Direct anterior approach (DAA) vs. conventional approaches in total hip arthroplasty: A RCT meta-analysis with an overview of related meta-analyses

Objectives Several systematic reviews and meta-analyses on short-term outcomes between total hip arthroplasty (THA) through direct anterior approach (DAA) compared to THA through conventional (including anterior, anterolateral, lateral transgluteal, lateral transtrochanteric, posterior, and posterolateral) approaches (CAs) in treatment of hip diseases and fractures showed contradicting conclusions. Our aim was to draw definitive conclusions by conducting both a fixed and random model meta-analysis of quality randomized controlled trials (RCTs) and by comparison with related meta-analyses. Design We performed a systematic literature search up to May 2020 to identify RCTs, comparing THA through DAA with THA through CAs and related meta-analyses. We conducted risk of bias and level of evidence assessment in accordance with the Cochrane’s Risk of Bias 2 tool and with the guidelines of the Centre for Evidence-Based Medicine. We estimated mean differences (MD) with 95% confidence intervals (CI) through fixed and random effects models, using the DerSimonian and Laird method. Heterogeneity was assessed using tau-square (τ2). Our conclusions take into account the overall results from related meta-analyses. Results Nine studies on THA through DAA met the criteria for final meta-analysis, involving 998 patients. Three studies were blinded RCTs with a level I evidence, the other 6 studies were non-blinded RCTs with a level II evidence. We came to the following results for THA through DAA compared to THA through CAs: operation time (I2 = 92%, p<0.01; fixed: MD = 15.1, 95% CI 13.1 to 17.1; random: MD = 18.1, 95% CI 8.6 to 27.5); incision length (I2 = 100%, p<0.01; fixed: MD = -2.9, 95% CI -3.0 to -2.8; MD = -1.1, 95% CI -4.3 to 2.0); intraoperative blood loss (I2 = 87%, p<0.01; fixed: MD = 51.5, 95% CI 34.1 to 68.8; random: MD = 51.9, 95% CI -89.8 to 193.5); VAS 1 day postoperatively (I2 = 79%, p = 0.03; fixed: MD = -0.8, 95% CI -1.2 to -0.4; random: MD = -0.9, 95% CI -2.0 to 0.15); HHS 3 months postoperatively (I2 = 52%, p = 0.08; fixed: MD = 2.8, 95% CI 1.1 to 4.6; random: MD = 3.0, 95% CI -0.5 to 6.5); HHS 6 months postoperatively (I2 = 0%, p = 0.67; fixed: MD = 0.9, 95% CI -1.1 to 2.9; random: MD = 0.9, 95% CI -1.1 to 2.9); HHS 12 months postoperatively (I2 = 0%, p = 0.79; fixed: MD = 0.7, 95% CI -0.9 to 2.4; random: MD = 0.7, 95% CI -0.9 to 2.4). We compared our findings with 7 related meta-analyses. Conclusions Considering the results of our meta-analysis and the review of related meta-analyses, we can conclude that short-term outcomes of THA through DAA were overall better than THA through CAs. THA through DAA had a shorter incision length, a tendency towards a lower pain VAS 1 day postoperatively and better early postoperative functional outcome than THA through CAs. The intraoperative blood loss showed indifferent results. THA through DAA had a longer operation time than THA through CAs.

Over the past decade, several systematic reviews and meta-analyses were conducted to reveal differences in outcomes between THA through DAA compared to THA through CAs [26][27][28][29][30][31][32][33][34]. Unfortunately, their conclusions were partly contradicting. While some of those meta-analyses found an advantage of THA through DAA compared to other approaches [27-29,31-33], the other meta-analyses came to different conclusions [26,30,34]. Furthermore, only one of those systematic reviews and meta-analyses considered the TT utilization in THA through DAA [28]. Science needs further high quality research on this subject in order to be able to draw a definitive conclusion.
Our first objective was to compare short-term outcomes of THA through DAA and THA through CAs in treatment of hip diseases and fractures by performing systematic literature review and both a fixed and random model meta-analysis of quality RCTs. Our second objective was to compare our results with related meta-analyses in order to draw definitive conclusions.

Reporting guidelines and protocol registration
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis-Protocols (PRISMA-P) guidelines [35]. The review protocol was registered retroactively with the International Prospective Register of Systematic Reviews (PROSPERO) on 28 January 2021 and finally approved on 28 February 2021 (CRD42021233481) at http://www.crd.york.ac.uk/ PROSPERO/

Search methodology
We built a BOOLEAN search strategy (see S1 Appendix) and adapted it to the syntax of the used databases. Results of the searches were exported to a reference management software [36]. Our search was performed in the following databases: PubMed, Google Scholar, The Cochrane Library, Clinical trials. Furthermore, we checked citations of screened studies and reviews for additional records. The search continued up to May 2020. We searched those databases for related meta-analyses and checked their references for relevant studies.

Study selection
The process was performed in two stages. Two independent reviewers (NR and PL) screened titles and abstracts to identify articles for further consideration. The full text of the selected articles were obtained and screened again by the two reviewers (NR and PL) according to inclusion criteria. Disagreements were resolved by consensus. Kappa coefficient was used to measure the agreement between the reviewers.

Inclusion criteria and outcomes
Inclusion criteria were: randomized controlled trials (RCTs) with no restriction to language and publication date; studies which compared outcomes in THA through DAA and THA through CAs. We included human participants with hip disease or hip fracture. We did not include studies comparing outcomes in THA through DAA and THA through mini-incision approaches as well as surgical techniques using of a computer navigation system. In the studies concerned, we relied on the authors' assessment that a mini-incision approach was used. The types of measured outcomes were: 1. Surgical outcomes: • The operation time in min. was defined as the period of time from the beginning of skin incision to surgical closure.
• The incision length in cm was measured on graduated scale.
• The intraoperative blood loss in ml was the total amount of blood from the suction device.

Pain Visual Analogue Scale
• The pain Visual Analogue Scale (VAS) was an instrument for measuring pain intensity, providing a range of scores from 0 to 10 points [37,38].

Functional outcome:
• The Harris Hip Score (HHS) was developed for assessment of the results of hip surgery [39]. The hip joint function was evaluated at periodically time intervals after operation. The score collects points from the assessment of four aspects: pain, function, degree of deformity and range of motion of the hip. The higher the added score, the better the results, providing a range of added scores from 0 to 100 points.

Radiological outcomes:
• The prosthesis cup abduction angle and the anteversion angle have ideal values for positioning: abduction angle from 40˚to 50˚and anteversion angle from 10˚to 25˚ [40]. Especially, the ideal cup anteversion is of great importance, since a too large angle often leads to anterior dislocation and a too small angle often leads to posterior dislocation.

Data extraction and analysis
We extracted data on study characteristics, methods, quality assessment, on characteristics of participants, on details of the interventions, and on measured outcomes into a standard electronic spreadsheet and the Cochrane software program Review Manager Version 5.3 [41]. We contacted the authors for missing data. In some cases relevant data was still missing, so the corresponding study was excluded in order to guarantee a high quality inclusion of RCTs.

Risk of bias and level of evidence
Our risk of bias and level of evidence assessment were performed in accordance with the Cochrane's Risk of Bias 2 (RoB 2) tool [42] and with the guidelines of the Centre for Evidence-Based Medicine (Oxford, UK) [43]. Furthermore, we considered the publication year in risk of bias assessment, since it was shown that publication bias is smaller in meta-analyses of more recent studies [44].

Statistical analysis
Measures of treatment effect. DAA represented the "experimental group" and CAs represented the "control group". Mean differences (MDs) with 95% confidence intervals (CIs) were estimated through fixed and random effects models for all outcomes. A common τ 2 was assumed for calculation of the random effects estimates, using the DerSimonian and Laird method. Study weighting was performed by inverse variance [45]. We evaluated the results and analysed them on basis of the Cochrane Handbook for Systematic Reviews of Interventions [46], R packages meta [47] and metafor [48].
Assessment of heterogeneity. We did not pool study data that were clinically too diverse. Heterogeneity was assessed using tau-square (τ 2 ), which followed a distribution with k-degrees of freedom (p value < 0.10 is indicative of heterogeneity), and a Higgins' test I 2 (low heterogeneity, < 25%; moderate heterogeneity, 25-75%; and high heterogeneity, > 75%) [49].
Interpretation of our results and comparison between our study and related meta-analyses. Our meta-analysis offers both a fixed and a random effects model which obviously increases the statistical value and enriches our study. Nevertheless, in interpretation of our results we stuck to the common scientific understanding that random effects models are more conservative and provide better estimates with wider confidence intervals [50,51]. Wherever both effects models offered different results, the random effects model was preferred. In addition, a simple comparison of our results with the results of the related meta-analyses was performed. In case that overall results of all meta-analyses cumulated in the same direction, we tried to draw a definitive conclusion. So, our conclusions take into account the overall results from all meta-analyses.
TT-subgroup analysis. We performed a TT subgroup-analysis in order to examine whether the utilization of a TT in THA through DAA leads to different short-term outcomes.

Study identification and selection
We searched PubMed, Google Scholar, The Cochrane Library, Clinical trials and found 3,248 studies. Additional records were not found during manual searches of reference lists. We removed 324 duplicates, resulting in a total of 2,924 studies in initial literature search. 38 studies were assessed for eligibility after first screening procedure by title and abstract (κ = 0.95) with disagreement between the reviewers concerning 2 studies. The remaining 38 studies were read in full, and 29 were excluded according to inclusion criteria (κ = 1.0). Four of those studies were excluded because they did not provide any information on standard deviation of the outcome parameters examined [52][53][54][55]. A total of 9 studies on THA through DAA met the criteria for final meta-analysis [56][57][58][59][60][61][62][63][64]. Details of study identification, screening, and selection are given in a PRISMA flow diagram (Fig 1). Table 1 gives an overview of the main characteristics of the 9 included RCTs. The main preoperative diagnoses were osteoarthritis, femoral neck fracture and avascular necrosis of the femoral head. The 9 studies, comparing T'HA through DAA with THA through CAs, were published between 2009 and 2019, altogether involving 998 patients (with 1002 operated hip joints). 440 were operated through DAA and 558 through CAs. The sample size of the 9 studies ranged from 46 to 169 patients and they were published in English language. Of the 9 studies, 4 included conventional THA through posterolateral approach [56,57,63,64], 5 through lateral transgluteal approach [58][59][60][61][62]. Two studies reported to have used TT in THA through DAA [56,57]. Only one study included patients with bilateral THA [64].

Risk of bias and level of evidence
Three of 9 studies were rated with a low risk of bias [57,60,64], 3 studies with a moderate risk of bias [59,61,62] and 3 studies with a high risk of bias [56,58,63]. Table 2 shows the summarized risk of bias assessment. Three out of 9 studies were blinded RCTs with a level I evidence [57,60,64], the other 6 studies were non-blinded RCTs with a level II evidence [56,58,59,[61][62][63].

Related systematic reviews and meta-analyses
The characteristics of the corresponding systematic reviews and meta-analyses are listed in Table 3. They were published from 2015 to 2019, they included 9 to 88 studies. One of them was a systematic review and network meta-analysis [30], 6 systematic reviews and meta-analyses [26-29,31,32], and 2 systematic reviews [33,34]. The results of short-term outcomes of THA through DAA compared to other approaches by the related meta-analyses are listed in Table 4.   Fig 3). The incision length of THA through DAA was 2.9 cm shorter than the incision length of THA

PLOS ONE
Direct anterior approach (DAA) vs. conventional approaches in total hip arthroplasty Table 4. Short-term outcomes of THA through DAA compared to other approaches by related meta-analyses.

Operation time Results
Our study fixed effects model 15 Fig 4). The intraoperative blood loss of THA through DAA was 51.5 ml higher than the intraoperative blood loss of THA through CAs, using a fixed effects model (MD = 51.5, 95% CI 34.1 to 68.8). There was no difference in intraoperative blood loss, using a random effects model (MD = 51.9, 95% CI -89.8 to 193.5).

PLOS ONE
Direct anterior approach (DAA) vs. conventional approaches in total hip arthroplasty
Traction table subgroup analysis. A TT was used in two [56,57] out of 9 studies on THA through DAA. The utilization of TT in THA through DAA showed no influence on the overall effect of the DAA group.
Discussion of the main findings. We performed a direct comparison on short-term outcomes between THA through DAA and CAs. The meta-analysis was conducted using a fixed and a random effects model for all outcomes. Our meta-analysis included 9 RCTs and 998 patients. We assessed the risk of bias of the RCTs included using the Cochrane's Risk of Bias 2 (RoB 2) tool [42]. Furthermore, we considered the publication year in risk of bias assessment [44]. Three out of 9 studies were RCTs with a low risk of bias [57,60,64], 3 out of 9 were RCTs with a moderate risk of bias [59,61,62], 3 out of 9 were RCTs with a high risk of bias [56,58,63]. We assessed the level of evidence according to the guidelines of the Centre for Evidence-Based Medicine (Oxford, UK) [43]. Three out of 9 studies were blinded RCTs with a level I evidence [57,60,64], and the other 6 studies were non-blinded RCTs with a level II evidence [56,58,59,[61][62][63].
THA through DAA showed a longer operation time, a shorter incision length and a higher intraoperative blood loss than THA through CAs. THA through DAA showed a better pain VAS score 1 day postoperatively than THA through CAs. THA through DAA showed a higher HHS 3 months postoperatively compared to THA through CAs. The subsequent HHS 6 and 12 months postoperatively showed no difference. Both approaches showed overall sufficient results in acetabular cup positioning.
Overview of related meta-analyses on THA through DAA. All meta-analyses on THA through DAA had their strengths and weaknesses. The differences between them helped us to get an overall fuller picture of the comparison between THA through DAA and THA through CAs.
The 2015 meta-analysis by Higgins et al.
[26] compared THA through DAA with THA through posterior approach. This meta-analysis included 17 studies with a total of 2,302 patients. Unfortunately, this meta-analysis included only 2 RCTs, 10 retrospective studies, and 5 non-randomized prospective studies. The inclusion of studies with a high risk of bias and a The 2019 meta-analysis by Kucukdurmaz et al.
[28] compared THA through DAA with THA through other approaches, including only RCTs. This meta-analysis included 18 RCTs with a total of 1,661 patients. This meta-analysis found overall more RCTs than our meta-analysis. Unfortunately, navigated THA and THA through mini-incision approaches were pooled in one group with THA through CAs. Furthermore, two relevant recent RCTs were not included in final meta-analysis [57,62].
The 2018 meta-analysis by Miller et al.
[29] compared THA through DAA with THA through posterior approach. This meta-analysis included 13 prospective studies with 524 patients treated with THA through DAA and 520 patients treated with THA through posterior approach. Unfortunately, this meta-analysis included only 7 RCTs and a short follow-up period of 90 days postoperatively. Furthermore, some of the outcome parameters were pooled from studies that measured those parameters at different times. In this meta-analysis [29] a sensitivity analysis was performed which showed that the overall results remained unchanged when assessing only RCTs. Nevertheless, the same first author published a further meta-analysis two months later [65], including exactly the same 7 RCTs that were used in their sensitivity analysis before [29,65].
The Total hip arthroplasty through direct anterior approach compared to conventional approaches-trying to draw a definitive conclusion. Operation time. Our meta-analysis found that the operation time through DAA was 15.5 min. longer than the operation time through CAs, using a fixed effects model and 19.1 min. longer, using a random effects model. The 2015 meta-analysis by Higgins et al. [26] found no differences in operation time between THA through DAA and THA through posterior approach. The meta-analysis by Jia et al. [27] found that THA through DAA had a 13 min. longer operation time than THA through posterior approach. The meta-analysis by Kucukdurmaz et al. [28] found that THA through DAA had a 7 min. longer operation time than THA through other approaches. The meta-analysis by Miller et al. [29] did not investigate operation time. The meta-analysis by Wang et al. [31] showed no difference between THA through DAA and posterior approach in operation time. The meta-analysis by Yue et al. [32] showed that THA through DAA had an 8 min. longer operation time than THA through lateral approach. In summary, we can conclude with the following: THA through DAA had longer operation time than THA through CAs, but a similar operation time compared to THA through posterior approach.
Incision length. Our meta-analysis showed a 2.9 cm shorter incision length of THA through DAA than THA through CAs, using a fixed effects model. There was no difference in incision length, using a random effects model. In summary, we can conclude with the following: THA through DAA had an approximately 3.0 cm shorter incision length than THA through CAs.
Intraoperative blood loss. Our meta-analysis showed that the intraoperative blood loss of THA through DAA was 51.5 ml higher than the intraoperative blood loss of THA through CAs, using a fixed effects model. There was no difference in intraoperative blood loss, using a random effects model. The meta-analysis by Higgins et al. [26] showed no difference in intraoperative blood loss between THA through DAA and THA through posterior approach. [32] showed no difference in postoperative blood transfusion rates between THA through DAA and THA through lateral approach. In summary, we can conclude that there is no relevant difference in intraoperative blood loss of THA through DAA and THA through CAs.
VAS 1 day postoperatively. Our meta-analysis showed a 0.8 point lower pain VAS 1 day postoperatively in THA through DAA than THA through CAs, using a fixed effects model. There was no difference in pain VAS 1 day postoperatively, using a random effects model. [32] did not meta-analyse pain VAS postoperatively. In summary, the overall results regarding pain VAS postoperatively were very contradictive. So, we can conclude with the following: THA through DAA showed a tendency towards lower pain VAS 1 day postoperatively than THA through CAs.
Harris Hip Score. Our meta-analysis showed a 2.8 points higher HHS 3 months postoperatively in THA through DAA than THA through CAs, using a fixed effects model. There was no difference in HHS 3 months postoperatively, using a random effects model. The subsequent HHS 6 and 12 months postoperatively showed no difference, using a fixed and a random effects model. showed that THA through DAA had a 5.6 points higher HHS 1.5 months postoperatively than THA through other approaches. Further, this meta-analysis compared the WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) Score 1.5 months postoperatively. In this score a lower result is interpreted as better and a higher result as worse in contrast to the HHS. Kucukdurmaz et al. [28] found that THA through DAA had a 3.1 points lower WOMAC Score 1.5 months postoperatively than THA through other approaches. The meta-analysis by Miller et al. [29] showed an overall 0.3 point higher HHS for THA through DAA than THA through posterior approach. The HHS was measured at different times in a 90 days postoperative follow-up. The network meta-analysis by Putananon et al. [30] showed that THA through DAA had a better functional outcome than THA through lateral, posterior and posterior-2 incision approach. THA through DAA had a 2.6 points higher HHS 1-1.5 months postoperatively than THA through lateral approach, a 4.8 points higher HHS 1-1.5 months postoperatively than THA through posterior approach and a 10.8 higher HHS 1-1.5 months postoperatively than THA through posterior-2 incision approach. The HHS of THA through DAA, measured at the latest time in the different follow-up period of the included studies, was 6.9 points higher than THA through lateral approach, 2.4 points higher than THA through posterior approach and 4.4 higher than THA through posterior-2 incision approach. The meta-analysis by Wang et al. [31] found that THA through DAA had a 7.4 and 6.8 points higher HHS, respectively, 0.5 and 1.5 months postoperatively than THA through posterior approach. [32] found no difference in acetabular cup inclination and anteversion angles between THA through DAA and THA through lateral approach. The acetabular cup anteversion angle varied from 21.8˚-24˚degrees in THA through DAA and the acetabular cup inclination angle varied from 37.6˚-47˚in THA through DAA. Analysing the acetabular cup anteversion angle, we have to emphasize that this is a very questionable outcome parameter since this angle was measured in almost every study in conventional radiographs. The acetabular cup anteversion angle can only be measured reliably with a CT-scan [66]. In summary, we can conclude that THA through DAA showed sufficient results in acetabulum cup positioning with a slight tendency towards a too flat inclination angle.
Postoperative complications. In our meta-analysis we could not include enough data on postoperative complications such as dislocation, infection, periprosthetic fracture, pulmonary embolism, infection, wound healing problems and heterotopic ossifications. Nevertheless, this is a very important outcome parameter and it was interesting what other meta-analyses had found so far. The meta-analysis by Higgins et al. [26] showed better results in postoperative dislocation for THA through DAA than THA through CAs. The meta-analysis by Jia et al. [27] investigated the following postoperative complications: dislocation, lateral cutaneous nerve of the thigh neuropraxia, intraoperative fractures. This meta-analysis found that THA through DAA had worse results in lateral cutaneous nerve of the thigh neuropraxia and intraoperative fractures than THA through posterior approach. The meta-analysis by Kucukdurmaz et al.
[28] investigated the following postoperative complications: infection, wound healing problems, neurovascular damage, fracture, thrombosis, dislocation, component malpositioning, heterotopic ossification and death. This meta-analysis found no difference in the overall complication rates between THA through DAA and THA through other approaches. The metaanalysis by Miller et al. [29] investigated the following postoperative complications over 90 days of follow-up: dislocation, fracture, hematoma, infection, thromboembolic event, and reoperation. This meta-analysis found no differences in postoperative complication between THA through DAA and THA through posterior approach. The network meta-analysis by Putananon et al.
[30] compared the following postoperative complications in THA through DAA, lateral, posterior and posterior-2 incision approaches: dislocation, infection and fracture. This meta-analysis showed better results for THA through posterior approach followed by THA through DAA, lateral and posterior-2 incision approaches. The meta-analysis by Wang et al. [31] showed no differences between THA through DAA and THA through posterior approach in terms of intraoperative fracture, postoperative dislocation, and heterotopic ossification. The meta-analysis by Yue et al. [32] showed that no differences between THA through DAA and THA through lateral approach in terms of dislocation, intraoperative fracture, superficial infection, deep infection, and postoperative hematoma. This meta-analysis found that THA through DAA had a higher risk of lateral cutaneous nerve of the thigh palsy. More quality RCTs and meta-analyses are required to draw a conclusion regarding postoperative complications. So far we can conclude that THA through DAA had a higher risk of damage of the lateral cutaneous nerve of the thigh.
Traction table utilization. In subgroup analysis our study showed that there was no influence of TT utilization on outcomes in THA through DAA. A recent 2020 systematic review on DAA by Sarraj included 44 studies with a total of 26,353 patients [67]. The study found no relevant difference in outcome between TT versus standard table THA through DAA.

Limitations
We identified the following limitations to our study: First, the long-term outcomes in THA were not considered. Second, due to insufficient data, important outcome parameters such as hospitalization time, postoperative drainage volume and postoperative complications could not be considered. Third, we did not consider the possible influence of the surgeon operating skills, the utilization of tranexamic acid and anticoagulants, bone cement or the types of implants for hip replacement. Fourth, our findings include THA as both elective and traumatic surgery. Fifth, we excluded studies comparing DAA with mini-incision approaches. The point behind our intention in excluding mini-incision approaches was to enable a clear comparison between a purely minimally invasive DAA and purely conventional approaches. Nevertheless, this might have influenced our findings. Sixth, our meta-analysis did not include enough studies to assess publication bias tests and plots [68], which present a major limitation to our study. Lastly, CAs were summarized in one group, although they differ greatly from one another.

Conclusions
Considering the results of our meta-analysis and the review of related meta-analyses, we can conclude that short-term outcomes of THA through DAA were overall better than THA through CAs. THA through DAA had shorter incision length, a tendency towards lower pain VAS 1 day postoperatively and better early (up to 3 months) postoperative functional outcome than THA through CAs. The intraoperative blood loss showed indifferent results. THA through DAA had a longer operation time than THA through CAs.