Updated Meta-Analysis of Randomized Controlled Trials Comparing External Fixation to Intramedullary Nailing in the Treatment of Open Tibial Fractures

Background: The purpose of this study was to collect all available randomized controlled trials (RCT) on the treatment of open tibial fractures with an external fixator (EF) and intramedullary nailing (IMN) for meta-analysis to provide reliable evidence-based data for clinical decision-making. Material and methods: The systematic review was undertaken in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and AMSTAR (Assessing the Methodological Quality of Systematic Review). An electronic search of PubMed, Cochrane Library, and Web of Science was performed until 1 March 2023 to identify RCTs which compared either IMN or EF to fix the open tibial fracture. Outcome measures were: postoperative superficial and deep infection, time to union, delayed union, malunion, nonunion and hardware failure. In addition, pain and health-related quality of life were evaluated after 3 and 12 months of follow-up. Results: Sixteen publications comprising 1011 patients were included in the meta-analysis. The pooled results suggested that the IMN technique had a lower postoperative superficial infection and malunion rate (RR = 3.56, 95%CI = 2.56–4.95 and RR = 1.96, 95%CI = 1.12–3.44, respectively), but higher hardware failure occurrence in contrast to EF (RR = 0.30; 95%CI = 0.13–0.69). No significant differences were found in the union time, delayed union or nonunion rate, and postoperative deep infection rate between the treatments. Lower levels of pain were found in the EF group (RR = 0.05, 95%CI = 0.02–0.17, p < 0.001). A difference in quality of life favoring IMN after 3 months was found (RR = −0.04, 95%CI = −0.05–0.03, p < 0.001), however, no statistical difference was found after 12 months (RR = 0.03, 95%CI = −0.05–0.11, p = 0.44). Conclusions: Meta-analysis presented reduced incidence rates of superficial infection, malunion, and health-related quality of life 3 months after treatment in IMN. However, EF led to a significant reduction in pain and incidence rate of hardware failure. Postoperative deep infection, delayed union, nonunion and health-related quality of life 12 months following therapy were similar between groups. More high-quality RCTs should be conducted to provide reliable evidence-based data for clinical decision-making.


Introduction
The most common type of open fracture of the long bones in the extremities is the open tibial fracture, which is frequently observed in traffic accidents [1,2]. For patients suffering from this type of fracture, emergency wound debridement, exploration of vascular and nerve damage, early soft tissue coverage, and fracture stabilization are agreed-upon clinical trials, and one clinical trial failed. The study selection process using the PRISMA flow diagram is shown in Figure 1.

Characteristics of Eligible Studies
Characteristics of all 16 publications included in the meta-analysis are presented in detail in Table 1. The studies were published between 1994 and 2022, with a minimum sample size of 29 [23] and a maximum of 221 [24]. Five eligible studies were conducted from Asian countries, three from African countries, four from European countries, three from the USA, and one eligible study was conducted from South America ( Figure 2). The average age varied from 25 years to 46 years, and studies comprised a total of 1011 patients (811 male and 200 female). The average duration of follow-up ranged from 4.5 to 46.5 months, and in eligible studies fracture types ranged from I to IIIb according to the Gustilo-Anderson classification.

Characteristics of Eligible Studies
Characteristics of all 16 publications included in the meta-analysis are presented in detail in Table 1. The studies were published between 1994 and 2022, with a minimum sample size of 29 [23] and a maximum of 221 [24]. Five eligible studies were conducted from Asian countries, three from African countries, four from European countries, three from the USA, and one eligible study was conducted from South America ( Figure 2). The average age varied from 25 years to 46 years, and studies comprised a total of 1011 patients (811 male and 200 female). The average duration of follow-up ranged from 4.5 to 46.5 months, and in eligible studies fracture types ranged from I to IIIb according to the Gustilo-Anderson classification.

Quality Assessment of the Eligible Studies
Sixteen studies which reported the randomization method [8,[23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] were assessed for the risk of bias according to the Cochrane Handbook. Six studies reported that the method of randomization in their study was based on even or odd medical record number [8,23,25,31,33,34]. In studies conducted by Kisitu et al. [27], Ramos et al. [28], Li [29], Braten et al. [32], and Rodrigues [35], sealed opaque envelopes were used for randomization; Li et al. [29] stated that the randomization was performed by computer allocation and that sequentially numbered opaque envelopes were assigned to included patients prospectively. However, Kisitu et al. [27] reported in their study that computer randomization was not logistically viable, therefore opaque envelopes were sorted in random sequence. A centralized web-based electronic randomization tool was the randomization method used in the study by Rohilla et al. [26], Haonga et al. [24], and Frihagen et al. [36]. Garg et al. [30] reported that the randomization chit box was used, whereas Esan et al. [37] randomized patients with simple random sampling using a balloting process. In the study conducted by Kisitu et al. [27], it was stated that patients and the treating staff were not blinded to their study group allocation. All studies showed a low risk of bias due to missing outcome data and the measurement of the outcome. Detailed information about the quality assessment of the eligible studies is shown in Figure 3a,b.

Quality Assessment of the Eligible Studies
Sixteen studies which reported the randomization method [8,[23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] were assessed for the risk of bias according to the Cochrane Handbook. Six studies reported that the method of randomization in their study was based on even or odd medical record number [8,23,25,31,33,34]. In studies conducted by Kisitu et al. [27], Ramos et al. [28], Li [29], Braten et al. [32], and Rodrigues [35], sealed opaque envelopes were used for randomization; Li et al. [29] stated that the randomization was performed by computer allocation and that sequentially numbered opaque envelopes were assigned to included patients prospectively. However, Kisitu et al. [27] reported in their study that computer randomization was not logistically viable, therefore opaque envelopes were sorted in random sequence. A centralized web-based electronic randomization tool was the randomization method used in the study by Rohilla et al. [26], Haonga et al. [24], and Frihagen et al. [36]. Garg et al. [30] reported that the randomization chit box was used, whereas Esan et al. [37] randomized patients with simple random sampling using a balloting process. In the study conducted by Kisitu et al. [27], it was stated that patients and the treating staff were not blinded to their study group allocation. All studies showed a low risk of bias due to missing outcome data and the measurement of the outcome. Detailed information about the quality assessment of the eligible studies is shown in Figure 3a
OR PEER REVIEW 8 of 16
x FOR PEER REVIEW 8 of 16
x FOR PEER REVIEW 10 of 16
x FOR PEER REVIEW 10 of 16

Pain
Data on pain were provided by four studies [25,26,28,32] included in the meta-analysis. Due to low heterogeneity (I 2 = 28%) among included studies, fixed-effect analysis was performed, and the results showed that statistical difference was found in the presence of pain between EF and IMN groups, favoring EF (RR = 0.05, 95%CI = 0.02-0.17, p < 0.001) ( Figure 11).

Health-Related Quality of Life Measured after 3 Months
Three studies [24,27,28] with a total of 334 cases (EF = 165, IMN = 169) provided data about the quality of life related to health. The fixed-effects model was adopted due to no heterogeneity among studies (I 2 = 0%). The results of the meta-analysis showed a significant difference in quality of life between EF and IMN groups after 3 months of procedure, favoring IMN (RR = −0.04, 95%CI = −0.05-0.03, p < 0.001) ( Figure 12).

Health-Related Quality of Life Measured after 12 Months
Data on health-related quality of life measured after 12 months were provided by three studies [24,27,28] included in the meta-analysis. Due to high heterogeneity (I 2 = 89%) Figure 10. Forest plot presenting the comparison of hardware failure occurrence between EF and IMN; Blue squares-the individual study effect; Diamond-pooled effect [8,23,27,29,32,33].

Pain
Data on pain were provided by four studies [25,26,28,32] included in the meta-analysis. Due to low heterogeneity (I 2 = 28%) among included studies, fixed-effect analysis was performed, and the results showed that statistical difference was found in the presence of pain between EF and IMN groups, favoring EF (RR = 0.05, 95%CI = 0.02-0.17, p < 0.001) ( Figure 11).

Pain
Data on pain were provided by four studies [25,26,28,32] included in the meta-analysis. Due to low heterogeneity (I 2 = 28%) among included studies, fixed-effect analysis was performed, and the results showed that statistical difference was found in the presence of pain between EF and IMN groups, favoring EF (RR = 0.05, 95%CI = 0.02-0.17, p < 0.001) ( Figure 11).

Health-Related Quality of Life Measured after 3 Months
Three studies [24,27,28] with a total of 334 cases (EF = 165, IMN = 169) provided data about the quality of life related to health. The fixed-effects model was adopted due to no heterogeneity among studies (I 2 = 0%). The results of the meta-analysis showed a significant difference in quality of life between EF and IMN groups after 3 months of procedure, favoring IMN (RR = −0.04, 95%CI = −0.05-0.03, p < 0.001) (Figure 12).

Health-Related Quality of Life Measured after 12 Months
Data on health-related quality of life measured after 12 months were provided by three studies [24,27,28] included in the meta-analysis. Due to high heterogeneity (I 2 = 89%) among included studies, random-effect analysis was performed, and the results showed that no statistical difference was found in the presence of quality of life measured after 12

Health-Related Quality of Life Measured after 3 Months
Three studies [24,27,28] with a total of 334 cases (EF = 165, IMN = 169) provided data about the quality of life related to health. The fixed-effects model was adopted due to no heterogeneity among studies (I 2 = 0%). The results of the meta-analysis showed a significant difference in quality of life between EF and IMN groups after 3 months of procedure, favoring IMN (RR = −0.04, 95%CI = −0.05-0.03, p < 0.001) (Figure 12).

Pain
Data on pain were provided by four studies [25,26,28,32] included in the meta-analysis. Due to low heterogeneity (I 2 = 28%) among included studies, fixed-effect analysis was performed, and the results showed that statistical difference was found in the presence of pain between EF and IMN groups, favoring EF (RR = 0.05, 95%CI = 0.02-0.17, p < 0.001) ( Figure 11).

Health-Related Quality of Life Measured after 3 Months
Three studies [24,27,28] with a total of 334 cases (EF = 165, IMN = 169) provided data about the quality of life related to health. The fixed-effects model was adopted due to no heterogeneity among studies (I 2 = 0%). The results of the meta-analysis showed a significant difference in quality of life between EF and IMN groups after 3 months of procedure, favoring IMN (RR = −0.04, 95%CI = −0.05-0.03, p < 0.001) (Figure 12).

Health-Related Quality of Life Measured after 12 Months
Data on health-related quality of life measured after 12 months were provided by three studies [24,27,28] included in the meta-analysis. Due to high heterogeneity (I 2 = 89%) among included studies, random-effect analysis was performed, and the results showed that no statistical difference was found in the presence of quality of life measured after 12 months between EF and IMN groups (RR = 0.03, 95%CI = −0.05-0.11, p = 0.44) (Figure 13).

Health-Related Quality of Life Measured after 12 Months
Data on health-related quality of life measured after 12 months were provided by three studies [24,27,28] included in the meta-analysis. Due to high heterogeneity (I 2 = 89%) among included studies, random-effect analysis was performed, and the results showed that no statistical difference was found in the presence of quality of life measured after 12 months between EF and IMN groups (RR = 0.03, 95%CI = −0.05-0.11, p = 0.44) ( Figure 13).

Discussion
The findings of this meta-analysis showed that EF was superior in the treatment of open tibia fractures as it was associated with lower rates of hardware failure and pain, but it was inferior in terms of higher rates of superficial infection and malunion, as well as lower health-related quality of life 3 months after treatment. Furthermore, our findings demonstrated no significant difference in deep infection, delayed union and nonunion rates, time to union, and health-related quality of life 12 months following therapy between EF and IMN groups.
The debate over the best surgical treatment technique for open tibia fractures has still been challenging and a subject of discussion for a long time. In 2015. Foote et al. [6] published a network meta-analysis with the objective to compare the relative effects of various treatment options on the risk of requiring unscheduled revision surgery after open fractures of the tibial diaphysis. Independent of the Gustilo classification of the fracture, Foote et al. [6] discovered that the unreamed nail fixation was associated with a lower risk of reoperation in contrast to EF. Results of the meta-analysis conducted by Foote et al. [6] further validated current knowledge that EF of open fractures was associated with increased infection rates [38], many of which may require reoperation. As secondary study endpoints were to quantify differences in malunion, deep infection, and superficial infection in unreamed nail fixation and EF, Foote et al. [6] were unable to identify any significant differences between treatment options due to the small number of reported events. Not long after, in 2016. a research paper reflecting the current status of both IMN and EF for treating Gustilo grade IIIB open tibial fractures was published [15]. This study s metaanalyses demonstrated similar results to Foote et al. [6], moreover that unreamed IMN was superior in the treatment of Gustilo IIIB tibial fractures, with shorter time to union and decreased incidence of superficial infection and malunion, while without increasing the risk of delayed union, non-union, deep infection, and fixation failure compared to EF.
As meta-analyses of RCTs are regarded to provide the strongest evidence for clinical interventions, as compared to observational studies and solitary randomized trials, it is of immense importance to mention Fu et al. s [13] 2018. meta-analysis of RCT, aimed to compare EF with tibial nailing treatment modality of open tibial fractures. Six RCTs with a total of 407 cases were included in the Fu et al. s [13] meta-analysis and their results suggest that nailing treatment is superior to EF in preventing post-fixation complications, such as superficial infection and malunion. Consistent with previously published re-

Discussion
The findings of this meta-analysis showed that EF was superior in the treatment of open tibia fractures as it was associated with lower rates of hardware failure and pain, but it was inferior in terms of higher rates of superficial infection and malunion, as well as lower health-related quality of life 3 months after treatment. Furthermore, our findings demonstrated no significant difference in deep infection, delayed union and nonunion rates, time to union, and health-related quality of life 12 months following therapy between EF and IMN groups.
The debate over the best surgical treatment technique for open tibia fractures has still been challenging and a subject of discussion for a long time. In 2015. Foote et al. [6] published a network meta-analysis with the objective to compare the relative effects of various treatment options on the risk of requiring unscheduled revision surgery after open fractures of the tibial diaphysis. Independent of the Gustilo classification of the fracture, Foote et al. [6] discovered that the unreamed nail fixation was associated with a lower risk of reoperation in contrast to EF. Results of the meta-analysis conducted by Foote et al. [6] further validated current knowledge that EF of open fractures was associated with increased infection rates [38], many of which may require reoperation. As secondary study endpoints were to quantify differences in malunion, deep infection, and superficial infection in unreamed nail fixation and EF, Foote et al. [6] were unable to identify any significant differences between treatment options due to the small number of reported events. Not long after, in 2016. a research paper reflecting the current status of both IMN and EF for treating Gustilo grade IIIB open tibial fractures was published [15]. This study's meta-analyses demonstrated similar results to Foote et al. [6], moreover that unreamed IMN was superior in the treatment of Gustilo IIIB tibial fractures, with shorter time to union and decreased incidence of superficial infection and malunion, while without increasing the risk of delayed union, non-union, deep infection, and fixation failure compared to EF.
As meta-analyses of RCTs are regarded to provide the strongest evidence for clinical interventions, as compared to observational studies and solitary randomized trials, it is of immense importance to mention Fu et al.'s [13] 2018. meta-analysis of RCT, aimed to compare EF with tibial nailing treatment modality of open tibial fractures. Six RCTs with a total of 407 cases were included in the Fu et al.'s [13] meta-analysis and their results suggest that nailing treatment is superior to EF in preventing post-fixation complications, such as superficial infection and malunion. Consistent with previously published research, their meta-analysis revealed a higher incidence of superficial infections among EF patients, however, Fu et al. [13] pointed out that with effective debridement, stringent wound management, and prudent antibiotic use the risk of superficial infections among these patients can be reduced to a manageable level. In contrast to superficial infection, deep infections and osteomyelitis are more concerning complications of open tibia fractures treatment. Due to exposure of the marrow cavity, the risk of its further contamination and even amputation is increased in patients treated by IMN. However, in EF patients, osteomyelitis is more often presented after early infection and massive soft-tissue defects and is rarely developed from the pin tract infection. Antibiotic-coated nails have recently been introduced to address this issue since they deliver a high concentration of local antibiotic elution in addition to providing stability at the fracture site. A recent study carried out by Rohilla and colleagues [26] found that both the antibiotic-coated tibial interlocking nail and ring fixator obtained equivalent rates of union and similar complications. The findings of this study showed that, although ring fixation is a well-established alternative for treating open tibial fractures, an antibiotic-coated intramedullary nail may also be a viable choice.
EF patients, on the other hand, had a much lower incidence of hardware failure compared to patients in whom nailing was the treatment modality. With a frequency of up to 3-16%, hardware failure remains the most often reported nailing complication across studies, due to the breakage of locking screws [13]. A total of 22 cases of nailing hardware failure were reported in the study by Fu et al. [13], compared to the six EF cases. Additionally, results of the meta-analysis conducted by Fu et al. [13] showed similar treatment effects for postoperative deep infection, delayed union, and nonunion between groups. The pooled results of the most recent meta-analysis including 9 RCTs with 733 cases [16], suggested that the IMN technique had a lower postoperative superficial infection and malunion rate (RR = 2.84, 95%CI = 1.83-4.39 and RR = 3.05, 95%CI = 2.06-4.52, respectively), but higher hardware failure occurrence in contrast to EF (RR = 0.38; 95%CI = 0.17-0.83). No significant differences were found in the union time, delayed union or nonunion rate, and postoperative deep infection rate between the treatments. These findings are in accordance with the results of our meta-analyses based on 16 RCTs and 1011 cases, where patients treated with IMN had higher rates of hardware failure, however lower rates of complications such as superficial infection and malunion. It is worth mentioning that although several RCTs compared health-related quality of life and pain between EF and IMN patients, no meta-analysis has been performed in this regard. In 2014, a randomized prospective study comparing the Ilizarov circular fixator and locked IMN methods in patients with tibia shaft fractures was published [28]. Results of Ramos et al. [28] demonstrated that despite the fact that the Ilizarov group had more open fractures, the absolute number of severe complications was greater in the IMN group. With respect to assessing major complications between groups, such as compartment syndrome, deep infection, hardware failure, delayed union, pseudarthrosis, and malunion, several self-report measures of pain and functionality were included in this study additionally. There was a statistically significant difference between the Ilizarov and IMN groups of patients on both pain and satisfaction at 1 year, showing that patients in the Ilizarov group scored better than patients in the IMN group. Ramos et al. [28] concluded that diaphyseal tibial fractures may be effectively treated utilizing the Ilizarov method, a minimally invasive procedure that allows for immediate weight-bearing, tend to decrease anterior knee discomfort and benefit patients with leaving no implant behind.
In addition to assessing the composite primary event of reoperation or mortality for deep infection, nonunion, or malalignment at 1 year, Haonga et al. [24] assessed the quality of life between uniplanar EF and IMN patients. The results of this RCT conducted at a tertiary orthopedic center in Tanzania showed significant early differences in quality of life in favor of IMN, however, these differences did not persist at 12 months. Haonga et al. [24] stated that given the inconveniences of an external fixator, the improved early quality of life following intramedullary nailing is not unexpected. Despite variations in radiographic healing and ultimate alignment, quality of life in the treatment groups equilibrated between 6 and 12 weeks, the time during which external fixators were removed. Our meta-analysis, to our knowledge, is the first to evaluate both pain and health-related quality of life, as important outcome measures after open tibia fracture treatment. Pooled results of our metaanalysis revealed a significant difference in quality of life 3 months after the procedure between the EF and IMN groups, favoring IMN. However, the assessment of quality of life evaluated 12 months following interventions presented findings comparable to Haonga et al. [24]. Given that only four studies assessing pain and three studies assessing health-related quality of life were eligible for meta-analysis, more high-quality RCTs should be conducted to provide reliable evidence-based data for clinical decision-making.
There are several limitations of this meta-analysis. The first limitation of this study was that out of the 16 studies included, 7 studies showed a high overall risk of bias. Next, six studies had the method of randomization based on even or odd medical record numbers, not meeting the strict randomization criteria. Furthermore, only 3 studies provided data of pain and health-related quality of life, therefore more RCTs are needed to generate more convincing evidence.

Conclusions
This meta-analysis presented reduced incidence rates of superficial infection, malunion, and health-related quality of life 3 months after treatment in IMN. However, EF led to a significant reduction in pain and incidence rate of hardware failure. Postoperative deep infection, delayed union, nonunion and health-related quality of life 12 months following therapy were similar between groups. More high-quality RCTs should be conducted to provide reliable evidence-based data for clinical decision-making.