Frozen Shoulder (Adhesive Capsulitis): What It Is and How to Prevent and Treat It

Michael Greger M.D. FACLM · October 19, 2026 ·

When it comes to frozen shoulder, less is sometimes more.

Discuss

Below is an approximation of this video’s audio content. To see any graphs, charts, graphics, images, and quotes to which Dr. Greger may be referring, watch the above video.

Frozen shoulder, also known as adhesive capsulitis, is a common, debilitating disorder characterized by shoulder pain and loss of movement. What happens is the lax fibrous sheath (the “capsule”) that encloses the joint becomes inflamed and thickens as it starts to scar over, restricting shoulder movement. It most commonly strikes people in their mid-fifties and has an estimated lifetime prevalence of 2 to 5 percent.

Frozen shoulder is described as a “mystery syndrome.” Although some cases are precipitated by trauma or surgery, suggesting the freezing is a protective response by our body to defend the joint with the greatest range of motion, some cases appear to come on spontaneously. I had it once. I wrenched my shoulder trying to move my treadmill desk.

Uncertainties about its cause, about its diagnosis, and about its treatment all contribute to frozen shoulder’s reputation as a “medical enigma.” Although there are no formally recognized diagnostic criteria, loss of both active and passive range of motion is a hallmark of the disease. Active movement is when you move the joint yourself; passive movement is when you relax and let someone else, like a therapist, try to move your joint. In rotator cuff disease, active movement can be painful, but, classically, there is full range of motion with passive movement. Sometimes, what may seem like a loss of passive range may just be a patient locking up due to pain or fear, in which case an injection of lidocaine within the joint can help differentiate the two.

One of the most important things doctors can do is offer reassurance, communicating to patients that frozen shoulder is a benign condition that tends to get better on its own. Unfortunately, it has a long road to recovery, taking one to three years to resolve. Thankfully, the pain eases first, but that is followed by pronounced stiffness that can last for months before it starts to clear up.

Traditionally, the natural course of the condition was conceptualized as three overlapping stages: the “freezing” stage characterized by pain and loss of movement, followed by a “frozen” stage of stiffness, and finally a “thawing” stage of recovery. The painful freezing stage might last 10 to 36 weeks. The frozen stage may last four to 12 months, and the final thawing stage might run five to 12 more months. However, more recent longitudinal follow-up data suggest the greatest recovery in function occurs within the first few months, followed by diminishing returns. Eventually, though, full range of motion should be reclaimed. The same shoulder is rarely re-affected, but as many as one in seven patients develop it in their other shoulder within five years.

Frozen shoulder, through the impairment of sleep and everyday activities, can markedly affect quality of life. When in bed, the affected arm can be rested on pillows to help stop you from rolling onto it in the middle of the night. Some find hot packs to be helpful. Beyond self-care, treatment options commonly include physical therapy, steroid injections, and a variety of more invasive options, like surgery. Does anything actually help shorten the disease course?

Joint mobilization is a common treatment for frozen shoulder. It’s a hands-on technique in which a physical therapist attempts to break up some of the scar tissue through controlled movements and stretches of the shoulder joint. It can improve objective range-of-motion measures but worsens pain and disability. The discomfort caused by the procedure and the subsequent increase in pain and decrease in subjective function might be worth it if physical therapy actually shortened the course of the illness, but such a benefit has never been demonstrated.

What about injections of hyaluronic acid? No clinically significant benefits have been found, and platelet-rich plasma injections may even increase the risk of developing a frozen shoulder in the first place, by either causing inflammation or accelerating the buildup of scar tissue.

Steroid injections appear to help symptomatically for up to three months compared to no injections, but they may not offer any clear long-term benefit over getting injected with placebo fluids, like saline. If one were to try steroid injections, there appears to be no difference in clinical outcomes if delivered freehand or via ultrasound guidance, and low-dose steroids appear to be just as effective (or ineffective) as higher doses.

Hydrodilation involves the injection of a large volume of saline into the joint to distend the capsule. This may provide a transient improvement in disability of unclear clinical relevance but no significant improvement in pain.

Suprascapular nerve block involves the ultrasound-guided injection of an anesthetic to temporarily block pain transmission from the frozen shoulder. Thankfully, only about four percent of patients are harmed by the procedure (including about a one in a thousand chance of a lung being punctured), but, arguably, any risk is too great for a so-called treatment that fails to offer clinically meaningful improvements in pain and disability.

What about needles with no injection? Acupuncture. There are a number of ways to create placebo groups in acupuncture trials. One is to use sham acupuncture points, and another is to use sham acupuncture needles. A sham point trial for frozen shoulder showed no significant difference in pain relief between those getting the electroacupuncture in the actual acupuncture points versus needle insertion one to two centimeters away. An innovative sham acupuncture trial allowed for a double-blind study design that involved taped-on press tacks with or without penetrating needles. Surprisingly, neither the acupuncturist nor the patient could tell which was which. No clinically significant benefit was found, similar to what’s been found in a systematic review of acupuncture trials for frozen shoulder. Even though the average reduction in pain wasn’t much, twice as many in the needle group had an immediate, clinically relevant reduction of pain compared to the placebo group. Only one treatment was given, though, with no follow-up.

The more invasive approaches to treat frozen shoulder are manipulation under anesthesia and arthroscopic capsular release. Manipulation under anesthesia involves ripping the internal scar tissue by forcibly stretching the joint while the patient is under, and arthroscopic capsular release involves surgically cutting away the scar tissue. No clinically significant difference in long-term efficacy has been found between the two approaches, though ACR appears to carry quadruple the odds of severe complications. The real question, though, is do they work compared to doing neither?

When manipulation under anesthesia was finally put to the test in a randomized controlled trial, it failed to provide any apparent benefit for pain or disability at any time period. When arthroscopic capsular release and manipulation under anesthesia, both with physical therapy, were pitted head-to-head against physical therapy alone in the famous FROST study, which is meant to stand for FROzen Shoulder Trial, all three groups ultimately improved at a similar rate, demonstrating that these invasive techniques may do more harm than good. Though serious complications were rare, there were ten serious adverse events in the two surgical groups, including a deep vein blood clot, joint infection, and stroke, while there were zero serious events in the non-operation group.

There are dozens of randomized controlled trials of all manner of treatments for frozen shoulder, from lasers to “shockwave” therapy. Overall, even trials showing significant results have been described as “statistically fragile.” The Fragility Index was developed to identify the number of outcome events that, if flipped, would reverse the statistical significance of a study. For example, in the sports medicine literature, on average, if four fewer subjects had positive outcomes in the treatment groups of statistically significant studies, then the studies would lose statistical significance. That’s how close to the edge treatment effects are to being dismissed as chance findings. But they didn’t have four fewer positive outcomes, so they were considered to be statistically significant effects. However, the average number of participants lost to follow-up was eight, meaning we don’t know what happened to them. If they dropped out because the treatment wasn’t working for them, then that would flip the result of the whole study.

It’s even worse in the frozen shoulder literature, which has a Fragility Index of 2, meaning the reversal of just a few outcome events would change the significance of the findings. In other words, even the studies of frozen shoulder treatments that suggest benefit should be interpreted with caution. The bottom line is that no treatment has been proven to consistently shorten the course of the disease. A systematic review and meta-analysis comparing treatments for frozen shoulder concluded: “Because of the paucity of robust evidence, no firm recommendations exist for clinical practice.”

The best treatment may actually be no treatment at all. Unexpectedly, the best clinical outcomes ever published in the medical literature on frozen shoulder came from effectively doing nothing. A landmark study randomized seventy-seven patients with frozen shoulder to intensive physical rehabilitation treatment, including manual stretching and manipulation, versus “supervised neglect,” that is, letting the shoulder heal on its own, being encouraged to resume all activities as tolerated, and only doing painless exercises. A year later, 64 percent of those in the supervised neglect group achieved normal or near-normal shoulder function without pain. Guess how many achieved that in the physical therapy group? None. Zero percent. At the end of two years, 89 percent of individuals in the supervised neglect group were totally pain free with nearly normal range of motion, compared to 63 percent in the physical therapy group. So, the quote-unquote “do nothing” group healed at the same rate in half the time! Think of all the time, pain, and hassle of getting intensive physical therapy for years when you are in effect paying to slow your own recovery.

After all, if a frozen shoulder is your body’s way of protecting your joint, it’s thought that putting additional stress on it just contributes to further inflammation and scarring, exacerbating the condition and slowing down the healing process. Sometimes, less is more.

Is there anything we can do to prevent frozen shoulder the first place? A systematic review and meta-analysis found that blood cholesterol levels were significantly higher in those with frozen shoulder, leading the accompanying commentary to ask “What’s Good for the Heart Is Good for the Shoulder?” Those with high cholesterol may have a 50 percent higher risk of developing the condition, perhaps due to inflammation caused by LDL lodging in the shoulder joint. Statins don’t appear to have a protective effect, though, and most damning for the theory are Mendelian randomization studies that show that those born with high lifelong cholesterol levels regardless of what they eat do not appear to be at higher risk. This suggests that the cholesterol itself isn’t causing problems, but rather something else about cholesterol-raising diets.

The strongest association with frozen shoulder is diabetes, which may offer a clue. Diabetics endogenously produce more advanced glycation end-products, which would otherwise mostly come from meat-rich diets. Capsular tissues taken from frozen shoulder patients were indeed found to have significantly higher levels of AGEs than tissues from people with other shoulder pathologies, such as rotator cuff tears. We know AGEs trigger inflammation, so maybe AGEs play a role in the development of the disease. To reduce exposure to the pro-aging toxins one could reduce the intake of the food group that contain the highest levels of AGEs, meat.

Please consider volunteering to help out on the site.

Motion graphics by Avo Media

Doctor's Note

That video was 13 minutes long! Our videos are typically about five minutes or so—prevention in one video, then treatment in another—but we decided to give you the whole story in one video this time.

This is the last video in a six-part series. If you missed the others, check out:

You may be interested in my new book How Not to Hurt, available now for preorders. (All proceeds received from the book are donated directly to charity.)

If you haven't yet, you can subscribe to our free newsletter. With your subscription, you'll also get notifications for just-released blogs and videos. Check out our information page about our translated resources.

Pin It on Pinterest

Share This