Corticosteroid injections in the knee are VERY commonly performed for any knee pain, but particularly for osteoarthritis. They don’t provide significant benefit to people, and they cause harm.
1. They don’t help
This Cochrane review from 2015 used data from 27 trials. The quality of studies was generally low, there was evidence of publication bias, and yet the effect on pain and function (compared to doing nothing or placebo) was smaller than the minimum important difference for pain and for function. And the maximum effect was in the first two weeks, trailing off after that. At best, these injections might help 10% of patients a little bit, for a few weeks. Given the likely biases present in the studies, the real effect is likely to be even less.
2. They harm
Studies show increased risk of infection after knee replacement if the patient has had steroid injections previously (here). A randomised trial just out from the US (here) showed no difference in pain but greater cartilage loss over two years in patients treated with repeated injections of steroids into the knee.
The bottom line
Steroid injections are widely used but either not effective or so marginally effective that they are probably not worth the bother. This is the case in the spine (see previous post) and for the knee and very likely for everywhere else they are used.