Study Characteristics
Table 1 shows the characteristics of the 9 studies and their patients. 4 of the 9 studies were conducted in the US, 1 in Canada, 1 in China, 1 in Denmark and 1 in the UK. All the studies were published during the last 20 years, i.e., from 1990 to 2019. The studies included 3 prospective cohort studies and 7 retrospective cohort studies and enrolled a total of 507 patients with ankle arthritis. No randomised controlled trial was included. A total of 303 patients underwent arthroscopic arthrodesis, and 214 patients underwent open arthrodesis. In all cases, arthroscopy was used with standard anteromedial and anterolateral portals.[21, 22]
Using the Newcastle-Ottawa scale, 2 analyses had a moderate risk of bias for study participation. One analysis had a high risk of bias. The graph in Figure 2 demonstrates the summary and results of the methodological quality assessment. Regarding the MINORS, 3 articles had a score of 22. Five analyses had scores between 18 and 20. One article had a score of 16. Supplemental Table 1 contains a summary of the results.
Overall, methodologic limitations were examined in the studies with an average Coleman methodologic score (mean: 61; SD: 10.42; range: 43-75). The reasons with the low score were a short period of follow up, high percentage coverage of retrospective studies and unclear procedure for assessing outcomes. (Supplemental Table 2)
Primary outcome
Fusion Rate
Seven studies reported the fusion rate, and the pooled data showed a significantly higher rate of fusion during arthroscopic arthrodesis compared with open surgery (odds ratio 0.25, 95% CI 0.11 to 0.57, p = 0.0001) in 128 of 167 patients. In addition, there was no significant heterogeneity between these two groups. (Figure 3a) (for fusion rate assessing in arthroscopic arthrodesis with no three cannulated screws subgroup, 167 participants, odds ratio 0.26, 95% Cl 0.13 to 0.55, p = 0.0004; arthroscopic arthrodesis with three cannulated screws subgroup, 200 participants, odds ratio 0.33, 95% Cl 0.12 to 0.88, p = 0.03)
Days to Union
Three studies assessed the days to union, and there was no difference between arthroscopic and open surgery (WMD 1.62, 95% CI -5.97 to 59.08, p = 0.11). There was no significant heterogeneity among these three studies. (Figure 3b)
Surgical Outcomes
Operation Time
Four studies reported the operation time. The level of heterogeneity was low (chi-square = 4.65, df = 3, I2 = 35%, p = 0.20), and the pooled data from the four studies did not show a notable difference between arthroscopic arthrodesis and open surgery (WMD 3.72, 95% CI -5.31, 12.76, p = 0.42). (Figure 4a)
Estimated Blood Loss
Two studies assessed the estimated blood loss during surgery, and the pooled data showed significantly less blood loss during arthroscopic arthrodesis compared with open surgery (WMD 52.04, 95% CI 14.14 to 89.94, p = 0.007). Also, there was no significant heterogeneity between the two groups. (Figure 4b)
Tourniquet Time
Four studies reported the tourniquet time, and the pooled data showed remarkably shorter tourniquet times during arthroscopic arthrodesis compared with open surgery (WMD 22.68, 95% CI 1.92 to 43.43, p = 0.03). There was significant heterogeneity among these four studies. (Figure 4c)
Length of Hospital Stay
Six studies reported the length of stay in the hospital, and the pooled data showed markedly less hospitalisation time for patients undergoing arthroscopic arthrodesis compared with open surgery (WMD 1.62, 95% CI 0.97 to 2.26, p < 0.00001), with a low level of heterogeneity (chi-square = 8.21, df = 4, I2 = 51%, p = 0.08). (Figure 4d) (for length of hospital stay assessing in arthroscopic arthrodesis with no three cannulated screws subgroup, 173 participants, odds ratio 2.01, 95% Cl 1.53 to 2.49, p < 0.00001; arthroscopic arthrodesis with three cannulated screws subgroup, 180 participants, odds ratio 2.56, 95% Cl 2.22 to 2.91, p < 0.00001)
Complications
Overall Complication Rate
Eight studies assessed the overall complication rate, and there was no difference between arthroscopic arthrodesis and open surgery (WMD 1.70, 95% CI 0.84 to 3.43, p = 0.14). There was no obvious heterogeneity among these eight studies. (Figure 5a) (for overall complication rate assessing in arthroscopic arthrodesis with no three cannulated screws subgroup, 231 participants, odds ratio 1.35, 95% Cl 0.70 to 2.61, p = 0.37; arthroscopic arthrodesis with three cannulated screws subgroup, 297 participants, odds ratio 1.81, 95% Cl 0.96 to 3.42, p = 0.07)
Rate of infection
Seven studies, including 365 patients, reported the rate of infection, and the pooled data showed no significant difference between patients who underwent arthroscopic arthrodesis and those who underwent open surgery (odds ratio 1.58, 95% CI 0.60 to 4.16, p = 0.36). There was no significant heterogeneity among these seven studies. (Figure 5b) (for overall complication rate assessing in arthroscopic arthrodesis with no three cannulated screws subgroup, 206 participants, odds ratio 2.82, 95% Cl 0.79 to 10.12, p = 0.11; arthroscopic arthrodesis with three cannulated screws subgroup, 159 participants, odds ratio 0.72, 95% Cl 0.16 to 3.15, p = 0.66)
Postoperative Outcomes
One Year Post-Surgery
Two studies reported the 1-year postoperative recovery with the AOS score. The pooled data showed markedly better recovery for the patients who underwent arthroscopic arthrodesis compared with those who underwent open surgery (WMD 14.73, 95% CI 6.66 to 22.80, p = 0.0003) with a low heterogeneity (chi-square 0.49, df = 1, p = 0.48, I2 = 0%). (Figure 6a)
Two Years Post-Surgery
Two studies assessed the 2-year postoperative recovery using the AOS scale and reported that patients who underwent arthroscopic arthrodesis had no notably greater recovery than those who underwent open surgery (WMD 8.13, 95% CI -3.40 to 19.99, P = 0.48). In addition, there was no significant heterogeneity between these two groups (I2 = 40%, P = 0.20). (Figure 6b)
Publication Bias
Visual inspection of the Begg funnel plots for the fusion rate, overall complication and infection rate revealed symmetry (Figure 7). To ensure that there was not publication bias, Egger’s test was also conducted (Supplemental Table 3). There was not statistically significant publication bias for any of the three results (95% Cl -1.58 – 2.88, p = 0.75; 95% Cl -0.96 – 2.97, p = 1.19; 95% Cl -2.55 – 2.78, p = 0.92, respectively).