Use of Surgical Approach Is Not Associated With Instability After Primary Total Hip Arthroplasty

A Meta-Analysis Comparing Direct Anterior and Posterolateral Approaches

Elina Huerfano, MD; Maria Bautista, MD, MSc; Manuel Huerfano, MD, MSc; Juan M. Nossa, MD


J Am Acad Orthop Surg. 2021;29(22):e1126-e1140. 

In This Article


According to the results of this meta-analysis, the risk of dislocation is not different between DAA and PA (RD = −0.00; P = 0.92). Likewise, no differences in dislocation rates were identified when posterior soft-tissue repair, learning curve, acetabular cup positioning, and LLD were analyzed independently.

Several meta-analyses comparing outcomes and complications between direct anterior and posterior approaches for THA have been published. However, they were not designed to test differences in the risk of dislocation regarding other variables related to prosthetic instability[7,38–41] and did not include the latest RCTs[4,30] nor the retrospective cohort study published by Aggarwal et al[15] (2,986 subjects). The meta-analysis performed by Kunutsor et al[42] with over 4 million patients from 125 studies and registry analyses reported mixed results of RR of dislocation by analyzing anterior and posterior approach modifications separately, discounting that variations among them are not notable. Furthermore, studies including national joint registries should be interpreted with caution given the limitations and bias in registry and database information.[7]

Prosthetic hip dislocations can occur up to 6 months after the index procedure, with a very discrete increase in the cumulative rate over time;[2] consequently, Barrett et al[4] hypothesized that late dislocations might not be attributable to the surgical approach. Therefore, our meta-analysis focused on analyzing other factors contributing to instability after THA.[2]

The effect of the learning curve on postoperative outcomes of THA through DAA has been widely described;[23,24] however, the number of procedures required to reach the plateau has not been clearly established. Some studies included in this meta-analysis assessed the learning curve in the transition from PA to DAA and described that surgeons require from 50[32] to 100[23,33] cases to learn the surgical technique. We did not find a difference in the rate of prosthetic hip dislocation when analyzing the publications in which surgeons have completed the learning curve (20 in DAA versus 21 in PA; RD = 0.00 [95% CI: −0.00 to 0.01]; Figure 6), with significant homogeneity among them (I2 = 0%). Thereby, our results might suggest that surgeons with expertise in either surgical approach could produce similar results in the prevalence of this complication.

Similarly, to determine whether posterior soft-tissue repair after PA is comparable with DAA, we analyzed data from 14 studies reporting posterior capsule or external rotators repair but did not find statistically significant differences in the number of cases of dislocation in either group (8 in DAA versus 11 in PA) (Figure 5). Unfortunately, it is not possible to discriminate whether dislocation cases in the DAA group occur anteriorly or posteriorly. According to previous reports, the dislocation direction might not be determined by the surgical approach;[43] it cannot be assumed that only anterior stability is compromised in DAA. McLawhorn et al,[44] using magnetic resonance imaging, described that only 75% of piriformis tendons and 38% of conjoined tendons were intact after DAA, which might cause minor residual posterior instability. We conclude that the direction of the instability might be a confounding factor in assessing the risk of dislocation in either surgical approach.

Regarding accurate acetabular cup positioning in the coronal plane, we found a difference in the risk of reaching Lewinnek's safe zone, favoring DAA (RR = 1.17, 95% CI: 1.03 to 1.33; P = 0.01). However, these estimates are associated with high heterogeneity (I2 = 93%) limiting the ability to conclude in favor of any approach, similar to that described by other authors.[38,41] Furthermore, a recent publication has questioned the significance of this "safe zone" as a predictor of instability.[45] In the assessment of cup anteversion, our results demonstrate a tendency toward a less anteverted implant in the DAA group, but this difference did not reach statistical difference either.

We hypothesize that these small discrepancies in acetabular orientation are the result of several factors. Rodriguez et al[24] described that in DAA cups were intentionally positioned with less anteversion because of concerns about anterior instability, and Gonzalez Della Valle et al[46] demonstrated that in PA the surgeon tends to overestimate intraoperative cup anteversion because of a progressive anterior pelvic roll. Further studies are required to determine whether the surgical approach is an independent factor for implant positioning.

Leg length discrepancy after THA is an independent risk factor for postoperative dislocation,[47] explained because of the decrease in overall myofascial tension.[2] If the use of intraoperative fluoroscopy and the supine position reduces the likelihood of LLD after THA, DAA would overcome PA. However, the current meta-analysis demonstrates no statistically significant differences in the magnitude of LLD between surgical approaches (Figure 10). This might be related to the several surgical techniques and methods that have been implemented to improve leg length equalization accuracy.[48]

Given that prosthetic dislocation is associated with multiple variables,[2] the use of a certain approach might not be sufficient to prevent this complication. To our knowledge, this is the first meta-analysis that evaluates the differences in the risk of instability after THA assessing the variables that might be influenced by the surgical approach with the purpose of isolating its effect.

The main strength of our study is that we aimed to broadly include all the evidence available, both RCTs and non-RCTs, and the high quality of most studies included. In addition, we excluded studies based on joint registries and administrative claim databases, reducing the pitfalls related to the use of these data as the lack of detailed clinical information and the risk for coding bias.[7] On the other hand, we acknowledge several limitations. First, in most studies, the adequate concealment of allocation and the blinded assessments of the results were unclear, which might influence the validity of the outcomes. Second, we were not able to conclude definitively toward either approach given the high heterogeneity identified for the pooled estimates of acetabular cup positioning. This might be related to the lack of a standardized tool to accurately measure cup orientation, allowing for the misinterpretation of the results. Furthermore, some studies reported the means of continuous data but did not present SD. Finally, despite the relevance of the femoral position in the overall prosthetic hip instability, due to the limitation in the assessment of this variable, the analysis of the femoral orientation in DAA or PA was not performed.