A meta-analysis on RCTs of direct anterior and conventional approaches in total hip arthroplasty

Nikolai Ramadanov; Simon Bueschges; Philip Lazaru; Dobromir Dimitrov

doi:10.1038/s41598-021-00405-4

. 2021 Oct 25;11:20991. doi: 10.1038/s41598-021-00405-4

A meta-analysis on RCTs of direct anterior and conventional approaches in total hip arthroplasty

Nikolai Ramadanov ^1,^✉, Simon Bueschges ², Philip Lazaru ³, Dobromir Dimitrov ⁴

PMCID: PMC8546071 PMID: 34697357

Abstract

To conduct a systematic review and meta-analyses on short-term outcomes between total hip arthroplasty (THA) through direct anterior approach (DAA) compared to THA through conventional approaches (CAs) in treatment of hip diseases and fractures. We performed a systematic literature search up to March 1, 2021 to identify RCTs, comparing THA through DAA with THA through CAs. We calculated mean differences (MDs) with 95% confidence intervals (CIs) for continuous outcomes, using the DerSimonian and Laird method and a random effects model. We calculated odds ratios (ORs) with 95% CIs for dichotomous outcomes, using the Mantel–Haenszel method and a random effects model. Ten RCTs met the criteria for final meta-analysis, involving 1053 patients. Four studies were blinded RCTs with a level I evidence, the other 6 studies were non-blinded RCTs with a level II evidence. DAA had a longer operation time than CAs (MD = 17.8, 95% CI 4.8 to 30.8); DAA had similar results compared to CAs for incision length (MD = − 1.1, 95% CI − 4.1 to 1.8), for intraoperative blood loss (MD = 67.2, 95% CI − 34.8 to 169.1), for HHS 3 months postoperatively (MD = 2.4, 95% CI − 0.7 to 5.5), for HHS 6 months postoperatively (MD = 0.8, 95% CI − 1.9 to 3.5), for HHS 12 months postoperatively (MD = 0.9, 95% CI − 0.7 to 2.5), for pain VAS 1 day postoperatively (MD = − 0.9, 95% CI − 2.0 to 0.15), for acetabular cup anteversion angle (MD = − 4.3, 95% CI − 5.2 to − 3.5), for acetabular cup inclination angle (MD = − 0.5, 95% CI − 2.1 to 1.1) and for postoperative complications (OR = 2.4, 95% CI 0.5 to 12.4). Considering the overall results of our meta-analysis, we can conclude that THA through DAA showed similar short-term surgical, functional, radiological outcomes and postoperative complications compared to THA through CAs.

Subject terms: Rehabilitation, Fracture repair

Introduction

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 approaches (CAs)^1–9 came to contradicting conclusions. While some of them found an advantage of DAA^2–4,6–8, others showed indifferent results^1,5,9. Most of those meta-analyses had relevant limitations. The 2015 meta-analysis by Higgins et al.¹, comparing DAA with posterior approach, included 17 studies (2 RCTs, 10 retrospective, and 5 non-randomized prospective studies) with a total of 2302 patients. The 2019 meta-analysis by Jia et al.², comparing DAA with posterior approach, included 20 studies (4 RCTs, 3 prospective studies and 13 retrospective studies) with a total of 7377 patients. The 2015 meta-analysis by Yue et al.⁷, the first meta-analysis comparing DAA with lateral approach, included 12 studies (2 RCTs and 10 non-RCTs) with 2,991 cases of THA through DAA and 1,910 cases of THA through lateral approach. Unfortunately, the inclusion of studies with a low level of evidence in these meta-analyses^1,2,7 affects the validity of the end point estimates and overall results. The 2019 meta-analysis by Kucukdurmaz et al.³, comparing DAA with other approaches, included 18 RCTs with a total of 1,661 patients. Only this meta-analysis considered the influence of traction (fracture) table utilization in THA through DAA³. Unfortunately, navigated THA and THA through mini-incision approaches were pooled in one group, which presents a severe confounding factor. Furthermore, three relevant recent RCTs^10–12 were not included in final meta-analysis. The 2018 meta-analysis by Miller et al.⁴, comparing DAA with posterior approach, included 13 prospective studies (7 RCTs and 6 non-RCTs) with 1,044 patients. This meta-analysis had a very short follow-up period of 3 months. Furthermore, some outcome parameters were grouped together, but measured at very different time points. The 2018 meta-analysis by Wang et al.⁶, the first meta-analysis comparing DAA with posterior approach, included 9 RCTs with a total of 854 patients. It was performed with high quality methods. The 2018 meta-analysis by Putananon et al.⁵, the first network meta-analysis comparing outcomes of different surgical approaches in THA, included 14 RCTs with a total of 1,201 patients. It investigated the outcomes of THA through DAA, lateral, posterior and posterior-2-incision approaches. This network meta-analysis limited their investigation to three outcome parameters: VAS, HHS and postoperative complications. Posterior and posterior-2-incision approaches were pooled based on only two studies. The systematic reviews by Kyriakopoulos et al. and Meermans et al. did not perform a meta-analysis^8,9. In order to overcome these limitations of the related previous systematic reviews and meta-analyses^1–9, a new meta-analysis was required with a more extensive literature search, with restriction to RCTs and with implementation of high-quality statistical methods.

Our aim was to compare short-term outcomes of THA through DAA and THA through CAs in treatment of hip diseases and femoral neck fractures by performing a systematic literature review and meta-analysis of RCTs.

Methods

Literature search and study selection

Our review protocol was registered and approved with PROSPERO on 28 February 2021 (CRD42021233481). We followed the PRISMA-P guidelines to report our meta-analysis. Our search was performed up to March 1, 2021 in the following databases: PubMed, Embase, Web of Science, The Cochrane Library, SCOPUS, CINAHL, Clinical Trials. We used a BOOLEAN search strategy [(DAA) OR (direct anterior approach)], adapted to the syntax of the used databases. Furthermore, we checked citations of screened studies and Google Scholar for additional records. Titles and abstracts were screened to identify articles for further consideration, full text of the selected articles were screened again according to inclusion criteria. The study selection was determined by the consensus of two independent reviewers (NR and PL). Kappa coefficient was used to measure the agreement between the reviewers.

Inclusion/exclusion criteria and outcomes

Inclusion criteria were as follows: RCTs with no restriction to language and publication date, comparing outcomes in THA through DAA and THA through CAs. We included human participants with the following hip pathologies: osteoarthritis, avascular necrosis of the femoral head, dysplasia, femoral neck fracture. We excluded studies on THA through mini-incision approaches as well as surgical techniques, using a computer navigation system. The types of measured outcomes were the following: 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. The pain Visual Analogue Scale (VAS)^13,14 ranges from 0 to 10 points, with the pain intensity increasing with the number of points. Hip pain was evaluated at day 1 after surgery. The Harris Hip Score (HHS)¹⁵ ranges from 0 to 100 points, with hip function increasing with the number of points. It was evaluated at periodic time intervals after surgery. The acetabular cup anteversion and the inclination have ideal values for positioning: anteversion angle from 10° to 25° and inclination angle from 35° to 45°.

Data extraction and study quality assessment

We extracted data on study characteristics, methods, quality assessment, characteristics of participants, details of the interventions, and measured outcomes into a standard electronic spreadsheet. We contacted corresponding authors for missing data. In case that relevant data was still missing, the concerning study was excluded in order to guarantee a high quality inclusion of RCTs. Risk of bias and level of evidence assessment were performed according to the Cochrane’s Risk of Bias 2 (RoB 2) tool¹⁶ and the guidelines of the Centre for Evidence-Based Medicine¹⁷.

Statistical analysis

DAA represented the “experimental group” and CAs represented the “control group”. We tested both fixed and random effects models. The random effects model provided more reliable results, so we proceeded as follows: We calculated mean differences (MDs) with 95% confidence intervals (CIs) for continuous outcomes, using the DerSimonian and Laird method and a random effects model. We calculated odds ratios (ORs) with 95% CIs for dichotomous outcomes, using the Mantel–Haenszel method and a random effects model. A common τ² was assumed for calculation of the random effects estimates. Study weighting was performed by inverse variance¹⁸. We evaluated the results and analysed them on basis of the Cochrane Handbook for Systematic Reviews of Interventions¹⁹. We did not pool study data that were clinically too diverse. Heterogeneity was assessed using a test on Cochrane’s Q statistic, which followed a distribution with k-degrees of freedom (p value < 0.10 is indicative of heterogeneity), and a Higgins’ test I² (low heterogeneity, < 25%; moderate heterogeneity, 25–75%; and high heterogeneity, > 75%)²⁰. We performed a traction table subgroup-analysis in order to examine the influence of traction table utilization in THA through DAA.

Results

Study identification and selection

After removal of 824 duplicates, a total of 7,024 studies were found in initial literature search. Forty-one studies were assessed for eligibility after first screening procedure by title and abstract with disagreement between the reviewers concerning one study (κ = 0.98). Thirty-one studies were excluded after full-text screening 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^21–24. A total of 10 studies on THA through DAA met the criteria for final meta-analysis^{10–12,25–31}. Details of study identification, screening, and selection are given in a PRISMA flow diagram (Fig. 1).

PRISMA flow diagram of the search results and selection according to our inclusion criteria. *DAA* direct anterior approach, *RCTs* randomized controlled trials.

Characteristics of the RCTs

Table 1 gives an overview of the main characteristics of the 10 included RCTs, published between 2009 and 2020, altogether involving 1,053 patients with 1,057 operated hip joints. Of these, 468 were operated through DAA and 585 through CAs. The sample size of the 10 studies ranged from 46 to 169 patients, all published in English language. Of the 10 studies, 4 included conventional THA through posterolateral approach^10,25,30,31, one through posterior approach¹², 5 through lateral transgluteal approach^11,26–29. Three studies reported to have used traction table in THA through DAA^10,12,25. Only one study included patients with bilateral THA³¹.

Table 1.

Main characteristics of RCTs included in network meta-analysis.

Study	Sample size, n		Surgical approach		Mean age, y (SD or range)		Gender (M/F), n		BMI, kg/m² (SD or range)		Hip pathology				Follow-up period, months	Origin of study
Study	Pts	Hips	DAA	CAs	DAA	CAs	DAA	CAs	DAA	CAs	Osteoarthritis	ANFH	Dysplasia	Femoral neck fracture	Follow-up period, months	Origin of study
Barrett et al. 2013	87	87	43TT	44 pl	61,4 ± 9,2	53,2 ± 7,7	29/14	19/25	30,7 ± 5,4	29,1 ± 5	87	–	–	–	12	USA
Bon et al. 2019	100	100	50 TT	50 pl	67,26 ± 10	68,98 ± 7,93	21/29	23/27	26,46 ± 3,58	26,69 ± 3,12	100	–	–	–	3	France
D'Arrigo et al. 2009	169	169	20	149 l	64 ± 8	65 ± 9,8	12/8	81/68	22,7 ± 1,5	28 ± 1,8	169	–	–	–	1,5	Italy
De Anta-Diaz et al. 2016	99	99	49	50 l	63,5 ± 12,5	64,8 ± 10,1	26/23	26/24	26,9 ± 3,1	26,6 ± 3,9	99	–	–	–	12	Spain
Mjaaland et al. 2015	163	163	83	80 l	67,2 ± 8,6	65,6 ± 8,6	25/58	30/50	3,6 ± 1,9	27,6 ± 3,9	163	–	–	–	< 1	Norway
Moerenhout et al. 2020	55	55	28 TT	27 p	70,4 ± 9,1	68,9 ± 8,8	11/28	18/27	26,5 ± 4,3	27,6 ± 4,4	55		–	–	55	Canada
Nistor et al. 2017	70	70	35	35 l	67	64	26/9	16/19	27,45 ± 3,76	38,63 ± 3,12	70	–	–	–	3	Romania
Reichert et al. 2018	148	148	77	71 l	63,2 ± 8,2	61,9 ± 7,8	45/32	71/0	28,1 ± 3,7	28,3 ± 3,4	148	–	–	–	12	Germany
Rykov et al. 2017	46	46	23	23 pl	62,8 ± 6,1	60,2 ± 8,1	8/15	11/12	29 ± 5,6	29,3 ± 4,8	46	–	–	–	1,5	Netherlands
Zhao et al. 2017	116	120	60	56 pl	64,9 ± 12,1	62,2 ± 14,7	24/36	22/34	24,35 ± 3,1	25,58 ± 2,83	81	26	13	–	6	China

Open in a new tab

DAA direct anterior approach, TT traction table, CAs conventional approaches, pl posterolateral approach, p posterior approach, l lateral approach, Pts patients, ANFH aseptic necrosis of the femoral head.

Risk of bias and level of evidence

Four out of 10 studies were rated with a low risk of bias^10,12,28,31, 4 studies with a moderate risk of bias^11,26,27,29 and 2 studies with a high risk of bias^25,30 as shown in Table 2. Four out of 10 studies were blinded RCTs with a level I evidence^10,12,28,31, the other 6 studies were non-blinded RCTs with a level II evidence^{11,25–27,29,30}.

Table 2.

Risk of bias assessment.

Study	Random sequence generation	Allocation concealment	Blinding	Complete outcome data	No selective reporting	No other sources of bias	Overall risk of bias
Barrett et al. 2013	Y	N	N	U	Y	U	High RB
Bon et al. 2019	Y	Y	Y	Y	Y	Y	Low RB
D'Arrigo et al. 2009	Y	Y	U	Y	Y	U	Moderate RB
De Anta-Diaz et al. 2016	Y	Y	U	Y	Y	Y	Moderate RB
Mjaaland et al. 2015	Y	Y	Y	Y	Y	Y	Low RB
Moerenhout et al. 2020	Y	Y	Y	Y	Y	Y	Low RB
Nistor et al. 2017	Y	Y	U	Y	Y	U	Moderate RB
Reichert et al. 2018	Y	Y	U	Y	Y	U	Moderate RB
Rykov et al. 2017	Y	Y	N	Y	Y	U	High RB
Zhao et al. 2017	Y	Y	Y	Y	Y	Y	Low RB

Open in a new tab

DAA direct anterior approach, Y Yes, U Unclear, N No, RB risk of bias.

Outcomes

Operation time

Data on 672 patients were pooled from 7 RCTs (I² = 90%, p < 0.01, Fig. 2). The operation time of THA through DAA was 17.8 min. longer than the operation time of THA through CAs (MD = 17.8, 95% CI 4.8 to 30.8).

Incision length

Data on 372 patients were pooled from 4 RCTs (I² = 100%, p < 0.01, Fig. 2). There was no difference in incision length (MD = − 1.1, 95% CI − 4.1 to 1.8).

Intraoperative blood loss

Data on 418 patients were pooled from 4 RCTs (I² = 87%, p < 0.01, Fig. 2). There was no difference in intraoperative blood loss (MD = 67.2, 95% CI − 34.8 to 169.1).

HHS

Data on 505 patients were pooled from 5 RCTs (I² = 10%, p = 0.35, Fig. 3). There was no difference in HHS 3 months postoperatively (MD = 2.4, 95% CI − 0.7 to 5.5). Data on 406 patients were pooled from 4 RCTs (I² = 0%, p = 0.82, Fig. 3). There was no difference in HHS 6 months postoperatively (MD = 0.8, 95% CI − 1.9 to 3.5). Data on 389 patients were pooled from 4 RCTs (I² = 0%, p = 0.80, Fig. 3). There was no difference in HHS 12 months postoperatively (MD = 0.9, 95% CI − 0.7 to 2.5).

Pain VAS 1 day postoperatively

Data on 250 patients were pooled from 2 RCTs (I² = 79%, p = 0.03, Fig. 3). There was no difference in pain VAS 1 day postoperatively (MD = -0.9, 95% CI − 2.0 to 0.15).

Acetabular cup positioning

Data on 203 patients were pooled from 2 RCTs (I² = 0%, p = 0.34, Fig. 4). The acetabular cup anteversion angle of THA through DAA was 4.3° lower than the acetabular cup anteversion angle of THA through CAs (MD = − 4.3, 95% CI − 5.2 to − 3.5). Data on 576 patients were pooled from 6 RCTs (I² = 83%, p < 0.01, Fig. 4). There was no difference in acetabular cup inclination angle (MD = − 0.5, 95% CI − 2.1 to 1.1).

Postoperative complications

Data on 394 patients were pooled from 4 RCTs (I² = 53%, p = 0.10, Fig. 4). There was no difference in postoperative complications (OR = 2.4, 95% CI 0.5 to 12.4).

Traction table subgroup analysis

A traction table was used in three^10,12,25 out of 10 studies on THA through DAA. The utilization of traction table in THA through DAA showed no influence on the overall effect of the DAA group.

Discussion

Main findings

In order to overcome the limitations of the related previous systematic reviews and meta-analyses^1–9, we conducted our meta-analysis with a more extensive literature search, with restriction to RCTs and with implementation of high-quality statistical methods, testing both fixed and random effects models. Our direct comparison in meta-analysis on short-term outcomes between THA through DAA and CAs included 10 RCTs and 1,053 patients. Four out of 10 studies were blinded RCTs with a level I evidence^10,12,28,31, the other 6 studies were non-blinded RCTs with a level II evidence^{11,25–27,29,30}. THA through DAA showed overall similar surgical, functional, radiological outcomes and postoperative complications compared to THA through CAs. Both approaches showed overall sufficient results in acetabular cup positioning. THA through DAA showed a longer operation time compared to THA through CAs. We found no influence of traction table utilization on outcomes in THA through DAA.