Home›Conditions›Shoulder pain
Conditions
Shoulder pain is slow to settle and can wreck your sleep. This page sets out what is likely going on in yours, what the evidence supports, and where I would start if you came to see me for it.
I am a chiropractor in Hillcrest and shoulders are among the most common things I see, from swimmers and racquet players around the Upper Highway to people who have spent thirty years working with a tool above their head. Most of them walk in having been told a version of the same story about a tendon, a bone spur and, sometimes, surgery.
Fortunately, a good deal of that story has not survived the last ten years of research. Several sections below will contradict advice you may have already been given, including advice given by other chiropractors and health care professionals. Every claim on this page is referenced at the bottom if you would like to read further.
91%
of people with a shoulder problem are woken by it at night
65%
of rotator cuff tears found in a general population caused no symptoms at all
Nil
clinically important benefit of decompression surgery over a placebo operation
Questions this page answers
The short answer
Rotator cuff related shoulder pain is the umbrella term now used for the group of presentations previously labelled subacromial impingement, subacromial pain syndrome, rotator cuff tendinopathy and subacromial bursitis. It is the most common cause of shoulder pain. Symptoms are typically felt at the front and outside of the shoulder, with difficulty in lifting and rotating.
The diagnostic name we use to describe a shoulder problem has changed twice in the last fifteen years. Impingement used to describe a mechanical theory, that the tendon was being pinched under the bony arch above it. When that theory was tested by removing the bone, the surgery performed no better than a placebo surgery. The name we use now describes where the pain is instead of what is supposedly causing it.
Underneath the label though sits a group of four tendons that in most cases has been asked to do more than it was able to do at the time. Sometimes the trigger is obvious, a weekend of painting or a new gym programme, but often it is not, and the load tends to creep up over months without anything feeling like an injury. Either way the tissue is often not the biggest issue but rather the sensitised nervous system, and that responds really well to being loaded.
Two other conditions are also commonly mistaken for causing shoulder pain. A frozen shoulder restricts movement in every direction, including when somebody else moves the arm for you, and we will discuss this in its own section further down. And pain referred from the neck usually changes when you move your neck rather than your arm and tends to be a sharper pain that you can't seem to find relief from, no matter what you do to the shoulder.
The short answer
The tendons of the rotator cuff carry their heaviest load through the middle of the range, as the arm passes shoulder height. Pain therefore tends to appear somewhere between 60 and 120 degrees and settle again above it. If someone else lifts the arm for you, it usually hurts a good deal less.
The painful arc is one of the more useful things you can observe about your own shoulder. Raise the arm out to the side slowly and if pain appears as the hand passes chest height, peaks near shoulder height, then fades as the arm continues overhead, that pattern points at the cuff and the tissue around it.
The second observation separates two different problems. Lie on your back and let somebody lift the arm for you while you stay completely relaxed. If it moves much further and hurts much less that way, the restriction is most likely coming from the effort of the muscles rather than from the joint. If it is just as stiff either way, something is limiting the joint itself and a frozen shoulder becomes more likely.
Neither observation is a diagnosis but both will tell me a great deal in the first few minutes of an assessment.
What your shoulder does through the middle of the range tells me more in five minutes than most scan reports do.
The short answer
Rotator cuff tears are common in shoulders that do not hurt. A general population screening study found full thickness tears in 22 per cent of people, and two thirds of them caused no symptoms. Finding one on your scan can be a starting point for the assessment but is not a reason to rush off to the specialist.
One of the findings that changed how I read a scan report came out of a shoulder clinic where 588 people arrived with pain in one shoulder and both shoulders were scanned. Among those with a full-thickness tear on the painful side, the shoulder that felt perfectly fine also carried a full-thickness tear 35.5 per cent of the time, and a partial tear in a further 20.8 per cent. In that study fewer than half of the pain-free shoulders were normal.
Having a tear and having a painful shoulder are two things that often work separately. Past fifty a good many people have a rotator cuff tear and never even know about it. Whether this tear explains this shoulder comes down to what the shoulder does on examination, how the pain behaves through the day, and how it responds when we load it.
Scans still have their place and none of the above means that we shouldn't have one done. It means the report needs reading against your age and against what your shoulder is doing, something we will go through together during your consultation. For what it is worth, in 452 people with a full-thickness tear who followed a structured loading programme, fewer than a quarter went on to choose surgery, and those who did generally decided on it inside the first three months of feeling the pain.
There is more on how scan findings and pain relate to each other in Does Pain Mean Damage.
Full-thickness rotator cuff tears found on ultrasound, by age
| Age | Proportion with a full-thickness tear |
|---|---|
| 20s to 40s | 0% |
| 50s | 10.7% |
| 60s | 15.2% |
| 70s | 26.5% |
| 80s | 36.6% |
Both shoulders scanned in 664 residents of one village. Overall prevalence 22.1 per cent. Of every tear found, 65.3 per cent caused no symptoms.
The short answer
Most people with a painful shoulder sleep badly. In a study of 343 people with shoulder problems, 91 per cent reported pain at night and the majority scored as poor sleepers on a validated scale. How badly you sleep turns out to bear almost no relationship to what a scan shows.
Ninety-one per cent is close to everybody. Between seven and nine in ten of those people scored as poor sleepers depending on their diagnosis. So if being woken at two in the morning feels like a signal that something serious is going on, it is worth knowing that it can be the ordinary presentation of an ordinary shoulder problem.
What's important to remember and what reassures people most is that night pain carries little information about severity. Across 209 people with full-thickness tears, no feature of the MRI correlated with how well they slept and tear characteristics accounted for only four per cent of the difference in sleep quality between people. A separate series of 324 patients found that people with small tears tended to sleep worse than people with larger ones.
Sleep loss is not a passive consequence either. In a controlled crossover study, a single night without sleep lowered pain thresholds and switched off the body's own descending pain inhibition altogether. Your shoulder wakes you, and the lost sleep then makes the shoulder hurt more the next day. That loop is something we want to interrupt as soon as possible, and it is why sleep is part of the treatment plan rather than a symptom to wait out.
So you might be asking then what do I do about it. No sleeping position has ever been tested in a trial for shoulder pain, and the largest study of the question, 761 workers, found no relationship between the position people sleep in and rotator cuff problems. Here is a short list of what patients report helps, worth trying especially if you are battling on the sleep side of things.
Worth trying
01
Sleep on the other side with the sore arm resting forward onto a pillow in front of you, so the arm is supported instead of hanging across your body.
02
On your back, put a folded towel or a thin pillow under the upper arm so the shoulder is not dropping backwards into the mattress.
03
A spell propped up, on a wedge or in a recliner, is worth trying through the worst two or three weeks.
04
Do your loading earlier in the day. A shoulder worked hard at nine at night is often the one that wakes you at two in the morning.
The short answer
About half of new shoulder problems have fully settled by six months and around 59 per cent by twelve. Shoulders can unfortunately be slow. They are also one of the areas of the body where doing nothing tends to be worse than doing something, so your time waiting is better spent loading than resting.
Those figures come from 349 consecutive episodes followed through general practice and from a systematic review of sixteen cohorts. At twelve months, 41 per cent of people still had something going on in the shoulder, either persisting or having come back. That is unfortunately more sobering than most people expect when they book, but I would rather you heard it from me at the start than concluded at week ten that you were failing with conservative management.
There is some really good news though when it comes to conservative management of shoulder pain. Pooled across nineteen studies of subacromial shoulder pain, people receiving usual care improved by about 32.5 points out of 100 at three months. People receiving nothing improved by 5.3. Roughly forty per cent of the three-month gain was seen inside the first six weeks whereas shoulders left alone mostly stay sore.
The factors that predict who is still sore at a year are not the ones most people expect. A study of 1030 people starting physiotherapy for shoulder pain measured seventy-one baseline variables. The psychological ones predicted the outcome more strongly than any clinical examination test did.
What predicts a slower recovery
A long time between the pain starting and getting it looked at
ModifiableHigh pain at the first visit
Partly modifiableHigh disability at the first visit
ModifiableLow confidence that you can manage the pain yourself
ModifiableExpecting slight improvement rather than full recovery
ModifiableNeck pain at the same time
TreatableStill painful at six weeks
ReassessSome more good news is that five of those seven are things we can work on directly and two of them are beliefs about what is going to happen, which is an important finding. In the same dataset, somebody with low pain but low confidence did about as badly at six months as somebody with high pain and high confidence.
If your shoulder has been sore for longer than three months, Exercise for Persistent Pain covers the loading side of this in more detail.
The short answer
Yes, but maybe not in the way that you would expect. We use hands-on work to reduce pain and make movement possible sooner, reducing fear and avoidance. Progressive loading then changes the outcome over the next few months. Importantly, we use the first to make room for the second.
01
The first job is working out which of the shoulder problems you might have. Rotator cuff pain, a frozen shoulder and pain referred from the neck will tend to get three different plans and can routinely be confused with one another so it is important that your shoulder is examined first before we rush off for a scan.
If you have a report from a recent scan we will go through it together. One frightening word left unexplained can do a lot of damage to how willing a person is to move an arm.
02
Tissue adapts to the load you give it. Too much too soon flares a shoulder up, and nothing at all leaves it less tolerant than it was before. The work is finding the dose in between and moving it up as you improve, on a graded basis rather than a hard and fast weekly increase.
Manipulation, dry needling, ischaemic compression and IASTM all sit inside this step. They open a window in which movement becomes easier and less threatening and what you do inside that window tends to ultimately change the outcome. In one trial of people already sitting on a surgical waiting list, a specific exercise strategy took the proportion who went ahead with the operation from 63 per cent down to 20.
03
Shoulder dysfunction is measured by what you can no longer do. Reaching the top shelf, fastening a bra, swimming a length, sleeping on that side, throwing a ball for the dog. We identify the one you want back first, then build towards it in steps small enough that you keep going through the ordinary bad weeks.
The short answer
Start with one exercise that you fear the least, rather than ten. A trial of 86 people found a single loaded exercise, three sets of ten to fifteen twice a day, matched a full supervised physiotherapy programme at three, six and twelve months. Pain up to five out of ten during it is acceptable.
Most shoulder programmes hand you eight exercises and most people only end up doing two of them, badly. In that trial the intervention was one exercise, progressed over about twelve weeks from an isometric hold against a wall, to a band, to a light weight. It held its own against a full course of supervised physiotherapy at every follow-up point.
Across 46 exercise programmes in 22 trials, the doses that worked ran from one to three sets and from four to thirty repetitions, mostly either daily or three times a week, over six to twelve weeks. Nobody has ever compared frequencies or programme lengths against each other, so anybody quoting you an exact schedule is quoting convention. Heavier is not better either. A trial of 100 people comparing a programme progressing to a six-repetition maximum against one held at twenty to twenty-five found no difference between them.
The rule for how much soreness is acceptable also comes straight out of those trials. Pain up to about five out of ten during the exercise is fine, provided it settles over the following hours and is no worse over the following days. It should also not climb from one week to the next.
There is one caveat on that worth noting. Exercising into pain has not been shown to beat exercising without pain. Sixteen trials found no difference between the two, and a shoulder-specific trial of 43 people comparing four to seven out of ten against nought to two saw both groups improve by the same amount. The five out of ten rule is really just permission to carry on when it is a bit sore and should not be read as an instruction to make it hurt when exercising.
Where to start this week
01
Pick one movement that reproduces your pain in a mild, controlled way. Most often that is lifting the arm out to the side against a light band.
02
Three sets of ten to fifteen, twice a day. Choose a resistance where the last two repetitions are harder.
03
Judge it the next morning rather than during. The same or better means hold the dose. Worse two mornings running means drop back a notch.
04
Add resistance once a week at most, and only if the week behind you was comfortable.
05
Give it a fortnight before deciding it has not worked. Very little about a shoulder changes inside a week.
There are shoulder and upper back options on the Stretches page, though for this problem loading will do more for you than stretching will.
The short answer
Keep using it, within a dose that settles. Complete rest stiffens a shoulder quickly and does not speed anything up. The workable rule is that soreness during and shortly after activity should settle by the next morning and should not climb from one week to the next.
Complete rest in a conservative management programme is the one option with a track record of failing. A shoulder that is not moved stiffens quickly, and the pooled data on doing nothing puts three-month improvement at roughly five points out of a hundred. Rest also does nothing for confidence or expectation, and those are the two factors that most reliably predict a slow recovery.
What you are aiming for is a level of activity the shoulder can absorb and then adapt to and one which doesn't increase the fear of making things worse. That level tends to be lower than you would like in the first fortnight, progressing thereafter. The traffic light below is a nice guide when it comes to the question of how you should feel and what to do at each level.
Green
Soreness during or shortly after activity that settles within 24 hours and is not climbing session on session. Carry on. This is what adaptation feels like.
Amber
Soreness still up after 24 hours, or creeping upward week on week. Hold the dose where it is for a week instead of increasing it. Do not stop.
Red
You cannot actively lift the arm, new weakness or numbness appears, the shoulder is hot and swollen, or any of the symptoms in the section below. Stop and get assessed.
Two specific things are worth avoiding during a flare. Repeated overhead work late in the day when the shoulder is already tired, and sleeping deliberately on the sore side to check whether it still hurts. It will, so let's stay away from poking the bear.
The short answer
Probably not. Two placebo controlled trials compared the standard shoulder decompression operation against a sham operation and found no clinically important difference, at two years and again at five. A Cochrane review rates that evidence high certainty. A guideline panel also issued a strong recommendation against the procedure in 2019.
The first trial randomised 313 people into three groups. One had the decompression operation. One had an arthroscopy in which the surgeon looked inside and did nothing else, with the patient unaware which of the two they had received. One had no treatment at all. At six months the real operation scored 32.7 on a 48-point shoulder score and the placebo operation scored 34.2.
The second trial set out in advance how big a difference would count as meaningful, at 15 points on a 100-point pain scale, then randomised 193 people to decompression, to a sham arthroscopy or to exercise. At two years the operation beat the sham by 4.6 points on rest pain and 9.0 points on pain with activity. At five years the gap was 2.0 and 8.0. It never reached the threshold the researchers had set in advance, and it never separated from exercise either.
Cochrane pooled the placebo-controlled evidence and rated it high certainty, a grade that doesn't often turn up in musculoskeletal medicine. A guideline panel writing in the BMJ issued a strong recommendation against surgery for this particular group. That is about as clear as this field gets.
Repair of an actual tear is a different question and the evidence there conflicts. One trial of older patients with small tears found repair no better than conservative care at six years. Another, in younger and more active patients, found repair ahead by about twelve points at fifteen years, with the gap widening rather than closing. If you sit in the second group then the assessment is even more important and the conversation may end up with a shoulder surgeon.
Four questions worth taking to a surgeon
On the last one, the best available review of surgical delay found no rise in retear rates at three to six months, though that work covers traumatic tears and does not settle the question for degenerative ones. The rehabilitation guideline's own referral trigger is a maximum of twelve weeks of appropriate non-surgical care.
None of those four questions is adversarial. A good surgeon will have an answer to all of them, and I refer to several who do.
Where the problem is calcific rather than a tear, shockwave has been shown to be a really good option. Its defined role and the detail sit on Calcific Tendinitis of the Shoulder and on the Shockwave Therapy page.
The short answer
That decision belongs with your doctor. The evidence shows a real but short benefit. Pooled trial data found meaningful pain relief up to about eight weeks and none by twelve, and repeat injections performed no better than a single one at any point measured.
Eleven trials and 726 patients were pooled. Against placebo, cortisone produced a clear benefit under four weeks and a slightly larger one between four and eight weeks, with a number needed to treat of about five. Between eight and twelve weeks the effect no longer reached significance.
Across three of those studies, covering 292 patients, repeat injections did no better than a single one at any timepoint measured. A separate database study of more than a hundred thousand patients found that one injection in the year before rotator cuff surgery carried no increased revision risk, while 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 it does not establish cause, though it is the reason I get uneasy when somebody is on their third.
Compared against doing the work, the picture is modest. In a trial of 708 people, adding an injection to exercise improved shoulder scores by 5.6 points at eight weeks, against a threshold of 8 points for a difference a patient would notice, and by 1.1 points over the full year. In a separate trial of 104 people, injection and manual therapy both produced roughly a fifty per cent improvement held to one year, although the injection group needed more follow-up visits and more further injections.
The decision belongs with your doctor and so does the injection. My job is making sure that if you have one, the window it opens gets used.
It is also worth noting that some groups carry a higher risk profile with cortisone, people with diabetes among them, because the injection can push blood glucose up for several days afterwards. And if you are heading towards surgery, an injection given in the weeks beforehand has been linked to a higher rate of infection after the operation. So if you are thinking about a cortisone injection, ask your doctor about the risks that apply to you before taking the plunge.
The short answer
Frozen shoulder restricts movement when someone else moves your arm for you, not only when you move it yourself. The clearest sign is a disproportionate loss of outward rotation with the elbow at your side. Rotator cuff pain hurts on resisted effort. Frozen shoulder hurts at the end of range.
Sit down, let the arm hang at your side with the elbow bent to a right angle, and turn the forearm outwards away from your body without letting the elbow leave your ribs. Compare the two sides. A frozen shoulder loses that movement badly and out of proportion to everything else, and it loses it whether you do the turning or somebody else does it for you. Rotator cuff pain usually keeps most of that range and hurts instead when you push against resistance.
Published thresholds vary, from a thirty per cent loss across two directions to a fifty per cent loss of outward rotation compared with the other side, so treat this as a guide rather than a test.
The standard line is that frozen shoulder is self-limiting and clears within one to three years. The largest long-term follow-up tracked 269 shoulders for an average of 4.4 years. Fifty-nine per cent had recovered to normal or near normal. Forty-one per cent still had symptoms and six per cent had both pain and functional loss. Broader synthesis puts persisting pain or stiffness somewhere between twenty and fifty per cent, and the systematic review of natural history concluded that the idea of phases leading to complete recovery without treatment is unfounded.
The freezing, frozen and thawing description is a convention rather than a finding. Different guidelines attach different timelines to it and they contradict each other.
Corticosteroid injected into the joint works considerably better early. Pooled across 65 trials and 4097 people, an injection inside the first twelve weeks produced clear improvement in both pain and function, and the benefit largely faded when it was given later. In a frozen shoulder of less than a year's duration, early injection combined with a home exercise programme is what the meta-analysis supports.
Stretching a frozen shoulder hard is the most common self-inflicted setback I see. Where the shoulder is highly irritable, pain at seven out of ten or above, work at an intensity that adds no extra pain. Where it is quieter, below three out of ten, longer holds with some discomfort are appropriate. Stretching beyond painful limits in a highly irritable shoulder tends to produce worse outcomes.
Frozen shoulder is about five times more common in people with diabetes, and around thirty per cent of people who develop it have diabetes. Thyroid disease is often listed alongside it, though the evidence is inconsistent and I would not put much weight on it.
As for what the larger interventions achieve, a trial of 503 people compared manipulation under anaesthetic, arthroscopic capsular release, and early structured conservative care with an injection. At twelve months no approach was clinically better than another. It's worth knowing that before booking theatre time because once you have gone under anaesthesia you can't undo it.
The short answer
Almost certainly not on its own. A review of 923 athletes found shoulder blade movement patterns carried no significant increase in the risk of future shoulder injury, and a separate review found no difference in upper back posture between people with shoulder pain and people without it.
This is one of the most common things patients arrive believing, and it comes from us. Manual therapists, chiropractors included, have spent thirty years telling people their shoulder blade sits wrong and their upper back is too rounded and the research has not been kind to either claim.
An early review of 419 athletes suggested that abnormal shoulder blade movement raised the risk of a future shoulder injury by 43 per cent, though its confidence interval only just cleared no effect at all. A larger review followed, 923 athletes and 212 injuries, and found a relative risk of 1.07 with a p value of 0.59. Abnormal shoulder blade movement turned up in 46 per cent of the athletes and it did not predict who got hurt.
A separate review asked whether people with shoulder pain hold the shoulder blade differently from people without it. It screened 7445 abstracts and found ten usable trials. Some reported reduced upward rotation. Some reported the opposite. Some found no difference. The authors state plainly that the evidence does not justify rehabilitating a shoulder blade towards an idealised normal position.
On the upper back, ten studies gave moderate evidence of no difference in thoracic curvature between people with shoulder pain and people without. Sitting up straight does increase how far you can lift the arm in that moment, in sore and pain-free people alike, so it is a useful thing to do while you are loading. As an explanation for why the shoulder started hurting, it has not held up.
One more, and this one is aimed squarely at my own profession. Isometric holds are widely taught as a way of switching shoulder pain off. The finding that started that belief came from six volleyball players with knee tendon pain. A larger crossover in 21 people found a reduction of 0.9 points on a ten-point scale, gone again by 45 minutes, and no different from ordinary exercise. A review of ten trials found isometrics no better than isotonic work at any timepoint. They are a good way to start loading a shoulder that cannot tolerate movement yet, and I use them for exactly that.
Danger signals
The short answer
See a doctor the same day if you cannot actively lift the arm after a fall or a wrench, if the shoulder is hot and swollen with a fever, or if you notice an unexplained lump or visible wasting. Night pain that never eases with any position also needs review.
Go to an emergency department or see a doctor the same day for an inability to actively lift the arm following a fall or a wrench, an obviously deformed shoulder after trauma, or a hot and swollen shoulder alongside a fever or feeling unwell in yourself.
See a doctor promptly for an unexplained lump or swelling around the shoulder or collarbone, visible wasting of the muscle over the shoulder blade or the upper arm, new weakness or numbness in the arm or hand, unexplained weight loss or night sweats alongside the shoulder pain, or a history of cancer.
See a doctor promptly too for night pain that is unrelenting and does not change with position or movement. Night pain that eases when you shift or get up is the ordinary presentation of a sore shoulder.
All of this is uncommon. It is set out here so that you know what does need urgent attention, which makes it easier to stop worrying about the rest.
Although I have an extensive knowledge of pain pharmacology, questions about medication are better aimed at your doctor or pharmacist rather than at me. I am not allowed to script any medication.
Next step
If your shoulder has been sore for weeks, if it is waking you at night, or if somebody has shown you a scan and used a word that frightened you, then that is what the first consultation is for. It runs 45 to 60 minutes and no referral is needed.
Phone 031 035 1165 · WhatsApp 064 820 5203 · A1a Meyrickton Park, 2 Meyrickton Pl, St Helier, Hillcrest, 3610
This page is general information and is not a substitute for personalised medical advice. Always consult a healthcare provider about your specific condition.
