A meta‐analysis evaluating risk factors for compound free flaps for upper extremity defect reconstruction comparing complications and functional outcomes of compound free flaps with and without bone components

Compound flaps offer the advantage of one stage defect reconstruction respecting all relevant tissues and early functional recovery by optimal vascularity of all components. Due to its specific vascular anatomy and the three‐dimensional donor site, compound flaps with bone components may result in higher complication rates compared to soft tissue compound flaps. The meta‐analysis summarizes the available evidence and evaluates whether bone components are a risk factor for periprocedural complications in upper extremity multidimensional defect reconstruction.


| INTRODUCTION
Compound flaps for microsurgical upper extremity defect reconstruction are regarded to be both indispensable and often superior to alternative techniques (Kremer et al., 2007;Sauerbier et al., 2012;Wang et al., 2013).
In the current literature, the incidence of venous thrombosis in compound flaps with bone components is reported ranging between 12 and 25%, and addressed in six studies (Heitmann et al., 2002;Jupiter et al., 1997;Kremer et al., 2007;Lin et al., 2005;Liu et al., 2015;Noaman, 2013). Following by the flap loss rates of compound flap range from 0% (dorsalis pedis compound free flap) to 14% (osteocutaneous fibular free flap) (Eo et al., 2008;Li et al., 2000;Lin et al., 2005;Noaman, 2013). Partial flap loss rates range from 0% to 21% (dorsalis pedis compound free flap) (Eo et al., 2008;Ju & Hou, 2012), and nonunion rates of graft bone components vary from 0% (Liu et al., 2015) to 27% (osteoseptocutaneous fibular bone graft) (Heitmann et al., 2002;Liu et al., 2015); hematoma was reported by two studies ranging from 5% (Sauerbier et al., 2012) to 14% (Lin et al., 2005). In view of the available literature, surgeons seem to not often choose compound flaps with vascularized bone for upper extremity reconstruction, and the literature reflects a more based on ideas and innovations driven indication of compound flaps with vascularized bone for upper extremity reconstruction, than a standardized study approach (Kremer et al., 2007). For upper extremity reconstruction, no systematic analysis of outcome and safety of compound flaps is available yet, which may contribute to the evidence of compound free flaps by summarizing the limited data as one of the common disciplines in microsurgery. Specifically, this study was conducted to make clear the risks of compound flaps and further elucidate postoperative complications. A systematic study of these unique flaps will better allow the surgeon to communicate the perioperative risks to the patient.
From the authors' clinical experiences, compound bone flaps have higher complication rates, especially venous thrombosis, than compound soft tissue flaps. The hypothesis of this study is that vascularized bone components in compound flaps for the upper extremity defect reconstruction are a risk factor for perioperative complications.

| MATERIALS AND METHODS
This meta-analysis was conducted in accordance with the methodology of the PRISMA Statement Guidelines (Hutton et al., 2015). Papers published in PubMed and Embase were searched, language was restricted to "English", "German" and "Chinese". Included studies were published between January 1988 and May 2018. Manual search was performed with the following search terms: "free flap" "compound flap", "composite flap", "conjoined flap", "chimeric flap", "osteocutaneous flap", "myocutaneous flap", "neurofasciocutaneous flap", "tendinocutaneous flap", "upper extremity", "arm", "forearm", "elbow", "wrist", "palm", "hand", and other individual corresponding terms. All cited papers have been reviewed for further potential studies, as also similar studies suggested by PubMed have been reviewed.
The study flow diagram is shown in Figure 1.
Inclusion criteria were studies with consecutive cases of compound free flaps for upper extremity defect reconstruction (soft tissue compound flaps including nerve or tendon fascia and a muscle component or compound flaps with soft tissue and bone components), with a sample size of free flaps equals to or larger than 5 and extractability of clinical data on outcomes and complications.
Exclusion criteria were studies with inconsecutive cases, overlapping articles, articles with a sample size of free flaps and articles without the opportunity to extract clinical data on outcomes and complications. Furthermore, reviews, abstracts, or letters were excluded.  Due to its specific harvesting technique, the specific flap anatomy (tight, undissected connections of the soft and bone components) as well as special factors characterizing the three-dimensional wound bed and assessment of the donor site, compound toe flaps for toe-tothumb transfer/finger reconstruction were not included in the current data analysis.
Two independent reviewers (first and second author) extracted the data from eligible studies with predefined inclusion and exclusion criteria. The results were screened for their titles and abstracts, followed by proofreading of the full text article to apply inclusion criteria. A third reviewer was consulted in case of disagreement for inclusion/exclusion of studies (senior author) and discussion on evidence levels.
The following data from all included studies were extracted when available: first author, year of publication, country of origin, sample size, patients' characteristics, indication, defect localization, follow-up time, and type of included compound flap (Table 1).
The Extractable outcome parameters such as the DASH (Disability of the Arm, Shoulder, and Hand) score, two-point discrimination, Semmes-Weinstein monofilament, cold intolerance, and range of motion were planned to be extracted and analyzed.
Intentionally, only cases from publications with compound soft tissue flaps containing bone as well as soft-tissue compound flaps containing tendon or nerve, or those which principally can be harvested with bone as a comparative group was extracted.
Studies were rated on methodological quality based on the American Society of Plastic Surgeons Evidence Rating Scale for Therapeutic Studies (Sullivan et al., 2011).
Susceptibility of the systematic review to publication bias was assessed with the Egger linear regression test (Egger et al., 1997).

| STATISTICAL ANALYSIS
All data were analyzed with the software R GUI 3.3.1 (The R foundation for statistical computing, https://www.r-project.org/). The statistical heterogeneity among included studies was evaluated using I 2 statistic and Q statistic P-values. Heterogeneity was considered significant if I 2 value was greater than 50% or P < .05.
Meta-analysis of single proportions was performed to estimate the summarized complication rates and corresponding correction intervals on a per flap basis. Logit transformation was set as the summary measure. In case of zero value in a study, a continuity correction would be applied. Radom effects models were used to weight the individual studies, in order to cover the variation between and within studies. I 2 statistic and Q Statistic P values were calculated.
Categorical variables were compared using Chi 2 , Fisher's exact test. All statistical analyses were 2-sided end, P < .05 was considered significant. Statistical analyses were performed using SPSS 22.0 (SPSS Inc., Chicago, USA).

Society of Plastic Surgeons Evidence Rating Scale for Therapeutic
Studies (Sullivan et al., 2011).
For the present meta-analysis, 159 patients with 159 flaps were included from 12 studies. One hundred six out of 144 (74%) included patients from 11 studies with extractable data were male. The average patient age of all studies with extractable data was 35.5 (7-78) years.
One hundred fourteen (72%) defects were caused by trauma, followed by infection in 23 (15%) patients (Figure 2(a)). The size of flaps for defect reconstruction ranged from 4 to 435 cm 2 . Fifty-eight (36%) out of 159 patients' donor sites were closed by skin grafts.
In accordance with the aim of the study, compound flaps with and without bone were compared and subgroups were created.
One hundred seven (67%) of 159 flaps included bone components. In

| Subgroup analyses and publication biases
Subgroups were chosen to compare the flap type specific risk of compound flaps either with (Subgroup 1) or without bone components (Subgroup 2) for defect reconstruction (Table 2, Figure 3 and Figure 4). Publication bias was depicted in funnel plot of all kinds of complication rates ( Figure 5).

| Functional results
No study evaluated the DASH score, the two point's discrimination score or other functional scores, such as Semmes-Weinstein monofilament, Hand outcomes Questionnaire or cold intolerance. There are some limitations which should be taken into consideration: Firstly, there were no randomized-controlled studies or comparative studies eligible for this meta-analysis, all included studies achieved a Level of evidence of III. Secondly, the total number of eligible studies (n = 12) addressing upper extremity defect reconstruction with compound bone flaps is rather small for a meta-analysis but improves the level of evidence by summarizing all single studies.
Thirdly, the sample size of each study was relatively small, with an average of 13 cases (range: 5 to 23), which resulted in a smaller sample size for subgroup analyses and made it statistically impossible to decrease the rate of heterogeneity. Finally, the reported parameters of each study were measures poorly and without comparative data, thus, hampering further summarization of the outcome data.
In view of the reported perioperative outcome of free compound flaps with bone components, we have to anticipate a selection bias in patients with vascularized compound bone grafts which are often indicated in more complex defects, accompanied by more soft tissue trauma and thus reduced vascularity, increased contamination and more complex or large bone defects.
As discussed recently (Xiong et al., 2016;Zhang et al., 2019), it is crucial to add standardized patient and procedure related parameters (e.g., differentiation between type of flap, type and number of anastomosis) as well as outcome data (type of thrombosis, etc.) and especially functional outcome parameter for all future studies on microsurgical reconstruction to enable outcome assessment between studies for consecutive meta-analysis with decreased heterogeneity (e.g., DASH score (Hudak et al., 1996)). In addition, randomized studies on compound flaps would be helpful to further answer relevant questions, for example, by directly comparing vascularized bone grafts included in compound flaps with nonvascularized bone grafts, with and without bone segment transfer with modern nails and tradition external systems. In addition, further comparative designs may include alternative bone substitutes instead of nonvascularized bone grafts, especially in the treatment of infected pseudarthrosis.

ACKNOWLEDGMENT
Open access funding enabled and organized by Projekt DEAL.