Comparison between Minimally Invasive Plate Osteosynthesis and Conventional Open Plating for Midshaft Clavicle Fractures: A Systematic Review and Meta-Analysis

Objective The aim of this study was to compare the functional outcome and complications in midshaft clavicle fractures receiving minimally invasive plate osteosynthesis and conventional open plating. Methods Relevant studies were searched in the databases of Medline, EMBASE, Cochrane Library, Ovid, and Web of Science from inception to March 1, 2019. Pooled data were analyzed with Cochrane Collaboration's Review Manager 5.3. Results A total of 7 studies were included, of which 2 were randomized controlled trials, 3 were retrospective cohort studies, and 2 were prospective cohort studies including 316 patients. No statistical differences in functional outcome (weighted mean difference [WMD] = 0.99, P = 0.12), operation time (WMD = −10.44, P = 0.07) and time to bone union (WMD = −0.23, P = 0.70) were observed between the two groups. However, minimally invasive plate osteosynthesis reduced rates of skin numbness (odds ratio (OR) = 0.25, 95% CI : 0.13 to 0.48; P < 0.0001) and complications (OR = 0.33, 95% CI : 0.16 to 0.71; P = 0.005) compared with conventional open plating. Conclusion This systematic review and meta-analysis found no differences in terms of functional outcomes, operation time, and fracture healing time between minimally invasive plate osteosynthesis and conventional open plating. However, minimally invasive plate osteosynthesis had apparent advantages in rates of skin numbness and complications.


Introduction
e clavicle fractures constitute an estimated 2% to 5% of all fractures in adults [1]. ese fractures are most common in younger patients, and are o en associated with direct clavicle injuries such as contact sports and motor vehicle accidents. It is estimated that approximately 80% occur in the middle third of the clavicle, of which half are displaced [2]. Traditionally, midsha clavicle fractures were treated nonoperatively by closed reduction, sling or figure-of-8 harness, and physical therapy, with a low rate of nonunion less than 1% [3]. However, several studies illustrated that the nonunion rate with nonsurgical management is between 15% and 20% [4][5][6]. Furthermore, a current meta-analysis found that surgical treatment of midsha clavicle fractures presented a significantly lower nonunion rate compared with nonoperative treatment [7]. erefore, surgical treatment is o en preferred for midsha clavicle fractures today.
Open reduction and plate fixation is one of most commonly performed surgical techniques for midsha clavicle fractures. However, conventional open plating (COP) may compromise blood supply, so tissues, and cause several adverse events especially anterior chest numbness or hypoesthesia [8]. e minimally invasive plate osteosynthesis (MIPO) technique was widely used for lower extremity fractures owing to its role in protecting periosteal blood supply of the fracture area [9]. Recently, MIPO technique has been utilized for the treatment of midsha clavicle fractures with satisfactory clinical outcomes [10,11]. To date, several clinical studies indicated that MIPO could achieve similar results with fewer complications compared with COP [12,13]. However, the optimal surgical approach for midsha clavicle fractures remains controversial. Based on the information all above, this systematic review and meta-analysis aims to compare functional outcome and complications of MIPO with COP in the treatment of midsha clavicle fractures.

Search Strategy and Article Selection.
e literature searches were performed in the following databases: Medline, EMBASE, Cochrane Library, Ovid, and Web of Science from inception to March 1, 2019. e key words used were "clavicle/ collarbone/clavicular", "midsha /mid-sha ", "fracture/ broken", "plate", "open", and "minimally invasive/MIPO" in combination with the Boolean operators "AND" or "OR". Search the reference lists of selected articles manually as a secondary source. Articles were not restricted by languages and publication type.
Two reviewers (Enzhe Zhao, Rui Zhang) screened title and abstract of the search results independently, and removed duplicate articles. Both reviewers screened potentially relevant articles in full for evaluation. Disagreements were resolved by discussion with a third reviewer (Dou Wu).

Eligibility Criteria.
e inclusion criteria were as follows: (a) study design: randomized controlled study (RCT) or nonRCT; (b) patients with midsha clavicle fractures (15.2 according to AO/OTA classification [14] or type 2 according to the Robinson classification [15]) ( Figure 1); (c) intervention: MIPO and COP; (d) at least one of following data were reported: functional outcomes, operative time, union time, and complications. e exclusion criteria were as follows: (a) studies without controlled groups; (b) duplicate publication; (c) pathological fractures; (d) unavailable relevant data.

Quality Assessment. Modified version of the Cochrane
Collaboration's tool was applied to assess the risk of bias in RCTs [16]. e Methodological Index for Nonrandomized Studies (MINORS) methodology was used to evaluate other nonRCTs [17]. According to the Cochrane Collaboration recommendations, two reviewers (Enzhe Zhao and Yao Guo) assessed the methodological quality of each included study independently, and a third reviewer (Dou Wu) solved any possible inconsistency.

Data Extraction and Outcome Measurement.
A spreadsheet for data extraction was created prior to this study. Two researchers independently extracted the baseline study data as follows: the first author's name, study design, year of publication, interventions, sample size, mean age, follow-up time, operative time, complications, time to bone union, and functional outcomes. 15  e primary outcome of this meta-analysis was Constant-Murley Shoulder score. Secondary outcomes were operative time, time to bone union, skin numbness/hypoesthesia, and the other complications (e.g., infection, hypertrophic scaring, nonunion, re-fracture, implant failure, skin irritation, and painful shoulder). Fracture union was assessed using radiographic evidence, such as callus formation and bony bridging across the fracture site. Nonunion defined as a lack of complete osseous bridging a er 6 months.

Statistical Analysis.
e Review Manager so ware (RevMan 5.3, e Nordic Cochrane Center, e Cochrane Collaboration, Copenhagen, Denmark) was used for data analysis [18]. e weighted mean differences (WMDs) and odds ratios (ORs) were used to represent continuous and dichotomous outcomes, respectively. Data were pooled using the inverse-variance method for continuous outcomes and Mantel-Haenszel method for dichotomous outcomes. All data were reported with WMD or OR and the 95% confidence intervals (CI). Heterogeneity between studies was assessed using Chi-square test and I-squared test ( 2 ). Fixed-effects model was used if no significant heterogeneity existed (푃 > 0.1, 퐼 2 < 50%). If significant heterogeneity was present (푃 < 0.1 or 퐼 2 > 50%), data were rechecked first, then a random-effects model was used when heterogeneity persisted. Sensitivity analysis was evaluated by sequentially removing outlier studies, one at a time.

Search Results.
A systematic search strategy was created and a total of 114 relevant articles were identified. A er removal of duplicates, 64 articles were screened based on title and abstract for eligibility, and 7 articles were selected. A er reading the full text of these 7 articles, no article was excluded based on the selection criteria. erefore, 7 articles were included in the systematic review [12,13,[19][20][21][22][23] (Figure 2).

Quality Assessment.
e quality of RCTs was evaluated by modified version of the Cochrane Collaboration Risk of Bias Tool (Figure 3), and MINORS methodology was used to evaluate nonRCTs (Table 1).

Study Characteristics.
A total of 316 patients with midsha clavicle fractures were involved, including 158 patients treated by COP and 158 patients treated by MIPO. Of the 7 included studies, 2 were randomized controlled trials, 3 were retrospective cohort studies, and 2 were prospective cohort studies. e full characteristics of the included studies are listed in Table 2.

Functional Outcome.
Five studies [12,13,20,21,23] reported Constant-Murley scores at one-year follow-up. However, one study [12] was excluded from this analysis, Records excluded by title and abstract (n = 57) Full-text articles assessed for eligibility (n = 7) Full-text articles excluded (n = 0) Studies included in qualitative synthesis (n = 7) Studies included in quantitative synthesis (meta-analysis) (n = 7) Identi cation Screening Eligibility Included F 2: Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of study selection process.

Discussion
e present systematic review and meta-analysis, comparing MIPO versus COP for the treatment of midsha clavicle fractures, found no differences in terms of long term functional outcomes, operation time, and time to bone union between MIPO and COP. However, MIPO had apparent advantages in rates of skin numbness and complications. ese results suggested that MIPO is a safe surgical treatment of midsha clavicle fractures with fewer complications.
Postoperative functional recovery, one of the most crucial outcomes, is closely related to the quality of life of patients. Clinical scores such as American Shoulder and Elbow Surgeons (ASES) score, Constant-Murley score and Disabilities of the Arm, Shoulder, and Hand (DASH) were o en used to assess the postoperative functional recovery. We also employed the Constant-Murley score, including pain, range of motion and activities of daily living, to evaluate postoperative functional recovery. Although several studies [10,11,[24][25][26] have reported the good clinical outcomes of MIPO, this meta-analysis found no difference in terms of long term functional outcomes at one-year follow-up between MIPO and COP (푃 = 0.12).
Two studies [20,23] showed mean operation time of MIPO was shorter than COP. However, Zehir et al. [22] reported the operation time of COP was shorter. Our meta-analysis found no significant difference in operation time between MIPO and COP (푃 = 0.07). ere was a significant heterogeneity between the two groups (푃 < 0.00001, 퐼 2 = 94%), and sensitive analysis showed the total pooled effect size was greatly affected by the study of Zehir [22]. Operation time of COP was shorter than MIPO in the study of Zehir [22] might be due to repetitive fluoroscopy use for fracture reduction and placement of plate and high loss to follow-up (10/32) in the MIPO group.
MIPO technique, used to stabilize acute fractures without extensive so tissue dissection, was believed to promote bone healing by preserving the enveloped so tissue and periosteal circulation [10]. However, this meta-analysis found no significant difference in time to bone union between MIPO and COP groups (푃 = 0.70). is might be due to low sample size of for reason that both standard deviation and standard error were not available from the full text. A fixed-effects model was used without heterogeneity (푃 = 0.31, 퐼 2 = 17%). Finally, it was found that the Constant-Murley scores at one-year follow-up did not differ between two groups (WMD = 0.99, 95% CI : −0.25 to 2.23; 푃 = 0.12) (Figure 4).

Time to Bone
Union. Time to bone union was reported in all seven studies. However, one study [19] only reported the maximum and minimum values of healing time without mean values and standard deviation, thus a total of six studies [12,13,[20][21][22][23] were included in this analysis. A random-effects model was applied due to the remarkable heterogeneity across studies (푃 = 0.03, 퐼 2 = 60%). ere was no significant statistical difference between MIPO and COP regarding time to bone union (WMD = −0.23, 95% CI : −1.42 to 0.96; 푃 = 0.70) ( Figure 6). In addition, a sensitive analysis by excluding outlier study [22] showed that the result was robust (퐼 2 = 15%; WMD = 0.20, 95% CI : −0.62 to 1.02; 푃 = 0.63). F 3: Risk of bias assessment summary of randomized controlled trials. "+" = risk of bias not present, "−" = risk of bias present, and "?" = insufficient information to judge risk of bias. [28,29]. Several investigations reported MIPO technique significantly reduced the anterior chest wall numbness compared with COP [19,21]. Similarly, our meta-analysis found that MIPO was superior to OCP in skin numbness (푃 < 0.0001). Anatomically, no branch of supraclavicular nerve was found within 2.7 cm of the sternoclavicular joint or within 1.9 cm of the acromioclavicular joint [30]. e medial and lateral incision selected by the MIPO technique was precisely within these two areas with no neural branches. Although an additional central incision on the fracture site is needed for anatomical reduction in MIPO (the technique used in studies of included studies and the careful dissection during COP to avoid damaging blood supply. Although there was a significant heterogeneity between the two groups (푃 = 0.03, 퐼 2 = 60%), these findings were reliable because the result of sensitive analysis did not alter significance by excluding the outlier study [22] (퐼 2 = 15%; WMD = 0.20, 95% CI : −0.62 to 1.02; 푃 = 0.63).

Skin
Previous studies have shown that anterior chest wall numbness is one of the most common complications in the treatment of clavicle fracture with plate, by reason of damage to branches of supraclavicular nerve [27,28]. e incidence of skin numbness a er COP has been reported to be 12%-83%  Conflicts of Interest e authors declare that they have no conflicts of interest.
Jiang [12], Beirer [19] and Zehir [22]), the incision length and so tissue forcible retraction in MIPO were still less than COP. ese might explain the reason why skin numbness occurred more o en in the COP group.
In this meta-analysis, complications such as infection, hypertrophic scaring, nonunion, refracture, implant failure, skin irritation, and painful shoulder were evaluated also. e reason why these complications were pooled together was that the sample size would be smaller if each complication was assessed separately. Pooled results indicated that patients receiving COP had more complications than MIPO, and this was found to be significant (푃 = 0.005). Five of the included studies reported no major complications such as nonunion, re-fracture, and implant failure in both groups [12,[19][20][21]23]. is might be due to the low sample size of included studies. To our knowledge, this is the first systematic review and meta-analysis comparing MIPO and COP for the treatment of midsha clavicle fractures. However, there were several limitations that should be noted. First, there were only two RCTs that met the eligibility criteria, and the sample sizes were small in most studies. Second, the follow-up duration was relatively short and might underestimate the complications.
ird, due to the lack of relevant data, we could not perform subgroup analysis according to fracture classification. Finally, the existence of bias, owing to different surgeons and surgical technologies, might have been inevitable in our research.

Conclusion
is systematic review and meta-analysis found no differences in terms of functional outcomes, operation time, and fracture healing time between MIPO and COP. However, skin