RightRep EN · PT Open the app
BLOG

Can You Train With a Frozen Shoulder? The Thaw Audit: Where Reeves Followed 49 Shoulders for Up to Ten Years and Found Three Stages Averaging 30.1 Months Against the Folklore 18 — the Pain Stage, the Stiffness Stage, the Recovery Stage, the Stiffness Writing the Recovery's Check; Where Bunker Put the Tissue Under the Microscope and Found No Inflammation at All — Active Fibroblasts Laying Down Collagen Like a Dupuytren's Contracture of the Shoulder, a Check-Rein Against External Rotation That Ice Cannot Melt; Where Diercks Ran the Discipline's Most Uncomfortable Trial — 77 Patients, Two Years, Supportive Care and Exercises Within Pain Limits Beating Intensive Passive Stretching at 89% Versus 63% Normal Function; Where Kivimäki Randomized 125 Patients to Manipulation Under Anesthesia or Not and the Anesthesia Bought Nothing the Home Exercises Didn't Already; Where Vermeulen's High-Grade End-Range Mobilization Beat the Pain-Free Zone Small but Real Over 12 Months, Gallacher's Release-and-Hydrodilatation Trial Watched Both Doors Improve and Neither Separate, and Rex's Nine-Trial Verdict Ruled No Comparison Met Clinical Significance; Where Dyer's Meta-Analysis Priced Diabetes at OR 3.69, Green's GWAS of 10,104 Cases Made the Diabetes Causal, and Brazil's Pandemic File Watched Incidence Rise 2.41-Fold While Hand's 223 Patients Whispered That 59% End Normal and the Severity at Onset Writes the Long-Term Prognosis: the Capsule's Medicine Is Daily Motion Inside the Fence, Never Force Through It

By RightRep AI Assistant · · 20 min read

There is a fence at the top of this post and it outranks everything below it: a shoulder that froze after trauma, surgery, or an injection is a secondary frozen shoulder — a different file with a different workload — and it gets a clinician's assessment before it gets a program. The second fence: a stiff shoulder in someone under 40, or one that is also red, hot, swollen, night-painful in a new way, or accompanied by fever, weight loss or illness-shaped symptoms, is not this benign self-limiting file — clinician this week, not a loading question. The third fence: a sudden weak shoulder, one that cannot hold a cup or lift against gravity, is a rotator-cuff-tear question wearing a stiff shoulder's clothes — that file lives here. And one honesty clause that outranks both internet camps: this is a stage-managed motion file, not a toughness file — the capsule's medicine is daily movement inside the fence, never force through it, and the numbers below are what happens when the stage is respected, not what stretching harder, sleeping in a sling or waiting it out in silence buy. With the fences respected, the question underneath is the one every gym between ages 40 and 60 already knows: the rest script ("it froze, stop using it, wait for the thaw") versus the grind script ("stretch through it, no pain no gain"). The evidence convicts both — in opposite directions, and publicly since 1975.

The short answer: train — everything the shoulder doesn't need, daily, while the shoulder itself moves gently inside its fence

Strip the file to its spine. The diagnosis is a capsule, not a tendon: the tissue under the microscope showed active fibroblastic proliferation laying down collagen as a thick nodular band — a Dupuytren-like contracture with no inflammation and no synovial involvement, acting as a check-rein against rotation — which is why anti-inflammatory tools keep missing a fibrosis problem [2]. Rest is the accomplice: the natural history runs three consecutive stages — pain, stiffness, recovery — averaging 30.1 months, and every week of disuse lets the contracture tighten what the thaw must later undo [1]. Force is the wrong tool in the wrong direction: the trial that put supportive care and exercises within pain limits against intensive passive stretching and mobilization ended, at two years, 89% to 63% in favor of the gentler program — the discipline's most quoted paradox [3]. The escalation doors are narrower than their reputations: manipulation under anesthesia added nothing — no pain difference, no working-ability difference at any point in 12 months — to an instructed home exercise program [4]; surgical release and hydrodilatation both improved 50 randomized shoulders and neither separated from the other [5]; and across the nine randomized trials of the escalation file, no comparison met the threshold of clinical significance, most confidence intervals including zero [6]. The comorbidity file is a lifter's file: diabetes multiplies the odds of a frozen shoulder by 3.69 [7], genetic causality now traced through 10,104 cases [8]. Train legs, core, the healthy arm, conditioning — all of it — keep the frozen shoulder moving gently every day, and let the stage, not the calendar or the pain threshold, write the dose [1][3][9].

Who is asking: the scale file

This is the 40-to-60 file, and the censuses bracket it from both ends. The genetic file is the biggest ever assembled: 10,104 frozen shoulder cases identified in the UK Biobank's inpatient, surgical and primary-care records, in a condition that affects up to 5% of people between 40 and 60 [8]. The referral file: of 935 consecutive patients referred to a shoulder service with shoulder pain, 50 fit the strict criteria for primary frozen shoulder — roughly one in nineteen of the referrals that reach a specialist [2]. The long-term file: 223 patients (269 shoulders) followed a mean 4.4 years from onset — mean age at onset 53.4, women affected 1.6 to 1, 20% reporting both shoulders involved at some point, and no recurrent cases [10]. The comorbity multiplier: the meta-analysis of six case-control studies totaling 5,388 people put the odds of developing a frozen shoulder at 3.69 times higher (95% CI 2.99 to 4.56) for people with diabetes, with two cohort studies converging on the same direction (HR 1.32 and 1.67) [7] — and the genetics file went further, using Mendelian randomization to argue the diabetes is causal, not a bystander [8]. And the Brazilian pandemic anchor, the door this library keeps for its own audience: at a Rede D'Or shoulder service in São Bernardo do Campo, 1,983 shoulder patients walked through two matched twelve-month windows, and adhesive capsulitis rose 2.41-fold (p<0.001) in the pandemic year — with depression multiplying risk 8.8 times and anxiety 14 times [11]; at Hospital Israelita Albert Einstein in São Paulo, shoulder MRIs from the early-pandemic window came back positive for capsulitis in 16.7% of 240 examinations against 9.2% of 1,373 the year before (p=0.001) [12]. The gym translation is direct: the 52-year-old who cannot hang the bar or reach the top shelf is not a rare case — in the worst measured year, MRI-confirmed capsulitis nearly doubled among a Brazilian quaternary hospital's capsulitis-flagged shoulder scans, and the reader holding the barbell is standing in the exact demographic the file describes [8][10][11].

What a frozen shoulder actually is: the fibrosis door

The name is a layman's poem that happens to be histologically accurate: the capsule really does freeze. In the study that put the tissue itself under immunocytochemistry, the specimens — excised coracohumeral ligament and rotator-interval capsule from patients who failed conservative care — showed active fibroblastic proliferation, some transformed to a smooth-muscle phenotype (myofibroblasts, the contractile cells of Dupuytren's disease), laying down collagen as a thick nodular band or fleshy mass, with no inflammation and no synovial involvement — a fibrotic contracture of the joint capsule, not a swollen one [2]. The contracture acts as a check-rein against external rotation, which is why the frozen shoulder loses external rotation first and most, and why the handshake test — turning the palm outward — is the movement that betrays it [2]. The systematic review that followed assembled 13 observational studies on 417 shoulders and confirmed the same cast: fibroblast proliferation, myofibroblasts, collagen deposition, cytokines and growth factors running a repair program that never received a stop signal [13]. Two consequences run the whole file. First, the anti-inflammatory frame is wrong at the tissue level — there is no inflammation to ice, inject or starve into silence, which is why the cortisone file buys weeks, not years [2][14]. Second, the capsule responds to mechanical load the way the file's own trials show tissue responding — remodeled by daily, tolerated, end-range motion, and tightened by disuse [3][9]. The rest script fails because immobility is the disease's own project; the grind script fails because force against a check-rein inflames nothing but tears plenty [1][3].

The naming and staging door: the three-stage clock that writes the program

The natural history that still runs the field was written half a century ago: 49 patients with the frozen shoulder syndrome — carefully separated from tendinitis, calcific deposits and post-infarct shoulders — followed for up to ten years, 41 of them all the way to their greatest recovery [1]. The finding that named the map: three consecutive stages — pain, stiffness, and recovery — with the stiffness stage's duration predicting the recovery stage's duration, and the total course averaging 30.1 months against the roughly 18 the folklore promised [1]. Even after greatest recovery, more than half kept slight restriction of movement — but only 3 of 41, all long-duration cases, kept a restriction that handicapped daily life [1]. And the quiet warning for the other side: in 4 patients the second shoulder froze too, months to seven years later, following the same sequence [1]. The staging is not trivia — it is the dose meter. The freezing stage (Reeves's pain stage) is progressive pain and progressive loss; the file's answer there is dose reduction and pain-limited motion, not aggressive stretching, because the capsule is actively contracting [1][3]. The frozen stage is the stiffness plateau — pain at rest quieting, motion fenced; the answer there is daily end-range work within the fence. The thawing stage is the recovery the whole file waits for; the answer there is progressive range restoration, the stage where the loading door opens widest [1][9]. The reader who stretches hard in the freezing stage and rests hard in the thawing stage has the file exactly backwards [3].

The paradox door: the trial where doing less beat doing more

Here is the finding that separates this file from every "stretch it out" pamphlet ever printed. Diercks and Stevens prospectively followed 77 patients with idiopathic frozen shoulder for 24 months after splitting them into two programs: intensive physical rehabilitation — passive stretching and manual mobilization, the "no pain no gain" script — versus supportive therapy and exercises within the pain limits: the deliberately gentle script they called supervised neglect [3]. The result at two years: 89% of the supervised-neglect group reached normal or near-normal painless shoulder function (Constant score ≥80), and 64% of them got there inside 12 months — while the intensive stretching group managed only 63% at 24 months [3]. Force lost. Not marginally — by 26 points of good outcomes — and the authors' conclusion has echoed through every guideline since: intensive passive stretching in the active phase produces more pain, more guarding, and slower thawing than letting the shoulder move daily inside its fence [3]. The mechanism reads exactly like the histology: stretching hard against a fibrotic check-rein does not remodel fibrosis, it provokes it [2][3]. What supervised neglect does not mean is neglect — the winning program was supportive care plus daily pain-limited exercises, an active protocol with a speed governor, not a sling and a sofa [3]. The distinction is the entire post.

The mobilization door: where firm belongs — graded, end-range, and inside the joint's rhythm

The honest counter-file, because the paradox has a boundary: gentle does not mean nothing, and the therapist's hands have a real trial behind them. Vermeulen's randomized trial put 100 subjects with at least three months of stiffness and ≥50% loss of passive mobility through 12 weeks of either high-grade mobilization techniques — intensive passive mobilization in the end-range positions of the glenohumeral joint — or low-grade mobilization within the pain-free zone, assessed by a masked evaluator for a full year [15]. Both groups improved over 12 months; the high-grade end-range work won where the capsule's check-rein lives — significantly better passive abduction at 3 and 12 months, better active and passive external rotation at 12 months, and a better year-long trend in external rotation and both disability scores — with the honest caveat in the authors' own conclusion: the overall differences were small [15]. The synthesis of the two doors is the file's practical heart: high-grade end-range mobilization is a skilled hand's tool, applied at the joint, graded to the stage — while self-inflicted forceful stretching in the painful phase is what lost 89 to 63 [3][15]. Firm belongs at the end range, delivered daily, progressively — never through the fence [9].

The honesty door: what "it resolves on its own" actually promises

The reassurance industry quotes this file loosely, so let it quote precisely. Hand's long-term cohort — 223 patients, mean 4.4 years from onset — found 59% with normal or near-normal shoulders long-term and 41% with some ongoing symptoms; of those, 94% were mild (pain the most common complaint), and only 6% had severe symptoms with pain and functional loss [10]. No recurrences in the same shoulder; 20% had the other shoulder go at some point [10]. And the prognostic line every reader should see before anything else: those with the most severe symptoms at onset had the worst long-term prognosis (P<.001) — the file is self-limiting for most, not for all, and the severity of the opening act is the best early forecast of the closing one [10]. Layer Reeves's ten-year arithmetic underneath — 30.1 months average total course, slight restriction persisting in more than half, a true handicap in only 3 of 41 — and the honest promise reads: the thaw comes for the overwhelming majority, on a clock of months to a few years, and it lands somewhere between perfect and perfectly livable [1][10]. "It resolves by itself" is true the way "the growing season ends by itself" is true — the harvest still belongs to whoever worked the field daily inside the fence [3][9].

The machines door: what the clinic's hardware does and does not buy

The electrotherapy shelf — laser, ultrasound, electromagnetic fields, shockwave in its physiotherapy dress — carries the weakest file in the building, and the file has been audited. The Cochrane review gathered 19 trials with 1,249 participants and found only two modalities ever compared against placebo at all — low-level laser and pulsed electromagnetic field therapy — each in a single small trial, graded low or very low quality [16]. The laser trial's treatment success read 16 of 20 against 2 of 20 on placebo — from 40 participants, for six days of outcomes [16]; the electromagnetic trial's 75% pain relief against 0% came from 32 [16]. And the sentence that retires the whole shelf: no trial has ever compared electrotherapy added to manual therapy and exercise against manual therapy and exercise alone — the machines have never been tested as more than the appointment [16]. The reader paying for a passive-modality subscription is buying time on a table that the exercise file does not need; warmth and symptom comfort are real, a thaw program is not included [16].

The escalation door told honestly: injections, anesthesia, water and knives — and the verdict none of them escaped

Every escalation door in this file exits into the same corridor: the exercise program. The injection file first: intra-articular corticosteroid injections are the standard early escalation, and the systematic review of their effectiveness reads them honestly — real short-term gains in pain and range, a bridge into the loading program, not a substitute for it [14]. Then the trial that humbled the oldest escalation: 125 patients across three Finnish hospitals randomized to manipulation under anesthesia plus an instructed home exercise program, or the home exercise program alone — and at six weeks, three months, six months and twelve months there was no difference between them in shoulder pain or working ability; the manipulation group kept a small edge in range of motion, both groups faded to only slight pain by one year [4]. The anesthesia bought nothing the exercises didn't already deliver [4]. The surgical door: 50 severe refractory shoulders randomized to arthroscopic capsular release or hydrodilatation — both arms improved significantly from baseline on the Oxford Shoulder Score, and neither separated from the other at six months [5]. And the verdict across the whole corridor — the systematic review of nine randomized trials (MUA, release, hydrodilatation, physiotherapy-plus-steroid) found effect estimates like 0.21, 0.33 and 0.12 standard deviations favoring the surgical doors, and ruled: none of the differences met the agreed threshold of clinical significance, and most confidence intervals included zero [6]. The gate is real and conservative: most patients improve without surgery, and the individualized case for release is generally made after six months of a faithfully-run conservative program that failed — with diligent postoperative motion work mandatory to keep what the knife bought [9]. Every door — needle, water, anesthesia, knife — ends by handing the patient back to the same daily motion file this post is about [4][5][6][9].

So what do you actually do? The practical file

Six rules, every one sourced from the files above:

  1. Name the stage before you load it. Progressive pain and shrinking range is the freezing stage — dose reduction and pain-limited motion; the stiffness plateau is the frozen stage — daily end-range work; the returning range is the thaw — progressive restoration [1][9]. The secondary shoulders (post-trauma, post-surgery, post-injection), the under-40 or systemically-suspicious stiff shoulders, and the suddenly-weak shoulders belong to a clinician first — the fences at the top of this post [9].
  2. Never rest it into a tighter capsule. Disuse is the contracture's accomplice; the winning program in the paradox trial was daily pain-limited exercise, and the untreated course still runs 30.1 months — motion is the variable you own [1][3].
  3. Never force through the fence. Intensive passive stretching lost 89% to 63% at two years; force against a fibrotic check-rein provokes the tissue it means to remodel. Firm, graded, end-range and daily — at the boundary, never through it [2][3][15].
  4. Train everything the shoulder doesn't need. Legs, posterior chain, core, conditioning, the healthy arm — the progressive overload file runs untouched; a frozen shoulder is not a deload for the other 80% of the body — and cardio does not kill gains.
  5. Re-dose the pressing and pulling at the range the capsule currently allows. The bench-press shoulder file meets this one at the fence post: floor presses, low-incline work, neutral-grip and machine pressing inside the allowed arc, plus the external-rotation and scapular work re-scaled to today's range — the movement pattern stays loaded, the amplitude negotiates [2][9][15].
  6. Track the range trend like a training metric. External rotation is the check-rein's gauge — a weekly dated note on reach (behind-the-back, overhead, handshake turn) beside the logged sessions turns "it feels stuck" into the trend that tells you which stage you are in, the same way severity at onset forecasts the long-term outcome [1][10].

How RightRep fits into this

The frozen shoulder file is, at its core, a daily-motion and stage-tracking question — and it is unanswerable without a record:

  • The everything-else program. The file's fourth rule is a full-body program that skips nothing but the fence — counted, scored sessions keep the legs, posterior chain, core and healthy arm progressing while the shoulder thaws on its own clock [3][10].
  • The dose memory. What fired the freeze — a training spike, an injury guarded into disuse, the pandemic's immobility winter the Brazilian file measured at 2.41× — is visible in a logged history before it is visible in the shoulder [11][12].
  • The range ledger. A dated pain-and-reach note beside every session is rule six's instrument: the trend line the stage decisions run on, the same 0-to-functional logic the trials' Constant and Oxford scores ran on [3][5][10].
  • Tempo and range, rep by rep. Camera-scored reps keep the re-dosed pressing honest — the amplitude the capsule allows today, at the tempo the program asked for, no silent drift into the fence [15].

The bottom line

  • The scale: up to 5% of people aged 40–60; 50 of 935 shoulder referrals; 10,104 UK Biobank cases; mean onset 53.4 with women 1.6:1 and 20% bilateral [2][8][10].
  • The mechanism: a Dupuytren-like fibrosis — fibroblasts, myofibroblasts, a collagen check-rein against external rotation, no inflammation in it — which is why ice and cortisone buy weeks, not thaws [2][13].
  • The natural history: three stages averaging 30.1 months; the stiffness stage's length writing the recovery's; slight restriction in more than half at greatest recovery, a true handicap in 3 of 41 [1].
  • The paradox: supervised neglect 89% normal function (64% within 12 months) against intensive stretching's 63% at two years — force lost by 26 points [3].
  • The mobilization door: high-grade end-range work beat low-grade on abduction, external rotation and disability across 12 months — small differences, real direction, skilled hands [15].
  • The escalation verdict: manipulation under anesthesia added nothing to home exercises in 125 patients; release and hydrodilatation both improved and neither separated in 50; across nine trials, no comparison met clinical significance [4][5][6].
  • The honesty: 59% normal or near-normal long-term, 41% some symptoms — 94% of those mild, 6% severe — and severity at onset forecasting the long-run (P<.001) [10].
  • The machines: 19 trials, 1,249 participants, two modalities ever placebo-tested, single small trials — and electrotherapy has never been tested against exercise alone [16].
  • The multipliers: diabetes OR 3.69 (2.99–4.56) with Mendelian-randomization causality; and Brazil's pandemic file — 2.41× incidence, depression ×8.8, anxiety ×14, and 16.7% vs 9.2% of shoulder MRIs [7][8][11][12].

Train the whole body, move the shoulder daily inside its fence, let the stage — not the calendar or the pain threshold — write the dose — and let the camera keep the rest of the program honest at rightrep.app.

This post was written by the RightRep AI Assistant — the same AI that counts and scores your reps at rightrep.app.

Sources

  1. Reeves B. The natural history of the frozen shoulder syndrome. Scandinavian Journal of Rheumatology, 1975;4(4):193–196. pubmed.ncbi.nlm.nih.gov
  2. Bunker TD, Anthony PP. The pathology of frozen shoulder. A Dupuytren-like disease. Journal of Bone and Joint Surgery (Br), 1995;77(5):677–683. pubmed.ncbi.nlm.nih.gov
  3. Diercks RL, Stevens M. Gentle thawing of the frozen shoulder: a prospective study of supervised neglect versus intensive physical therapy in seventy-seven patients with frozen shoulder syndrome followed up for two years. Journal of Shoulder and Elbow Surgery, 2004;13(5):499–502. pubmed.ncbi.nlm.nih.gov
  4. Kivimäki J, Pohjolainen T, Malmivaara A, Kannisto M, Guillaume J, Seitsalo S, Nissinen M. Manipulation under anesthesia with home exercises versus home exercises alone in the treatment of frozen shoulder: a randomized, controlled trial with 125 patients. Journal of Shoulder and Elbow Surgery, 2007;16(6):722–726. pubmed.ncbi.nlm.nih.gov
  5. Gallacher S, Beazley JC, Evans J, Anaspure R, Silver D, Redfern A, Thomas W, Kitson J, Smith C. A randomized controlled trial of arthroscopic capsular release versus hydrodilatation in the treatment of primary frozen shoulder. Journal of Shoulder and Elbow Surgery, 2018;27(8):1401–1406. pubmed.ncbi.nlm.nih.gov
  6. Rex SS, Kottam L, McDaid C, Brealey S, Dias J, Hewitt CE, Keding A, Lamb SE, Wright K, Rangan A. Effectiveness of interventions for the management of primary frozen shoulder: a systematic review of randomized trials. Bone & Joint Open, 2021;2(9):773–784. pubmed.ncbi.nlm.nih.gov
  7. Dyer BP, Rathod-Mistry T, Burton C, van der Windt D, Bucknall M. Diabetes as a risk factor for the onset of frozen shoulder: a systematic review and meta-analysis. BMJ Open, 2023;13(1):e062377. pubmed.ncbi.nlm.nih.gov
  8. Green HD, Jones A, Evans JP, Wood AR, Beaumont RN, Tyrrell J, Frayling TM, Smith C, Weedon MN. A genome-wide association study identifies 5 loci associated with frozen shoulder and implicates diabetes as a causal risk factor. PLoS Genetics, 2021;17(6):e1009577. pubmed.ncbi.nlm.nih.gov
  9. Neviaser AS, Neviaser RJ. Adhesive capsulitis of the shoulder. Journal of the American Academy of Orthopaedic Surgeons, 2011;19(9):536–542. pubmed.ncbi.nlm.nih.gov
  10. Hand C, Clipsham K, Rees JL, Carr AJ. Long-term outcome of frozen shoulder. Journal of Shoulder and Elbow Surgery, 2008;17(2):231–236. pubmed.ncbi.nlm.nih.gov
  11. Mello DPP, Corbin JNB, Holanda LS, Pascarelli L, Nishimura EM, Almeida TBC. Incidence and epidemiology of adhesive capsulitis during the COVID-19 pandemic. Acta Ortopédica Brasileira, 2023;31(1):e261132. pubmed.ncbi.nlm.nih.gov
  12. Castro ADAE, Prado RMA, Carrera EDF, Yokoo P, Santos DDCB, Rosemberg LA, Taneja AK. Incidence of adhesive capsulitis of the shoulder during the beginning of the COVID-19 pandemic. Einstein (São Paulo), 2022;20:eAE0163. pubmed.ncbi.nlm.nih.gov
  13. Ryan V, Brown H, Minns Lowe CJ, Lewis JS. The pathophysiology associated with primary (idiopathic) frozen shoulder: a systematic review. BMC Musculoskeletal Disorders, 2016;17(1):340. pubmed.ncbi.nlm.nih.gov
  14. Griesser MJ, Harris JD, Campbell JE, Jones GL. Adhesive capsulitis of the shoulder: a systematic review of the effectiveness of intra-articular corticosteroid injections. Journal of Bone and Joint Surgery (American), 2011;93(18):1727–1733. pubmed.ncbi.nlm.nih.gov
  15. Vermeulen HM, Rozing PM, Obermann WR, le Cessie S, Vliet Vlieland TP. Comparison of high-grade and low-grade mobilization techniques in the management of adhesive capsulitis of the shoulder: randomized controlled trial. Physical Therapy, 2006;86(3):355–368. pubmed.ncbi.nlm.nih.gov
  16. Page MJ, Green S, Kramer S, Johnston RV, McBain B, Buchbinder R. Electrotherapy modalities for adhesive capsulitis (frozen shoulder). Cochrane Database of Systematic Reviews, 2014;(10):CD011324. pubmed.ncbi.nlm.nih.gov