Literature searching
All literature screening processes were performed with Endnote X8. After literature searching, a total of 5974 citations from PubMed, 6587 citations from Embase, 23 citations from Cochrane and 18 citations from Clinicaltrial.gov were obtained. We excluded 6897 duplicate citations by using Endnote duplicate citations finding function. After initial title and abstract screening, 5593 citations were excluded and disagreement would be resolved by routine meeting of research group. During full text screening, a total of 97 citations not compliant to the criteria were excluded and 13 citations of studies were included in this meta-analysis eventually[11, 25-36]. The PRISMA flowchart of this meta-analysis was displayed in Figure 1.
Baseline characteristics
There were a total of 1164 patients included in this study, with 603 patients in surgical group and 561 patients in non-surgical group. The mean age of enrolled patients was around 40 years old ranging from 18 to 63 years old, which conformed to the regular ATR population. Overall, male and female patients consisted of 84% and 16% population included in the study, respectively. For the time period between injury and treatment, 2-day was the shortest period reported by Twaddle et al.[29] while 21-day was the longest period reported by Nistor et al.[25]. In addition, surgical techniques of each included studies were extracted for better interpretation of baseline characteristics and, end-to-end Bunnell-type was the most adopted technique for ATR repair. Last but not least, different follow-up period could be a significant factor affecting the results so that it was recorded as well. One and two year were the widely accepted follow-up period among the included RCTs. The detailed information of baseline characteristics of each RCTs was shown in Table 2.
Risk of bias assessment
Two independent authors strictly assessed the risk of bias across studies under the instruction of Cochrane Collaboration Tool[17] and the visualization of results was displayed in Figure S1. Risk of bias was relatively low owing to the characteristics of RCTs. However, assessment of unclear risk was occurred in several studies. Regarding to selection bias about random sequence generation, Moller et al.[27], and Keating et al.[33], did not state clearly about the situation and, unclear risk was assessed in Nistor et al.[25], Fischer et al.[36], and Cetti et al.[26], respectively. When it comes to blinding of participants and personnel in performance bias, unclear risk occurred in Nistor et al., and cetti et al. while high risk was assessed in Fischer et al.. Inadequate blinding of assessment was not clearly declared in Nistor et al. and Fischer et al. so that unclear risk was obtained.
Publication bias was assessed by administrating Revman software and egger’s test was adopted. Each outcome measure was assessed individually and visualization of results were shown in Figure S2(A) to (L). Inspection of symmetry was obtained, indicating no publication bias among each outcome measure.
Primary outcomes
Re-rupture rate
All included 13 studies reported the result of re-rupture rate and we divided it into re-rupture in accelerated functional rehabilitation and re-rupture not in accelerated functional rehabilitation as subgroup analysis. In the subgroup of re-rupture occurred in accelerated functional rehabilitation, no significant difference between surgical and conservative treatment could be observed (3 studies, 289 participants, Z=1.04, P=0.30, I2=0%, RR: 0.59, 95% CI: 0.22 to 1.59). In contrast, compared with conservative group, significant reduction in re-rupture rate not in accelerated functional rehabilitation could be observed in surgical treatment group (10 studies, 850 participants, Z=3.90, P < 0.0001, I2=0%, RR: 0.34, 95% CI: 0.19 to 0.58). Collectively, the overall result showed that surgical treatment was associated with significant reduction in re-rupture rate (13 studies, 1139 participants, Z=3.97, P < 0.0001, I2=0%, RR: 0.38, 95% CI: 0.24 to 0.41). Detailed information about re-rupture rate was shown in Figure 2(A).
Return to sport
There were 8 studies reported the result of return to sport among patients receiving ATR repair. Cetti et al.[26] and Costa et al.[28] reported the favorable outcome of surgical treatment in recovering the sport capacity of ATR patients compared with conservative management, while Manent et al.[35] reported the opposite result favoring conservative treatment. Collectively, the overall result indicated that no significant difference between surgical and conservative treatment in sport capacity recovery (8 studies, 567 participants, Z=0.35, P=0.73, I2=75%, RR: 1.09, 95% CI: 0.67 to 1.77). Detailed information about return to sport was shown in Figure 2(B).
Secondary outcomes
Complication rate
We defined complication rate as complication occurred after ATR treatment other than re-rupture and it was reported in 12 of included studies. The overall result indicated that complication rate after treatment in conservative treatment group was significantly lower than that in surgical treatment group (12 studies, 1107 participants, Z=2.56, P=0.01, I2=69%, RR: 2.62, 95% CI: 1.25 to 5.46). Main complications occurred after ATR treatment were deep vein thrombosis, adhesion of scar to underlying tendon, sural nerve injury, superficial and deep infection. Detailed information about overall complication rate was shown in Figure 3(A).
Deep vein thrombosis
Deep vein thrombosis, a severe complication usually occurred after ATR treatment owing to plaster casting immobilization[28], was reported in 8 of included studies. The overall result showed that no significant evidence could be obtained to distinguish better management strategy to avoid deep vein thrombosis (8 studies, 777 participants, Z=1.12, P=0.26, I2=0%, RR: 0.58, 95% CI: 0.22 to 1.51). Detailed information about deep vein thrombosis was displayed in Figure 3(B).
Adhesion
Adhesion of scar to underlying tendon was reported in 3 of included studies and it might lead to secondary surgery. The overall result revealed that surgical process might lead to an increasing incidence of adhesion of scar to underlying tendon (3 studies, 294 participants, Z=2.08, P=0.04, I2=55%, RR: 8.77, 95% CI: 1.13 to 67.99). Detailed information about adhesion of scar to underlying tendon was displayed in Figure 3(C).
Sural nerve injury
Disturbance in sensation of ATR patients after treatment due to sural nerve injury was reported in 6 of included studies. The overall results showed that significantly increasing incidence of sural nerve injury occurred in patients with surgical treatment compared with conservative management in ATR (6 studies, 603 participants, Z=3.71, P=0.0002, I2=0%, RR:6.77, 95% CI: 2.47 to 18.56). Detailed information was shown in Figure 3(D).
Infection
Wound infection was a common complication of surgical treatment in ATR repair and it could be divided into superficial and deep infection. For superficial infection, compared with surgical treatment group, conservative management showed significant evidence to prevent infection after treatment (7 studies, 659 participants, Z=3.28, P=0.001, I2=0%, RR: 7.34, 95% CI: 2.23 to 24.17). Detailed information was shown in Figure 3(E).
In contrast, regarding to deep infection, there was no significant difference between surgical treatment and conservative treatment group, even though no case of deep infection in conservative group was reported (8 studies, 653 participants, Z=1.88, P=0.06, I2=0%, RR: 3.85, 95% CI: 0.95 to 15.65). Detailed information about deep infection was shown in Figure 3(F).
Period absence from work
ATR results in loss of motor ability as well as absence from patients’ occupation so that different time period is an essential assessment index. The pooled result showed that neither surgical treatment nor conservative management was in possession of shorter period absence from work (3 studies, 330 participants, Z=0.10, P=0.92, I2=77%, RR: -0.22, 95% CI: -4.32 to 3.89). Detailed information was displayed in Figure 4(A).
ATRS functional score
ATRS functional score, with high reliability, validity, and sensitivity for quantifying functional outcome of patient receiving ATR treatment, is a indispensable index to determine the better treatment[16]. According to pooled result, there was no significant difference between surgical and conservative treatment regarding ATRS assessment (3 studies, 207 participants, Z=1.86, P=0.06, I2=0%, RR: 4.27, 95%CI: -0.24 to 8.77). Detailed information about ATRS assessment was displayed in Figure 4(B).
Flexion
Range of motion is a reflection of joint motor ability and, dorsiflexion as well as plantarflexion are suitable index to the assessment. For mean dorsiflexion, surgical treatment group was similar with conservative treatment group (2 studies, 204 participants, Z=0.32, P=0.75, I2=51%, RR: 0.62, 95% CI: -3.23 to 4.46). Detailed information about dorsiflexion was shown in Figure 5(A).
Similarly, there was no significant difference could be observed regarding to pooled result of mean plantarflexion (4 studies, 349 participants, Z=1.08, P=0.28, I2=92%, RR: 2.43, 95% CI: -1.97 to 6.83). Detailed information about plantarflexion was shown in Figure 5(B).