Maxillofacial defects that result from congenital abnormalities, traumatic injuries, or the surgical removal of tumors pose significant challenges for both patients and healthcare providers. These defects disrupt essential functions such as mastication, speech, and swallowing, and they also significantly affect patients’ psychosocial well-being and overall quality of life^1,2. In recent years, various reconstructive techniques have emerged to address these challenges, and microvascular fibula flap (FFF) reconstruction stands out for its potential for successful osseointegration³.

Despite significant advances in reconstructive techniques, the intricacy of FFF reconstruction procedures and patient-specific factors can occasionally result in failures⁴. A critical component of these procedures is the anastomosis, or surgical connection, of separate or severed vessels, which can be categorized broadly into venous and arterial anastomosis, each with its own unique set of challenges and considerations⁵.

Venous anastomosis, which connects the vein of the free flap to a vein in the recipient site, is pivotal in ensuring the success of the procedure. As an essential aspect of postoperative flap survival, venous anastomosis allows the outflow of blood from the flap, preventing a buildup of blood that could lead to venous congestion. Failures primarily occur due to flap congestion, often secondary to venous thrombosis. Thrombosis is a significant concern because it can lead to vessel blockage, thereby compromising the viability of the flap. This problem is often exacerbated by patient-specific factors such as underlying coagulation disorders or systemic conditions that affect vascular health⁶.

In contrast, arterial anastomosis involves connecting the artery in the free flap to an artery in the recipient site, providing an inflow of blood to the flap. Although arterial thrombosis is less common than venous thrombosis, it remains a serious potential complication. Failures in arterial anastomosis can lead to insufficient perfusion to the flap, potentially resulting in tissue necrosis. This situation is often influenced by the caliber and condition of the recipient vessels, so careful selection and preparation of the recipient site is necessary⁷.

Beyond venous and arterial anastomosis, other factors can contribute to the failure of FFF reconstruction. For example, postoperative hematoma can lead to vascular compression, which can in turn compromise blood flow and lead to flap failure. Similarly, wound dehiscence, the separation of the wound edges, can put tension on the anastomotic site, possibly leading to vessel thrombosis or tearing^8,9.

Various strategies and methods have been proposed to mitigate these complications, such as the use of venous coupler devices instead of hand-sewn sutures. These devices, which were introduced in the 1960s, can simplify the anastomosis process, potentially reducing the time the tissue is without a blood supply and thus lowering the risk of thrombosis⁷. However, although preliminary results from retrospective studies are promising, these interventions need further validation through robust, prospective research to ensure their effectiveness and safety^10,11. In that context, our systematic review and meta-analysis evaluates the failure rate and identifies factors that affect anastomosis in FFF reconstruction. We critically analyze the available literature, including randomized controlled trials (RCTs), cohort studies, and case series, to gather comprehensive evidence about the vascular complications and determinants of success in FFF reconstruction.

This systematic review and meta-analysis was meticulously designed and executed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines and the Cochrane Handbook for Systematic Reviews of Interventions to ensure a comprehensive and rigorous investigation¹².

1. Research question formulation

We developed well-structured research question based on the PICO framework, as follows:

• Population: Patients undergoing maxillofacial reconstruction using microvascular fibula flaps.

• Intervention: Implementation of arterial and venous anastomosis.

• Comparison: Comparison of anastomosis techniques, such as couplers, and the use of different arteries and veins for anastomosis.

• Outcome: Failure rates pertinent to vascular compromise and its underlying causes.

2. Literature search strategy

We performed an exhaustive, multi-platform literature search in the MEDLINE, Web of Science, Embase, Scopus, and Cochrane’s CENTRAL databases from the inception of each source to April 2023. We further probed the gray literature, encompassing trial registrations, conference proceedings, dissertations, and other non-peer-reviewed sources, to ensure a holistic approach. Specific search terms were tailored for each database and included “maxillofacial reconstruction” and “fibula flap.” High impact factor (>1) journals specializing in maxillofacial reconstruction were also scrutinized manually. A comprehensive tabulation of our search strategies and results is provided in Table 1.

3. Inclusion and exclusion criteria

Our stringent selection criteria targeted studies with retrospective or prospective cohort and clinical designs and human subjects that focused on graft anastomosis in microvascular fibula flap maxillofacial reconstructions. We excluded case-control studies, research involving other types of reconstruction methods, and animal-based studies.

4. Study selection process

Two independent authors (H.M. and S.S.) screened the titles and abstracts of potentially relevant articles according to the PICO framework and the predetermined inclusion and exclusion criteria. Any disagreements were resolved by consulting a third author (R.T.). A comprehensive full-text review of the selected articles ensured strict adherence to the research criteria, with rationales documented for excluded studies.

5. Data extraction and management

One researcher (R.T.) meticulously extracted pertinent data from the selected articles, and that work was cross-verified by another researcher (S.S.). We collected the study details, patient demographics, clinical presentation, anastomosis techniques, assessment methods, and outcome measures, among other data points. This information was systematically catalogued using piloted data extraction forms. A comprehensive summary of the extracted data is presented in Table 2.

6. Risk of bias assessment

To ensure the credibility of our findings, we assessed the risk of bias in the included studies. For the retrospective and prospective cohort studies, we used the Newcastle–Ottawa scale to evaluate the quality of their selection, comparability, and outcomes. For RCTs, we used the Cochrane’s risk of bias tool. The risk of bias assessment results are presented in Table 3.

7. Data synthesis and analysis

For the data to be suitable for a meta-analysis, we had to ensure the comparability of the therapeutic interventions and measured outcomes across studies. The overall failure rates caused by vascular compromise and specific failures were pooled by calculating standard errors for each study with consideration of the failure rate and the study size. Inverse-variance random-effects meta-analyses were conducted for all of the aforementioned outcomes. We found several sources of heterogeneity: differences in treatment, treated populations, study design, and data analysis methods. In the absence of heterogeneity, estimates are considered to differ only due to random sampling errors. Heterogeneity is important, and investigating sources of heterogeneity in meta-analyses involves the identification of study-level characteristics associated with differences in outcomes. Methods commonly used to do that are subgroup analyses and meta-regressions.

Heterogeneity between studies was tested using Cochrane’s Q test and quantified using I² statistics, which measure the inconsistency in pooled calculations caused by study heterogeneity. The pooled failure rates were computed using STATA 17 (Stata Corp.). We want to provide accurate and reliable information, so we take concerns about heterogeneity in the results seriously, particularly the incidence and causes of vascular insufficiency across studies. Although it is true that heterogeneity can cause variability in pooled estimates, we have made every effort to analyze and interpret the available data with care. Additionally, we relied on established scientific methods to account for heterogeneity, such as performing subgroup analyses and using random effects models.

1. Screening and selecting studies

The initial search strategy yielded 769 unique articles. Following a preliminary assessment of titles and abstracts, 738 were ruled out due to their lack of relevance, leaving 31 for further evaluation. An additional four studies were identified through our scrutiny of the gray literature, and six were found through citation searches. Those 40 full-text articles (10 found by hand and 30 through digital searches) were screened according to the inclusion and exclusion criteria. After that, 15 studies were included, and 25 studies were excluded. The process of study identification, inclusion, and exclusion, along with the reasoning behind each decision, is comprehensively illustrated in Fig. 1.

2. Descriptive analysis of studies

This systematic review comprises 15 studies: 13 retrospective cohort studies^10,11,13-23, one prospective cohort study²⁴, and one randomized clinical trial²⁵. The studies were conducted from 2007 through 2022 and collectively included 2,284 patients, 1,346 males, 719 females, and 219 whose sex was not reported. The sample sizes across studies were diverse, ranging from a minimum of nine in Gaggl et al.¹⁵ to a maximum of 601 in Assoumane et al.¹¹, with a higher prevalence of male subjects in all the studies.

The duration of the studies also varied considerably, with Abramowicz et al.¹³ reporting the longest duration of ten years. Defect locations were primarily in the head and neck regions, with a particular focus on mandibular defects, but they also extended to other regions such as the oral floor and oropharynx.

The anastomoses encompassed the facial, superior thyroid, and lingual arteries and the external jugular, common facial, and anterior jugular veins. The assessment methodologies ranged from Doppler devices to clinical observation and radiographs.

The number of grafts performed varied from study to study, with a minimum of six in Gaggl et al.¹⁵ and a maximum of 854 in Assoumane et al.¹¹. Vascular complications led to graft failures in several studies, with common causes including thrombosis, hematoma compression, dehiscence, and a twisted or kinked vascular pedicle. Some studies reported no instances of failure (Abramowicz et al.¹³, Ritschl et al.²⁰). A detailed representation of these characteristics is provided in Table 2.

3. Assessment of bias

The risk of bias in the 13 retrospective and one prospective cohort study ranged from low to high on the Newcastle–Ottawa scale (scores 5 to 9). Senthil Murugan et al.’s RCT²⁵ also had some bias-related concerns. Therefore, the interpretation of our results must take into account the diverse methodological quality of the included studies. Detailed risk of bias assessments for all studies are presented in Table 3.

4. Synthesis of data and meta-analysis

Despite the diverse set of studies, we constructed a meta-analysis to find overarching trends in vascular-related failures and discern the primary etiologies of those failures. Unfortunately, given the considerable variation in intervention strategies and outcome measures across studies, it was unfeasible to conduct a meta-analysis comparing different host veins and arteries, double versus single anastomoses, and the use of couplers versus hand suturing techniques.

To ascertain the overall rate of vascular failure, we pooled data from 11 studies that collectively account for 2,405 free fibula grafts. The pooled vascular failure rate was approximately 6% (confidence interval [CI] 0.03-0.09). However, that estimate should be interpreted with caution because the underlying heterogeneity was quite high (I²=93.5%), indicative of significant variability across the contributing studies.(Fig. 2)

To further break down the etiologies of vascular failures, we computed pooled prevalence rates for three primary causes: venous thrombosis, arterial thrombosis, and vein compression due to hematoma. This part of the meta-analysis incorporated eight studies that cumulatively represent 1,272 grafts.

The pooled prevalence of venous thrombosis was 3% (CI 0.01-0.05). Similar to the overall failure rate, this result was subject to substantial heterogeneity (I²=77.7%), reflecting considerable disparity in the rates reported across studies.(Fig. 3)

In contrast, arterial thrombosis accounted for a smaller proportion of vascular failures, with a pooled failure rate of 1% (CI 0.00-0.03). Notably, the level of heterogeneity was also lower (I²=56.5%), suggestive of less variability in the rates of arterial thrombosis among the studies.(Fig. 4)

Lastly, we calculated the pooled failure rate for vein compression due to hematoma and found that it contributed to less than 1% of the overall failures, reinforcing the relatively small role of this specific complication in graft failures.

1. Summary of results

Our meta-analysis found an overall vascular failure rate of 6% across 11 studies involving 2,405 free fibula grafts. The substantial heterogeneity among those studies (I²=93%) suggests significant variability in the reported rates of vascular failure. We also identified specific causes of vascular failure, and venous thrombosis and arterial thrombosis were the leading causes at 3% and 1%, respectively. The incidence of vein compression due to hematoma was found to be less than 1%. Venous and arterial thrombosis pose significant challenges to the successful execution of free fibular grafts in reconstructive surgeries²⁶. Venous thrombosis, often due to mechanical obstructions such as compression, kinking, or twisting, can lead to congestion of the graft, compromising its viability²⁷. Rapid correction of the mechanical obstructions generally leads to high salvage rates²⁸. However, arterial thrombosis, predominantly associated with pre-existing vascular diseases such as atherosclerosis or calcified plaques but sometimes the result of intraoperative damage or vessel size discrepancies, can lead to more severe complications. Arterial compromise is harder to identify and correct than venous issues; a swift, significant change in graft color is usually the primary indicator of a venous problem, whereas arterial problems can go unnoticed until severe ischemia occurs²⁹.

Meta-analyses to compare different host veins and arteries, double versus single anastomoses, and the use of couplers versus hand suturing techniques were not possible.

Another issue that can introduce vascular deficiencies and lead to flap failure is blood vessel kinking, which can originate from multiple sources, such as the creation of an acute angle between the fibula axis and the pedicle, undue tension in the intraoral space, mechanical compression from hardware, inappropriate neck flexion, and erroneous pedicle positioning^14,30. Each of those factors can potentially induce blood vessel kinking, thereby impairing blood flow, inducing graft congestion, and ultimately precipitating flap failure. A mitigation strategy proposed by Aksoyler et al.¹⁴ involves the strategic placement of a small piece of adipose tissue around the pedicle, which serves as both a cushion to prevent vessel kinking and a guard against twisting. Our results are consistent with the literature. It has the longest vascular flap available for mandibular reconstruction. This flap is 20 to 30 cm in length, depending on the patient’s size. In addition, this is a unique technique that can reconstruct bone defects from corner to corner, after several osteotomies have been performed to make the model³¹. Its vascular pedicle, which has a relatively stable anatomy, exhibits two vascular systems (periosteal and endosteal). This flap allows flexible molding and assembly and can be accompanied by a skin island for intraoral and extraoral reconstruction. Therefore, it supports osseointegrated implants for dental arch reconstruction and has relatively low morbidity at the donor site³². In 100% of cases in which a reconstructive plate was used, the plate was exposed to the oral cavity during the postoperative period. When small plates were used, only 4 (9.3%) of them were exposed²⁶. The microvascular fibrous flap remains the gold standard in mandibular reconstruction because of its low loss and complication rates. The advent of small hard disks has increased the benefits of using this technique because they reduce the risk of complications²⁶. Additionally, a case series and meta-analysis showed that the results of end to side venous anastomosis in the limbs were positive and that this technique was effective in treating differences in vein size; however, its superiority over end-to-end anastomosis could not be demonstrated²⁷.

2. Factors influencing the success rate

During the surgical procedure, the anastomosis technique plays a paramount role in the success of the operation. The critical junction between the donor and recipient vessels can be achieved through couplers or hand suturing. In several studies, the coupler technique demonstrated advantages over hand suturing⁷. Couplers have shown a lower incidence of thrombosis, leave no foreign material in the vascular intima that could precipitate thrombosis, are less susceptible to inaccuracy and mechanical failures than hand suturing, and can overcome discrepancies up to a ratio of 3:1, thereby providing a mechanically stronger anastomosis than hand suturing^10,23.

Nevertheless, it’s crucial to state that couplers have not unequivocally surpassed sutures in all studies. For instance, the study by Senthil Murugan et al.²⁵ indicated a marginally higher flap success rate in the coupled anastomosis group, but the difference was not statistically significant (P>0.5). Some retrospective studies have shown similar trends, with the success rates of coupled and sutured anastomoses being almost the same^10,11. Moreover, the time-saving advantage of couplers is questionable. In the studies of Senthil Murugan et al.²⁵ and Assoumane et al.¹¹, the time difference of approximately 11 and 15 minutes, respectively, between sutured and coupled anastomoses was deemed clinically insignificant due to the lack of correlation between the time taken to perform anastomosis and the flap success rate.

One significant drawback of couplers is their considerable cost, which might not always be justified by the reduction in operating time¹⁰. Only in instances in which sutured anastomoses failed and re-exploration was necessary did the additional cost of the coupler balance out³³. Thus, although couplers do possess certain advantages over conventional hand-sewn sutures, their effectiveness is heavily reliant on the surgeon’s experience and the cost-benefit analysis for each individual patient. On the other hand, couplers are cost effective in the reconstruction of long bone defects because they achieve a higher union rate than hand-sutured fibular grafts.

The diameter of the coupler used in vein anastomosis can significantly affect the success or failure of the procedure. In the studies of Senthil Murugan et al.²⁵ and Vernier-Mosca et al.²³, vein anastomosis using a coupler with a diameter of less than 2 mm introduced a high risk of venous thrombosis. For anastomosed vessels with a diameter of less than 2 mm, the recommendation is to perform the anastomosis manually, which allows the vessels to dilate. The selection of an appropriate ring size is crucial to prevent complications, such as the constriction or tearing of the vein wall or a reduction in the lumen size, which can lead to thrombosis.

Both single vein and double vein anastomosis methods have been used for free fibular grafts. Single vein anastomosis, which involves the connection of one vein from the graft to one vein at the recipient site, is often favored when there is a good match in vessel sizes and the risk of venous congestion is minimal. In contrast, double vein anastomosis, involving the connection of two veins from the graft to two veins at the recipient site, is typically used when there’s a risk of venous congestion due to size mismatch or when the venous outflow is anticipated to be high^17,34. The intent behind double vein anastomosis is to escalate outflow capacity and mitigate the risk of venous thrombosis. Nevertheless, the studies of Assouman et al.¹¹ and Han et al.¹⁷ revealed no significant difference in the vascular failure rate between free fibular grafts anastomosed using the double vein and single-vein methods. Although flap failure was comparable between flap anastomoses performed using a venous coupler and those performed using a conventional suture technique, findings revealed that the total number of venous thrombosis cases was higher in the coupler-anastomosis group. Therefore, microsurgeons should remember that these devices involve specific techniques and challenges not normally seen in hand joining. Especially at the early stages of the learning process, their application must be questioned critically and limited to clearly defined situations³⁵. Using couplers for venous anastomosis in head and neck free flap reconstruction influences surgical planning and procedures. Research results show that using a vein graft device in head and neck free flap reconstruction does not change the rate of flap necrosis. However, the use of an implant device can reduce overall procedure time³⁶.

The task of reconstructive surgery, especially in the maxillofacial region, can be significantly complicated by the limited availability of recipient blood vessels due to factors such as tumor invasion and previous surgeries. The choice of veins and arteries for anastomosis is influenced by factors such as the size of the graft blood vessel, its proximity to the excision area, and the quality of the graft vessels^37,38. In the studies we reviewed, the most commonly used graft artery for free fibular grafts was the peroneal artery, and the most common recipient arteries were the superior thyroid artery and the facial artery. In terms of veins, the peroneal vein in the graft and branches of the internal and external jugular veins in the recipient site were the most commonly used.

Despite the methodical selection of recipient vessels, several challenges persist. In particular, previously dissected and irradiated zones can restrict anastomosis possibilities. Prior exposure to radiation therapy and cervical dissections can lead to radicular fibrosis and scar tissue, which can significantly limit the options for successful anastomosis and contribute to a host of complications³⁹. In the study conducted by Li et al.¹⁰, two out of three reported cases of vein thrombosis occurred in patients who had undergone radiotherapy prior to the anastomosis procedure.

The success of an anastomosis procedure can be significantly affected by the quality of the recipient vessels. One major cause of failure, venous thrombosis, is often linked to issues such as poor quality of recipient vessels due to prior irradiation, the thinness of the recipient vessels, and a mismatch in the diameter of the vessels³⁹. The latter necessitates a high level of skill and time to prevent fluctuations in blood flow and intraluminal protrusions of the wall, which can cause thrombosis, and ensure successful anastomosis. The use of microvascular coupling can be a viable strategy to compensate for this discrepancy, prevent vasospasm, and reduce anastomosis time.

In addition to those complications, some patients present with systemic blood clotting disorders, such as thrombophilia, which can dramatically increase the risk of vein thrombosis post-anastomosis⁴⁰. Hereditary (e.g., Factor V Leiden mutation, prothrombin gene mutation) or acquired (e.g., antiphospholipid syndrome) thrombophilic conditions can predispose patients to hypercoagulability, leading to a high risk of clot formation that can occlude the anastomosis site. Similarly, conditions such as von Willebrand disease can affect clotting by causing deficiencies in clotting factors that make it challenging to achieve hemostasis after surgery and can lead to anastomotic leakage⁴¹. Therefore, a comprehensive preoperative evaluation, including a thorough review of the patient’s medical history, previous treatments, and possible coagulation disorders, is vital to the success and durability of an anastomosis procedure.

3. Pre- and postoperative evaluations of the anastomosis procedure

In the postoperative phase, the monitoring and care of anastomoses is vitally important. In the studies we reviewed, multiple methodologies were used to ensure the success and viability of the procedure, including monitoring blood flow changes based on transit-time flow measurements and the use of color Doppler ultrasound^20,42. However, flap perfusion rates can vary due to factors such as the type of flap, the donor and recipient sites, the patient’s general health condition, vascular diseases, nicotine abuse, and previous surgeries. Therefore, it’s recommended that preoperative examinations be conducted, including computed tomography or magnetic resonance angiography of the head, neck, and donor site, to ensure a comprehensive understanding of the vascular situation^43,44.

Postoperatively, the heart rate and mean arterial blood pressure should be closely monitored, but it is worth noting that these measurements do not always correlate with the flap’s arterial perfusion. Furthermore, the effects of variations in arterial perfusion caused by factors such as nicotine consumption, type II diabetes, and transplant loss need to be acknowledged⁴⁵.

Given the dynamic nature of the circulatory system, each anastomosis case presents unique challenges. The patient should keep their head elevated on two pillows whenever they lie down for two weeks to help reduce swelling. The patient can also gently apply an ice pack to the wound for 10 minutes, three or four times an hour for the first 24 hours. This will help reduce swelling, but it is not effective after 24 hours. The frequency of clinical monitoring during the early postoperative period varies between centers, but clinical flap checks are typically performed every 2 to 4 hours during the first 2 to 3 days after surgery. To date, the gold standard for FFF monitoring includes clinical examination (i.e., flap color, capillary refill, tissue tension, temperature) and handheld acoustic Doppler ultrasound. Postoperative care and monitoring should reflect a detailed understanding of flap physiology, individual patient factors, and previously determined reference values. By adopting this meticulous and comprehensive approach, a stable vascular supply, successful healing, and optimal patient outcomes following anastomosis can be ensured.

In summary, clinical follow-up remains the gold standard for postoperative free flap evaluation. Other adjuvants might be useful to surgeons and medical teams, but there is no clear consensus regarding their appropriate use. Like many postoperative treatment regimens, free flap follow-up technique is surgeon-dependent and can vary significantly between institutions.

4. Limitations and suggestions for further research

This review has several limitations. The meta-analyses in our review displayed a significant level of heterogeneity, indicating considerable variation in vascular failure rates among the studies. Therefore, the pooled estimates might not accurately represent the true prevalence due to differences among the individual studies in sample size, study methodology, or other factors. Evaluating and addressing bias is crucial for ensuring research credibility and generalizability. Although it was unfortunate to find a range of bias across the studies we found in the literature, it is a valuable discovery in itself that highlights the need for further investigation and improvements in the methodologies used in future studies. Also, due to insufficient data, we were unable to compare the effects of different host veins and arteries, double versus single anastomoses, and the use of couplers versus hand suturing techniques.

Moreover, the results might be subject to various forms of bias, particularly in the retrospective studies, where a lack of randomization and blinding could also bias the results. We did not consider patient-level factors, such as underlying health conditions, previous surgeries, the extent of tissue damage, and patient lifestyle, which can all affect the success rate of anastomosis. Our focus solely on free fibula grafts might also limit the generalizability of the results to other types of grafts or reconstructive surgeries. Our inclusion criteria targeted retrospective or prospective cohort studies and clinical studies investigating dental and functional rehabilitation outcomes in human subjects undergoing maxillofacial reconstruction using microvascular flaps. We excluded case-control studies, studies of alternative reconstruction methods, and animal investigations.

Given those limitations, future research should include more comprehensive comparative studies that assess different techniques, host veins, arteries, and grafts. Economic evaluations considering the significant cost difference between couplers and hand suturing techniques would also be beneficial. Additionally, future studies should incorporate individual patient factors, including coagulation disorders, previous radiotherapy, the extent of tissue damage, and lifestyle factors that could potentially affect the outcomes. Further long-term follow-up research of patients undergoing anastomosis could provide valuable insights into the durability of different techniques and the prevalence of late complications. By addressing these limitations and incorporating these suggestions, future research can contribute significantly to improving the success rates of anastomosis in reconstructive surgery.

This systematic review and meta-analysis presents an in-depth evaluation of vascular failure rates in microvascular reconstruction surgeries involving free fibula grafts. We found a 6% overall vascular failure rate across the reviewed studies, with substantial heterogeneity suggesting significant variability in the reported rates of vascular failure. The findings indicate that vein thrombosis (3%) and arterial thrombosis (1%) were the leading specific causes of vascular failure. Our analysis highlights the complex interplay of multiple factors such as the technique of anastomosis, the quality of recipient vessels, the use of couplers versus hand suturing, and the diameter of the coupler in influencing the success of anastomosis.

Despite the potential advantages of couplers in most situations, our study suggests that their superiority over sutures is not consistently demonstrated across all studies, underlining the importance of considering individual patient characteristics and procedural specifics. Similarly, although both single and double vein anastomoses have merits and drawbacks, their efficacy seemed to be similar in terms of vascular failure rates in free fibular grafts.

Our study also underscores the importance of comprehensive preoperative evaluation, meticulous operative techniques, and thorough postoperative care and monitoring for optimal patient outcomes following anastomosis.

H.M. and S.S. participated in data collection and wrote the manuscript. M.H.K.M. and R.T. participated in the study design and performed the statistical analysis. E.T. and S.H. participated in the study design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.

No potential conflict of interest relevant to this article was reported.