancements have been made, the level of fracture care still
seems to depend on the judgment and experience of the individual treating orthopaedic surgeon, attempting to match the best type
and timing of care to each open fracture.
I have long believed that if you hypothetically handed each member of the OTA the exact same open fracture of both
bones of the forearm, the postoperative radiographs would look remarkably similar, but the amount of surgical debridement
performed for that injury would differ markedly. Only when we can accurately classify open fractures, can we subdivide injuries
into like groups and, perhaps, learn to optimally treat these subdivisions differently. A reproducible classification of open
fractures should logically lead to a reproducible measure of the accuracy and effectiveness of debridement, which, I believe, has
a direct effect on infectious outcomes. So I envision standardization and acceptance of the classification of open fractures as
a necessary first step to the standardization of open fracture treatment and results.
All good science raises further questions and that remains true here. There is still unimpressive reliability and agreement in
2 key areas of injury, muscle damage and contamination. Certainly, these are 2 important factors in analyzing outcomes of open
fractures. The authors recognized this and are to be commended for honestly concluding that these areas need further
adjustments of their tested criteria so that greater agreement can occur. It is apparent that certain injuries have significant
agreement in all areas, whereas others seem less so.

The author reports no conflict of interest.

Copyright © 2013 by Lippincott Williams & Wilkins

384 | www.jorthotrauma.com Ó 2013 Lippincott Williams & Wilkins