Você está na página 1de 7

Journal of Plastic, Reconstructive & Aesthetic Surgery (2013) 66, 531e537

Delayed amputation following trauma increases

residual lower limb infection
Abhilash Jain a,b, Graeme E. Glass a,*, Hootan Ahmadi b, Simon Mackey b,
Jon Simmons b, Shehan Hettiaratchy b, Michael Pearse c,
Jagdeep Nanchahal a,b

Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences (NDORMS), Kennedy Institute of
Rheumatology, University of Oxford, ARC Building, 65 Aspenlea Road, Hammersmith, London W6 8LH, UK
Department of Plastic and Reconstructive Surgery, Imperial College Healthcare NHS Trust, Charing Cross Hospital
Campus, Fulham Palace Road, London, UK
Department of Orthopaedic Surgery, Imperial College Healthcare NHS Trust, Charing Cross Hospital Campus,
Fulham Palace Road, London, UK

Received 13 August 2012; accepted 19 November 2012

KEYWORDS Summary Introduction: Residual limb infection following amputation is a devastating

Trauma; complication, resulting in delayed rehabilitation, repeat surgery, prolonged hospitalisation
Amputation; and poor functional outcome. The aim of this study was to identify variables predicting
Lower limb; residual limb infection following non-salvageable lower limb trauma.
Infection; Methods: All cases of non-salvageable lower limb trauma presenting to a specialist centre over
Stump 5 years were evaluated from a prospective database and clinical and management variables
correlated with the development of deep infection.
Results: Forty patients requiring 42 amputations were identified with a mean age of 49 years
(19.9, 1SD). Amputations were performed for 21 Gustilo IIIB injuries, 12 multi-planar deglov-
ing injuries, seven IIIC injuries and one open Schatzker 6 fracture. One limb was traumatically
amputated at the scene and surgically revised. Amputation level was transtibial in 32, through-
knee in one and transfemoral in nine. Median time from injury to amputation was 4 days (range
0e30 days). Amputation following only one debridement and within 5 days resulted in signif-
icantly fewer stump infections (p Z 0.026 and p Z 0.03, respectively, Fishers exact test).
The cumulative probability of infection-free residual limb closure declined steadily from day
5. Multivariate analyses revealed that neither the nature of the injury nor pre-injury patient
morbidity independently influenced residual limb infection.

* Corresponding author. Tel.: 44 0208 383 4763; fax: 44 0208 383 4499.
E-mail address: graeme.glass@ndorms.ox.ac.uk (G.E. Glass).

1748-6815/$ - see front matter 2012 British Association of Plastic, Reconstructive and Aesthetic Surgeons. Published by Elsevier Ltd. All rights reserved.
532 A. Jain et al.

Conclusion: Avoiding residual limb infection is critically dependent on prompt amputation of

non-salvageable limbs.
2012 British Association of Plastic, Reconstructive and Aesthetic Surgeons. Published by
Elsevier Ltd. All rights reserved.

Wound infection is a major complication of post-traumatic performed during the initial hospital admission by senior
extremity amputation and reported to occur in 27e48% of plastic surgeons with input from senior orthopaedic
cases.1e3 Infections delay healing time, prolong hospital surgeons. The decision to amputate was made jointly.
stay and can necessitate revision surgery.1,2 A protracted Twelve cases were excluded as they presented via the
period of rehabilitation is necessary in these cases.4,5 tertiary referral system with established soft-tissue infections
Identifying avoidable factors implicated in post- or deep infection (osteomyelitis), having previously under-
amputation infection following non-salvageable extremity gone definitive reconstruction elsewhere. Two were repatri-
trauma should help to improve patient outcomes. ated from overseas. A summary of the 40 cases is presented in
Indications for primary amputation following severe Table 1. Nineteen patients presented directly through our
extremity trauma include an almost completely severed emergency department or through our allied trauma centre.
limb with significant distal trauma, extensive crush, severe The remainder were referred from other hospitals.
nerve dysfunction, an ischaemic limb with >4 h warm
ischaemic time,6,7 segmental bone loss (>1/3 length of the Statistical methods
tibia), muscle loss in more than two compartments (espe-
cially the posterior compartment) and severe open foot Fishers exact test was used to calculate all contingencies.
injuries.7 Transtibial amputees have better functional Linear regression analysis was used to analyse the rela-
outcomes than either through-knee or transfemoral ampu- tionship between the number of operative procedures
tees.8,9 Whilst there are now clear guidelines on how to performed and the delay to amputation. A KaplaneMeier
achieve optimal outcomes for patients undergoing limb survival plot was used to evaluate cumulative probability of
reconstruction,7 it is unclear whether the time from injury infection-free residual limb closure.
to amputation contributes to residual limb infection.
The aim of this study was to identify factors implicated
in post-amputation residual limb infection. Results

Patients and methods
Forty patients with 42 non-salvageable lower limb injuries
A retrospective chart review of 40 patients with severe were included for analysis. The mean age was 49 years (18e96
lower limb trauma necessitating eventual limb amputation years). Fourteen patients were smokers and eight patients had
presenting to a single level-1 trauma centre between a pre-injury history of diabetes mellitus. One smoker was
January 2007 and December 2011 was undertaken. Patients diabetic. Twenty-eight of 40 patients suffered non-
referred from other centres with established deep infection salvageable lower limb trauma as a result of a vehicular
were excluded. Data were collected on injury mechanism, collision. Of the remaining 12 injuries, four occurred as a result
severity, timing of operative procedures, whether the of a high-velocity fall; four as a result of a low-velocity fall (of
patient was a smoker or diabetic, subsequent infection and which one was identified as a pathological fracture); three as
organism involved. a result of a heavy object falling onto the patient; and one as
All patients were managed according to the joint British a result of a blast wave secondary to an explosion. In two
Association of Plastic, Reconstructive and Aesthetic cases, deliberate self-harm was identified as a precipitant
Surgeons/British Orthopaedic Association (BAPRAS/BOA) factor. Eight patients presenting following pedestrian vs. car
standards for the management of lower limb trauma.7 The collisions suffered nine non-salvageable lower limb injuries of
protocol included: immediate transfer to the specialist which five were Gustilo IIIB tibial fractures. Ten patients
centre where practicable, and antibiotic administration presenting following pedestrian vs. heavy goods vehicle
according to our local policy based on the BAPRAS/BOA (HGV)/coach or train collisions suffered 11 non-salvageable
standards and assessment of viability and debridement lower limb injuries of which seven were multi-planar degloving
jointly by consultant plastic and orthopaedic surgeons on injuries. Two patients had concomitant Gustilo IIIB fractures of
the next available day time list, unless there was a life- or the contralateral limb (in addition to the non-salvageable
limb-threatening injury, when the patient was taken injury) that were successfully reconstructed.
immediately to the operating theatre. Immediate ampu- In total, 21 of 42 injuries resulting in amputation were
tation was performed for obviously non-salvageable Gustilo IIIB fractures. Additionally, seven injuries were open
injuries. Injuries initially deemed salvageable or where fractures with an ischaemic limb (Gustilo IIIC), 12 were multi-
consent to proceed to amputation was not obtained planar degloving injuries or crush injuries and the remaining
underwent subsequent debridement as necessary and two included one Schatzker 6 tibial plateau fracture and one
application of negative pressure wound therapy until the traumatic amputation at the scene of injury. Excluding the
decision to amputate was made. All amputations were Schatzker 6 fracture, 23 of the remaining 28 fractures
Delayed amputation following trauma
Table 1 Details of 40 patients who presented with 42 non-salvageable lower limb injuries over 5 years.
Patient Age Direct Smoker DM Mechanism Injury No. Day of Day of Level of Stump Notes
(years) admission of ops 1st op amputation amputation infection
1 83 Y N N Pedestrian vs. HGV IIIB femur 1 0 0 AKA N
2 38 Y N N Crushed by advertising IIIC tibia 1 0 0 BKA N
3 76 Y N Y Pedestrian vs. HGV MPD thigh 2 0 0 AKA N
4 36 Y Y N Pedestrian vs. train Bilateral IIIC 2 0 0 AKA (R) N Psychiatric Hx
femur AKA (L)
5 53 Y N Y Pedestrian vs. coach Traumatic 3 0 0 BKA N
amputation tibia
6 43 N N N Blast injury Crush tibia 1 1 1 BKA N
7 32 Y N N HGV ran over foot Crush foot 2 1 2 BKA N
8 27 N N N Cyclist vs. HGV IIIC femur 6 0 2 AKA N IIIB right tibia
9 29 N Y N Baggage cart vs. MPD leg 1 2 2 BKA N
10 50 Y N N Motorbike vs. car IIIB tibia 3 0 2 BKA Y
11 39 Y Y N Pedestrian vs. car IIIB tibia 2 0 2 TKA Y
12 36 Y N N Pedestrian vs. car IIIB tibia 2 0 3 BKA Y Multi-system trauma
13 54 N N N Pedestrian vs. car IIIC tibia/knee 3 0 3 AKA N
14 32 Y N N Fall from height (DSH) IIIB ankle 2 0 3 BKA N Pelvic, sacral and
lumbar fractures
15 78 Y Y Y Low velocity fall IIIB tibia 2 1 3 BKA N Alcoholic
16 72 Y N Y Bus ran over foot MPD foot 2 0 3 BKA N
17 65 N Y N HGV ran over foot MPD foot 4 2 4 BKA Y
18 42 Y Y N Pedestrian vs. HGV MPD leg 3 1 4 BKA N
19 32 N N Y Pedestrian vs. car IIIB tibia (R) 2 0 4 AKA (L) N
Schatzker 6 (L) BKA (R)
20 41 Y Y N Pedestrian vs. train MPD foot 2 0 4 BKA Y Psychiatric Hx
21 52 Y Y N Crushed by steel plate IIIB ankle 2 0 4 BKA N
MPD foot
22 41 N N Y Fall from motorbike IIIB tibia 2 1 5 BKA N IIIB left tibia
23 73 N N Y Low velocity fall IIIB tibia 3 0 5 BKA Y
24 36 N N N Pedestrian vs. car IIIB tibia 3 0 6 BKA N
25 31 Y N N Motorbike vs. car IIIC tibia 3 0 6 BKA Y
26 53 N N N Cyclist vs. van IIIB tibia 3 1 8 BKA Y
27 41 N N N Motorbike vs. van IIIB tibia 4 1 8 BKA Y Femoral shaft
and neck #
28 67 Y N N Cyclist vs. car MPD leg 3 0 9 BKA Y
(continued on next page)

534 A. Jain et al.

involved the tibia and/or the ankle with the remaining five

Abbreviations: DM (diabetes mellitus); HGV (heavy goods vehicle); DSH (deliberate self harm); MPD (multi-planar degloving); AKA (above knee amputation); BKA (below knee amputation);
Patient initially refused BKA
cases involving the femur. Hence, in our series, tibial/ankle
fractures were approximately 4 times more common than
femoral fractures. Interestingly, of the five femoral frac-

Multi-system trauma

Multi-system trauma
tures, four presented with an ischaemic limb due to vascular
injury. By contrast, 3 of 23 tibial/ankle fractures involved an
ischaemic vascular injury (p Z 0.008, Fishers exact test).

Multi-level #
Hence, open femoral fractures, while less common than open
tibial fractures, were more likely to present with ischaemic

vascular compromise.

Operative procedures

The 40 patients underwent a mean of 2.7 operative



procedures (1.1, 1SD). The median time to first operation

was 0 days (within 24 h of injury, range 0e2 days). The

Level of


median time from injury to amputation was 4 days (range

0e30 days). There was no correlation between the type or



the mechanism of injury and the number of operative
procedures performed.
Day of



The level of the amputation was transtibial in 32 limbs,

Day of
1st op

through-knee in one limb and transfemoral in nine limbs.

One limb initially amputated through-knee was later revised


to above-knee after failure of the stump to heal. As ex-

of ops

pected, the type of injury reflected the level of the ampu-


tation. There was no correlation between the number of



operative procedures or the delay in amputation and the

level of the amputation. Hence, the level of the amputation

was dependent only on the nature of the injury.

Crush foot

IIIC femur
IIIB ankle
MPD foot

MPD foot
IIIB tibia

IIIB tibia

IIIB tibia
IIIB tibia
IIIB tibia

IIIB tibia
IIIB tibia

Residual limb infection


Infection occurred in a total of 16 residual limbs. Twenty-

Crushed by steel plate

five of 42 amputations occurred within 5 days of injury. Of

these, six developed infections in the residual limb. The
Pedestrian vs. car

Pedestrian vs. car

Pedestrian vs. car

Motorbike vs. van
Motorbike vs. car

remaining 17 amputations were performed between 6 and

Low velocity fall

Low velocity fall

Fall from height

Fall from height

Cyclist vs. van
Moped vs. car

30 days following injury. Of these, 10 developed residual

limb infection (p Z 0.03, Fishers exact test). Twenty-three

TKA (through knee amputation); Hx (History); # (fracture).

of 42 limbs underwent one debridement prior to

Table 2 Summary of the factors evaluated for their

contribution to infection of the residual limb.


Relative 95% CI P value


odds ratio
>1 Debridement 4.95 1.3e19.0 0.026




Amputation >5 days 4.5 0.2e3.0 0.03


post injury

Ischaemic vascular injury/ 0.6 0.2e2.3 NS

multi-planar degloving
Table 1 (continued )



Above-knee amputation 0.6 0.1e3.5 NS


Tertiary referral 0.9 0.2e3.0 NS


Smoker 1.8 0.5e7.1 NS



Diabetes mellitus 0.4 0.1e2.5 NS


Abbreviations: CI (confidence interval).

Bold values are the significant results.

Delayed amputation following trauma 535

more than one debridement was performed and when

amputation was delayed beyond 5 days (Table 2). However,
the number of debridements was not an independent
variable as it correlated highly (p < 0.001) with delay in
amputation. The cumulative probability of infection-free
residual limb with time is shown in Figure 1.
The presence of an ischaemic vascular injury, multi-
planar degloving, necessity for transfemoral amputation,
referral from another centre, smoking status or diabetes
mellitus was not independently predictive of a stump
infection (Table 2).
Figure 1 KaplaneMeier plot demonstrating cumulative Microbiological analyses of tissue samples from residual
probability of infection-free residual limb with time from limbs stumps revealed that 7 of 16 samples were positively
injury to amputation. attributed to nosocomial organisms. In one additional case,
the microbiological profile was suggestive of nosocomial
amputation (initial debridement and assessment followed infection. In only two cases did the microbiological profile
by immediate or planned amputation) and of these, five of the wound (coagulase-negative Staphylococcus and S.
developed infection in the residual limb. The remaining 19 aureus, respectively) match that of the patients own
limbs underwent more than one debridement prior to commensal skin flora. In six further cases, the microor-
amputation. Of these, 11 developed subsequent residual ganisms were not identified. The microbiological profile
limb infection (p Z 0.026, Fishers exact test). Thus, and the antibiotic therapy for the 16 residual limb infec-
residual limb infection was significantly more likely when tions are summarised in Table 3.

Table 3 Microbiological summary of 16 residual limb infections.

Patient How identified Microbiology Nosocomial Abx
10 Cellulitis Coliforms Yes Cefuroxime
Pseudomonas Ciprofloxacin
Enterobacter Clindamycin
11 Cellulitis Acinetobacter Yes Vancomycin
Pus expressed from wound Enterococcus Meropenem
13 Cellulitis No growth Unknown Meropenem
17 Cellulitis No growth Unknown Cefuroxime
20 Cellulitis Pseudomonas Unknown Tazocin
23 Cellulitis Citrobacter Yes Augmentin
Coliforms Vancomycin
25 Wound breakdown MRSA Yes Rifampicin
Bacillus Vancomycin
26 Cellulitis No growth Unknown Vancomycin
Sepsis Ciprofloxacin
27 Pus expressed from wound Pseudomonas Yes Cefuroxime
Enterococci Ciprofloxacin
28 Cellulitis Staphylococcus aureus No Augmentin
Bacillus Tazocin
29 Cellulitis Enterococci Yes Vancomycin
Pus expressed from wound Pseudomonas Ciprofloxacin
Anaerobes Metronidazole
33 Pus expressed from wound No growth Unknown Flucloxacillin
37 Pus expressed from wound Coliforms Yes Tazocin
Enterobacter Flucloxacillin
38 Cellulitis Coag. Neg. Staph No Vancomycin
39 Wound breakdown No growth Unknown Benzylpenicillin
40 Cellulitis No growth Unknown Vancomycin
536 A. Jain et al.

Discussion represents the point of divergence in therapeutic strategy in

most cases, when, for example salvage was considered
Severe extremity trauma may necessitate amputation. The achievable or when the patient refused to consent to
rationale for attempting reconstruction is guided by patient amputation following the consultation process.
factors, technical considerations and predicted functional It may be considered surprising that our infection
outcomes and performed when the outcomes are expected proportion was as high as 16 of 42. However, grossly
to exceed or at least equal those of amputation.4,7,8,10,11 contaminated wounds lead to colonisation of the veno-
These are difficult decisions and a reluctance to commit lymphatic system and our infection rate is similar to the
to amputation often results in multiple debridement and 34% reported by the LEAP (Lower Extremity Assessment
dressing changes. These problems are exacerbated by Project) group following review of 149 amputations for non-
delays in transfer to level-1 trauma centres and lack of salvageable lower limb injuries.16
exposure to such injuries among surgical trainees who Our data show that 7 of 16 residual limb infections could be
receive these patients acutely.12 Occasionally some definitely attributed to nosocomial infections and in only 2 of
patients are transferred back to the referring centre for 16 patients could the infection be ascribed to skin commen-
amputation as it is perceived that rehabilitation may be sals. As has previously been reported by our group and others,
more easily achieved in a hospital close to their home, surveillance cultures were not predictive of the infecting
despite evidence for improved functional outcomes and organism.2,18 Rather, colonisation of unhealthy tissues is
cost-effectiveness when definitive treatment is undertaken more likely to occur in patients with a longer duration in
in a specialist environment.13,14 hospital before excision of all the non-viable tissue.18 It is
Our most significant finding was that infection was extremely difficult to adequately debride a limb with a non-
directly related to a delay in performing the amputation. salvageable injury without proceeding to an amputation.
While it has previously been reported that delayed ampu- The finding that a delay of up to 5 days can be tolerated
tation following severe lower limb trauma is associated without negatively impacting on the risk of post-amputation
with a high probability of infectious complications,15,16 our deep-tissue infection of the stump is useful in that it allows
study differs in that we have stratified our results from the time to prepare the patient for their amputation physiologi-
initial hospital admission only. The proportion of infections cally and psychologically. To this effect we enlist the services
was significantly higher in patients where the amputation of a multidisciplinary limb rehabilitation team including
was delayed for more than 5 days following the injury, patients who have completed their rehabilitation.19,20
irrespective of when the first debridement was performed The LEAP study group found that smoking contributed to
(within 24 h of admission in most cases). While Pollak infection and osteomyelitis in salvaged open tibial fractures.21
et al.17 reported no correlation between delay from injury In our series of amputations, the fact that neither patient
to debridement and subsequent infection among 307 factors such as smoking and diabetes mellitus nor the pattern
patients with salvageable high-energy trauma, the delay of the tissue injury appears to have influenced the rate of
was usually <24 h. Our series represents a different pop- stump infection emphasises the importance of the surgical
ulation as the injuries were non-salvageable and the management and prompt decision making in these cases.
definitive debridement (amputation) was delayed for Our results suggest that infection-free amputation is
a number of clinical and patient-related reasons. Clearly, critically dependent on an expeditious decision-making
however, adequacy of debridement is more important than process. Once a limb has been deemed to be non-
the urgency with which it is performed. salvageable, the patient should be counselled that
Deep infection was significantly more likely when more prompt amputation should be performed in order to ach-
than one debridement was performed and when amputation ieve optimal outcomes. Repeated debridement and
was delayed for more than 5 days. These data convey similar dressing changes, including the use of negative pressure
messages as in all but two cases only one debridement fol- wound therapy, do not permit extension of this window.
lowed by amputation was performed by day 5 (in these two
cases a second debridement was performed at day 5). It
could be argued that, as infection increases with delay to Financial disclosures
amputation and the number of operative procedures per-
formed, we should advocate only one procedure. However, JN: Project no F-09-23N was supported by the AO
in practice we have found it is often appropriate to take the Foundation.
patient to theatre for an initial assessment and then plan an GG is in receipt of a research grant from the Academy of
amputation as a semi-elective second procedure following Medical Sciences.
a discussion with the patient and their family. This is our
preferred option for cases when the decision is based on Conflict of interest statement
functional and rehabilitative grounds. We perform imme-
diate amputation only for clinical reasons when appropriate.
Moreover, patients with severe extremity trauma are often
not able to consent to amputation within 24 h of injury. The
medico-legal implications may be considerable. We do not References
wish to discourage a period of reflection and consultation
and we feel that it would be unwise to imply that immediate 1. Akinyoola AL, Ojo OD, Oginni LM. Microbiology of amputation
amputation in all cases is justified. Hence, the distinction wound infection in a Nigerian setting. J Wound Care 2008;
between two procedures and three is important as this 17(202):204e6.
Delayed amputation following trauma 537

2. Burns TC, Stinner DJ, Mack AW, et al. Microbiology and injury 12. Eardley WG, Taylor DM, Parker PJ. Amputation and the
characteristics in severe open tibia fractures from combat. assessment of limb viability: perceptions of two hundred and
J Trauma Acute Care Surg 2012;72:1062e7. thirty two orthopaedic trainees. Ann R Coll Surg Engl 2010;92:
3. Huh J, Stinner DJ, Burns TC, et al. Infectious complications and 411e6.
soft tissue injury contribute to late amputation after severe 13. MacKenzie EJ, Weir S, Rivara FP, et al. The value of trauma
lower extremity trauma. J Trauma 2011;71:S47e51. center care. J Trauma 2010;69:1e10.
4. Bosse MJ, MacKenzie EJ, Kellam JF, et al. An analysis of 14. Prada SI, Salkever D, Mackenzie EJ. Level-I trauma center
outcomes of reconstruction or amputation after leg- effects on return-to-work outcomes. Eval Rev 2012;36:
threatening injuries. N Engl J Med 2002;347:1924e31. 133e64.
5. MacKenzie EJ, Bosse MJ, Pollak AN, et al. Long-term persis- 15. Bondurant FJ, Cotler HB, Buckle R, et al. The medical and
tence of disability following severe lower-limb trauma. Results economic impact of severely injured lower extremities.
of a seven-year follow-up. J Bone Jt Surg Am 2005;87:1801e9. J Trauma 1988;28:1270e3.
6. Glass GE, Pearse MF, Nanchahal J. Improving lower limb 16. Harris AM, Althausen PL, Kellam J, et al. Complications
salvage following fractures with vascular injury: a systematic following limb-threatening lower extremity trauma. J Orthop
review and new management algorithm. J Plast Reconstr Trauma 2009;23:1e6.
Aesthet Surg 2009;62:571e9. 17. Pollak AN, Jones AL, Castillo RC, et al. The relationship
7. Nanchahal J, Nayagam S, Khan U, et al. Standards for the between time to surgical debridement and incidence of
management of open fractures of the lower limb (published on infection after open high-energy lower extremity trauma.
behalf of the British Association of Plastic, Reconstructive and J Bone Jt Surg Am 2010;92:7e15.
Aesthetic Surgeons and British Orthopaedic Association). RSM 18. Glass GE, Barrett SP, Sanderson F, et al. The
Press; 2009. microbiological basis for a revised antibiotic regimen in
8. MacKenzie EJ, Bosse MJ, Castillo RC, et al. Functional high-energy tibial fractures: preventing deep infections by
outcomes following trauma-related lower-extremity amputa- nosocomial organisms. J Plast Reconstr Aesthet Surg 2011;64:
tion. J Bone Jt Surg Am 2004;86-A:1636e45. 375e80.
9. Penn-Barwell JG. Outcomes in lower limb amputation following 19. Darnall BD, Ephraim P, Wegener ST, et al. Depressive symptoms
trauma: a systematic review and meta-analysis. Injury 2011; and mental health service utilization among persons with limb
42:1474e9. loss: results of a national survey. Arch Phys Med Rehabil 2005;
10. Busse JW, Jacobs CL, Swiontkowski MF, et al. Complex limb 86:650e8.
salvage or early amputation for severe lower-limb injury: 20. McCarthy ML, MacKenzie EJ, Edwin D, et al. Psychological
a meta-analysis of observational studies. J Orthop Trauma distress associated with severe lower-limb injury. J Bone Jt
2007;21:70e6. Surg Am 2003;85-A:1689e97.
11. MacKenzie EJ, Bosse MJ. Factors influencing outcome following 21. Castillo RC, Bosse MJ, MacKenzie EJ, et al. Impact of smoking
limb-threatening lower limb trauma: lessons learned from the on fracture healing and risk of complications in limb-
lower extremity assessment project (LEAP). J Am Acad Orthop threatening open tibia fractures. J Orthop Trauma 2005;19:
Surg 2006;14:S205e10. 151e7.