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

Disclosures

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

In This Article

Results

Of the 6,723 studies identified, 6,618 were unrelated to the topic of research. In the first screening based on titles and abstracts, 34 studies were rejected. After the first screening process and removal of duplicates, 71 potentially eligible studies were identified. We further filtered these studies according to the inclusion and exclusion criteria and methodologic quality. Finally, 25 studies were selected for inclusion in the analysis; these included data from 3,500 patients in the DAA group and 3,672 patients in the PA group (Figure 1).

Figure 1.

Flow diagram of selection of studies included in the analysis.

Barrett and colleagues performed an RCT including 43 subjects in the DAA group and 44 in the PA group and presented the outcomes in two separate articles.[4,14] Both articles were evaluated in the present analysis, but only one was considered when reporting in the total number of articles and patients, as they are different reports of the same patients.

All studies were published between 2009 and 2019. Sample sizes ranged from 40 to 2,986. The baseline parameters of included studies were comparable between groups and are summarized in Table 1.

Five of the 25 studies included in the meta-analysis were RCTs.[4,5,29,30,33] All the studies reported clear inclusion and exclusion criteria and suggested a methodology of randomization. Four of them[4,5,29,33] reported that the allocation concealment was achieved by sealed envelopes or other techniques. Participants and researchers were blinded preoperatively; however, it was not possible or planned for either the patient or the surgeon to be blinded after the procedure because of the obvious difference in the surgical incisions; therefore, there was a potentially high risk of performance bias. Only one study[33] had attempted to blind researchers who collected all the preoperative and postoperative data. All RCTs demonstrated complete outcome data, and the radiographic review was blinded. The risk of bias summary and graph are shown in Figures 2 and 3.

Figure 2.

Risk of bias graph for randomized controlled trials included studies.

Figure 3.

Risk of bias summary for randomized controlled trials included studies.

Twenty included studies were non-RCTs,[15–28,31,32,34–37] including 16 retrospective comparative studies[15,17–21,23,25–28,31,32,35–37] and 4 prospective comparative studies.[16,22,24,34] Seven studies received a score for 7 of 9 stars of the NOS,[16,17,20,25,26,31,32] nine studies had a score of 8,[18,19,21,22,27,28,34–36] and four articles had a score of 9.[15,23,24,37] Scores on the NOS are shown in Table 2.

Postoperative Dislocation

Five RCTs[4,5,29,30,33] and 16 of 20 non-RCTs,[15–28,31,32] adding up 6,493 patients and 62 events, provided data on dislocation rates. Follow-up duration ranged from 6 weeks to 5 years. Overall, the meta-analysis showed that there was no statistical difference in the risk of dislocation rates between the DAA group and the PA group (RD = −0.00, 95% CI: −0.01 to 0.00; P = 0.92). No significant heterogeneity was detected between the studies (P = 1.00; I2 = 0%) (Figure 4). Results were similar among the subgroup analysis performed in only RCT studies[4,5,29,30,33] (RD = −0.00, 95% CI: −0.03 to 0.02; P = 0.98; I2 = 0%) (Figure 4).

Figure 4.

Postoperative dislocation: Forest plot comparing the risk difference of dislocation after anterior approach versus PA. CI = confidence interval, DAA = direct anterior approach, M-H = Mantel-Haenszel, PA = posterolateral approach, RCT = randomized controlled trial

An analysis to determine the effect of the posterior soft-tissue repair on the dislocation rate was performed: 14 studies reported capsule or external rotators repair in the PA group.[5,16,17,19,21–24,28–33] Dislocation occurred in 8 of 1,355 patients in the DAA group and 11 of 1,233 patients in the soft-tissue repair PA group. This difference was not significant (RD = −0.00, 95% CI: −0.01 to 0.01; P = 0.50); heterogeneity was low (P = 1.00; I2 = 0%) (Figure 5).

Figure 5.

Dislocation and the effect of the posterior soft-tissue repair: Forest plot comparing the risk difference of dislocation after anterior approach versus PA with soft-tissue repair. CI = confidence interval, DAA = direct anterior approach, M-H = Mantel-Haenszel, PA = posterolateral approach

Finally, the effect of the learning curve on dislocation rates was analyzed. We include in this subanalysis 12 studies that only include THAs in which the surgeon completes the learning curve phase in the DAA group[4,5,15,19,22–24,28–30,32,33] (roughly defined by each study as >50 cases, >100 cases, or extensive experience). There was no significant difference between the groups regarding the dislocation rate (RD = 0.00, 95% CI: −0.00 to 0.01; P = 0.77; I2 = 0%) (Figure 6).

Figure 6.

Dislocation and the effect of the learning curve: Forest plot comparing the risk difference of dislocation after anterior approach in which the surgeon completes the learning curve phase versus PA. CI = confidence interval, DAA = direct anterior approach, M-H = Mantel-Haenszel, PA = posterolateral approach

Acetabular Implant Positioning

Acetabular cup inclination was reported in 18 studies.[5,14,17,18,20,21,23,24,26,28,30,33,35,37] Significant heterogeneity was observed; therefore, the random-effects model was used (I2 = 95%; P < 0.00001). The meta-analysis revealed that there was no statistical difference in cup inclination between the DAA group and the PA group (MD = −0.15, 95% CI: −1.42 to 1.11; P = 0.81) (Figure 7).

Figure 7.

Acetabular cup inclination: Forest plot comparing the mean difference of acetabular cup inclination after anterior approach versus PA. CI = confidence interval, DAA = direct anterior approach, IV = inverse variance, PA = posterolateral approach

Fourteen studies reported the cup anteversion.[5,14,17,18,20,21,23,24,26,28,30,33,35,37] Heterogeneity was significant between the studies, and the random-effects model was used (I2 = 99%; P < 0.00001). The pooled results showed no significant difference between the groups regarding the cup anteversion (MD = −2.12, 95% CI: −4.64 to 0.40; P = 0.10) (Figure 8).

Figure 8.

Acetabular cup anteversion: Forest plot comparing the mean difference of acetabular cup anteversion after anterior approach versus PA. CI = confidence interval, DAA = direct anterior approach, IV = inverse variance, PA = posterolateral approach

Data from 12 studies were available to examine the implant positioning within the ranges defined as a safe zone by Lewinnek, between 30° and 50° (40° ± 10°) of inclination and between 5° and 25° (15° ± 10°) of anteversion; data were either reported directly in the text[5,14,18,21,26,28,31,35] or extracted from published scatter plots.[17,23,24,33] The acetabular cup was positioned correctly between the safe zone in 904 of 1,355 THA in the DAA group and 744 of 1,025 THA in the PA group. The pooled results showed that the acetabular implant was better positioned in the DAA group (RR = 1.17, 95% CI: 1.03 to 1.33; P = 0.01), and heterogeneity was significantly high (P < 0.00001; I2 = 93%); therefore, the random-effect model was performed (Figure 9).

Figure 9.

Acetabular cup positioning within the Lewinnek safe zone: Forest plot comparing the risk ratio of acetabular cup positioning within the Lewinnek safe zone after anterior approach versus PA. CI = confidence interval, DAA = direct anterior approach, M-H = Mantel-Haenszel, PA = posterolateral approach

Postoperative Leg Length Discrepancy

Eight studies reported the LLD after THA.[20,24–27,30,36,37] The difference was not significant between groups (MD = −0.15, 95% CI: −0.63 to 0.33; P = 0.54) (Figure 10).

Figure 10.

Postoperative leg length discrepancy (LLD): Forest plot comparing the mean difference of LLD after anterior approach versus PA. CI = confidence interval, DAA = direct anterior approach, IV = inverse variance, PA = posterolateral approach

processing....