Steroid injection for osteoarthritis of the hip: A randomized, double-blind, placebo-controlled trial




To determine the efficacy of fluoroscopically guided corticosteroid injection for hip osteoarthritis (OA) in a randomized, double-blind, placebo-controlled trial.


Fifty-two patients with symptomatic hip OA were randomly allocated to receive placebo (10 mg bipuvicaine, 2 ml saline) (n = 21) or corticosteroid treatment (10 mg bipuvicaine, 40 mg triamcinolone hexacetonide) (n = 31). Patients were followed up for 1, 2, 3, and 6 months. The primary outcome measure was the pain improvement response, defined as a 20% decrease in the Western Ontario and McMaster Universities OA Index (WOMAC) pain score (on 5 100-mm visual analog scales [VAS]) (WOMAC20) from baseline to 2 months postinjection. Secondary outcomes were a 50% decrease in the WOMAC pain score (WOMAC50), changes in other WOMAC subscale scores, patient's global assessment of health (on a 100-mm VAS), and Short Form 36 (SF-36) quality of life indices. Analyses were based on the intent-to-treat principle.


The mean WOMAC pain score fell 49.2% (decreasing from 310.1 mm to 157.4 mm) at 2 months postinjection in patients receiving corticosteroid, compared with a decrease of 2.5% (from 314.3 mm to 306.5 mm) in the placebo group (P < 0.0001). The proportion of WOMAC20 responders at 2 months' followup was significantly higher in the corticosteroid group (67.7%) compared with the placebo group (23.8%) (P = 0.004); similar proportions of WOMAC50 responders were observed between groups (61.3% in the corticosteroid group versus 14.3% in the placebo group; P = 0.001). Response differences were maintained at 3 months' followup (58.1% responders in the corticosteroid group versus 9.5% responders in the placebo group; P = 0.004). Significant differences in the WOMAC stiffness and physical function scores (P < 0.0001), patient's global health scores (P = 0.005), and SF-36 physical component scores (P = 0.04) were observed, with patients in the corticosteroid group showing greater improvements. There were no differences in the frequency of adverse events between groups.


This placebo-controlled trial confirms that corticosteroid injection can be an effective treatment of pain in hip OA, with benefits lasting up to 3 months in many cases. Future studies should address questions related to the benefits of repeated steroid injection and the effects of this treatment on disease modification.

Osteoarthritis (OA) affects all vertebrates, and the prevalence of the disease increases exponentially with age. Estimates of the prevalence of hip OA in humans range from 2% at age 35 years to 35% at age 85 years or older (1), and autopsies have shown that cartilage degeneration can be detected in patients as early as the second decade of life (2). A disease prevalence estimate of 3% among North Americans older than age 55 years has been commonly quoted, but this may well be an underestimate of the true burden of hip OA in the general population (3, 4). The prevalence of the disease has significant fiscal implications, since it has been shown that populations with OA incur significantly increased direct health care costs when compared with a similar cohort without arthritis (5).

Management of OA is generally focused on symptomatic OA, and if a cause can be identified, it should be treated, when possible. Because the main symptoms are pain and restriction of activity, medical therapy for OA is directed toward alleviation of these symptoms. Patient education, weight loss, and unloading of weight-bearing joints are recommended approaches that are often helpful, and analgesia, antiinflammatory drugs, and/or physiotherapy are usually prescribed.

The role of intraarticular (IA) injection of corticosteroids (henceforth referred to as steroids) in the treatment of OA is controversial. The main indication for IA steroid use is to provide pain relief and suppression of synovitis in patients whose condition remains unresponsive to or intolerant of oral systemic medication. The major objective of this therapy is to aspirate any effusion and instill the steroid preparation that provides the most effective relief for the longest period of time.

Only 3 randomized, controlled trials of steroid injection in hip OA have been reported, with mixed results (6–8). In one trial, only a dichotomous outcome was used (6), which limits interpretation of the results; in the 2 other trials, peak effects of the steroid were observed by 2 or 3 weeks postinjection (7, 8). Despite these observations, the use of IA steroids is common in clinical practice. In 1995, the American College of Rheumatology (ACR) issued guidelines for the medical management of OA (9), which included specific recommendations for the routine use of IA steroids for knee OA. However, the hip joint is more difficult to inject than the knee, and the ACR has recommended against routine use of IA steroids in the hip, with the caveat that if this technically difficult procedure is proposed, it should be performed under fluoroscopic guidance (10).

The hip is a joint that is difficult to inject, and therefore entry of the therapeutic agent into the synovial space cannot be ensured without fluoroscopic guidance. Placement of the IA steroid (without imaging guidance) is frequently inaccurate, and this may have a significant effect on clinical response (11). Controversy persists regarding the claim that good results can be achieved without fluoroscopy (12). In addition, the European League Against Rheumatism has proposed treatment recommendations for the management of hip OA in which the use of IA injection of steroids constitutes the eighth of their 10 propositions (13). It was acknowledged that evidence in support of this proposition was not robust, and that placebo-controlled trials would be required.

The purpose of the present study was to resolve the controversy with respect to the efficacy of IA steroid injection in the management of hip OA. To accomplish this, we conducted a randomized, placebo-controlled trial in patients with hip OA whose disease was previously unresponsive to conventional medical therapy.



This randomized, double-blind, placebo-controlled clinical trial was approved by the ethics committee of the Faculty of Medicine and Dentistry at the University of Alberta. Patients with OA of the hip were referred mostly by rheumatologists and orthopedic surgeons from the Capital Health Region of Alberta. All patients provided their written informed consent.

The inclusion criteria were 1) a diagnosis of primary OA of the hip according to the ACR criteria, including radiologic evidence of OA (14), 2) age >40 years, 3) symptomatic disease for at least 6 months prior to enrollment, 4) persistent pain despite receiving the maximum tolerated doses of conventional medical therapy, including acetaminophen (4 gm/day) and/or a nonsteroidal antiinflammatory drug (NSAID), with persistent pain defined as a minimum mean score of 40 mm on the 5 visual analog scales (VAS) for pain (0–100-mm range for each) that are the first 5 questions on the Western Ontario and McMaster Universities OA Index (WOMAC) (15) comprising the WOMAC composite pain subscale, 5) daily pain during the month prior to study enrollment, 6) stable NSAID dosage for 2 weeks prior to enrollment, and 7) ability to attend followup appointments. The exclusion criteria included secondary causes of OA, local or systemic infection precluding injection, diabetes mellitus, systemic arthritis, allergy to anesthetic agent or contrast material, coagulopathy, anticoagulant therapy, previous IA steroid injection into the index hip, and avascular necrosis of bone.

Treatment groups.

Patients were randomly allocated to 1 of 2 treatment groups, either placebo or steroid treatment. Patients in the placebo group were administered a fluoroscopically guided IA injection of anesthetic and saline (10 mg [2 ml at 0.5%] bipuvicaine and 2 ml normal saline). Patients in the steroid treatment group were administered a fluoroscopically guided IA injection of anesthetic (10 mg bipuvicaine) and corticosteroid (40 mg triamcinolone hexacetonide [2 ml at 20 mg/ml]). Triamcinolone hexacetonide was chosen as the active IA treatment because its formulation is the most insoluble of the corticosteroids (16) and has been found to have the most prolonged duration of effect (17). A steroid dose of 40 mg in conjunction with a local anesthetic was the regimen chosen, in accordance with the ACR recommendations for IA use in large weight-bearing joints (9).

Randomization and blinding.

Randomization and blinding of treatment were conducted using computer-generated numbers. The study drug and the placebo were prepared by an independent research coordinator, which ensured similarity between preparations. Syringes containing either the study drug or placebo were covered with masking tape to ensure complete blinding. The independent research coordinator maintained the randomization codes in sealed envelopes. The referring physician, the physician performing fluoroscopically guided injection, the research nurse, and the patients were blinded to group assignment. Occurrence of a flare after injection did not result in unblinding of the group assignment, since the IA contrast material and the corticosteroid were equally likely to be the cause of any flare. Patients continue to remain blinded to the original treatment allocation.

Study procedures.

Within 3 weeks of screening the patients, all injections were administered by a single fellowship-trained musculoskeletal radiologist (RGWL). The patient's groin was prepared with a solution of betadine, which was allowed to dry for 3 minutes and then washed with alcohol. The area was draped with sterile towels, and disposable sterile gloves, syringes, and needles were used. The skin and periarticular soft tissue were injected with 1% lidocaine. The intrasynovial space was entered with a 22-gauge, 3.5-inch needle, and aspiration of the joint was attempted. Intrasynovial flow was established with an injection of a small dose (≤1 cc) of meglumine iothalamate (radiographic contrast material) (Figure 1).

Figure 1.

Fluoroscopically guided injection of the left hip of a 74-year-old female patient with hip osteoarthritis who was referred to the trial but did not meet entry criteria. The location of the injection into the hip joint is sensitive to subtle changes in technique. A 22-gauge needle is inserted into the hip joint and, as shown, the position of the needle tip is maintained by gentle downward pressure on the needle, ensuring contact with the femoral neck. A, The initial test injection is extrasynovial (arrowhead), after which the needle is rotated and free intrasynovial flow of contrast material is established (arrows), with the bevel of the needle facing medially. B, The needle is deliberately rotated 90° with the bevel facing superiorly, thus establishing extrasynovial flow of contrast material (arrowhead) and leading to enlargement of the localized collection; no change in the resistance to injection is experienced and the intrasynovial contrast material remains unchanged (arrows). C, The needle is rotated back to the same position as in A and intrasynovial flow is reestablished, with increased volume of contrast material in the synovial space (arrows); the localized collection remains unchanged (arrowhead). D, Finally, the corticosteroid is injected. Because it is radiolucent, the corticosteroid washes away the contrast material from the synovial space, as reflected over the front of the femur (asterisk), which increases the size and density of the contrast material medially (arrow) and leaves a clearer view of the injected extrasynovial collection (arrowhead) as depicted in C.

The joint was injected with 10 mg of bipuvicaine and either 40 mg of triamcinolone hexacetonide or 2 ml normal saline. The maximum volume of fluid injected was 5 cc, which is within the range of tolerance for diseased hip joints. In both the steroid treatment group and the placebo group, each patient was assessed for pain (on a 100-mm VAS) by a clinician nurse in the radiology department, immediately prior to and 30 minutes following injection. Four pain scores on 100-mm VAS were recorded (for walking 20 meters, standing, sitting, and lying down with leg straight; static positions were held for 30 seconds before scoring). These assessments were designed to confirm that the pain was articular in nature (18) and to determine whether an early pain response might be a predictor of the response to steroid injection (19). Following injection of the hip, the patient was instructed to rest, preferably in the form of bed rest, for 3 days and to maintain at least minimal activity. After this 3-day period, activity was restricted for 1 week, during which the patient was asked to refrain from active exercise and, if possible, to refrain from work.

Study assessments.

All assessments were performed at screening and at baseline (immediately prior to injection). Each patient was reevaluated at 1, 2, 3, and 6 months postinjection, to assess the safety and efficacy of treatment. Clinical assessments included evaluation of hip mobility and administration of questionnaires. Hip mobility was assessed by having the patients sit on a high table with legs dangling and hip and knee flexed 90 degrees. Full passive internal and external rotations of the hip were measured with a 30-cm–long goniometer while ensuring stable position of the pelvis.

Patients were asked to maintain a constant NSAID dosage throughout the trial. All medications were monitored using a medication log. Intake of analgesics was assessed by pill count, and patients were asked to maintain constant analgesic intake for 48 hours prior to each evaluation. Patients were required to bring medications to each visit and, during the trial, could not change any medication without first informing the research nurse. All adverse events were recorded by asking patients whether they had noticed “anything different or unusual” since the last visit. The adverse event, as well as the start and stop date of the event, its relationship to the study drug, and the course of action taken, were recorded. All study assessments were performed by a trained research nurse.

Study end points and outcome measures.

The primary end point for assessment of efficacy was set at 2 months postinjection. The primary outcome measure was the pain improvement response to treatment, defined (a priori by expert consensus) as a 20% decrease from baseline to 2 months postinjection in the summed score on the WOMAC pain subscale (WOMAC20). Secondary outcomes were the WOMAC pain subscale 50% response criterion (WOMAC50), scores on the WOMAC stiffness and function subscales, patient's global assessment of health (obtained with the use of a 100-mm VAS, with higher scores indicating more severe disease), Short Form 36 (SF-36) quality of life indices (20), assessment of hip internal and external rotation, and use of analgesic medication (quantified by pill count).

We also conducted post hoc analysis of the percentage of responders in each treatment group as defined according to the Osteoarthritis Research Society International/Outcome Measures in Rheumatology Clinical Trials (OARSI/OMERACT) responder criteria (21). This was reported after patient recruitment had already been initiated. Patients were defined as responders if there was high improvement (≥50% improvement response, and absolute change of ≥20 mm on 0–100-mm VAS) in pain or function. If this level of improvement was not observed, patients could still be defined as responders if there was improvement in at least 2 of the 3 criteria as follows: 1) ≥20% improvement response in the pain score, and absolute change of ≥10 mm on 0–100-mm VAS, 2) ≥20% improvement response in the function score, and absolute change of ≥10 mm on 0–100-mm VAS, and 3) ≥20% improvement response in patient's global assessment of health, and absolute improvement of ≥10 mm on 0–100-mm VAS.

Outcome measures were assessed at each visit. After the primary end point was reached, patients were allowed to receive a steroid injection at any time, according to the patient's wishes. However, for the purposes of the intent-to-treat analysis, these patients were then removed from the blinded portion of the trial and classified as a treatment failure.

Statistical analysis.

Responders at 2 months posttreatment were compared using Fisher's exact test. All analyses were based on the intent-to-treat principle, so that patients lost to followup prior to assessment at the 2-month primary end point were considered treatment failures. Comparisons of mean WOMAC pain subscale scores between treatment groups were performed using t-tests. Imputation for missing values due to a patient dropping out of the study prior to the primary end point was done using baseline values for both the primary and secondary outcomes, as opposed to using a last observation carried forward approach. The former approach was taken because of the possibility that any response to therapy among patients who dropped out prior to 2 months would be short-lived, and therefore outcome values would most likely have returned to baseline levels by 2 months. For the analysis of secondary outcomes, Fisher's exact test was used for comparisons of categorical data, and t-tests were used to compare changes in continuous variables between groups.

Demographic and clinical factors at baseline in each group were compared using chi-square or t-tests. If a significant difference was found between groups at baseline, then that variable was used as a covariate and the analysis of covariance test was applied. All comparisons were 2-sided, and P values less than 0.05 were considered significant.


Characteristics of the study patients.

Two hundred eleven patients were referred to the trial from March 2000 to September 2003, but 101 chose not to participate after they were informed of the possibility of receiving placebo. Of the 110 patients screened, 52 met all entry criteria. In view of the delayed recruitment, the ethics committee subsequently raised concerns about continuation of the study, since obvious symptomatic responses had been noted in some patients; the committee recommended that an interim analysis be performed, when 52 patients had been recruited. Since treatment group differences were highly significant, it was deemed unethical to continue further recruitment to the study beyond 4 years.

There were no significant differences between patients enrolled in the study and those excluded, in terms of age, sex, and referring physician type (results not shown). The majority of patients were women, particularly in the group receiving steroid therapy (Table 1). Patients randomized to receive placebo were significantly younger, but there were no significant differences in the time (number of months) since diagnosis of OA, concomitant therapies, extent of either internal or external rotation of the hip, and Kellgren/Lawrence grading of radiographic appearance or radiographic findings (22) (Table 1). There were also no significant differences in the patient's global health scores, WOMAC pain, stiffness, and physical function scores, and SF-36 physical component scores.

Table 1. Baseline demographic and clinical characteristics of the patients with hip osteoarthritis randomized to receive either placebo or intraarticular (IA) injection of steroid*
 PlaceboIA steroid injection
  • *

    P < 0.0005 versus placebo.

  • NSAIDs = nonsteroidal antiinflammatory drugs.

No. of patients2131
Age, mean ± SD years56.9 ± 1165.6 ± 11*
No. (%) of female patients10 (48)21 (68)
No. (%) of patients receiving NSAIDs alone2 (9.5)3 (9.7)
No. (%) of patients receiving acetaminophen/codeine alone0 (0)8 (25.8)
No. (%) of patients receiving NSAIDs plus acetaminophen/codeine18 (85.7)20 (64.5)
No. of left index hips/no. of right index hips9/1317/14
Time since diagnosis, mean ± SD months51 (56)51 (39)
Kellgren/Lawrence grade  

Data from 3 patients, all of whom had been randomized to the placebo group, were not available at the primary end point, because 1 patient had undergone hip arthroplasty 1 week prior to this time point, the second patient was evaluated at 1 month and 3 months postinjection but was unavailable for assessment at month 2, and the third patient insisted on receiving a steroid injection at the 1-month followup assessment even though a WOMAC20 response had been attained, and was subsequently lost to followup. After the primary end point, 3 patients in the placebo group withdrew as a consequence of undergoing hip arthroplasty, 8 patients (2 in the placebo group and 6 in the steroid group) were lost to followup, and 20 patients (11 in the placebo group and 9 in the steroid group) chose to receive open-label steroid injection (Figure 2). Of these 20 patients who received open-label injection, 2 patients (1 in the placebo group and 1 in the steroid group) subsequently underwent hip arthroplasty, and 1 patient (in the original steroid group) was lost to followup (Figure 3).

Figure 2.

Distribution of the clinical trial patients with hip osteoarthritis who were randomized to receive either placebo or intraarticular injection of a corticosteroid, with followup over 6 months. Responders were patients who showed a 20% decrease in the summed score for the Western Ontario and McMaster Universities Osteoarthritis Index pain subscale.

Figure 3.

Distribution of the clinical trial patients in the open-label steroid injection phase of the trial.

Efficacy assessment.

A significant difference in the primary outcome (WOMAC pain subscale score improvement response from baseline to 2 months) was noted between treatment groups, with the WOMAC pain score falling from 310.1 mm at baseline to 157.4 mm at 2 months' followup (49.2% decrease) in the steroid treatment group, and from 314.3 mm at baseline to 306.5 mm at 2 months' followup (2.5% decrease) in the placebo group (P < 0.0001) (Table 2). When the comparison was adjusted for age as a covariate (due to a significant difference in the mean age of each group at baseline), the difference in the primary outcome between groups remained significant (results not shown).

Table 2. Primary and secondary outcomes in patients randomized to receive either placebo (n = 21) or intraarticular (IA) injection of steroid (n = 31)*
OutcomePlaceboIA steroid injectionP
  • *

    WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-36 = Short Form 36; NS = not significant.

  • P values represent comparison of change from baseline between groups, by t-test for continuous variables.

  • Summed scores (in mm) were assessed on a 0–100-mm visual analog scale. Pain subscale scores ranged from 0 to 500 mm, stiffness subscale scores ranged from 0 to 200 mm, and physical function subscale scores ranged from 0 to 1,500 mm.

  Baseline314.3 ± 76.2310.1 ± 54.6 
  Month 1276.4 ± 129.0149.6 ± 113.00.0005
  Month 2306.5 ± 121.2157.4 ± 127.2<0.0001
  Baseline124.5 ± 37.7137.4 ± 33.0 
  Month 1119.8 ± 43.879.6 ± 57.30.0004
  Month 2126.8 ± 48.475.6 ± 58.10.0001
 Physical function   
  Baseline970.9 ± 254.5969.3 ± 167.8 
  Month 1897.4 ± 369.3516.0 ± 388.1<0.0001
  Month 2949.1 ± 350.4538.5 ± 402.0<0.0001
Patients' global assessment of health   
 Baseline59.0 ± 19.664.7 ± 19.2 
 Month 159.7 ± 23.540.7 ± 24.20.0001
 Month 260.2 ± 20.244.5 ± 25.90.005
SF-36 (scale 0–100)   
 Physical component   
  Baseline25.50 ± 7.0725.73 ± 5.28 
  Month 126.88 ± 9.6232.17 ± 9.900.016
  Month 226.58 ± 6.7831.01 ± 8.590.042
 Bodily pain   
  Baseline27.52 ± 11.1527.19 ± 8.97 
  Month 133.95 ± 19.4846.37 ± 19.300.019
  Month 232.29 ± 14.7943.77 ± 21.400.019
 Physical functioning   
  Baseline23.25 ± 17.5820.16 ± 13.87 
  Month 126.43 ± 22.7036.17 ± 21.840.009
  Month 222.62 ± 19.3432.62 ± 19.330.008
 Social functioning   
  Baseline55.36 ± 24.8755.65 ± 26.39 
  Month 160.12 ± 29.2172.50 ± 24.65NS
  Month 253.57 ± 24.7366.94 ± 27.870.042
Hip internal rotation, degrees   
 Baseline15.9 ± 8.715.9 ± 7.8 
 Month 116.8 ± 7.420.4 ± 9.0NS
 Month 215.4 ± 7.920.2 ± 9.6NS
Hip external rotation, degrees   
 Baseline21.4 ± 5.520.4 ± 8.6 
 Month 121.2 ± 4.223.0 ± 6.9NS
 Month 220.3 ± 8.223.9 ± 7.1NS
Analgesic pill count   
  Month 147.5 ± 78.231.7 ± 49.7NS
  Month 260.3 ± 96.535.5 ± 58.9NS

There was also a significant difference in WOMAC stiffness and physical function scores and patient's global health scores, with patients receiving steroid treatment showing greater improvement in these outcomes at 2 months compared with patients receiving placebo. Significant improvement in the SF-36 quality of life indices was noted in the steroid treatment group, specifically for the physical component score and the bodily pain, physical functioning, and social functioning subscales. No significant improvement in the remaining subscales was noted.

The percentage of patients who achieved at least 20% improvement on the WOMAC pain subscale at 2 months after treatment was significantly higher in the steroid treatment group (21 [67.7%] of 31 patients) as compared with the placebo group (5 [23.8%] of 21 patients) (P = 0.004) (Figure 4). When the definition of a responder was set at 50% improvement (WOMAC50), there was still a significant difference in the pain improvement response between groups, with more patients showing improvement in the steroid treatment group (19 [61.3%] of 31 patients) than in the placebo group (3 [14.3%] of 21 patients) (P = 0.001). Therefore, most patients who were considered treatment responders experienced substantial improvement in their joint pain.

Figure 4.

Percentage of patients receiving either placebo or intraarticular corticosteroid injection who showed a response from baseline up to 6 months as defined by a 20% decrease in the summed score for the Western Ontario and McMaster Universities Osteoarthritis Index pain subscale.

Post hoc analysis also revealed that there were significantly more treatment responders, according to the OARSI/OMERACT criteria, in the steroid treatment group (22 [71.0%] of 31 patients) as compared with the placebo group (4 [19.0%] of 21 patients) (P = 0.0005). At 3 months posttreatment, there were still more WOMAC20 responders in the steroid treatment group (18 [58.1%] of 31 patients) compared with the placebo group (1 [4.8%] of 21 patients) (P = 0.004).

No association between the Kellgren/Lawrence grade of radiographic severity of hip OA and responder status was evident (results not shown). Analgesic intake over the first and second months possttreatment was higher in the patients receiving placebo than in those receiving steroid injection, although the differences were not statistically significant (Table 2).

Over the course of the trial, 21 patients elected to receive an open-label injection of the steroid and were followed up accordingly (Figure 3). Eleven patients in the placebo group, all of whom were WOMAC20 nonresponders at 2 or 3 months, elected to receive an open-label steroid injection at the corresponding 2- or 3-month time point. Nine (81.8%) of these patients were WOMAC20 responders at 1 month after this second injection. Nine patients in the steroid group, all of whom were WOMAC20 nonresponders at 2 or 3 months, elected to receive an open-label injection at the corresponding 2- or 3-month time point. Seven (77.8%) of these patients were WOMAC20 responders at 1 month after the second injection.

Safety assessment.

Treatment was well tolerated, with no withdrawals due to adverse events. There was no significant difference between groups in the frequency of adverse events, with 52% of the placebo group and 51% of the steroid group reporting at least 1 adverse event in the 2-month period; most of these events were either mild and/or considered unrelated to treatment. One patient in the steroid treatment group experienced deep vein thrombosis at 3 months postinjection.

There was no significant difference between the groups in the response to local anesthetic at 30 minutes postinjection. Most patients (90% in the placebo group and 71% in the steroid group) experienced at least 20% improvement in pain, although the response to local anesthetic was not predictive of the response at 1 month or 2 months postinjection in either group (results not shown). One patient in the placebo group and 3 patients in the steroid group reported an increase in pain after the procedure.


This study showed that the group of patients receiving IA steroid injection of the hip, administered under fluoroscopic guidance, not only had significantly better outcomes than the placebo group, but also demonstrated significant gains from baseline to 2 months in all measures of pain, stiffness, and physical function. In addition, an improvement of 20% in the WOMAC pain score was seen in 68% of the IA steroid group compared with 24% of the placebo group. Of more importance was the 50% improvement in the WOMAC pain scale observed in 61% of the steroid injection group compared with 14% of the placebo group, indicating that most of the elicited responses were clinically significant (23). Moreover, these benefits extended to measures of quality of life, and significant differences in treatment effects as compared with placebo were still evident 3 months after treatment.

Our experience in this study highlights the difficulties in recruiting patients to placebo-controlled trials of interventions that are even minimally invasive. The majority of patients had severe disease (clinically and radiographically), were receiving substantial daily doses of combination acetaminophen/codeine therapy, and were on waiting lists for hip arthroplasty, which may have further deterred patients from being included in a trial designed with a placebo treatment arm. This might also explain the observation that the placebo responses were somewhat lower than those recorded in previous trials in which disease severity was more moderate. The relatively low placebo response rate in this study might also be a reflection of the primary end point being at 2 months, as opposed to the earlier time points chosen in prior studies (7). Clinically significant responses were nevertheless seen in a majority of our patients. The finding of no correlation between treatment response and disease severity suggests that patients who experience treatment failure with oral medication should be offered steroid injection therapy, regardless of disease severity.

Decreases in analgesic intake and improvements in joint motion were also observed in the steroid injection group, although the differences in these outcomes compared with those in the placebo group were not statistically significant. This is likely a reflection of the small sample size, disease severity, and irreversibility of the disease in this patient population.

The OARSI/OMERACT responder criteria (21) were not prespecified outcomes for this trial, because the design of the trial and the initial recruitment of patients began prior to the publication of these criteria. Nevertheless, it is noteworthy that use of the WOMAC pain 20% improvement criterion identified virtually all patients who were also considered responders according to the OARSI/OMERACT criteria. One patient in the placebo group who was a responder by the WOMAC20 criterion was not defined as a responder according to the OARSI/OMERACT criteria. Conversely, 1 patient in the steroid treatment group was a responder by the OARSI/OMERACT criteria but not by the WOMAC20 criterion.

Nearly all patients in the placebo group who failed to achieve a response by the 2-month primary end point opted for an open-label injection of the steroid and were subsequently recorded as having achieved a WOMAC20 response 1 month thereafter. It is unclear why the nonresponders in the steroid injection group who opted for open-label injection were also recorded as responders 1 month thereafter. This could represent a placebo response to open-label injection, although the response rate (7 of 9 patients) appears somewhat high for a placebo response alone. It is possible that there was a cumulative effect from the second steroid injection that led to a clinical benefit. It may be worthwhile trying to address this in future studies by using serial magnetic resonance imaging to evaluate the inflammatory response.

Figure 1 illustrates the potential difficulty in ensuring that a hip injection is reliably administered into the synovial space. In this trial, the intrasynovial flow of contrast material and that of the steroid were separately confirmed in every case. However, in a severely diseased joint, it is not easy to prove that the entire injectate is intrasynovial, and some variability in patient response may be related to the degree of difficulty of the injection. Postinjection leakage of injectate may also account for any variability in patient response.

Previous placebo-controlled studies demonstrated mixed results from steroid injection into the hip. The earliest study reported used imaging and contrast injection to guide needle placement in patients who were on a waiting list for hip arthroplasty (6). Of 36 patients injected, a similar proportion of those who received saline, anesthetic alone, or a steroid (20 mg triamcinolone) plus anesthetic reported improvement in the first month, which rapidly declined by month 3. A major limitation of that study was that patients were told that they would receive priority for inclusion in the study based on the severity of their pain. A rather large volume of fluid (10 ml) was injected into the joint, which could have extravasated outside the synovial space. The incidence of postinjection flares was not reported. In addition, patients received active and passive physiotherapy 3 hours postinjection and on each of the following 2 mornings. Recent data indicate that immobilization for 24 hours after injection of weight-bearing joints confers longer-lasting benefit from steroid injection (24).

A second trial recruited 80 patients who were on a waiting list for hip replacement. These patients received either 80 mg triamcinolone acetonide or local anesthetic (1% mepivacaine), under fluoroscopic guidance (7). Patients were instructed to rest on the day of the injection and then to resume normal activities. Significant reductions in VAS scores for pain at rest and during weight bearing of the joint were noted in the steroid injection group at 3 weeks, which were maintained at 3 months, and the range of joint motion was increased in all directions. No significant improvement in pain or function was observed in the placebo group. A potential problem with this particular trial design was the omission of any local anesthetic from the injection for the steroid treatment group. Consequently, an immediate response to injection of local anesthetic would almost certainly unblind the assessor, and potentially the subject, and reveal the fact that steroid had not been administered.

In the third trial, 101 patients received 3 ultrasound-guided IA injections of steroid administered at 14-day intervals (8). Randomization was into 1 of 3 treatment groups: a single injection (1 ml) of steroid (40 mg depomedrol) followed by 2 sham injections, or 3 injections (2 ml) of sodium hyaluronate, or 3 injections (2 ml) of saline. All patients also received 1 ml of lidocaine. No bed rest was required postinjection. Most patients had Kellgren/Lawrence grades of <3, and patients who had previously received steroid injections were allowed into the trial, although responses to previous injections were not stated. A significant improvement in the primary outcome, pain on walking, was observed in the groups receiving either steroid or sodium hyaluronate compared with the placebo group, at 14 and 28 days but not at 3 months. No significant differences between treatment groups were observed in secondary outcomes such as WOMAC scores, although, after 28 days, 53%, 66%, and 44% of patients in the sodium hyaluronate, steroid, and placebo groups, respectively, were responders according to the OARSI/OMERACT responder criteria (21). The radiographic severity of OA did not influence the response to treatment. Differences between the study by those investigators and our study include their use of 3 ultrasound-guided use of a steroid preparation of lower potency, the absence of postinjection bed rest, and the inclusion of patients with less severe disease.

Investigations of the effects of steroids on cartilage metabolism suggest that a mechanism exists through which steroids may be inferred to have a positive effect in degenerative arthropathy (25–27), and specifically, triamcinolone has been shown to protect against cartilage fibrillation and osteophyte formation (28). The metabolic pathway of intrasynovial corticosteroids has not been fully elucidated. The rate of absorption and duration of action are related to solubility of the compound, and triamcinolone hexacetonide is the most insoluble preparation currently available (16). Depending on the formulation, a steroid can be detected in synovial fluid up to 14 days after injection (29).

Adverse effects of IA injection of steroids have been reported previously, and it is recommended that weight-bearing joints should not be injected more than once per month or more than 4 times per year (9). Judicious use of this approach would indicate that IA steroid injection appears to convey an extremely small risk of joint damage. Infectious arthritis is a rare complication resulting from joint injection; the number of reported cases is small and the incidence is as low as 1:50,000 (30). Postinjection flare or crystal-induced arthritis are more common, but symptoms are usually mild and rarely last more than 48 hours. Of course, postinjection flare may also occur following injection of radiographic contrast material. Overall, when steroid treatment is compared with alternative strategies such as NSAIDs, its safety profile can be considered favorable, particularly in the age group of patients in whom symptomatic OA develops.

In conclusion, the results of this randomized, placebo-controlled trial show that fluoroscopically guided IA injection of steroids is an effective treatment for the management of hip OA, with improvement lasting up to 3 months. Future studies should address questions related to the benefits of repeated injection. Most importantly, imaging studies should be conducted to investigate the effects of IA steroid injection on disease modification.


Dr. Lambert had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Lambert, Jhangri, Maksymowych.

Acquisition of data. Lambert, Hutchings, Maksymowych.

Analysis and interpretation of data. Lambert, Grace, Jhangri, Maksymowych.

Manuscript preparation. Lambert, Grace, Jhangri, Maksymowych.

Statistical analysis. Lambert, Grace, Jhangri, Conner-Spady, Maksymowych.

Review of protocol. Hutchings.


We would like to thank Tracey Clare for safe conduct of the trial, Shannon Erichsen for preparation of illustrations, Kamran Golmohammadi for assistance with manuscript preparation, and Medical Imaging Consultants (MIC) and the staff of MIC College Plaza for providing their support during the conduct of the trial.