Search results and the basic characteristics of the research
The flow chart of literature retrieval is shown in Figure 1. A total of 256 related articles were retrieved according to the search strategy. By reading the title, abstract and full text, a total of 8 studies met the inclusion criteria[28-35] and were published from 2011 to 2019. 654 patients were included in these studies, including 328 patients in DAA group and 326 patients in PLA group. Follow-up duration ranged from 2 weeks to 1 year. The basic characteristic of the included studies can be seen in Table 1. Randomized programs included in the study include randomization schedule[29] , computer generated random numbers[35] , random envelopes[32] , and unreported[28,30,31,33,34]. In the randomly generated sequence, 3 studies are of low risk bias[29,32,35] and 5 studies are of uncertain risk bias[28,30,31,33,34]. The risk of hidden grouping bias is low in 3 studies[32-34], high in 2 studies[28,30], and inaccurate in 3 studies[29,31,35]. In three studies[32,34,35], the researchers and subjects were blinded and the risk of bias was low; in five studies[28-31,33], the researchers and subjects were not blinded and the risk of bias was high. In the result evaluation blindness, 2 studies[32,33] had low risk bias, 5 studies[28-31,34] had high risk bias, and 1 study[35] had inaccurate risk bias. The results of 5 studies [28,29,32,34,35]were complete with low risk bias. Data in the 3 studies [30,31,33]were incomplete and the risk bias was high. The risk of selective reporting was low in seven studies[28-29,31-35], while the risk of selective reporting was inaccurate in one study[30]. Eight studies had a low risk of other bias[28-35]. Cochrane bias risk assessment is shown in Figures 2 and 3.
Results of meta-analysis
Operation time
Six studies[28-30,32-34] analyzed operative time and collected data on a total of 503 primary THAs, including 250 patients in the DAA group and 253 patients in the PLA group. Comparison of operative time for two different approaches. Because there was significant heterogeneity among the results of each study (I²>50%), heterogeneity could not be eliminated by sensitivity analysis and subgroup analysis, so a random effects model was used for the analysis. The results showed that the operative time of DAA and PLA were equivalent [MD = 7.49, 95%CI(-4.01 ~ 18.98), P = 0.2], the difference was not statistically significant, as shown in Figure 4.
Total blood loss
Four studies[28-29,32-33] analyzed blood loss, but one of them [32]described intraoperative blood loss specifically, so it was not included for analysis. A total of 190 patients with primary THA were included in the 3 studies[28-29,33], among which 95 patients received DAA and 95 patients received PLA. The difference in total blood loss between the two approaches was statistically significant [MD = 114.82, 95%CI(69.33 ~ 160.31), P < 0.00001], as shown in Figure 5. Due to the significant heterogeneity (I²> 50%) among the study results, sensitivity analysis revealed that the heterogeneity came from Barrett et al[29], after removing this study, found that the total bleeding in the DAA and PLA groups was comparable [MD = 50.04, 95%CI(-10.31 ~ 110.4), P = 0.1], the difference was not statistically significant. as shown in Figure 6.
Acetabular abduction angle
Six studies[28-30,32,34,35] analyzed the degree of acetabular abduction angle, including 557 primary THA patients, including 277 patients in DAA group and 280 patients in PLA group. There was no statistically significant difference in the degree of acetabular abduction angles between the two surgical approaches [MD=-0.45, 95%CI(-1.24 ~ 0.35) , P=0.27], as shown in Figure 7. Because there was significant heterogeneity (I²> 50%) among the study results, sensitivity analysis revealed that the heterogeneity came from Barrett et al[29], after removing this study, it was found that the degree of acetabular abduction angles in the DAA approach was smaller than that in the PLA approach [MD =-0.87, 95%CI(-1.69 ~-0.04), p = 0.04], and the difference was statistically significant, as shown in Figure 8.
Acetabular anteversion angle
Four studies[29,30,32,34] analyzed the degrees of acetabular anteversion angle and included 400 patients with primary THA, 198 patients in the DAA group and 202 patients in the PLA group. The difference of acetabular anteversion angle between the two surgical approaches is statistically significant [MD=-4.03,95%CI(-4.73~-3.33) , P<0.00001] , as shown in Figure 9. Due to the heterogeneity (I²>50%) between the results of the studies, sensitivity analysis revealed that the heterogeneity originated from Cheng et al.[34]. After removing this study, it was found that the acetabular anteversion angle degree of the DAA was still smaller than that of the PLA [MD =-4.25, 95%CI(-4.96 ~-3.54), p < 0.00001], and the difference was statistically significant, as shown in Figure 10.
Harris hip score at 6 weeks, 3 months, 6 months, 12 months
Harris hip scores were reported in 7[28-33,35] out of 8 studies[28-35], while 1[35] of the 7 studies lacked data and was therefore excluded.
Four studies[29-31,33] were recorded for Harris hip score at 6 weeks , two studies[29,32] were recorded for Harris hip score at 3 months , two studies[29,32] were recorded for Harris hip score at 6 months, and two studies[29,30] were recorded for Harris hip score at 12 months. A total of 5 studies were included. Meta-analysis was performed in subgroups according to the different of postoperative follow-up time, and 424 patients underwent primary THA, including 214 patients in DAA group and 210 patients in PLA group. Because of the heterogeneity among the results (I²> 50%), a random effects model was used for the Meta-analysis. 6 weeks after operation, the Harris hip score of DAA was higher than that of PLA [MD = 5.35, 95%CI(2.38 ~ 8.32), P = 0.0004], and the difference was statistically significant. 3 months after operation [MD = 2.82, 95%CI (-3.54 ~ 9.17), P = 0.39], 6 months after operation [MD = 0.67, 95%CI (-1.87 ~ 3.21), P = 0.6], 12 months after operation [MD = -0.42, 95%CI (-2.42 ~ 1.58), P = 0.68], the difference was not statistically significant. However, for the total Harris hip score in the follow-up of 4 periods of 5 studies, DAA is higher than PLA, and the difference is statistically significant [MD = 2.3, 95%CI (0.19 ~ 4.41), P = 0.03], as shown in Figure 11.
Postoperative dislocation rate
Four studies reported postoperative dislocation rates and included 380 patients with primary THA, including 188 patients in the DAA group and 192 patients in the PLA group. The incidence of postoperative dislocation was compared between the two approaches. As there is no heterogeneity (I²<50%) between the results of the two approaches, the fixed-effects model was used for Meta-analysis. The results showed that the postoperative dislocation rates of DAA was the same as that of PLA [RR = 0.80, 95%CI(0.20 ~ 3.19), P = 0.75], and the difference was not statistically significant, as shown in Figure 12.