Tibia fractures are common injuries that occur as a result of both high- and low-energy trauma. Healing is less reliable than that of other long-bone fractures, with reported nonunion rates as high as 14%.1,2 Multiple factors influence primary healing of tibial shaft fractures, including degree of bony apposition, mechanism of injury (MOI), and extent of associated soft tissue injury. Complex interactions between fracture, injury, and patient characteristics can lead to nonunion when the process of fracture repair is unable to overcome impairment of local biology, unfavorable mechanics, and/or challenging host factors.3–6 Tibia fractures with segmental defects and those with established nonunions require secondary procedures and represent a particularly challenging subset of this injury with regards to achieving union. Myriad factors contribute to the failure of primary healing in these cases making surgical treatment of the nonunited tibia technically challenging with unpredictable results.
Failure to achieve bony union in tibia fractures results in significant economic burden to both the patient and the health care system at large.7 Patients with tibia nonunions have increased pain, opioid use, and greater difficulty returning to work when compared with patients with healed tibia fractures.8 The ability to predict which fractures are most likely to require revision surgery to promote bone healing is desirable to help set patient expectations and guide decision making.4,9,10 Recent work by O'Halloran et al published a prediction model using easily defined patient and fracture variables to identify which tibia fractures, without segmental defects or established nonunion, are most likely to heal.6 However, despite the body of work trying to predict which factors predict need for surgery to promote bony healing, to our knowledge, no previous study has identified clinical risk factors associated with failure of secondary surgery designed to heal tibias with nonunions or segmental defects.
The primary aim of this study is to develop a clinically useful prediction model of success at the time of surgery to promote bone healing for established tibial nonunion or staged traumatic bone defects. Our hypothesis is that by using readily available and routinely collected data, we will be able to reliably predict a patient's risk of failure for these surgeries and potentially identify correctible risk factors to enhance success. Because our cohort includes both established nonunions and traumatic defects, a secondary aim is to determine whether traumatic bone loss represents an independent risk factor for surgical failure. We hypothesize that there will be no difference in failure rates between nonunion and traumatic gap patients.
This institutional review board approved, retrospective case–controlled study was performed at a single Level 1 trauma center in the United States. The study involved retrospective review of our prospectively collected trauma database. The database was queried to identify adult patients treated for established tibia fracture nonunion or traumatic bone defects from 2007 to 2016. For the sake of this study, this constitutes the “index surgery.”
Three hundred forty patients were identified who underwent surgery for tibia nonunions or segmental bone defects during the defined time period at our institution. Patients were excluded if they were younger than 18 years (3 patients), were managed without surgery (1 patient), sustained a pathologic fracture (1 patient), were treated with arthroplasty or amputation at the time of index surgery (5 patients), and had articular fractures of either the plafond or plateau that did not involve the diaphyseal region (28 patients). Minimum follow-up for inclusion in the study was 12 months. Patients with less than 12 months follow-up who had not demonstrated radiographic healing were excluded (99 patients). We did not exclude patients based on treatment modality and included patients with open reduction internal fixation, intramedullary nail fixation, and multiplanar external fixators.
Our cohort consisted of 203 patients who underwent surgery for nonunion (n = 143) or traumatic bone gaps (n = 60). There were 61 patients treated with open reduction internal fixation plate and screw constructs, 86 patients treated with intramedullary nails, and 56 patients treated with multiplanar external fixators. All 203 patients were included in the final chart review and statistical analysis (see Table 1, Supplemental Digital Content 1, http://links.lww.com/JOT/A123). A full chart review was performed by authors not involved in the clinical care of these patients.
The primary outcome measure was failure of the “index surgery” to promote bone healing that was defined as unplanned revision surgery for lack of bone healing or deep infection. No patients who had a primary outcome event were excluded.
Literature review yielded 25 potential independent variables thought to be associated with delayed healing of fractures and were therefore investigated (see Table 2, Supplemental Digital Content 2, http://links.lww.com/JOT/A124).4–6,11–13,18–19,23,26–27,30,32,35–50 There were 4 variables for which a primary literature source could not be identified. These included race, hepatitis C, insurance status, and type of tibia surgery (nonunion or traumatic bone gap). Although it has not been shown to be associated with delayed bone healing, race has been identified as a predictive factor in many medical studies.11–13 Hepatitis C has not been directly linked to bone healing but, because similar chronic diseases have, we included this variable in our review. Insurance status has not previously been linked with bone healing but is associated with other orthopaedic trauma outcomes and was therefore included in the analysis.14 Lastly, the secondary study aim of our study was to see if traumatic defects represented an independent risk factor for failure when examined against established tibia nonunions, and we therefore included the indication for the procedure, tibia nonunion, or traumatic bone defect, as a variable.
Radiographs of all study patients were reviewed by an independent evaluator blinded to outcome to assess preoperative mean cortical gap size and postoperative healing. Radiographs were reviewed using IMPAX 6.5 Client software (Agfa-Gevaert NV, Mortsel, Belgium). Mean cortical gap size was measured as described by Haines et al.15 Using this method, cortical defect size was defined as the mean cortical gap distance on the anterior, posterior, medial, and lateral tibia borders using the plain radiographs immediately preceding the index surgery. Radiographic healing after the surgery to promote bone healing was judged in the method described by Whelan et al.16 If the fracture had a radiographic union score for tibia fractures (RUST) of 8 or higher, it was considered healed and as stated above, these patients were included in the study despite less than 12 months follow-up. There is no current gold standard RUST score equating to fracture healing. The cutoff RUST score of 8 was chosen, as this would indicate bridging callus on all 4 cortices.
Chi-square tests were used to assess differences in distribution of categorical covariates by revision surgery status, and Student t tests were used to assess differences in mean by revision surgery status for continuous variables.
The predictive model was developed in 3 stages. The initial model included measures from the preinjury, initial hospitalization, and follow-up stages before nonunion surgery. A multiple variable logistic regression model was developed with this covariate set using backward stepwise regression—an approach in which all variables are included in a first model and then the least significant variables are removed until no covariates with P values greater than 0.2 remain in the model. The model was then expanded, using the same approach, to include measures collected at the time of surgery and expanded a second time to include measures collected during and following surgery. Finally, backward stepwise regression was applied to all measures in a single model to confirm that no additional measures should be included in the fully adjusted model.
Risk scores were calculated based on the rounded odds ratios from the prediction model. This score included 0, 17, 28, and 22 points for fall, high-energy blunt, ballistic/gun shot wound (GSW), and industrial/other injury mechanisms, respectively, with an additional 1 point for every 10 units of flap size, cortical gap distance, and body mass index (BMI), as well as 2 points for Medicaid or no insurance and 3 points for Medicare.
The first surgery to promote bone healing resulted in successful union in 68.0% of the cases (137/203), providing an adequate distribution of control and experimental patients. Results of the χ2 and Student t test for categorical and continuous variables are presented in Supplemental Digital Content 3 (see Table 3, http://links.lww.com/JOT/A125). Of note, the rate of major revision surgery was not significantly different when patients who underwent treatment of traumatic segmental defects were compared with those with fixation of established tibia fracture nonunion (P = 0.41), so these 2 groups were combined for our primary analysis (see Table 3, Supplemental Digital Content 3, http://links.lww.com/JOT/A125).
Multivariate logistic modeling identified 5 significant (P < 0.05) risk factors for failure of the surgery to promote bone healing: (1) Mechanism of injury (MOI), (2) increasing BMI, (3) cortical defect size (mm), (4) flap size (cm2), and (5) insurance status. Using the 5 significantly associated risk factors, a prediction scoring model was created. Within this prediction model, MOI was afforded the highest point totals: 0 points for fall, 17 points for high-energy blunt trauma (OR = 17; 95% CI, 1–286, P = 0.05), 22 points for industrial/other (OR = 22; 95% CI, 1–4, P = 0.04), and 28 points for ballistic injuries (OR = 28; 95% CI, 1–605, P = 0.04). One point is given for every 10 cm2 of flap size (OR = 1; 95% CI, 1–1.1, P < 0.001), 10 mm of mean cortical gap distance (OR = 1; 95% CI, 1–2, P = 0.004), and 10 units BMI, respectively (OR = 1.5; 95% CI, 1–3, P = 0.16). Two points are awarded for Medicaid or no insurance (OR = 2; 95% CI, 1–5, P = 0.035) and 3 points for Medicare (OR = 3; 95% CI, 1–9, P = 0.033) (see Table 4, Supplemental Digital Content 4, http://links.lww.com/JOT/A126).
When the risk score was regressed on the binary revision surgery measure, each 1-point increase in risk score was associated with a 6% increased chance of requiring at least 1 revision surgery (P < 0.001) with an Area Under the Curve (AUC) of 0.77 indicating reasonable performance of the prediction model (see Table 5, Supplemental Digital Content 5, http://links.lww.com/JOT/A127).
Chi-square testing indicated that the presence of postoperative infection after the surgery to promote bone healing was significantly associated with major revision surgery when compared with patients with no infection, established preoperative infection treated before index surgery, or patients with unanticipated positive intraoperative cultures (P < 0.001). Patients with unanticipated positive cultures or established preoperative infection did not have a significantly increased major revision rate compared with patients without infection. An association between postoperative infection and major revision surgery would be intuitively expected. Furthermore, by definition, postoperative infection can only be known to the patient and provider at some point after the index operation. The purpose of the model is to use preoperative information to predict failure risk before embarking on surgical intervention. For these reasons, we did not include postoperative infection in the final prediction model.
In this study, we were successful at creating a time zero model to predict healing in patients who underwent surgery for segmental defects and tibia nonunions. The prediction model uses 5 clinically relevant and readily available variables. Surgery for tibia nonunions and traumatic segmental defects is a sizeable undertaking for both clinicians and patients. Our prediction scoring system may provide a tool that surgeons can use to counsel patients based on personalized, quantifiable information to help frame treatment decisions.
In addition, our findings supported the hypothesis that treatment scenario, either traumatic bone loss or established nonunion, would not represent a predictor for failure of surgery to promote bone healing. Both groups were included in the model because both are unlikely to unite without intervention. Because patient type was not a predictor of success or failure, the factors influencing the success or failure of treatment should be similar for both groups. Despite different time courses and initial fracture characteristics, similar techniques are used to promote bone healing in both of these scenarios.17 Gap size was, however, a significant determinant of surgical failure with a 6% increase in revision probability for every 10-mm increase in gap distance.
In any prediction model, patient modifiable factors are of particular interest to clinicians. In this study, increased BMI was significantly associated with revision surgery. For every 10-point increase in BMI, there is a 6% increase in risk of surgical failure. This finding has been supported in other union prediction models and represents a potential avenue for outcome improvement.18,19 Preoperative BMI reduction has become a quality control mainstay in primary joint arthroplasty, resulting in reduced infection and revision rates,20–22 and the findings of our study could potentially be applied to this challenging clinical problem.
Cigarette smoking is a modifiable risk factor commonly associated with tibia nonunion with a majority of studies noting decreased healing rates in smokers.23–25 In a prospective cohort analysis, Schmitz et al noted a significant increase in time to tibia fracture union when patients smoked greater than 5 cigarettes per day. However, the exact relationship between smoking and bone healing remains unclear with several large series suggesting no association.26,27 Somewhat surprisingly, in our study, the rate of revision surgery was not significantly different between smokers (34%) and nonsmokers (31%) (P = 0.69). It is possible that poor outcomes attributed to smoking in other studies are due to confounding factors such as comorbid conditions or socioeconomic status that have not been rigorously accounted for. Insurance provider, for example, was highly correlated with revision rate in our study. Our study suggests that smoking status may not greatly influence healing rates after tibia nonunions and segmental defects. Therefore, the commonly applied policy of requiring patients to quit smoking before nonunion surgery may not be appropriate.
Multiple authors have advocated the use of bone graft in surgery to promote bone healing; however, in this study, we demonstrated a clinical trend toward lower failure rates in patients who underwent compression technique compared with patients who received bone graft (19% vs. 35%, P < 0.1).28–31 This difference is likely because of an unequal distribution toward bone grafting in our cohort (171 patients) and an inherent bias toward compression technique with smaller cortical gaps and simpler patterns. There was no significant difference in revision rate when different autograft sites (iliac crest, long-bone metaphyseal, and diaphyseal/RIA) were compared with allograft or no graft; however, the overall revision rate was lower if the patient received autograft (39% vs. 73%). Flierl et al also reported significantly lower revision rates in 182 patients with long-bone nonunions with autograft (17.1%) than allograft alone (47.4%) and combined allograft/autograft (25%) (P < 0.001).32 Although our results do not support or refute the autograft “gold standard,” the trend toward lower revision with autograft indicates that cases requiring bone graft may be better served with this modality rather than allograft.
Bone morphogenetic protein (BMP) is often used to treat difficult tibia fractures but, in this study, the use of BMP did not significantly impact the revision rate after surgery to promote bone healing (P = 0.769). There is no conclusive evidence to support BMP use in all patients.33 In a prospective, randomized controlled trial, there was no significant difference in healing rates when patients received either recombinant human BMP-7 or fresh autograft for tibial nonunion (81% vs. 85%, P = 0.524).33 Furthermore, the cost to a hospital system is greater when treatment includes BMP rather than autograft.34 BMP may provide a benefit when used to promote tibia union, but better evidence is needed to justify the significant cost incurred with routine use.
Infection is another commonly cited cause of the failure of surgery to promote bone healing and is closely associated with fracture nonunion.35,36 Long bones with an infected nonunion requiring bone resection greater than 4 cm can be especially difficult to heal, even with autografting the defect.35 As noted in this study, postoperative infection was significantly associated with need for major revision surgery to achieve union (P < 0.001). However, because our prediction model is designed to be used at the time of index surgery to promote bone healing, the postoperative infection variable is of little help to clinicians and patients. Furthermore, postoperative infection intuitively leads to failure requiring revision. For these reasons, postoperative infection was left out of the final prediction model.
Even in the absence of clinical symptoms or elevated inflammatory markers, cultures of the nonunion site can be positive for bacteria.36 Olszewski et al reported a 20% “surprise” positive culture rate at the time of definitive nonunion surgery.37 Twelve of 91 patients with surprise positive cultures failed to achieve union after nonunion surgery.37 In this study, we observed a 27.5% unanticipated positive culture rate. Although surprise positive cultures did not meet criteria for prediction model inclusion, our rate of revision surgery in this group was 32%, similar to that of the preoperative infection group (34%) and higher, although not statistically significant, than the negative culture group (24%). This finding supports the findings of Olszewski et al and suggests that surprise positive cultures warrant attention and appropriate treatment.
The study is subject to the limitations inherent to all retrospective study designs. We excluded 99 patients who had less than 12 months of postoperative follow-up. Complete data on these patients may have altered the variables included in the final prediction model. The study population was predominantly young (mean age 41 years), and all patients were treated at a single Level 1 trauma center. This could limit the generalizability to other populations. Grouping nonunions and segmental defects makes our study population inherently heterogeneous. As mentioned previously, the treatment modalities used for both patient groups are similar, and our study is designed to identify risk factors that can predict surgical failure. By including the patient subsets as variables within our model, we mitigated some of the bias introduced by a heterogeneous study population.
This study introduces a clinical score to predict the likelihood of success after surgery for tibia fracture nonunions or traumatic bone defects and may help clinicians better determine which patients are likely to fail these procedures and require further surgery. It may provide an opportunity to address modifiable risk factors such as BMI before undertaking surgical treatment. The data support our hypothesis that when compared with a cohort of established nonunion, surgery for traumatic tibia segmental defects is not an independent risk factor for failure of bone healing. The high rate of “surprise” positive cultures was not anticipated, and the lack of effect of smoking on outcome is also interesting and merits further investigation. This score might be used in clinical practice to help guide surgical decisions and help set appropriate patient expectations.
1. Court-Brown CM, McBirnie J. The epidemiology of tibial fractures. J Bone Joint Surg Br. 1995;77:417–421.
2. Zura R, Xiong Z, Einhorn T, et al. Epidemiology of fracture nonunion
in 18 human bones. JAMA Surg. 2016;151:e162775.
3. Sanders DW, Bhandari M, Guyatt G, et al. Critical-sized defect
in the tibia
. J Orthop Trauma. 2014;28:632–635.
4. Bhandari M, Tornetta P, Sprague S, et al. Predictors of reoperation following operative management of fractures of the tibial shaft. J Orthop Trauma. 2003;17:353–361.
5. Fong K, Truong V, Foote CJ, et al. Predictors of nonunion
and reoperation in patients with fractures of the tibia
: an observational study. BMC Musculoskelet Disord. 2013;14:103.
6. O'Halloran K, Coale M, Costales T, et al. Will my tibial fracture heal? predicting nonunion
at the time of definitive fixation based on commonly available variables. Clin Orthop Relat Res. 2016;474:1385–1395.
7. Antonova E, Le TK, Burge R, et al. Tibia
shaft fractures: costly burden of nonunions. BMC Musculoskelet Disord. 2013;14:42.
8. Tay WH, de Steiger R, Richardson M, et al. Health outcomes of delayed union and nonunion
of femoral and tibial shaft fractures. Injury. 2014;45:1653–1658.
9. Metsemakers WJ, Handojo K, Reynders P, et al. Individual risk factors for deep infection and compromised fracture healing after intramedullary nailing of tibial shaft fractures: a single centre experience of 480 patients. Injury. 2015;46:740–745.
10. Westgeest J, Weber D, Dulai SK, et al. Factors associated with development of nonunion
or delayed healing after an open long bone fracture: a prospective cohort study of 736 subjects. J Orthop Trauma. 2016;30:149–155.
11. Nejim B, Obeid T, Arhuidese I, et al. Predictors of perioperative outcomes after carotid revascularization. J Surg Res. 2016;204:267–273.
12. Flexman AM, Charest-Morin R, Stobart L, et al. Frailty and postoperative outcomes in patients undergoing surgery for degenerative spine disease. Spine J. 2016;16:1315–1323.
13. Goodman SM, Mandl LA, Parks ML, et al. Disparities in TKA outcomes: census tract data show interactions between race and poverty. Clin Orthop Relat Res. 2016;474:1986–1995.
14. Schoenfeld AJ, Belmont PJ, See AA, et al. Patient demographics, insurance status, race, and ethnicity as predictors of morbidity and mortality after spine trauma: a study using the National Trauma Data Bank. Spine J. 2013;13:1766–1773.e1.
15. Haines NM, Lack WD, Seymour RB, et al. Defining the lower limit of a “critical bone defect
” in open diaphyseal tibial fractures. J Orthop Trauma. 2016;30:e158–e163.
16. Whelan DB, Bhandari M, Stephen D, et al. Development of the radiographic union score for tibial fractures for the assessment of tibial fracture healing after intramedullary fixation. J Trauma Inj Infect Crit Care. 2010;68:629–632.
17. Giannoudis PV, Harwood PJ, Tosounidis T, et al. Restoration of long bone defects treated with the induced membrane technique: protocol and outcomes. Injury. 2016;47:S53–S61.
18. Claessen FMAP, Braun Y, Peters RM, et al. Plate and screw fixation of bicolumnar distal humerus fractures: factors associated with loosening or breakage of implants or nonunion
. J Hand Surg Am. 2015;40:2045–2051.e2.
19. Rodriguez EK, Boulton C, Weaver MJ, et al. Predictive factors of distal femoral fracture nonunion
after lateral locked plating: a retrospective multicenter case-control study of 283 fractures. Injury. 2014;45:554–559.
20. Wagner ER, Kamath AF, Fruth KM, et al. Effect of body mass index on complications and reoperations after total hip arthroplasty. J Bone Joint Surg Am. 2016;98:169–179.
21. Electricwala AJ, Jethanandani RG, Narkbunnam R, et al. Elevated body mass index is associated with early total knee revision for infection. J Arthroplasty. 2017;32:252–255.
22. Werner BC, Higgins MD, Pehlivan HC, et al. Super obesity is an independent risk factor for complications after primary total hip arthroplasty. J Arthroplasty. 2017;32:402–406.
23. Scolaro JA, Schenker ML, Yannascoli S, et al. Cigarette smoking increases complications following fracture: a systematic review. J Bone Joint Surg Am. 2014;96:674–681.
24. Patel RA, Wilson RF, Patel PA, et al. The effect of smoking on bone healing: a systematic review. Bone Joint Res. 2013;2:102–111.
25. Beyth S, Mosheiff R, Safran O, et al. Cigarette smoking is associated with a lower concentration of CD105(+) bone marrow progenitor cells. Bone Marrow Res. 2015;2015:914–935.
26. Lack WD, Starman JS, Seymour R, et al. Any cortical bridging predicts healing of tibial shaft fractures. J Bone Joint Surg Am. 2014;96:1066–1072.
27. Schemitsch EH, Bhandari M, Guyatt G, et al. Prognostic factors for predicting outcomes after intramedullary nailing of the tibia
. J Bone Joint Surg Am. 2012;94:1786–1793.
28. Sheerin DV, Turen CH, Nascone JW. Reconstruction of distal tibia
fractures using a posterolateral approach and a blade plate. J Orthop Trauma. 2006;20:247–252.
29. Stafford PR, Norris BL. Reamer-irrigator-aspirator bone graft and bi Masquelet technique for segmental bone defect
nonunions: a review of 25 cases. Injury. 2010;41:S72–S77.
30. Masquelet AC, Fitoussi F, Begue T, et al. Reconstruction of the long bones by the induced membrane and spongy autograft. Ann Chir Plast Esthet. 2000;45:346–353.
31. Ryzewicz M, Morgan SJ, Linford E, et al. Central bone grafting for nonunion
of fractures of the tibia
: a retrospective series. J Bone Joint Surg Br. 2009;91–B:522–529.
32. Flierl MA, Smith WR, Mauffrey C, et al. Outcomes and complication rates of different bone grafting modalities in long bone fracture nonunions: a retrospective cohort study in 182 patients. J Orthop Surg Res. 2013;8:33.
33. Friedlaender GE, Perry CR, Cole JD, et al. Osteogenic protein-1 (bone morphogenetic protein-7) in the treatment of tibial nonunions. J Bone Joint Surg Am. 2001;83-A(suppl 1):S151–S158.
34. Giorgio Calori M, Capanna R, Colombo M, et al. Cost effectiveness of tibial nonunion
treatment: a comparison between rhBMP-7 and autologous bone graft in two Italian centres. Injury. 2013;44:1871–1879.
35. Jain AK, Sinha S. Infected nonunion
of the long bones. Clin Orthop Relat Res. 2005;431:57–65.
36. Mills L, Tsang J, Hopper G, et al. The multifactorial aetiology of fracture nonunion
and the importance of searching for latent infection. Bone Joint Res. 2016;5:512–519.
37. Olszewski D, Streubel PN, Stucken C, et al. Fate of patients with a “surprise” positive culture after nonunion
surgery. J Orthop Trauma. 2016;30:e19–e23.
38. Heppenstall RB, Brighton CT, Esterhai JL, et al. Prognostic factors in nonunion
of the tibia
: an evaluation of 185 cases treated with constant direct current. J Trauma. 1984;24:790–795.
39. Audigé L, Griffin D, Bhandari M, et al. Path analysis of factors for delayed healing and nonunion
in 416 operatively treated tibial shaft fractures. Clin Orthop Relat Res. 2005;438:221–232.
40. Hernandez RK, Do TP, Critchlow CW, et al. Patient-related risk factors for fracture-healing complications in the United Kingdom General Practice Research Database. Acta Orthop. 2012;83:653–660.
41. Brinker MR, Bailey DE. Fracture healing in tibia
fractures with an associated vascular injury. J Trauma. 1997;42:11–19.
42. Choudry U, Moran S, Karacor Z. Soft-tissue coverage and outcome of gustilo grade IIIB midshaft tibia
fractures: a 15-year experience. Plast Reconstr Surg. 2008;122:479–485.
43. Ollivier M, Gay AM, Cerlier A, et al. Can we achieve bone healing using the diamond concept without bone grafting for recalcitrant tibial nonunions? Injury. 2015;46:1383–1388.
44. DeLee JC, Heckman JD, Lewis AG. Partial fibulectomy for ununited fractures of the tibia
. J Bone Joint Surg Am. 1981;63:1390–1395.
45. Copuroglu C, Calori GM, Giannoudis PV. Fracture non-union: who is at risk? Injury. 2013;44:1379–1382.
46. Dodwell ER, Latorre JG, Parisini E, et al. NSAID exposure and risk of nonunion
: a meta-analysis of case-control and cohort studies. Calcif Tissue Int. 2010;87:193–202.
47. Chrastil J, Sampson C, Jones KB, et al. Postoperative opioid administration inhibits bone healing in an animal model. Clin Orthop Relat Res. 2013;471:4076–4081.
48. Gaston MS, Simpson AHRW. Inhibition of fracture healing. J Bone Joint Surg Br. 2007;89:1553–1560.
49. Moghaddam A, Zimmermann G, Hammer K, et al. Cigarette smoking influences the clinical and occupational outcome of patients with tibial shaft fractures. Injury. 2011;42:1435–1442.
50. Gardner ROE, Bates JH, Ng'oma E, et al. Fracture union following internal fixation in the HIV population. Injury. 2013;44:830–833.