Arthroscopic hip surgery offers better early patient-reported outcome measures than targeted physiotherapy programs for the treatment of femoroacetabular impingement syndrome: a systematic review and meta-analysis of randomized controlled trials.

Targeted physiotherapy programs (TPP), and surgery, using either open surgical hip dislocation or hip arthroscopy (HA), are the treatment modalities available for femoroacetabular impingement syndrome (FAIS). Randomized controlled trials have recently been performed to compare these treatment options. This review was performed to provide a focused synthesis of the available evidence regarding the relative value of treatment options. A systematic search was performed of Medline, Embase, Cochrane Library and ClinicalTrials.gov databases. Inclusion criteria were randomized controlled trials comparing treatment methods. The Cochrane Risk of Bias assessment tool (RoB2) was used to assess the selected studies. A meta-analysis was performed between homogenous studies. Four trials were identified including 749 patients (392 males). The mean ages of the cohorts ranged between 30.1 and 36.2 years old. Three hundred thirty-five patients underwent HA by 46 surgeons among all trials. Fifty-two patients crossed over from the TPP to the HA group. One of the trials was found to have a high risk of bias, while the other three were between low risk and some concerns. The iHOT-33 was the most commonly used patient-reported outcome measure followed by the HOS ADL and EQ-5D-5L. Others scores were also identified. Scores from two trials could be pooled together for meta-analysis. Apart from SF-12 and GRC, all other scores have shown significantly better outcomes with HA in comparison to TPP at 8- and 12-months follow-up points. HA offers better patient-reported outcomes than TPP for management of FAIS at 8- and 12-months follow-up.

Introduction

The term femoroacetabular impingement (FAI) was initially coined by Myers et al. in 1999 as a complication secondary to periacetabular osteotomy surgery [1]. Primary FAI morphology was identified in asymptomatic individuals [2], and many theories have been proposed to provide an explanation for this anatomical variation. These included unrecognized mild slippage of the capital femoral epiphysis (SCFE) [3] and excessive stresses placed on the growing physis by sports participation [4].The term FAI Syndrome (FAIS) is used broadly to describe a triad of symptoms, signs and imaging in association with any abnormal structural conflict between the femoral head (and/or neck) and the acetabulum [5]. This can be due to (i) incomplete sphericity of the femoral head (cam-type morphology or pistol grip deformity), (ii) excessive coverage of the femoral head by the acetabular margin or the periacetabular bone (pincer-type lesion) or (iii) a combination of both (i and ii).

Patients usually present with symptoms, including, but not limited to, groin pain limiting many daily activities, decline in sports efficiency and painful hip flexion. Mechanical and/or painful limitation of hip flexion and internal rotation is the principal finding in clinical examination. Loss of adduction and internal rotation was confirmed by biomechanical studies [6].

Imaging studies of asymptomatic individuals have demonstrated cam and pincer morphology in between 10% and 67% [2, 7]. This may suggest that the pathology in patients with symptoms is not just about the bony shape but may include unsatisfactory muscle strength, control and movement patterns. There have been several reports [8-11] of targeted physiotherapy programs (TPP) that seek to improve these latter factors. In one study, half of FAIS patients benefited from TPP in the short term, but the presence of cam morphology was a predictor of poor outcome [12]. Most of these programs have included activity restriction [8, 10], but this is not always an acceptable treatment modality.

Surgical bone reshaping can correct the structural component of the pathology. Initially, open surgery was the only available option for the correction of impingement. However, the last two decades have witnessed a huge development in hip arthroscopic techniques. These have been employed in the treatment of FAIS and associated pathologies and reported good outcomes, return to sports and restoration of range of movement [13-16].

Until recently, few studies compared physiotherapy to hip arthroscopy (HA) for the treatment of FAIS [17-20]. The aim of this review is to provide a systematic review of the current level-I evidence comparing physiotherapy to the arthroscopic treatment of FAIS.

Methods

Our proposed research question was: ‘Does arthroscopic hip surgery produce better patient-reported outcomes than targeted physiotherapy in the treatment of patients with FAIS?’

A preliminary search was performed to not only examine the availability of articles that address the question but also to avoid duplication of previously published reviews. The review was registered on PROSPERO, the National Institute of Health Research-funded International prospective register of systematic reviews [21].

Search strategy

A systematic review was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [22]. Each of the online published databases of Medline and EMBASE was searched individually through the OVID Online research platform. The Cochrane Library and ClinicalTrials.gov were searched through their websites. Search dates for Medline were from 1946 and for Embase from 1974 to 20 August 2021. The MeSH terms searched were ‘Femoroacetabular’, ‘Impingement’ AND ‘Trials’. Search terms were ‘Exploded’ and relevant subheadings were included. The terms were then combined to produce lists of results. Finally, a search was performed using the Medline powered PICO (Population, Intervention, Comparison and Outcome) Linguist [23], and an additional concise list of studies was created on 25 August 2021.

Eligibility criteria

Inclusion criteria were based on the PICO approach [24] as P: Adult Humans with Femoroacetabular impingement Syndrome, I: Hip Arthroscopy, C: Physiotherapy (targeted physiotherapy programs) and O: Outcome Measures (e.g. iHot-12 and 33, HOS, etc.). The second consideration was the study design, being exclusive to Level-I Randomized Controlled Trials (RCTs) pertinent to the study question. Exclusion criteria were non-human studies, those published in a language other than English, all non-level-1 studies and published protocols of studies that were not executed by the date of the search.

Study selection and critical appraisal

Search results were presented in the form of lists of titles and abstracts. Two authors (SSSM and AT) scanned the lists independently to extract eligible studies and reported reasons for exclusion. The risk of bias of each of the included trials was assessed by the Cochrane risk-of-bias tool (RoB 2) [25]. In case of disagreement, the final decision was made by discussion with the senior author (JOD).

Data extraction

Full texts of eligible studies were downloaded, data extracted and used to synthesize results.

Results and Review

Search results

Figure 1 shows a diagrammatic presentation of the PRISMA guided search strategy and its results. The electronic search yielded a total of 456 records. Exclusion criteria were applied and only four published RCTs were identified, which provided specific answers to the question posed by the review [17-20]. We identified a published protocol of an RCT that is still recruiting. The study is comparing arthroscopic femoral and/or acetabular osteochondroplasty to sham surgery, and it was excluded [26].

Fig. 1.

PRISMA flow diagram showing the search strategy and results.

Centers, patients, surgeons and physiotherapists

Three of the included trials were multicenter studies: two from the United Kingdom [18, 19] and one from Australia [17]. One single-center study was American [20]. A total cohort of 749 patients were investigated, the majority of which were males with mean age ranging between 30.1 and 36.2 years old. Three hundred thirty-five patients underwent HA out of 370 originally allocated for the procedure. The largest difference between allocation and actual surgery was noted in one study [20]. In the physiotherapy groups, 46 patients did not receive their allocated treatment (46 out of 349 allocated). Two studies contributed to the majority of this difference in the TPP group [17, 20]. A single surgeon operated on all cases in the single-center trial, whereas 45 surgeons in total performed the surgeries in the other three. Those studies specified that the surgeons were specialist HA surgeons. TPP were delivered by multiple centers and physiotherapists in each of the trials apart from one where PT was delivered in a single center [20]. Numbers of physiotherapists involved in each trial are listed in Table I. Details of the TPP programs are listed in Table II.

Table I.

Study centres, location, patients and surgeons

Author/trial/year	Country	Journal	Sample size (n)	Mean age [SD] (years)	Male event (%)	Recruitment centres	Surgeons (n)	Physiotherapists (n)
Hunter et al. Australian FASHION. 2021	Australia	BMC Musculoskeletal	99	32.9[SD 10.5]	58	10 Australian sites	8	24 PT centres (number of physiotherapists: NA)
Palmer et al. FAIT, 2019	United Kingdom	BMJ	222	36.2[SD 9.7]	34	7 UK NHS sites	10	21
Griffin et al. UK FASHION, 2018	United Kingdom	Lancet	348	35.3[SD 9.6]	61.2	23 UK NHS sites	27	47
Mansell et al. 2018	United States	Am J Sp Med	80	30.1[SD 7.4]	58.8	Single Army Medical Centre	1	Single centre (number of physiotherapists: NA)

Table II.

Methodology: assessment tools, physiotherapy program details, allocations and follow-up

Study	Outcomes	PT program details	Study arms (allocated/intention-to-treat)- HA TPP	Last follow-up (months)
Hunter et al.	Primary: dGEMRICSecondary:- HOMAS Hip OA MRI- iHOT-33- HOOS- EQ-5D- SF-12- GIS- Modified UCLA activity score	- 6 PT sessions during the first 12 weeks of the trial - If needed, additional sessions between 12 weeks and 6 months, up to a total maximum of 10 sessions - Further PT sessions were allowed as a co-intervention. - Ultrasound guided intra-articular injection should the pain prevents engagement with the program	dGEMRIC:49/27 50/26PROMs:47/49 50/50	12
Palmer et al.	Primary: HOS ADLSecondary: - HOS sport - NAHS - HAGOS - OHS - iHOT-33 - EQ-5D-3L - PainDETECT - HADS	- Emphasis on muscle strengthening to improve core stability and movement control - Participants were encouraged to avoid impingement positions (extremes of hip flexion, abduction and internal rotation). - A maximum of eight sessions over a 5-month period	110/99 112/91	8 (at least 6 months following treatment)
Griffin et al.	Primary: - iHOT-33 Secondary: - EQ-5D-5L - SF-12 (version 2)	- 6–10 face-to-face PT sessions over 12–24 weeks - Muscle strengthening home program - One X-ray or ultrasound guided injection when the pain prevents performance of the exercise program	171/171 177/174	12
Mansell et al.	Primary: - HOS ADL Secondary: - iHOT-33 - Perception of improvement on GRC	- Supervised program, twice per week for 12 sessions - Typically, program will include hip mobilization and therapeutic exercises. - However, other interventions can be applied based on the discretion of the treating physiotherapist.	40/38 40/12	24

dGEMRIC: delayed Gadolinium-Enhanced Magnetic Resonance Imaging of Cartilage, HOMAS: whole Hip Osteoarthritis MRI Score, iHOT-33: International Hip Outcome Tool-33, HOOS: Hip disability and Osteoarthritis Outcome Score. GRC: Global Rating of Change,.

EQ-5D: European Quality of Life Five Dimension scores, SF-12: 12-Item Short Form Survey, GIS: Global Improvement Scale, Modified UCLA: Modified University Carolina Los Angeles activity score, HOS ADL: Activities of Daily Living domain of the Hip Outcome Score, NAHS: Non-arthritic hip score, HAGOS: Copenhagen hip and groin outcome score, OHS: Oxford hip score, HADS: Hospital anxiety and depression score.

Cross over

Fifty-two participants crossed over between the treatment groups among the trials reviewed. All incidents were for patients crossing over from TPP to HA except for four FAIT Study participants [19].

Critical appraisal

The results of RoB2 assessment [25] are listed in detail in Table III. It is expected that blinding of patients and personnel delivering interventions in such studies is impossible. Nevertheless, the post-intervention assessors were blinded in all trials.

Table III.

Risk of bias of the included studies

Study/bias	Arising from the randomization process	Due to deviations from intended interventions	Due to missing outcome data	In measurement of the outcome	In selection of the reported result	Overall risk of bias
Hunter et al. (primary outcome)	Low	High	High	High	Some concerns	High
Hunter et al. (secondary outcomes)	Low	Some concerns	Low	Low	Low	Some concerns
Palmer et al.	Low	Low	Low	Low	Low	Low
Griffin et al.	Low	Low	Low	Low	Low	Some Concerns
Mansell et al.	Low	Some concerns	High	Some concerns	High	High

Mansell et al. [20] was the first randomized controlled trial to address the question and concluded that patients did not benefit from surgery compared to a TPP; however, there was a high risk of bias on the RoB2 assessment. Firstly, the study is a single-center, single-surgeon trial. Secondly, the study participants were all military personnel, a specific patient group; hence, the results cannot be generalized to the public. In addition, there is an added factor of workers’ compensation, which is known to be associated with worse outcomes in HA [27], as military personnel are entitled to workers’ compensation. By the end of the follow-up interval, 45% (33 patients) of the patients had left the military service, of whom 72.7% (24 patients) had received a medical separation related to their hips. Thirdly, there was a very high crossover rate (70%) from the TPP group to the HA group leaving only 12 patients (of a total cohort of 80 patients) receiving TPP. The analysis was still performed according to the initial randomization on an intention-to-treat basis. In their discussion, the authors confirmed that the high crossover rate greatly increased the risk of type-II error (erroneous negative finding). Finally, there were 18 patients (22.5%) lost at the final follow-up. The reported results were based on two follow-up points for most of the patients; these could be 6 or 12 months. The author did not specify which follow-up points were used, and the analysis assumed the lost data to be at random.

Both the FAIT [19] and UK FASHION [18] trials showed a lower risk of bias and a lower percentage of crossover between groups of 5% and 8%, respectively.

In the FAIT trial, the end-point follow-up was at 8 months from randomization (or at least 6 months following the intervention). 80% of the TPP group completed their 8-months follow-up, while in the HA group, the percentage was 89%.

In the UK FASHION trial, the follow-up was at 6 and 12 months after randomization. The median waiting time to receive treatment was 122 days for the HA and 37 days for TPP. At 12-months follow-up, 84% of the HA group had undergone their procedure, and 95% had received their TPP.

In the most recently published Australian Fashion trial (Hunter et al.), the functional outcomes in the form of patient-reported outcome measures (PROMs) showed a lower risk of bias in all aspects. Cross over rate was only 6% (3 participants crossing from TPP to HA group). Ninety-one participants (100%) completed their 12-month follow-up scores for iHOT-33, EQ-5D-5L index and EQ-5D VAS scores, and 83 (91.2%) completed the HOOS questionnaire.

Outcome measures

All four studies employed various PROMs to assess patient’s progression. Only in the study by Hunter et al. did the investigators also use the changes in delayed Gadolinium-Enhanced Magnetic Resonance Imaging of Cartilage (dGEMRIC) as an outcome tool in order to try to compare the changes in d-GEMRIC scores between both treatment modalities at the final follow-up.

Patient-reported hip-related outcomes

iHOT-33

The iHOT-33 was the most commonly used hip outcome tool among all studies. The results of the FAIT demonstrated a significant improvement in the HA group compared to TPP group at 8-months follow-up [19].

In the study by Mansell et al. [20], there was no difference between the TPP and HA groups at any time point. However, upon analyzing patients as per-surgery, a significant improvement was noted from the baseline to 1 and 2 years follow-up in the HA group, unlike the TPP group. iHOT-33 scores from both FASHION trials were pooled together in our meta-analysis, as explained below.

On pooling the Australian and UK FASHION results, meta-analysis has shown that the HA group had significant improvement in comparison to the TPP group on both a fixed- and a random-effect model at 12 months.

HOS (ADL component)

HOS (ADL) was the second most commonly performed PROM. In the FAIT trial [19], the HA participants showed significantly higher scores than their TPP counterparts, and the author has further confirmed that the difference was clinically relevant.

The results of Mansell et al. [20] showed the HOS ADL scores of their cohort followed the same pattern of the iHOT-33 scores mentioned above.

General quality of life (EQ-5D-5L)

EQ-5D-5L was equally used by two trials. In the Australian FASHION, the scores were significantly better in the HA group (P = 0.007); however, in the UK FASHION, there was no difference (P = 0.397) between both groups. On pooling the results together, meta-analysis has shown that the HA group had significant improvement in comparison to the TPP group on a fixed-effect model at 12 months but not on a random-effect model.

Other outcome scores utilized by the trials, and their results are listed in Table IV.

Table IV.

Results of the outcome tools used by different trials

Score/study	Hunter et al. (12 months outcomes)	Palmer et al. (8 months)	Griffin et al. (12 months)	Mansell et al. (24 months)
iHOT-33	HA better than TPP (P = 0.002)	HA better than TPP (P < 0.001)	HA better than TPP (P = 0.0093)	No difference between HA and TPP
HOSADLSports		HA better than TPP (P < 0.001)HA better than TPP (P < 0.001)		No difference between HA and TPP
SF-12			No difference (P = 0.099)
NAHS		HA better than TPP (P < 0.001)
HAGOS		All scales:HA better than TPP (P < 0.001)
OHS		HA better than TPP (P < 0.001)
EQ-5D-3L		HA better than TPP (P = 0.003)
EQ-5D-5L	HA Better than TPP (P = 0.007)		No difference (P = 0.397)
HOOS - Pain - Symptom - ADL - Sport - QOL	- HA Better than TPP (P = 0.001) - HA Better than TPP (P = 0.007) - HA Better than TPP (P = 0.000) - HA Better than TPP (P = 0.003) - HA Better than TPP (P = 0.004)
Modified UCLA activity score		HA better than TPP (P = 0.01)
Perception of improvement on GRC				No difference between HA and TPP
dGEMRIC	Femoral: No difference (P = 0.240)Acetabular: No difference (P = 0.125)Combined: No difference (P = 0.137)
HOMAS	Worse cartilage score (P = 0.002) and higher number of worsened regions for labral score (P = 0.0009) for arthroscopy compared to PHT
PainDETECT		HA better than TPP (P = 0.03)
HADS - Anxiety - Depression		- No difference (P = 0.18) - HA better than TPP (P = 0.004)

Perceived improvement

On a further analysis of the HOS ADL scores, the FAIT investigators found that more HA participants reached acceptable symptomatic state (48% versus 19%) and achieved their expectations post-randomization (31% versus 15%). In the study by Mansell et al., there was no difference between both groups in their perceived current level of function on both domains of the HOS. However, a larger number of the HA group (45.2%) reported a GRC of 13 or more, in contrast to 25% of the TPP group. The Australian FASHION trial has shown significant improvement of all components of the HOOS score of the HA patients in comparison to the TPP patients. Together with the UK FASHION, they showed similar results on the EQ-5D-5L scores (HA is significantly better than TPP).

Range of movement

This was only examined by the FAIT trial, and it showed a significantly better range of hip flexion in the HA group in comparison to the TPP participants (P = 0.003).

Radiological (dGEMRIC)

The Australian FASHION Trial did not show a significant difference between HA and TPP at 12-months follow-up.

Meta-analysis

Due to the heterogeneity of the follow-up intervals and outcome tools, data could be pooled together from only two out of the four trials (Griffin et al. [18] and Hunter et al. [17]). iHOT-33 and EQ-5D scores could be pooled for analysis from both trials.

Heterogeneity between the FASHION trials was measured using the I2 statistic, with values of less than 25% usually viewed as low heterogeneity, between 25% and 50% as moderate and over 50% as high heterogeneity. Cochran’s Q test for homogeneity is underpowered when few studies have been included or when event rates are low, requiring the use of a significance level of above 5%. Since only two studies are included in this analysis, we have applied a 15% significance level in our test of homogeneity [28].

We conducted a meta-analysis using the {meta} package in R [29] assuming a fixed-effects model. Given the robust performance of this model for continuous outcome data, the restricted maximum likelihood estimator was used. For each study, the adjusted treatment effect estimates from the adjusted mixed-effects model (primary) analysis were used with the associated standard errors. An estimate of the between-study variance in a random-effects meta-analysis (known as tau-squared (τ2 or Tau2) was also carried out.

The results for the fixed-effects meta-analysis are presented in Table V and Fig. 2. For this model, no significant treatment differences in improvement were observed over 6 months. In contrast, at 12-months follow-up, significantly better improvements were observed in the HA group in comparison to the TPP participants on the iHOT-33 and EQ-5D 3L/5L Index scores (iHOT-33 95% CI = 4.30, 13.10 and EQ-5D 3L/5L 95% CI = 0.004, 0.083, P = 0.031). I2 statistics of between 40.4% and 72.5%, as well as the rejection of the Cochran’s Q test for homogeneity in most cases, suggested that a meta-analysis assuming a random-effects model should be conducted. When this analysis was performed, the HA group also performed better on the iHOT-33 scores at 12 months (95% CI= 2.31, 17.38). The results are presented in Table VI and Fig. 3.

Table V.

Meta-analysis fixed effects model results. results showing significant difference between HA and TPP are highlighted in bold

				Significance test		Adjusted difference (Standard error)
Measure	Change from baseline (months)	Overall effect	95% Confidence interval	z-value	P -value	United Kingdom	Australia
iHOT-33	6	0.708	(−3.10,4.51)	0.36	0.7155	−0.7 (2.225)	5.2 (3.975)
iHOT-33	12	8.70	(4.30,13.10)	3.88	0.0001	6.8 (2.575)	14.7 (4.575)
EQ-5D 5L VAS	6	−0.886	(−4.06,2.29)	−0.55	0.5845	−2.1 (1.775)	5.2 (3.975)
EQ-5D 5L VAS	12	3.35	(−.25,6.96)	1.82	0.0684	2.6 (1.9)	14.7 (7.375)
EQ-5D 3L/5L Index	6	−0.017	(−.053,0.019)	−0.92	0.3577	−0.042 (0.0233)	0.026 (0.0305)
EQ-5D 3L/5L Index	12	0.043	(0.004,0.083)	2.16	0.0307	0.020 (0.0235)	0.106 (0.0385)

Fig. 2.

Forest plot showing results for fixed-effects meta-analysis.

Table VI.

Meta-analysis random effects model results with significant results highlighted in bold

				Significance test		Test for heterogeneity
Measure	Change from baseline (months)	Overall effect	95% Confidence interval	z-value	p-value	I^2 (%)	tau^2	Q Statistic (df = 1)	p-value
iHOT-33	6	1.33	(−4.16, 6.82)	0.47	0.6350	40.4	7.03	1.68	0.20
iHOT-33	12	9.85	(2.31, 17.38)	2.56	0.0105	55.8	17.42	2.26	0.13
EQ-5D 5L VAS	6	0.68	(−6.27, 7.63)	0.19	0.8471	64.4	17.17	2.81	0.09
EQ-5D 5L VAS	12	6.55	(−4.57, 17.67)	1.15	0.2483	60.4	44.20	2.52	0.11
EQ-5D 3L/5L Index	6	−0.011	(−0.077, 0.056)	−0.32	0.7484	68.2	0.0016	3.14	0.08
EQ-5D 3L/5L Index	12	0.058	(−0.026, 0.141)	1.35	0.1770	72.5	0.0027	3.64	0.06

Fig. 3.

Forest plot showing results for random-effects meta-analysis.

In addition to the statistical meta-analysis, the clinical impact of the improvement in the PROMs scores should be emphasized (Table VII). Since the introduction of the iHOT-33 scoring system, its Minimal Clinically Important Difference (MCID) was found to be 6 points [30].

Table VII.

Actual values of individual scores for the iHOT-33, HOS ADL and the EQ-5D

Score/Study	Hunter et al. (Change in score at 12 months from baseline)		Palmer et al. (Change in score at 8 months from baseline)		Griffin et al. (Scores at 12 months)
	HA	PT	HA	PT	HA	PT
iHOT-33 (mean)	29.6	15.4			58.8	49.7
HOSADL (mean)			12.5(P < 0.001)	3.3
EQ-5D-5L	0.194(P = 0.0046)	0.101

In the Australian FASHION trial, the authors published the difference between 12 months follow-up and the baseline scores, while in the FAIT trial, Palmer et al. reported the end results of the HOS scores (8 months) together with the baseline values. The results above are the difference between those scores. The end results (12 months) of the iHOT scores of the UK FASHION are demonstrated above.

This threshold was exceeded in both FASHION trials [17, 18]. In the Australian FASHION trial. the authors showed improvement from baseline by 29.6 points in the HA group [17]. In the UK FASHION, Griffin et al. confirmed that MCID threshold was achieved, and they have published the final scores exceeding the Patient Acceptable Symptomatic State (PASS) of 58 [31] in the HA patient group. The PASS of the HOS ADL was calculated as 87 points for patients undergoing HA for FAIS [32]. In the FAIT trial, this value was achieved more in the HA group of patients than the TPP group. The difference was found to be statistically significant (48% in the HA group versus 19% for the TTP group) [19]. The MCID of the EQ-5D-5L (0.16) [33] was reached by the HA group, but not by the TPP group, in the Australian FASHION trial [17].

DISCUSSION

Several reviews have been performed previously, which assessed the first three of the four RCTs assessed in this review. The addition of the fourth RCT (Australian FASHION; Hunter, et al.) adds a significant cohort of 99 patients with a high follow-up rate, and very low treatment crossover rate, and therefore adds significantly to the overall data.

In addition, the very similar study protocols of the United Kingdom, and Australian FASHION studies allowed a meta-analysis to be performed for the first time using data from these published RCTs. These results were derived from a large number of surgeons and patients in two countries suggesting that the reported outcomes are likely to be generalizable.

This review demonstrates the overall superiority of hip arthroscopic surgery over a targeted physiotherapy program as the treatment strategy for patients with femoroacetabular impingement syndrome. This effect has been seen across a range of PROMs known to be sensitive and specific for the effect of hip surgery on young active individuals.

Whilst the role of surgery for this condition looks promising, there are some caveats:

The studies in this review measured outcomes at 8–12 months. This is short term in the context of young adults with most of their life still to live with the affected hip or hips. Observational studies have demonstrated good outcomes from HA at 5 and 10 years [28, 34], but we look forward to the long-term results of the randomized trials in this review.

This review has demonstrated the superiority of HA over TPP for FAIS on average. It is very unlikely that FAIS patients are homogeneous and that the effect of surgery is uniform. It may be that some groups of patients can be satisfactorily managed by TPP, and some groups should be encouraged to have early surgery. However, the published data do not yet identify the factors that influence in which group a patient lies, and so does not support individualized recommendations.

The idea that mechanical impingement is relieved by reshaping is logically consistent and is supported by the increased range of motion seen after surgery [13, 19]. However, the subsequent idea that cumulative damage to the labrum and articular cartilage is arrested or slowed has not yet been demonstrated. A difficulty has been in finding suitable imaging to measure cartilage damage. Among the possible candidates, dGEMRIC scores have been shown to predict the outcome of periacetabular osteotomy in hip dysplasia [35] and have been used to assess the condition of hips in patients with FAIS [36]. In one previously published observational study [37], dGEMRIC scans were performed in two groups of FAIS patients, treated operatively and non-operatively. The surgical group included a mix of HA and open surgery. The follow-up dGEMRIC scans showed a decline in the cartilage of both groups, more pronounced on the femoral side of the surgical dislocation group. The authors concluded that ‘The observed decline in dGEMRIC indices neither confirms the benefit of surgical treatment of FAI nor does it disprove the first reported favorable long- term results after FAI surgery in the literature. Longer term studies will be needed to determine whether the cartilage matrix changes seen here are permanent or reversible’. The Australian FASHION trial was the first attempt to use dGEMRIC in a randomized trial of FAIS treatment, but it did not show significant benefits of either arthroscopic or TPP treatments in terms of cartilage d-GEMRIC scores. This may have been due to the small number of patients who completed the two d-GEMRIC scans according to the study protocol, or perhaps the outcome measure was inappropriate as an assessment of cartilage health after surgery, or the time scale was too short, or the methodological issues overwhelmed the effect, as suggested by the previous study.

Despite these caveats, the evidence collated in this review shows that HA is a more effective strategy than prolonged TPP among FAIS patients who do not improve after initial physiotherapy assessment and treatment. HA was found to be successful in 81.1% of the FAIS patients at a minimum of 2-years follow-up [38]. Whilst targeted postoperative physiotherapy is a valuable part of a surgical strategy [39, 40], delayed surgery has been reported to be associated with significantly inferior outcomes [41] and worse intra-articular pathology [42]. This is important because FAIS is a risk factor for osteoarthritis [43-46], and so delayed surgical treatment may not only lead to prolonged symptoms but also result in earlier onset or worse degenerative change.

In conclusion, this review demonstrates that both HA and TPP lead to improved patient PROMs. On average, HA leads to significantly greater improvements than TPP in the treatment of FAIS in the short term. Future research should aim at investigating the long-term outcomes of HA for FAIS [47].