Low-energy Gunshot-induced Tibia Fractures: What Proportion Develop Complications?

Abstract

Background

Gunshot injuries of the extremities are common in the United States, especially among people with nonfatal gunshot wounds. Controversy persists regarding the proper management for low-energy gunshot-induced fractures, likely stemming from varying reports on the likelihood of complications. There has yet to be published a study on a large cohort of patients with gunshot-induced tibia fractures on which to base our understanding of complications after this injury.

Questions/purposes

(1) What percentage of patients with low-energy gunshot-induced tibia fractures developed complications? (2) Was there an association between deep infection and fracture location, injury characteristics, debridement practices, or antibiotic use?

Methods

This was a multicenter retrospective study. Between January 2009 and December 2018, we saw 201 patients aged 16 years or older with a gunshot-induced fracture who underwent operative treatment; 2% (4 of 201) of those screened had inadequate clinical records, and 38% (76 of 201) of those screened had inadequate follow-up for inclusion. In all, 121 patients with more than 90 days of follow-up were included in the study. Nonunion was defined as a painful fracture with inadequate healing (fewer than three cortices of bridging bone) at 6 months after injury, resulting in revision surgery to achieve union. Deep infection was defined according to the confirmatory criteria of the Fracture-Related Infection Consensus Group. These results were assessed by a fellowship-trained orthopaedic trauma surgeon involved with the study. Complication proportions were tabulated. A Kaplan-Meier chart demonstrated presentations of deep infection by fracture location (proximal, shaft, or distal). Univariate statistics and multivariate Cox regression were used to examine the association between deep infection and fracture location, entry wound size, vascular injury, intravenous (IV) antibiotics in the emergency department (ED), deep and superficial debridement, the duration of postoperative IV antibiotics, and the use of topical antibiotics, while adjusting for age, race/ethnicity, smoking status, and BMI. A power analysis for the result of deep infection demonstrated that we would have had to observe a hazard ratio of 4.28 or greater for shaft versus proximal locations to detect statistically significant results at 80% power and alpha = 0.05.

Results

The overall complication proportion was 49% (59 of 121), with proportions of 14% (17 of 121) for infection, 27% (33 of 121) for wound complications, 20% (24 of 121) for nonunion, 9% (11 of 121) for hardware breakage, and 26% (31 of 121) for revision surgery. A positive association was present between deep infection and deep debridement (HR 5.51 [95% confidence interval 1.12 to 27.9]; p = 0.04). With the numbers available, we found no association between deep infection and fracture location, entry wound size, vascular injury, IV antibiotics in the ED, superficial debridement, the duration of postoperative IV antibiotics, and the use of topical antibiotics.

Conclusion

In this multicenter study, we found a higher risk of complications in operative gunshot-induced tibia fractures than prior studies have reported. Infection, in particular, was much more common than expected based on prior studies. Consequently, surgeons might consider adopting the general management principles for nongunshot-induced open tibia fractures with gunshot-induced fractures, such as the use of IV antibiotics both initially and after surgery. Further research is needed to test and validate these approaches.

Level of Evidence

Level IV, therapeutic study.

Introduction

Gunshot extremity injuries are common in the United States, especially among patients with nonfatal gunshot wounds [5, 6]. Roughly half of all patients who are admitted to the hospital for gunshot injuries undergo fracture care, but a standardized protocol for treatment has not been well established [1, 2, 5, 6, 17]. Although initially classified as open fractures in the seminal work by Gustilo and Anderson [7], low-velocity gunshot-induced fractures have been reported to have similar infection rates as closed blunt injuries [3, 5, 7, 8, 13, 14, 18]. A 2018 survey of Orthopaedic Trauma Association (OTA) members revealed widely varying opinions regarding the need for antibiotics or surgical debridement for low-energy gunshot-induced fractures, as well as regarding the likelihood of nonunion [12].

Controversy about whether low-energy gunshot-induced fractures should be considered at elevated risk for infection or nonunion and thus should undergo preventative treatment and monitoring accordingly may stem from varying reports of complication proportions after these injuries. For gunshot-induced tibia fractures, deep infection has been reported in the range of 2.3% to 9.1% in a few small series [10, 13, 18]. Deep infection after low-energy gunshot-induced fractures overall has been reported to occur at a rate of 3.6% [13]. If the deep infection rates after low-energy gunshot-induced fractures are thought to be low, this may lead some surgeons to treat these injuries similarly to closed injuries regarding infection prophylaxis and other complication prevention principles. With respect to gunshot-induced tibia fractures, no large series on which to base our understanding of the complication proportions after this injury has yet been published.

We therefore asked (1) What percentage of patients with low-energy gunshot-induced tibia fractures developed complications? (2) Was there an association between deep infection and fracture location in the tibia, injury characteristics, debridement practices, or antibiotic use?

Patients and Methods

Study Design and Setting

This was a retrospective study performed at four academic Level I tertiary care referral centers. After obtaining institutional review board approval, an initial screen was performed to identify records by use of diagnostic and procedure codes, as well as demographic variables coded in the EMR.

Participants

Between January 2009 and December 2018, we saw 201 patients aged 16 years or older with a gunshot-induced tibia fracture who underwent operative treatment, including irrigation and debridement, external fixation, open reduction and internal fixation (ORIF), and/or intramedullary nail placement (IMN) (Fig. 1).

Fig. 1.

STROBE diagram demonstrating the selection of the cohort analyzed in this study; ORIF = open reduction internal fixation; IMN = intramedullary nailing.

Of those initially screened, patients were included if they had complete records (including whether deep debridement had been performed). Four patients did not have complete records and thus 98% (197 of 201) met this criterion. Patients were excluded if they had less than 90 days of follow-up available and did not reach a clinical endpoint before this (including deep infection, wound complication, nonunion, hardware breakage, or revision surgery). This excluded 39% (76 of 197), leaving 60% (121 of 201) of those originally screened in the final cohort, including seven patients with less than 90 days of follow-up but who reached an endpoint before 90 days.

Interventions

All included patients underwent operative treatments including debridement, external fixation, IMN, and/or ORIF. In general, surgical debridement was performed when foreign bodies, contamination, or necrotic tissue were present in the gunshot wound and were accessible without causing undue further trauma. Deep debridement was defined and differentiated from superficial debridement as the removal of bone fragments during the procedure. In general, temporizing external fixation was used when the soft tissue injury was felt to not be amenable to immediate internal fixation. In general, the definitive fixation of proximal and distal fractures (extending to the joint) was internal fixation with plates and/or screws and the definitive fixation of shaft fractures utilizing intramedullary nailing.

Descriptive Data

The final cohort of 121 patients had a median (interquartile range) follow-up of 9 months (5 to 15), had an average age of 33 ± 12 years, and was 90% male (109 of 121, Table 1). With regard to race and ethnicity, the cohort was identified as 79% Black (95 of 121), 17% white (21 of 121), 2% Hispanic/Latino (3 of 121), and 2% unknown (2 of 121). The average BMI was 27 ± 6 kg/m², and 48% (58 of 121) of patients were tobacco users. The cohort included 28% (34 of 121) proximal fractures (defined as extending to the knee), 62% (75 of 121) shaft fractures, and 10% (12 of 121) distal fractures (defined as extending to the ankle). Eighty-four percent (102 of 121) of the cohort underwent superficial debridement, 45% (54 of 121) underwent deep debridement, 22% (27 of 121) underwent external fixation, 36% (44 of 121) underwent ORIF, and 60% (73 of 121) underwent IMN. Deep debridement was performed in 55% (41 of 74) of proximal fractures, 29% (10 of 34) of shaft fractures, and 25% (3 of 12) of distal fractures. The definitive treatment was external fixation in five patients. The definitive treatment, after operative irrigation and debridement, was casting in two patients.

Table 1.

Patient demographic and injury characteristics

Demographic variable	Summary (n = 121)
Age in years	33 ± 12
Sex
Female	10 (12)
Race
Black	79 (95)
White	17 (21)
Hispanic	2 (3)
Unknown	2 (2)
BMI in kg/m2	27 ± 6
Smoker
Yes	48 (58)
Fracture location
Proximal	28 (34)
Shaft	62 (75)
Distal	10 (12)

Data presented mean ± SD or % (n).

Data Sources and Measurement

We reviewed inpatient and outpatient records and operative reports for patient demographics, initial diagnosis, antibiotic use and duration, surgical procedures performed, and complications. The observation period included all available outpatient follow-up with the respective institution. Complications included hardware breakage, revision surgery, deep surgical site infection, superficial wound complications, and nonunion. Nonunion was defined as a painful fracture with inadequate healing (fewer than three cortices of bridging bone) at 6 months after injury, resulting in revision surgery to achieve union. Superficial infection was defined as infection superficial to the fascia that was treated with local wound care and/or oral antibiotics. Deep infections were defined as fracture-related infections according to the Fracture-Related Infection Consensus Group [11]. This group defined fracture-related infection as involving the presence of one of the following: fistula, sinus, or wound breakdown (with communication to the bone or the implant); purulent drainage from the wound or presence of pus intraoperatively; phenotypically indistinguishable pathogens identified by culture from at least two separate deep tissue or implant (including sonification fluid) specimens taken during an operative intervention; and presence of microorganisms in deep tissue taken during an operative intervention, as confirmed by a histopathologic examination using specific staining techniques for bacteria or fungi. These criteria were applied by a review of operative reports and patient records by one of the four fellowship-trained orthopaedic trauma surgeons (CL, RVO, MH, JM) involved in this study.

Bias

Selection bias and assessment bias were concerns given the retrospective design of this study. Selection bias may have occurred due to surgeons at each center applying criteria for operative fixation and operative debridement in a manner that may have differed slightly from other centers. However, the primary outcome of aggregate complication proportions would not be affected by any potential differential selection since all patients were pooled (that is, there was no comparison between groups that would have been systematically biased by differential selection). Selection bias was mitigated in the test of association between deep infection and various baseline characteristics by performing a multivariate analysis and controlling for debridement, antibiotic use, wound size, and vascular injury, as well as age, race/ethnicity, smoking status, and BMI.

Assessment bias, wherein there was differential application of criteria for a given result such as infection, would also have been a possibility. This was mitigated by agreement among all outcome assessors (the fellowship-trained orthopaedic trauma surgeon on the study team at each site [CL, RVO, MH, JM]) on precise definitions for deep infection and nonunion, as indicated above.

Primary and Secondary Study Outcomes

Our primary study goal was to determine the proportions of patients developing deep infection, wound complications, nonunion, hardware breakage, or revision surgery after operative gunshot-induced fractures of the tibia. This was determined by compiling the study cohort as discussed above and calculating ratios and percentages of patients having each complication. Our secondary study goal was to determine whether an association was present between deep infection and location in the tibia, injury characteristics, debridement practices, or antibiotic use. This was determined by use of multivariate regression to test for associations between each potential predictor and the outcome of interest (deep infection) while controlling for other potential predictors as well as age, race/ethnicity, smoking status, and BMI.

Ethical Approval

Ethical approval for this study was obtained from the VCU Office of Research and Innovation (IRB number HM20014632), from the University of Mississippi Medical Center (IRB file #2017-0247), from the University of Iowa (IRB-01 DHSS Registration # IRB00000099), and from the University of Maryland, Baltimore Institutional Review Board (HP-00083674).

Statistical Analysis

Ratios and percentages were compiled for each complication in the final study cohort of 121 patients. We explored whether an association was present between deep infection and potential predictors of interest, including (1) fracture location (proximal, shaft, or distal), (2) the administration of intravenous (IV) antibiotics upon arrival in the emergency department (ED), (3) the use of deep debridement (defined as removal of bone fragments) during the initial operation, (4) the duration of postoperative IV antibiotics, (5) the size of the entry wound, (6) the use of topical antibiotics, and (7) the presence of a vascular injury. Some patients were missing data for certain predictors of interests, including 10% (12 of 121) for the variable of IV antibiotics in the ED, 44% (53 of 121) for entry wound size, 12% (14 of 121) for postoperative IV antibiotic duration, and 1% (1 of 121) for topical antibiotic use. Missing data were excluded from the summary for the relevant statistic. Proportions and percentages of patients with each of these potential predictors were compiled in the groups with and without detected deep infection. Univariate comparisons were made between groups with and without detected deep infection with the Fisher exact test for proportions and the Student t-test for continuous variables. Presentations of deep infection in the proximal fracture, shaft fracture, and distal fracture groups were visualized using a Kaplan-Meier chart. A multivariate Cox regression model was used to evaluate the association between deep infection and the above predictors, while controlling for age, race/ethnicity, smoking status, and BMI. Cox regression enabled the estimation of associations while accounting for loss to follow-up by considering observations to be right-censored at the final time-point of clinical follow-up. A power analysis for the outcome of deep infection using the current sample size and characteristics with observed infection rates demonstrated that we would have had to observe a hazard ratio of 4.28 or greater for shaft versus proximal locations to detect statistically significant results at 80% power and alpha = 0.05. All statistical analyses were performed with Stata version 16.1 (Stata Corp).

Results

Proportion of Patients who Developed Complications

In this cohort of 121 patients who underwent operative treatment for a gunshot-induced tibia fracture, 49% (59 of 121) developed complications. This included 14% (17 of 121) who developed deep infections, 27% (33 of 121) who developed wound complications, 20% (24 of 121) who developed nonunions, 9% (11 of 121) who developed hardware breakage, and 26% (31 of 121) who underwent revision surgery (Fig. 2). Deep infection occurred in 18% (6 of 34) of patients with proximal fractures, 15% (11 of 75) of patients with shaft fractures, and 0% (0 of 12) of patients with distal fractures.

Fig. 2.

Bar chart demonstrating complication proportions in this study.

Deep Infection as a Function of Injury Characteristics and Initial Management

Patients who developed and did not develop deep infections were compared with regard to injury characteristics and initial management practices (Table 2). Deep debridement had been performed in a 31-percentage-point higher proportion of patients who developed a deep infection compared with patients who did not develop an infection (p = 0.03 for the univariate comparison). Other injury and management characteristics, including fracture location, IV antibiotics in the ED, superficial debridement, postoperative IV antibiotic duration, entry wound size, topical antibiotic use, and vascular injury, did not differ between the groups in univariate comparisons.

Table 2.

Injury and initial management characteristics in patients with and without deep infection

	Deep infection
Management	Not detected (n = 104)	Detected (n = 17)	p value for univariate comparison
IV antibiotics in ED	80 (76 of 95)	12 of 14	> 0.99
Deep debridement	40 (42 of 104)	12 of 17	0.03
Duration postop IV antibiotics	2.6	1.8	0.17
Entry wound size	1.9	2.2	0.68
Topical antibiotics	6 (6 of 103)	3 of 17	0.12
Vascular injury	13 (14 of 104)	4 of 17	0.28
Superficial debridement	83 (86 of 104)	16 of 17	0.31

Data presented as % (n); the denominators for some proportions are adjusted to reflect missing data points.

Kaplan-Meier charts demonstrated visually the presentations of deep infection in proximal, shaft, and distal fracture groups (Fig. 3). After controlling for potential confounding variables including age, race/ethnicity, smoking status, and BMI, there was no association between deep infection and fracture location, IV antibiotic use in the ED, superficial debridement, postoperative IV antibiotic duration, entry wound size, topical antibiotic use, or vascular injury (Table 3). In the multivariate analysis, deep debridement was associated with a greater hazard for deep infection (HR 5.51 [95% CI 1.12 to 27.9]; p = 0.04).

Fig. 3.

Kaplan-Meier curve demonstrating presentations of deep infection in proximal, shaft, and distal fracture groups. Vertical black lines indicate censored observations (time points where clinical follow-up ended for one or more patients in the indicated group). A color image accompanies the online version of this article.

Table 3.

Cox model adjusted hazard ratios for deep surgical site infection

Variable	Hazard ratio (95% CI)	p value
Agea	1.02 (0.95-1.08)	0.64
Nonwhite vs. white	1.57 (0.18-13.8)	0.68
Smoking (yes vs. no)	0.97 (0.27-3.47)	0.96
BMIa	1.04 (0.95-1.13)	0.40
Location (proximal vs. shaft)	1.30 (0.34-5.17)	0.71
IV antibiotic administered in ED	0.96 (0.18-5.16)	0.96
Deep debridement performed	5.51 (1.12-27.9)	0.04
Superficial debridement performed	1.69 (0.20-14.0)	0.63
Duration postop IV antibiotic (per day increase)	0.92 (0.61-1.36)	0.67
Entry wound size (per cm increase)	0.96 (0.55-1.68)	0.88
Topical antibiotic given	2.38 (0.26-21.6)	0.44
Vascular injury	3.29 (0.81-13.4)	0.10

^aHR associated with 1 unit increase; ED = emergency department.

Discussion

Low-energy gunshot-induced extremity injuries have a high prevalence in the United States, placing a large economic burden on the healthcare system [1, 2, 5, 6, 17]. Despite the prevalence of these injuries, a standardized protocol for treatment has yet to be established, and there are wide variations in practices, such as antibiotic regimen and use of formal surgical debridement [2, 3, 8, 9, 17, 20]. This may be due in part to the absence of a perceived need for standardized treatment because previously reported complication proportions are similar to or lower than those for blunt injuries [13, 18]. Previous reports of low complication proportions in gunshot-induced tibia fractures, specifically, were derived from small case series. The present study represents the largest cohort published thus far of gunshot-induced tibia fractures. The proportion of complications was 49% (59 of 121), including a 14% rate (17 of 121) of deep infection. With the numbers available, fracture location, antibiotic administration in the ED, duration of postoperative antibiotic use, entry wound size, topical antibiotic use, vascular injury, and superficial debridement were not found to be associated with the occurrence of deep infection. Deep debridement was positively associated with deep infection, which may have been due to more severe injuries having a higher likelihood of being indicated for deep debridement and also having a higher likelihood of developing infection. These data suggest that low-energy gunshot-induced tibia fractures have higher proportions of complication than previously reported. With this information, surgeons may modify how they counsel patients with these injuries and may choose to change their management practices with regard to the use of antibiotics and surgical debridement. Further research is needed, however, to conclusively validate any new approaches to management and to establish appropriate protocols.

Limitations

The limitations of this study include biases related to its retrospective nature as well as the study being potentially underpowered to detect relationships of interest. Selection bias may have arisen from the decision to perform debridement and/or surgical fixation being made individually by the surgeon on the treatment team at each center. We believe that this risk for selection bias does not disqualify the findings because this is a realistic sample across multiple centers with indications applied according to the judgment of fellowship-trained orthopaedic trauma surgeons in the United States. There is no reason to believe there would be any systematic deviation in the application of these indications in how they would be applied elsewhere that would invalidate the generalizability of the findings.

An additional concern is transfer bias, which may have arisen due to loss to follow-up. Thirty-nine percent (76 of 197) of the eligible cohort was lost to follow-up before 90 days. Although it is possible that complications may have arisen in patients lost before 90 days or complications may have arisen in patients after the end of their documented follow-up in this study, the lack of detection of these additional complications means that the complication proportions reported here are, if anything, underestimates. Therefore, the interpretation that complication rates are higher than previously reported is not invalidated by this loss to follow-up as the reported rates would potentially be even higher if fewer patients were lost.

A third limitation is assessment bias. It is possible that the definitions of complications were applied differently at different centers. However, having a single fellowship-trained orthopaedic trauma surgeon apply standardized, agreed-upon criteria at each center helped to mitigate this bias.

A final limitation is that the study was underpowered to detect associations between deep infections and predictors of interest. We would have been able to detect an effect with an approximate HR of 4.28 or greater with the numbers available. In light of this, this study should be interpreted as calling for research targeted at testing different management approaches instead of providing conclusive evidence to change management practices.

Proportion of Patients who Developed Complications

The proportion of patients who developed complications in this series (49% [59 of 121]), and in particular deep infection (14% [17 of 121]), was higher than that reported by most other studies we found on this topic. Su et al. [18] published a retrospective review of 327 patients with tibial shaft fractures, including 43 that were gunshot-induced. In the gunshot-induced subgroup, they reported a 2.3% rate (1 of 43) of deep infection, as well as one malunion, no nonunions, and one revision operation (for deep infection). Metcalf et al. [10] published a retrospective review of 211 tibia fractures, including 23 gunshot-induced fractures. There was a 9% rate of infection and a 30% rate of nonunion in the gunshot-induced subgroup. Prather et al. [15] published a retrospective study that included 44 gunshot-induced tibia fractures. The infection rate was 9.1%, the nonunion rate was 11.4%, and the hardware failure rate was 4.7%. These data suggest that infection rates for gunshot-induced tibia fractures are in the 2.3% to 9.1% range.

In contrast, the present series, which is larger than those mentioned above by a factor of approximately 3 to 5, demonstrated an infection rate of 14%, similar to a 13% rate of deep infection recently demonstrated in 486 blunt open tibia fractures [19]. The infection rate in our series more closely approximates the previously published rate in Gustilo-Anderson Type II injuries (17%) rather than Type I injuries (1%) [19]. One possible explanation for the increased deep infection proportion in this study is that gunshot-induced tibia fractures represent a heterogeneous group of injuries, and this population may have been misrepresented in prior series with smaller cohorts. Though our study does not specifically validate any management approaches, it may serve to challenge the belief among some surgeons that gunshot-induced fractures can be treated similarly to closed blunt injuries. This belief was demonstrated indirectly as a 28% rate of agreement with use of no debridement and 24 hours of IV antibiotics for low-energy gunshot-induced fractures in a recent survey of OTA members [12], as well as a 15% proportion of OTA members in another survey not routinely providing antibiotic prophylaxis for these injuries [9]. Ideally, the present results will also spur further investigation of specific management protocols for low-energy gunshot-induced tibia fractures.

Deep Infection as a Function of Injury Characteristics and Initial Management

With the numbers available, and after controlling for potential confounding variables such as age, race/ethnicity, smoking status, and BMI, we found no association between the occurrence of deep infection and fracture location. The lack of a difference in infection rate by fracture location challenges the belief that articular involvement is an independent indication for antibiotic use in gunshot-induced fractures. This belief was demonstrated in a recent survey of OTA members wherein 50% (8 of 16) of those who do not routinely provide antibiotics said they would be more apt to administer them for intra-articular fractures. Though the present study was underpowered to detect differences in infection rate by fracture location, even the point estimates (18% for proximal, 15% for shaft, and 0% for distal) do not suggest a clinically meaningful higher infection proportion for articular injuries that would justify treating them differently than non-articular injuries with regard to infection prophylaxis.

There was a positive association between deep infection and deep debridement. This may have been due to more severe injuries being more likely to undergo deep debridement and also more likely to develop deep infections rather than deep debridement causing deep infection. This association between deep debridement and infection was also demonstrated by Donnally et al. [4], wherein the deep infection rate in 23 gunshot-induced tibia fractures treated with operative debridement was 26.1% compared with 6.25% in 16 similar fractures treated without debridement [4]. The association in our paper was similarly subject to selection bias as this was a retrospective study. The presence of this potentially confounded relationship in two observational studies (the current study and Donnally et al. [4]) suggests that a controlled experiment will be needed to determine whether there is a causal relationship between debridement and deep infection for gunshot-induced tibia fractures.

With the numbers available, there was no association in this study between deep infection and wound size, vascular injury, superficial debridement, IV antibiotics in the ED, postoperative IV antibiotic duration, or topical antibiotics. Forty-four percent (53 of 121) of patients were missing data for the entry wound size variable, which limits the interpretability of the finding of no association between wound size and deep infection. Again, the study was underpowered to detect these associations and the main takeaway is that a controlled experiment or large database study would be necessary to examine these associations more conclusively. A 2015 systematic review summarized the published results on antibiotic and debridement practices for gunshot-induced fractures and the authors stated they “cannot recommend for or against extensive debridement” and that the relatively few patients in published reports treated without antibiotics limits the conclusiveness of recommendations regarding antibiotics [16]. The authors similarly called for further high-quality research in this area. Still, we thought it was important to report the nonsignificant associations demonstrated in the present study because this is the largest published cohort of gunshot-induced tibia fractures we are aware of, and providing these findings highlights the inconclusive state of knowledge on how to treat these injuries.

Conclusion

The proportion of patients with gunshot-induced tibia fractures developing complications was 14% for deep infection, 27% for wound complications, 20% for nonunion, 9% for hardware breakage, and 26% for revision surgery. With the numbers available, there was no association between deep infection and fracture location, wound size, vascular injury, antibiotic practices, or superficial debridement. There was a positive association between deep debridement and deep infection, but this association may have been confounded by selection bias. The proportion of patients developing deep infections in this study is higher than that previously reported in smaller studies. This information should encourage surgeons to anticipate a relatively high rate of deep infection and other complications for this injury in both conversations with patients and in their management decisions. Experimental research is needed to validate causal associations between specific management practices such as debridement or extended antibiotic administration and infection or nonunion, with the aim of reducing complications after gunshot-induced tibia fractures.