The cortisone debate pops up in my rooms most weeks. Somebody has been offered one and wants to know whether to take it. Somebody has had three and is being offered a fourth. Somebody has been told to try everything else first and wants to know what everything else means. The good news is that the research on this is pretty good and that most practitioners are on the same page. Here is what the research currently says, so that when you sit down with your doctor or specialist you know what you are getting in to.
The short answer
A cortisone injection can be a good painkiller but it is not a repair or long term solution. Across most conditions it works well for four to eight weeks but fades by twelve, and in a few tendon problems leaves people worse at one year than a placebo injection does. In a small group of conditions it can be the right first step.
What cortisone does, and what it does not do
Cortisone is a powerful anti-inflammatory. That is important to know, because most of the stubborn tendon pain I see in practice is not an inflammatory problem. When you look at a chronically painful tendon under a microscope you find disorganised collagen, disordered cells and new blood vessels growing where they do not belong. You do not generally find the inflammatory cells. The drug is beautifully matched to an inflamed joint lining, an inflamed bursa or an inflamed tendon sheath but is poorly matched to a tendon that has been overloaded for years.
There is a second effect that is also super important to know about. Corticosteroid exposure reduces the tensile strength of collagen in laboratory work and suppresses the cells that build and maintain tendon. You may have heard about a friend or fellow athlete rupturing a tendon after repeated cortisone use and that's why. It's also part of the reason the pain recurrence rates look the way they do.
Timing is also worth knowing about before you have one. The local anaesthetic mixed into the syringe tends to wear off within a few hours. The cortisone itself usually takes three to seven days to start working and in-between, some people get a flare, a sharp rise in pain starting within a few hours of the injection and settling inside a day or two. A flare can be uncomfortable but it does not mean the injection has failed.
The shape of the benefit, week by week
Almost every trial on this subject produces the same curve, whatever the body part. If you understand the curve you can make a good decision about almost any injection you are offered.
Day one to day seven
Immediate relief from the anaesthetic, a possible flare, then the drug starts working somewhere between day three and day seven.
Week two to week eight
This is the important window. In the pooled shoulder trials, eleven studies and 726 patients, cortisone beat a placebo injection clearly under four weeks and slightly more strongly between four and eight. Roughly one person in five got meaningful pain relief they would not otherwise have had. That is a decent number for a five minute procedure.
Week eight to week twelve
In those same shoulder trials the benefit no longer reached statistical significance between eight and twelve weeks. In the largest analysis of knee injections, covering 57 trials and 22,795 people, there was no evidence of effectiveness at all beyond twenty-four weeks.
One year
Here the picture tends to split, mostly by condition rather than by dose or technique. In some conditions the injected group and the placebo group are indistinguishable in terms of reported pain and in one condition the injected group is measurably worse.
Where cortisone can be the right first choice
I am not going to spend an entire article arguing against something when there are conditions where I believe the person in front of me would benefit from early administration of a cortisone injection.
Frozen shoulder, inside the first twelve weeks
This seems to be the strongest case in the whole literature. Pooled across 65 randomised trials and 4,097 people, an injection into the joint within the first twelve weeks produced clear improvement in both pain and range of movement. Given later, the benefit largely disappeared. The same analysis supports pairing it with a home exercise programme rather than just using it alone.
The standard advice for frozen shoulder is to wait it out which also makes this one tricky because waiting it out wastes the window in which the injection works best. More on why the window is important later.
Trigger finger, De Quervain's and carpal tunnel
These are problems of a tendon sheath or a compressed nerve as opposed to a worn tendon, and the results tend to reflect that. For trigger finger, a national dataset of nearly 32,000 patients found the first injection worked about two thirds of the time and a second closer to eight times in ten. For De Quervain's, pooling sixteen studies and 1,206 patients, injection was 1.6 times more likely to resolve symptoms than a splint alone, and better again when the two were combined. For carpal tunnel, injection beats both placebo and oral cortisone in the short term, though a second injection adds little value to the first.
A joint that is inflamed
Inflammatory arthritis, or a swollen and hot joint, seems to be a case where the drug does shine, and can work really well. If, after the assessment, we find that we are dealing with a hot joint then my best position is to recognise that and refer you on to the right person for further evaluation.
Where the evidence runs against it
Tennis elbow
There is strong evidence against cortisone injections if you are looking for longer term management of tennis elbow.
A randomised trial gave 165 adults with tennis elbow either a cortisone injection or a saline injection. At one year, 96 per cent of the saline group had fully recovered or improved a great deal, against 83 per cent of the cortisone group. This is the important part though. Recurrence was 12 per cent after saline and 54 per cent after cortisone and adding conservative care to the injection did not change the result. Conservative care without the injection did however produce the lowest recurrence in the trial, under 5 per cent, and every single person in that group had recovered or was much improved at a year.
An earlier trial of 198 people found the same shape from the other direction. The injection was clearly the best treatment at six weeks. Then 47 of its 65 successes went backwards, and by fifty-two weeks the injection group was doing significantly worse than the conservative care group.
More on how I treat this on the tennis elbow page.
The Achilles, patellar and hamstring tendons
In athletic populations, tendon and fascial rupture is the most commonly reported serious complication of corticosteroid injection and these sites tend to be where these reports are most often found. Surveys of foot and ankle surgeons found that between 88 and 98 per cent never inject the Achilles at all. The International Olympic Committee position is that these injections are not appropriate for same-day return to play, or within seventy-two hours of competition, because of the effect on tendon strength.
If you are being offered an injection into or around a large weight-bearing tendon, that is probably a good time to pause and to ask for a second opinion.
The heel
An injection into the plantar fascia can relieve heel pain for roughly three months on average. The plantar fascia is however a load-bearing structure, and so the risks of rupture have structural implications in the foot and ankle. A review of 155 documented plantar fascia ruptures found that 130 of them had received an injection beforehand. Thinning of the heel fat pad and permanent lightening of the skin at the injection site have also both been described in the literature.
Head to head against shockwave, the two performed similarly at three months, with shockwave holding its result better longer term. More detail on this in the plantar fasciitis page and on Shockwave Therapy.
The side of the hip
The research seems to be pretty clear when it comes to cortisone injections in gluteal tendinopathy vs conservative care. Two hundred and four people were randomised to fourteen sessions of education and loading, a single injection, or wait and see. At eight weeks, 77 per cent of the education and exercise group were at least moderately better, against 58 per cent of the injection group and 29 per cent of those waiting. The advantage of education and exercise vs cortisone was still there at a year.
It's not to say that the injection is useless in this population, it is just not as good as conservative care. More on the gluteal tendinopathy page.
The two grey areas
The shoulder
Rotator cuff pain is where I tend to see the most injections being done and the research is on the fence when it comes to injection vs conservative care.
The pooled trials give a real benefit of cortisone injections up to eight weeks but nothing by twelve, with repeat injections performing no better than a single one at any point measured. Compared against doing the conservative work, the picture is modest. In a trial of 708 people, adding an injection to an exercise programme improved shoulder scores by 5.6 points at eight weeks, against the 8 points needed for a patient to notice a difference, and by 1.1 points across the full year. In a separate trial of 104 people, an injection and six sessions of manual therapy both produced about a fifty per cent improvement held to one year, though the injection group came back for more appointments and more injections.
The knee
The knee is the one place where I would push back against the position that cortisone may be useless. In the largest analysis of intra-articular treatments, covering 57 large trials and 22,795 people, cortisone was the only one of eighteen interventions to clear the threshold for a difference patients noticed, and those differences were still noted at two and six weeks. Almost nothing else produced the same improvement.
There is, however, also the structural question. A two-year trial injecting the knee every three months found more cartilage loss in the injected group than in the saline group, with no difference in pain. Larger observational cohorts have been less alarming, and one national study found no effect on the outcome of subsequent knee replacement. The 2025 multi-society guideline treats repeated injection as associated with a small amount of cartilage loss, with the effect growing as dose and duration grow, and recommends the lowest dose that works at the longest interval you can tolerate, if you are going to go down this road.
Side effects you should know about
Most side effects are minor and most tend to settle but it's worth knowing what they are and having the list in front of you anyway so that you know the risks involved before consenting.
- A flare of pain in the first day or two.
- Lightening of the skin and a dimple of fat loss at the injection site. This is more visible on darker skin and it is often permanent.
- A rise in blood sugar lasting several days, which is very important if you are diabetic.
- Facial flushing, and occasionally a few days of poor sleep or changed mood.
- Infection. Although uncommon, roughly one in a thousand after a large joint injection, it is still important knowing that it can appear as late as sixteen weeks afterwards and not only in the first two weeks.
- Rupture, concentrated in the tendons listed above.
- In the hip and knee, a small group of serious joint complications including accelerated arthritis and stress fracture under the cartilage.
There is actually no published guideline on how many injections you may have in a lifetime, or in a year. The three-a-year rule you may have heard about doesn't seem to be supported in the research.
The three month rule before surgery
If a joint replacement is on the cards, injection timing is super important. Analyses across hip, knee and shoulder replacement all find a raised infection rate when the cortisone injection was given within three months of the operation, with no raised rate beyond three months. That's important if an op is on the table and you decide to go ahead with a cortisone injection first.
The hip data is the strongest when it comes to cortisone injections pre surgery, pooling 28 studies and over 480,000 patients. The knee and shoulder findings are smaller but tend to point the same way. The findings don't necessarily mean that an injection is wrong but that the timing of it can be a deciding factor in the next steps after the injection and if the injection fails.
There is also another related finding for rotator cuff repair when it comes to multiple injections. A single injection in the year before surgery carried no increased revision risk. Two or more were associated with revision odds between two and a half and three and a quarter times higher. That is an observational finding and does not necessarily prove cause but is something to be aware of. It is however one of the reasons I get uneasy when somebody tells me they are booked for their third cortisone injection.
What I do with the window
The window is the part that can really change outcomes.
In the trials where the injected group did badly at one year, the drug had not visibly damaged anything in most of those people. What is more likely to have happened is that once the pain reduced, full load went straight back on to the shoulder but the tissue itself had not changed, and the problem returned with interest.
So the injection provides us with a window. What determines where you are in a year is what you do inside that window.
Knowing this, and how cortisone works, gives me three working rules.
Plan the loading programme before the needle goes in, not after the relief wears off, because in the two conditions where the combination has been tested directly it beat the injection results on its own.
Treat the first injection as the only one that you will have, because repeats do not perform better.
And treat a third injection request as a reason to re-evaluate the diagnosis and try to figure out what is actually going on before rushing off for the shot.
Having said all of this the decision itself to have a cortisone injection is not mine to make and neither is the administration of the needle. My work starts the day you walk into my rooms considering an injection, the day the pain drops if you decide to have one before seeing me, when the tissue can finally tolerate the loading that changes it and moves you in the right direction, or when you have had the cortisone and it did not produce the outcomes you were expecting. If you would like more about that, you can find it in graded exposure and in Understanding Pain.
Questions I get asked
Will an injection fix this?
It may reduce your pain, usually for four to eight weeks. Nothing in the research shows an injection changing the structure of a worn tendon or an arthritic joint for the better. It may buy you a period of comfort. What you do with that period tends to decide the outcome.
How many can I have?
There is no evidence-based limit, which is not the same as no limit. What the research does show is that repeat injections perform no better than a single one in rotator cuff pain, and that a poor response to two injections in the same site is a reason to stop and reconsider rather than to try a third.
Does cortisone damage the tendon?
It weakens collagen in laboratory conditions, and rupture is a documented complication that can be found in the Achilles, patellar, hamstring and plantar fascia groups. For most people having one injection into a shoulder or a joint, rupture is not the realistic risk, recurrence is.
Should I do the rehabilitation before or after?
Both, ideally. Start before, so that the programme is running when the pain drops and the window is not wasted. The two trials that tested the combination directly found it better than the injection alone at one year.
I have diabetes. Does that change anything?
Yes. The injection can push your blood glucose up for several days afterwards, so you need to know that in advance and monitor accordingly. Tell whoever is injecting you, and ask them what to expect. It is not a reason to refuse the injection but very important to address with the doctor.
Why did mine feel wonderful and then come back worse?
That is the commonest story I hear, and it is what the tennis elbow trials describe. Excellent at six weeks, worse at a year, with recurrence rates several times higher than placebo. The tissue had not changed, so when the pain relief faded the original problem was still there and had usually been loaded harder in the meantime.
Where this leaves you
A cortisone injection is a reasonable tool with a well-described benefit and a well-described cost. It is the wrong first move for tennis elbow, the side of the hip and the big weight-bearing tendons. It can be the right first move for a frozen shoulder caught early, for trigger finger, for De Quervain's, and for a joint that is truly inflamed. Everywhere else it is a decision about timing, and about what you have planned for the weeks the window may give you.
If you are weighing up whether to have a cortisone injection or not, then please ask me about it at your appointment. I will give you as much as I can on what the evidence says for your specific problem and what the alternatives may be, and then you can decide what the best route is for you and where you are, along with your doctor or specialist doing the injection.
