Outcomes of imaging-guided corticosteroid injections in hip and knee osteoarthritis patients: a systematic review

Abstract

Purpose

The purpose of this systematic review is to evaluate the current literature on the use of image-guided corticosteroid injections in the treatment of patients with knee and hip OA.

Evidence acquisition

We conducted a comprehensive literature search through June 30, 2022. Publication type, study design, imaging guidance modality, osteoarthritis severity, number of injections, steroid type and dose, anesthetic type and dose, the total number of patients, follow-up intervals, and measured outcomes were extracted from the included studies.

Evidence synthesis

There were 23 included studies (10 hips, 12 knees, 1 both hip and knee).

Hip injections were found to be effective in treating short- and long-term pain and more effective than hyaluronic acid, Mepivacaine, NSAIDs, and normal saline in terms of improvement in pain and/or function. There was less impact on QoL.

Knee injections were found either to have little or no impact or were similar or inferior to comparison injections (intra-articular hyaluronic acid, PRP, NSAIDs, normal saline, adductor canal blocks).

Study data could not be aggregated because the corticosteroid types and doses, methods of outcome assessment, and follow-up time points varied widely.

Conclusion

Our systematic review found generally positive outcomes for the hip, but overall negative outcomes for the knee, although hip injections may carry a risk of serious adverse outcomes. A larger trial with uniform methodology is warranted. Specific studies on the adverse effects of corticosteroid injections are also warranted.

Appendix

Critical appraisal and methodologic quality: hip

There were 5 RCTs (Jurgensmeier 2020 was included in both hip and knee analyses). One study had 3 deficiencies [15] and 4 studies with 1 or zero deficiencies [13, 16, 18, 19]. All three deficiencies for De Rezende 2020 were related to blinding (patients, providers, outcome assessors); overall, the methods section provided only limited details [15]. Three of the other four papers did not perform a power analysis [13, 18, 19].

There were 14 cohort studies. There was 1 study with 4 deficiencies [35], 3 studies with 3 deficiencies [22, 38, 40], 4 studies with 2 deficiencies [14, 21, 41, 42], and 6 studies with 1 deficiency [17, 23, 36, 37, 39, 43]. The most common deficiency was the lack of a sample size justification, which was only performed in 1 study [37]. The next most common deficiency was found in 7 studies, where the outcome assessors were not blinded to the exposure status of the participants. This deficiency was seen in 2 of the 4 (50%) prospective studies and 5 of the 10 (50%) of the retrospective studies. The other deficiencies were rare (2 or fewer papers).

There was 1 case control study [20]. The only deficiency was that concurrent controls were not used, but this would be difficult given that the study was retrospective.

Critical appraisal and methodologic quality: knee

There were a total of 10 RCTs (Jurgensmeier 2020 was included in both hip and knee analyses) [18]. One study had 5 deficiencies [28]; one study had 4 deficiencies [34], and 8 studies with 1 or zero deficiencies [18, 24, 26, 27, 30,31,32,33]. The most common deficiency was blinding. For example, in 4/10 (40%) papers, study participants and/or providers could not be blinded, because the comparison group was “No Injection,” or an additional procedure was performed (e.g., nerve injection in addition to a joint injection, blood draw for PRP) [28, 31, 32, 34]. One study explicitly stated that blinding did not occur due to safety concerns [32]. Two studies could not blind the treatment allocation or outcome assessors, for similar reasons [40, 41]. The other most common deficiency was power calculation, which was not described in 2 studies [18, 28]. In 2 additional studies, the authors claimed that the study was “adequately powered” but did not provide any additional information about how the power calculation was performed [33, 34]. The other deficiencies were rare (2 or fewer papers).

There were 4 cohort studies. There was 1 study with 3 deficiencies [25], 2 studies with 2 deficiencies [29, 44], and 1 study with 1 deficiency [9]. None of the studies performed a sample size justification. As for the hip papers, the next most common deficiency was found in 3 studies, where the outcome assessors were not blinded to the exposure status of the participants.

Adverse outcomes in hip corticosteroid injections

There were five papers focused on adverse outcomes in the hip joint: three focused on rapidly destructive osteoarthritis (RDOA), rapidly progressive osteoarthritis of the hip, or rapidly progressive idiopathic arthritis of the hip (different terms for the same entity), and two that focused on progression of hip osteoarthritis and articular surface collapse [35,36,37, 45, 46].

Hess et al. studied hip corticosteroid (triamcinolone 40 mg) injection patients over 13 years and found a 21% incidence of RDOA and a more rapid progression to total hip arthroplasty than control patients. They also found that RDOA was more likely to develop in older, white, and severe osteoarthritis patients [35]. Sanguino et al. found an RDOA incidence of 2.8% and found that RDOA patients were older and had a shorter duration of symptoms prior to the injection (p < 0.05). Half of the cases (13/26) developed within 3 months of the injection, with the remaining developing within 3 and 12 months [36]. Boutin et al. found an RDOA incidence of 7.2%, and patients who developed were older, had higher Croft score osteoarthritis grades, and narrower hip joints (p < 0.05) [38]. De Rezende et al. also reported 5 cases of RDOA with triamcinolone hexacetonide 20 mg [15]. None of the other studies reported this complication.

Simeone et al. found a significantly greater percentage of patients who had osteoarthritis progression after a corticosteroid injection compared with a control group that did not get an injection (44% vs 24%, p = 0.02) [39]. Patients were followed up between 3 and 12 months, with a mean of 6-month follow-up duration. In contrast, Abraham et al. did not find an increased risk of osteoarthritis progression or femoral head collapse [37]. The two studies differed in a few ways: 1, the Abraham et al. study only included patients who had an MRI prior to the injection; 2, the Abraham et al. cohort had predominantly moderate osteoarthritis (89%) prior to the injection, whereas the Simeone et al. cohort was more evenly spread out between mild (37%), moderate (46%), and severe (17%) osteoarthritis grades, and perhaps the grading systems or criteria were used slightly different between the two studies; and 3, the Simeone et al. cohort all received the same type and dose of corticosteroid (triamcinolone 40 mg) and anesthetic (ropivacaine 0.5% 4 mL), whereas the Abraham et al. cohort was injected with varying steroid and anesthetic types and doses. The most common steroid types and doses were triamcinolone 40 mg (31%) and methylprednisolone 40 mg (24%), and the most common anesthetic types and doses were lidocaine 1% in 1–6 mL (39%) and ropivacaine 0.5% in 2–6 mL (27%), and perhaps the risks are attributed to specific medication and doses.