Background Intra-articular injection is effective for osteoarthritis, but the best single injection strategy is not known, nor are there established predictors of response. The objectives of this study were to assess and predict response to a single ultrasound-guided injection in moderate to severe hip osteoarthritis.
Methods 77 hip osteoarthritis patients entered a prospective, randomised controlled trial, randomised to one of four groups: standard care (no injection); normal saline; non-animal stabilised hyaluronic acid (durolane) or methylprednisolone acetate (depomedrone).
Main Outcome Measures Numerical rating scale (NRS 0–10) ‘worst pain’, Western Ontario and McMaster Universities Arthritis Index (WOMAC) pain/function. Potential predictors of response (including radiographic severity, ultrasound synovitis and baseline symptom severity) were examined using univariate logistic regression analysis and Fisher's exact test.
Results NRS pain, WOMAC pain and function improved significantly for the steroid arm alone. Effect sizes at week 1 were striking: NRS pain 1.5, WOMAC pain 1.9 and WOMAC function 1.3. Outcome Measures in Rheumatoid Arthritis Clinical Trials—Osteoarthritis Research Society responder criteria identified 22 responders (intention-to-treat): steroid 14 (74%; number needed to treat, two); saline, four (21%); durolane, two (11%); and no injection, two (10%; χ2 test between groups, p<0.001). Corticosteroid arm response was maintained over 8 weeks (summary measures analysis of variance, p<0.002 for NRS pain). Synovitis was a significant predictor of response at weeks 4 and 8 (p<0.05, Fisher's exact test; week 4 OR 16.7, 95% CI 1.4 to 204).
Conclusions Ultrasound-guided corticosteroid injections are highly efficacious; furthermore synovitis on ultrasound is a biomarker of response to injection.
Statistics from Altmetric.com
Osteoarthritis is a common and increasing cause of disability worldwide, with a prevalence of 1–10%,1 and the number of total hip replacements (THR) for osteoarthritis has increased even more rapidly than expected.2 3 Recent guidelines have been published by the National Institute for Health and Clinical Excellence (NICE)4 5 and the Osteoarthritis Research Society International (OARSI)6 for the management of osteoarthritis and have demonstrated a lack of potent management options short of surgery for those refractory to first-line treatment. Hip injection is one potential option, but is not widely used; the reasons may include lack of local service provision and uncertainties regarding patient selection.
Hip joint injection is undertaken using guidance by either fluoroscopy or ultrasound. Although there are no published data on blind hip injection, more superficial joints (including knee or shoulder) have a high inaccuracy rate (29–48%) with blind injection,7 providing an upper estimate of the likely performance of blind hip injection. Ultrasound guidance has numerous advantages to fluoroscopic guidance. These include no requirement for ionising radiation or a contrast agent and the identification of diagnostic features, such as soft tissue structures or effusion.8 Technological advances have made portable machines more powerful and cheaper so these are now widely available and easily accessible to physicians for procedures such as central venous access.9 We have already shown that competence in performing ultrasound-guided injection can rapidly be achieved by a novice using a flexible training strategy.10
Recent trials have demonstrated benefits of hip injection with corticosteroid, and to a lesser degree hyaluronan.11,–,15 However, these studies used fluoroscopic guidance,11 12 or multiple injection strategies,13 with significant logistical and cost implications for routine use. Also, a single injection of a high molecular weight hyaluronan has not been assessed. Finally, there are no studies demonstrating clinical, radiographic or high resolution musculoskeletal ultrasound predictors of response to injection therapy in hip osteoarthritis, or indeed, in other joints.
Our objectives were, therefore, to assess the efficacy and duration of an objective improvement in pain, function and patient global assessment in patients with moderate to severe hip osteoarthritis after a single ultrasound-guided injection of corticosteroid or high molecular weight hyaluronan, with robust control groups; and to determine the clinical, radiographic and ultrasound predictors of response.
Approval for this study was obtained from the local research ethics committees. All participants in the study gave written informed consent. The study was conducted according to the principles of good clinical practice.
Patients with primary hip osteoarthritis were identified from primary and secondary care within the area served by Northumbria Healthcare NHS Foundation Trust. Inclusion criteria included age 50 years or greater, unilateral hip osteoarthritis, pain duration of more than a month, and either listed for elective THR or warranting consideration for THR (using New Zealand priority scoring16 with a minimum score of 20 as described in supplementary file 1, available online only). The exclusion criteria were: secondary hip osteoarthritis (different underlying pathology, eg, rheumatoid arthritis); total loss of joint space or collapse of femoral head on anteroposterior pelvic radiograph; co-morbid conditions resulting in gross lower limb asymmetry (eg, stroke, amputees, severe leg shortening) or mobility impairment; hip injection within 6 months; listing for bilateral THR; combination of hip and back pain with the primary source of the pain unclear. All participants met the American College of Rheumatology criteria for hip osteoarthritis.17
Randomisation and treatment groups
Randomisation was conducted using computer-generated numbers by the Research Randomizer system (http://www.randomizer.org). Participants were randomly assigned in a 1:1:1:1 ratio into one of four groups: standard care (non-injection group); normal saline (3 ml); non-animal stabilised hyaluronic acid (durolane, 3 ml/60 mg licensed for single injection) or methylprednisolone acetate (depomedrone, 3 ml/120 mg). All the ultrasound assessments and injections were performed by one researcher (IA), who was blind to the allocation group during the screening, recruitment and baseline assessments. Subjects were, of necessity, aware of allocation to injection, but blinded to the injection given throughout the study. All patients had ultrasound examination of the hip joint (Titan; Sonosite Inc, USA), before injection, if allocated to an injection group. Measurement of the bone to capsule distance was made in all: mean measurements of over 7mm are accepted as indicative of synovitis.18 In those allocated injection, the hip capsule was infiltrated with an aseptic technique, under direct ultrasound visualisation. Injections were performed using a 20G spinal needle, under local anaesthetic, with 2 ml 1% lidocaine and 3 ml active or control injection within the capsule—all groups therefore received the same total volume: 5 ml. Distension of the capsule provided evidence of adequate localisation. Patients were recommended to rest and avoid weight-bearing activities for 24 h but strict bed rest was not recommended.
Outcome measures and follow-up
All patients had a full musculoskeletal examination including range of movement of both hips and the New Zealand priority score performed at screening. All patients had anteroposterior pelvis radiographs within 6 months of recruitment into the study, which were scored for Croft grade and minimal joint space by a single experienced observer (FB) with established reliability.19 Information was obtained about other treatments and in particular medication use. There were no restrictions regarding medication use, but participants were requested to notify changes in medication during follow-up.
The primary outcome measure was the numerical rating scale (NRS) score for worst pain. The secondary outcome measures were the Western Ontario and McMaster Osteoarthritis Index (WOMAC) pain and function global scores,20 patient global assessment and Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT)–OARSI responder criteria.21 Adverse events were monitored throughout follow-up and safety monitoring continued after surgery for those listed for THR. Assessments were performed at baseline, 1 week, 4 weeks, 8 weeks and when possible, 16 weeks post-injection.
The power calculation, based on detecting a difference of 2.0 on the NRS 0–10 pain scale and assuming a SD of 1.9, indicated that 19 participants per group would be required (80% power, assuming type 1 error of 5%).
A prespecified summary measures analysis for area under the curve was performed to detect differences between the groups over 8 weeks.22 Between-groups comparisons were performed using parametric analysis of variance and non-parametric (χ2) techniques, as appropriate. Potential predictors of response (including radiographic severity, ultrasound evidence of synovitis and baseline severity of symptoms) were examined using univariate logistic regression analysis and Fisher's exact test. All analyses were performed using SPSS version 15.0 software package.
The study cohort consisted of 34 men and 43 women; mean age 69 years (SD 8). The demographics and other baseline characteristics are shown in table 1. The four groups were well matched for all key variables; 84% of participants had Croft grade 3 or 4 confirming moderate to severe hip osteoarthritis. The mean bone to capsule distance was 8.7 mm (accepted normal value up to 7 mm), but only three participants in the injection groups had any fluid that could be aspirated at the time of injection.
When recruitment for the study began, national government initiatives reduced the average time for elective THR locally to approximately 3 months. This both affected the recruitment rate and reduced the power of the week 16 analysis; therefore data are presented to week 8. The recruitment period was from August 2005 to March 2007, with follow-up of participants to August 2007.
The CONSORT flowchart for recruitment and follow-up of participants from primary and secondary care is illustrated in figures 1 and 2. Forty-three of the 77 participants were from secondary care. Follow-up rates at weeks 1 and 4 were very high, with only one drop out from the standard care arm. At week 8, there were three further drop outs from the standard care arm and one from each of the steroid and durolane groups.
Baseline and follow-up outcome measures values (mean with 95% CI) are shown in figure 3 and in supplementary table 1, available online only. The groups were matched for all key outcome measures at baseline. There were significant differences at the follow-up, with improvements in the steroid group for the primary outcome measures sustained over the 8-week period, using the prespecified summary measures analysis: p values of 0.002, 0.003 and 0.009 (analysis of variance) for NRS pain, WOMAC pain and WOMAC function, respectively.
The number of (OMERACT–OARSI) responders was significantly higher (χ2 test between groups, as shown in supplementary figure 1, available online only) in the steroid group sustained to 8 weeks, with 14 responders at week 1. The effect size (calculated as the mean change from baseline divided by the baseline SD) of the benefit from corticosteroid was large, becoming moderate by week 8: 1.5, 1.0 and 0.5 for NRS pain; 1.9, 1.1 and 0.6 for WOMAC pain; and 1.3, 0.9 and 0.4 for WOMAC function, at weeks 1, 4 and 8, respectively. There was a non-significant trend for improvement with saline injection at week 1 (effect size of 0.44 and 0.41 for NRS pain and WOMAC pain, respectively), which was not apparent at later assessments.
Univariate analysis of potential clinical, radiographic and ultrasound predictors of response revealed synovitis to be the single predictor of steroid response at weeks 4 and 8, as shown in table 2 (p<0.05, Fisher's exact test; week 4 OR 16.7, 95% CI 1.4 to 204); this predictive value was not explained by age and gender effects in an exploratory analysis, although group sizes were too small for multivariate logistic regression.
Ninety-eight per cent of patients were taking medications, with 48% taking non-steroidal anti-inflammatory drugs and 84% taking paracetamol or paracetamol/weak opioids. There were no differences between the groups at baseline, or changes reported by subjects during the trial. However, as formal table counts were not undertaken, or diaries completed at follow-up, changes in medications may have been underreported.
There was one case of femoral head collapse in a patient occurring in the standard care group (who therefore had not received an injection), confirmed by radiograph at week 6. There was one confirmed case of post-arthroplasty infection (durolane group). Four patients in the durolane group had significant flare of symptoms within a few days of the injection. In one case septic arthritis had to be excluded and a diagnostic aspiration was performed at week 1, which excluded infection. The symptoms settled within a few days of the aspiration. The ultrasound-guided hip injection procedure was very well tolerated by participants.
This study has demonstrated the benefit of a single injection of intra-articular steroid over an 8-week period for patients with advanced hip osteoarthritis on a waiting list/eligible for THR. The outcome measures are validated and patient centred. This is the first randomised controlled trial (RCT) comparing a single ultrasound-guided injection of steroid with high molecular weight hyaluronan (also known as viscosupplementation). Previous studies have also demonstrated the efficacy of steroid, some for up to 12 weeks. The magnitude (ES 0.6) and duration of effect was limited with a single injection of 40 mg methyl prednisolone,13 whereas there was evidence of dose-effect comparing 40 mg and 80 mg, in a non-randomised study14 and sustained effect to 3 months using 40 mg triamcinolone hexacetonide.12 This may be explained by the lower solubility of this corticosteroid preparation (available in the USA but not in the UK), which would be expected to prolong intra-articular half-life and thus the duration of action.
Ultrasound-guided injection was at least as effective as fluoroscopic-guided injection.12 Apart from the advantages to the patient, the ultrasound-guided technique is more likely to be adopted by rheumatologists or even primary care physicians with a musculoskeletal interest due to its suitability for a clinic setting, cost and diagnostic value. Indeed, an increasing number of physicians have access to ultrasound, used in particular for guidance of central venous access or injection.
The high prevalence of ultrasound evidence of synovitis and the high proportion of responders to corticosteroid injection suggest that inflammation is responsible for a significant and reversible burden of pain and function. If, as seems likely, this inflammation is a driver for structural progression, guided injections may have disease-modifying potential, although a repeated injection strategy has not been tested. Suppression of inflammation has been suggested as one of the most promising strategies for preventing progression in osteoarthritis.23
There is some support for this hypothesis from our analysis of predictors, which is the first demonstration of a biomarker predicting response to therapy in hip osteoarthritis, as far as we are aware. The bone to capsule distance is a relatively crude marker of synovitis, but the ultrasound equipment (Sonosite Titan), was chosen for portability and feasibility in a clinic setting and does not have the same resolution or facilities for power Doppler demonstration of inflammation available on state of the art scanners. Nonetheless, we have previously demonstrated the reproducibility of this measurement and the feasibility of a flexible, modular training strategy,10 and any misclassification from the limited resolution would be expected to reduce the ability of the study to demonstrate predictive value.
One study has looked at the predictive value of musculoskeletal ultrasound in knee osteoarthritis,24 in which the presence of effusion, synovitis or power Doppler were not predictive of response to injection. However, the severity of disease was not equivalent to this study and knee osteoarthritis may well be a different disease, which may not respond in the same way to steroid injection. The KIVIS study, comparing steroid injection with tidal irrigation in knee osteoarthritis, demonstrated that clinical evidence of knee effusion was a predictor of the duration of response but not the magnitude of response, but the reliability of clinical assessment for effusion is uncertain and ultrasound assessment was not included in the study.25
There was no demonstrable improvement from the injection of a high molecular weight hyaluronan in our study cohort. Meta-analyses of hyaluronans have failed to demonstrate a consistent benefit, predominantly based on studies of knee osteoarthritis patients. The studies demonstrating benefit of viscosupplementation in hip osteoarthritis were all open label and/or included early osteoarthritis.26,–,29 One RCT in hip osteoarthritis, of three ultrasound-guided injections of hyaluronic acid (hyalgan) demonstrated a possible small benefit, but did not measure the effect of multiple injections by including a non-injection arm.13 A 6-month RCT of hip osteoarthritis after a single fluoroscopically guided injection of a moderately high molecular weight hyaluronan (adant) demonstrated no improvement in the hyaluronan group over a control injection.15 Similarly, the lack of benefit in our randomised and robustly controlled study is likely to be real (even though our study was not powered to detect a response of small effect size) and suggests that even high molecular weight hyaluronans do not have a current role to play in moderate to severe hip osteoarthritis, although this study did not address early or mild disease. The magnitude of response from normal saline at week 1 is in accordance with published data on the placebo effect in hip osteoarthritis,30 and the short duration of response may be due to participants expecting their symptoms to worsen as they get closer to THR.
Potential limitations of our study include the selection of moderate to severe disease and recruitment from both primary and secondary care—if randomisation had led to differential allocation to the groups, this may have either underestimated or overestimated the treatment effects. However, allocation was matched and response found for both recruitment sources (data not shown), suggesting that internal validity was not compromised by this. The screening of a large number of these potential subjects and the inclusion of only a small proportion might also affect the external validity (or generalisability) of the findings; although the subject characteristics are consistent with the National Joint Replacement Registry data from the trust and England and Wales31 suggesting that the subjects were representative. Histological analysis of biopsies of the joint capsule might have been useful to confirm that the increased bone to capsule distance was not due to non-inflammatory thickening of the capsule. A double blind methodology would have been ideal, but a metrologist was not available, and the main outcomes are all self-report, and blinding of the patient to receiving an injection (rather than the substance injected) is not feasible. The inclusion of four arms has allowed an accurate estimate of both the specific and non-specific effects of each therapy, however, and the non-specific effects of injection were neither significant nor sustained. Finally, there are concerns about the risk of postarthroplasty infection following steroid injection, even though there is no clear evidence either confirming this risk or determining a time threshold.32 The practice in the unit presently is to wait 3 months for surgery following an injection.
In conclusion, musculoskeletal ultrasound can predict long-term response to corticosteroid in moderate to severe hip osteoarthritis and should therefore be used both to inform and guide effective therapy. This is the first RCT to demonstrate ultrasound as a biomarker of response to injection, in addition to a large, clinically important response to corticosteroid injection. This novel observation justifies the use of musculoskeletal ultrasound in both observational and interventional studies of osteoarthritis of the hip and other joints. It also provides further evidence of an inflammatory basis for moderate to severe hip osteoarthritis.
The authors acknowledge the help from Dr Nick Steen for statistical analysis, the orthopaedic surgeons Mr Mike R Reed, Mr Paul Partington, Mr Lester Sher, Mr John Leitch, Mr Simon Jones, Mr A Murty, Mr G Prasad, Mr Ian Carluke, at Northumbria Healthcare NHS Foundation Trust and General Practitioners who assisted recruitment.
Funding IA's fellowship was funded by Northumbria Healthcare NHS Foundation Trust. This work was supported by the UK NIHR Biomedical Research Centre for ageing and age-related disease award to the Newcastle upon Tyne Hospitals NHS Foundation Trust. Durolane for injection was supplied by Q-Med.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Newcastle and Northumberland local research ethics committees (Ref NLREC 54/2002, NNT SSA 05/Q0905/155).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.