Corticosteroid injections

Glucocorticoids suppress synovial inflammation through genomic and non-genomic pathways — the basis of the short-term pain relief, which is real. The same biology is chondrotoxic in the lab: at clinical concentrations, dose- and time-dependent chondrocyte apoptosis and matrix suppression. That laboratory backdrop matters because a randomized human trial has now found the structural signal it predicts.

The short-term pain effect

Cochrane's review (27 trials, 1,767 participants) found a pain benefit of about 1 cm on a 10 cm scale versus sham or no treatment — moderate at 1–2 weeks, small by 4–6 weeks, still detectable at 13 weeks, and gone at 26 weeks, with no quality-of-life benefit and evidence quality rated low.

That short-term efficacy is no straw man. In the 480-patient Mautner trial, a single corticosteroid injection was the control arm — and none of the marketed cell therapies (BMAC, SVF, umbilical-cord MSCs) beat it at 12 months. AAOS gives corticosteroids a moderate recommendation for short-term use.

The two-year randomized trial

McAlindon 2017 (JAMA) is the only long-term placebo-controlled RCT: 140 patients, triamcinolone 40 mg versus saline every 12 weeks for two years, double-blind. The steroid group lost significantly more index-compartment cartilage thickness (−0.21 vs −0.10 mm) with no difference in knee pain at any point. The authors concluded the findings do not support the treatment. Note the schedule: 40 mg every 3 months is the conventional ceiling many clinics treat as safe — the trial tested exactly that.

The observational data point the same way. In the Osteoarthritis Initiative, steroid initiation was associated with a 3-fold hazard of radiographic worsening and continuous use with 4.7-fold. And a Radiology special report described four structural adverse patterns after hip and knee steroid injections — accelerated OA progression, subchondral insufficiency fracture, osteonecrosis, rapid joint destruction — in roughly 8% of injected patients at one center.

The limits of the harm case

Each leg of the harm case has limits. Cartilage thickness is a surrogate; the McAlindon trial was not powered for arthroplasty or long-term symptoms. The cohort data face confounding by indication — worse knees get injected — that matching only partially handles. The 8% figure comes from an uncontrolled single-center report and is not a causal rate. Most importantly: occasional single injections have not been shown to cause measurable harm. The harm data concern repeated, scheduled use.

Practical framing

Typical dosing is triamcinolone or methylprednisolone 40 mg, by convention no more than every 3 months. A steroid shot buys weeks of relief — reasonable before a trip, or to break a flare — and a poor recurring strategy for a joint you are trying to preserve. If you find yourself on a standing schedule of injections, that is the exposure the trial data warn about.

Safety beyond cartilage

Transient glycemic spikes in diabetics, post-injection flare, skin depigmentation and fat atrophy, rare septic arthritis. Open questions the trials cannot yet answer: whether a strict once-or-twice-ever exposure carries any structural cost, and whether extended-release formulations change the structural picture — no RCT addresses either.