What the evidence can and cannot tell you

For many people living with knee osteoarthritis, the question arrives at some point: can I put off a knee replacement with injections? The honest answer is — possibly, for some time, for some patients — but the evidence behind that answer is uneven, and it matters which injection you are talking about.

Injection therapies are routinely used across the UK as part of knee OA management before surgery becomes the remaining option. However, no injection has been shown in a randomised controlled trial to modify the underlying disease or definitively prevent joint replacement. The more accurate framing is that certain injections are associated with a delay to surgery in large real-world studies — and that distinction is not a small one.

The evidence also has a clear hierarchy. Hyaluronic acid (HA, or viscosupplementation) carries the largest observational dataset and the most direct data on time to knee replacement. Corticosteroids have a strong short-term RCT base for pain relief, though that benefit fades within months. Orthobiologics — including platelet-rich plasma (PRP), bone marrow concentrate, and injectable scaffold materials — show early promise but lack large-scale trial evidence with surgery delay as the measured outcome.

One further point holds across all agents: injection therapies perform best in mild-to-moderate OA. In advanced, end-stage joint destruction, the biology leaves little room for them to work.

Hyaluronic acid — the strongest delay signal

The headline figure from the largest study to date is stark: in a U.S. claims database of 182,022 knee OA patients who eventually had a total knee replacement, those who had received at least one course of hyaluronic acid (HA) injection waited a median of 484 days before surgery. Patients who received no HA injections waited a median of 114 days. That gap — roughly a year — is clinically meaningful, not a statistical artefact. The same dataset documented a dose-response relationship: patients who completed five or more HA courses delayed surgery by a mean of 3.6 years, and the result held at p<0.0001 across every subgroup tested.

A separate French cohort of 14,782 patients followed between 2006 and 2013 reached the same conclusion by an independent route. Using restricted mean survival time — a measure of how long patients remained surgery-free — the HA group gained an extra 51 days at the one-year mark and 217 additional days at the 7.5-year horizon (all p<0.001). Both analyses used Cox proportional hazards models and achieved statistical robustness well below conventional significance thresholds.

The central caveat deserves plain statement: neither study is a randomised trial. Patients who pursue HA injections may not be the same kind of patient as those who do not — they may have milder disease, stronger engagement with their health, or better access to specialist care, and any of those differences would independently predict later surgery. This is confounding by indication, and it means the observed delay cannot be attributed to HA alone with certainty.

The finding is nonetheless the strongest real-world signal available for any injection agent. Within the HA category, product formulation also varies: higher molecular weight preparations have been associated with better treatment-completion rates and lower annual costs than five-injection low-molecular-weight regimens — a practical consideration when reviewing options with a clinician.

Corticosteroids — short-term relief, not a delay strategy

Corticosteroids tell a different story. A 2024 systematic review and meta-analysis of 11 randomised controlled trials involving 842 patients found that intra-articular corticosteroid injections reached the minimal clinically important difference for both VAS pain and WOMAC function scores only at six weeks or fewer. Beyond six months, there was no statistically or clinically significant advantage over placebo on either scale. The evidence is clear: corticosteroids work for flares, not for the long game.

That short window does carry genuine value. Among 321 patients studied ahead of total knee arthroplasty, those who received triamcinolone injections within six months of surgery reported significantly less preoperative pain (p=0.016) and better function (p=0.046) compared to those given ketorolac. By three months post-surgery, however, no meaningful difference remained across any KOOS subscale between the two groups. The practical implication is that corticosteroids can make the waiting period before an operation more comfortable — they are not a mechanism for avoiding it.

There is also a timing consideration patients may not be aware of. Research cited as the Insall Award study established that intra-articular injections given in the period before total knee arthroplasty increase the risk of periprosthetic joint infection. This makes repeated corticosteroid injections a poor long-term strategy for anyone whose osteoarthritis is already progressing toward replacement.

Clinically, this limitation is increasingly acknowledged. The MOTION study — a multicentre randomised trial of 264 patients — is currently comparing genicular artery embolisation against corticosteroid injection over 24 months. The fact that such a trial exists reflects a broader recognition that corticosteroids, while useful in the short term, are not a sustainable answer for patients trying to manage OA over years rather than weeks.

Orthobiologics — what early evidence shows

Bone marrow aspirate concentrate (BMAC), sub-chondroplasty, and platelet-rich plasma (PRP) represent the newer end of the injection landscape — agents with biological rationales and early signals worth noting, but evidence bases that are still developing.

The most direct data on BMAC comes from a retrospective review of 23 procedures in a veteran cohort: only five (21.7%) required any further intervention within two years, and just one patient requested a total knee arthroplasty. That is an encouraging signal, but a cohort of 23 cannot support broad conclusions, and no powered randomised trial has tested BMAC against a comparator with TKR delay as its primary outcome.

Sub-chondroplasty — the injection of calcium phosphate into subchondral bone defects — has the most structured evidence in this group. A stratified meta-analysis found reductions in pain, improved function, and delay to arthroplasty conversion in patients with advanced knee OA. Because it targets the bone layer beneath the cartilage rather than the joint space itself, it addresses a different component of the disease process from standard intra-articular approaches.

PRP works by concentrating growth factors from the patient's own blood to stimulate tissue repair — mechanistically distinct from both BMAC and inert fillers. Bannuru and colleagues' 2015 network meta-analysis, one of the more comprehensive comparative analyses in this area, ranked PRP favourably against HA for symptom relief in knee OA. That symptom signal is more than nothing; however, no study has used delay to knee replacement as a primary endpoint, and evidence of structural disease modification remains unconfirmed.

All three agents share the same ceiling: study populations are small, follow-up periods are limited, and none has been tested in a powered randomised trial with TKR delay as the central question. They may have a role for selected patients earlier in the OA trajectory, as part of a staged management plan, but the evidence is not yet at the level needed to recommend any of them specifically as a delay strategy.

Why 'associated with delay' is not the same as proven

Knowing where observational evidence ends and proven causation begins changes how any of the findings discussed so far should feed into a decision. The confounding and the RCT gap have already been named — what is worth adding here is what they mean in practice, at the point where a patient is weighing options.

Large, consistent, statistically significant associations across two independent national datasets — one American, one French — carry genuine evidential weight. That is not the same as noise. When two separate research groups, using different populations and different time horizons, arrive at the same directional finding through rigorous survival analysis, the pattern is meaningful. The appropriate response is not dismissal.

The appropriate response is calibrated expectation. "Associated with delay" means: among the many thousands of patients who received these injections and later had surgery, the surgery tended to come later than in those who did not receive them. It does not mean: this injection will delay your surgery by a defined period. The individual response depends on disease severity, activity level, and biological factors no database study can isolate.

The single most useful question to bring to any injection consultation is this: "Is the evidence for this treatment from a randomised controlled trial, or from observational data — and what does that mean for what I should expect?" A clinician who can answer that honestly — naming what the evidence does and does not establish — is giving you something more reliable than a headline number.

Choosing an approach: severity, timing, and next steps

The evidence surveyed above points to a consistent pattern: injections work best earlier in the OA trajectory. Viscosupplementation is indicated for mild-to-moderate disease — joints where synovial function is impaired but structural loss has not yet reached end-stage bone-on-bone contact. At that point, neither HA nor any other intra-articular agent is likely to add meaningful benefit, and the conversation appropriately shifts to surgical planning.

Timing carries a safety dimension that is easy to overlook. Because of the sequencing risk described in the section on corticosteroids — the elevated infection risk associated with pre-operative injections — any injection plan should be discussed with the treating team well before surgery is scheduled, not in the weeks immediately preceding it. If surgery is already under active consideration, the sequence matters as much as the choice of agent.

The pathway that reflects standard UK clinical practice runs in stages: conservative management first (physiotherapy, weight management, analgesics), then injection support where the clinical picture warrants it, then surgical consideration if function continues to deteriorate. Each stage is a decision point, not a fixed protocol, and the appropriate exit from one stage into the next depends on how the patient is responding.

For anyone uncertain about which stage they are at — whether their OA is still within the window where injections might help, or whether surgery has become the realistic next step — a specialist assessment is the most productive move available. That assessment, covering OA severity, functional impact, and overall fitness, is what determines whether any injection approach is worth pursuing and, if so, which one.

1. [1] Relationship Between Intraarticular Injections on Patient-reported Outcomes in Total Knee Arthroplasty. (2026). https://doi.org/10.5435/JAAOSGlobal-D-25-00064 https://doi.org/10.5435/JAAOSGlobal-D-25-00064
2. [2] Bone marrow aspirate concentrate chondroplasty to delaying need for total knee arthroplasty: a retrospective review of a veteran cohort.. (2023). https://doi.org/10.2217/rme-2023-0124 https://doi.org/10.2217/rme-2023-0124
3. [3] Sub-chondroplasty Reduces Pain, Improves Function and Delays the Conversion to Arthroplasty in Patients with Advanced Knee Osteoarthritis: A Stratified Meta-analysis and Quality Assessment.. (2023).
4. [4] Is Intra-Articular Injection of Synvisc Associated with a Delay to Knee Arthroplasty in Patients with Knee Osteoarthritis?. (2019). https://doi.org/10.1177/1947603518775792 https://doi.org/10.1177/1947603518775792
5. [5] Intra-articular corticosteroid injections provide a clinically relevant benefit compared to placebo only at short-term follow-up in patients with knee osteoarthritis: A systematic review and meta-analysis.. (2024). https://doi.org/10.1002/ksa.12057 https://doi.org/10.1002/ksa.12057
6. [6] Multicenter, PrOspective, Randomized, Controlled Trial Comparing Genicular Artery Embolization to Corticosteroid Injections for the Treatment of Symptomatic Knee Osteoarthritis: MOTION Study Protocol. (2025). https://doi.org/10.1007/s00270-025-03994-z https://doi.org/10.1007/s00270-025-03994-z
7. [7] A retrospective claims data analysis of health care utilization and cost among patients receiving multi-injection intraarticular hyaluronic acid.. (2024). https://doi.org/10.18553/jmcp.2024.30.10.1117 https://doi.org/10.18553/jmcp.2024.30.10.1117