What we don’t know
A register, not a disclaimer. Each entry is a specific missing piece of evidence, numbered so it can be cited from the pages it affects — and counted, so you can see how much of this site stands on it.
There is no randomized rehabilitation evidence for patellofemoral lesions.
Not weaker evidence — none. Every loading rule on this site for a kneecap or trochlear graft rests on mechanism, surveyed practice, or evidence from a different lesion. The weight-bearing trials that anchor the tibiofemoral side enrolled femoral condyle lesions only.
Affects 7 claims
Library · Active short-arc motion · Library · Walking in the brace, locked in extension · Library · Leg press, shallow arc · Library · Passive flexion toward full range · Library · Leg press, deepening arc · Library · Cycling, progressive · Library · Stairs, decline and landing · Multifocal disease
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
There is no published rehabilitation protocol for the high-density variant of this operation.
Every program such a patient follows, this site's included, is adapted from evidence built on different implantation densities and different lesions. No rehabilitation study has ever enrolled an HD-ACI cohort.
Nothing cites this entry claim-by-claim, because it conditions everything: for a high-density reader, the entire program is an adaptation.
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
Exercise selection, sets, reps and intensity have never been randomized at any stage.
The numbers on the program pages are this site's own judgement, not a reading of the literature. The trials in this field randomized how fast weight came back onto the leg and described their exercise content only as progressive lower limb and trunk strengthening — so there is no distribution of published sets and reps to sit in the middle of. Where a dose here is a convention repeated across protocols, the page says so; where it is a choice, that is what it is.
Affects the whole program — every dose column on every program page.
Stage III · Loading · Stage IV · Strength · Program · Stage III (patellofemoral) · Program · Stage IV (patellofemoral) · Library · Quadriceps set (isometric) · Library · Straight-leg raise, in the brace · Library · Straight-leg raises, all planes · Library · Active flexion, early and unloaded · Library · The weight-bearing ladder · Library · Ankle pumps and early mobilization · Library · Leg press, shallow arc · Library · Closed-chain quadriceps work begins · Library · Passive flexion toward full range · Library · Active flexion to full range · Library · Walking, unaided, level ground · Library · Cycling, recumbent then upright · Library · Leg press, deepening arc · Library · Leg press toward functional load · Library · Single-leg work · Library · Cycling, progressive · Library · Low-impact cardiovascular work · Library · Stairs, decline and landing · Library · Walking on uneven ground · Library · Heavy strength work, continued · Library · Maintenance strength
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
No study in this literature used a criterion-based battery to decide when a patient progressed.
Which is awkward, because this site publishes gates. Progression was time-based in every study; the batteries that exist were validated in ACL reconstruction. The gates here are assembled from described protocols and surveyed practice, and say so where they are used.
Affects 15 gates
Stage V · Impact · Program · Stage V (patellofemoral) · Library · Straight-leg raise, no lag · Library · Gait, made boring again · Library · Strength measurement · Library · Running progression · Library · Returning to sport · Library · Readiness check
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
Return-to-sport figures pool a 33–96% range, and most of the studies never defined "return".
A pooled 80.3% is arithmetic performed on incomparable definitions — 82% of the professional-athlete component studies omitted return criteria entirely. The range is printed wherever the pooled figure appears.
Affects 3 claims
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
A paper this site could not access, then obtained and read in full.
The canonical tibiofemoral rehabilitation guideline (E4) was cited from its abstract and from protocols that descend from it while the journal blocked retrieval. Kept its number and marked resolved rather than deleted: a site that changes its mind in public should keep the record of having done so.
What changed: Obtained via institutional access and read in full. No contradictions with the conventions previously attributed to it; the phase-by-phase summary now lives in R2 §1b, and the paper's own "Therapy, level 5" self-designation is recorded at every point of use. See it →
Resolved 2026-08-14
No supporting modality — CPM dose, bracing, NMES, hydrotherapy, blood-flow restriction, proprioceptive work — has ever been isolated as a variable in cartilage repair.
Most published protocols never recorded their CPM duration and no trial isolated it; bracing has no comparative trial; NMES and blood-flow restriction were studied in different operations entirely; hydrotherapy sits in the canonical protocol because its authors put it there. Whether any of them changes an outcome after MACI is unknown.
Quoted on 10 pages
Stage II · Protection · Program · Stage II (patellofemoral) · Library · Straight-leg raise, in the brace · Library · Neuromuscular electrical stimulation · Library · Passive motion, staged arcs · Library · Walking in the brace, locked in extension · Library · Hydrotherapy, deep water first · Library · Blood-flow-restriction cuff work · Library · Balance and neuromuscular work · Library · Anti-gravity treadmill
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
No cohort measured what patients actually did after discharge — so whether continued training causes durable outcomes is unanswerable.
The ten-year randomized cohort's own limitations state that long-term compliance and out-of-clinic load were never recorded. The association between staying strong and staying well is real and repeatedly observed; whether maintained strength protects a knee, or a good knee permits both, is exactly what Level IV evidence cannot answer.
Affects 3 claims
Stage VI · Long horizon · Program · Stage VI (patellofemoral) · Library · Walking, unaided, level ground · Library · Strength measurement · Library · Heavy strength work, continued · Library · Maintenance strength · Library · Holding impact capacity · Library · Annual strength review · Library · Years two to five · Library · Second-year vigilance
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
Nothing beyond about 12.6 years exists for patellofemoral ACI.
The longest patellofemoral series reports a mean under twelve years. Everything this site says about the second decade of a patellofemoral graft is extrapolation from tibiofemoral cohorts or from nothing.
Quoted on 2 pages
Stage VI · Long horizon · Program · Stage VI (patellofemoral)
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
The strength cut-offs in use were validated in ligament reconstruction, and limb symmetry can overestimate function.
The authors who reported limb-symmetry values in this literature cautioned themselves that the ratio improves when the uninvolved leg detrains too. The cut-off values clinics reach for were borrowed from a different operation and vary across the literature they were borrowed from.
Affects 2 claims
Library · Single-leg work · Library · Strength measurement · Library · Heavy strength work, continued · Library · Annual strength review
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
Under four per cent of the professional-athlete return-to-sport dataset was female.
The return-to-sport figures at the elite level describe men. The number has simply not been measured for women at that level, and this site reports the gap rather than smoothing over it.
Affects 1 claim
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
The interval between biopsy and implantation has no published guidance, and prehabilitation has never been trialled.
No protocol in existence describes the weeks between the two operations. The case for arriving strong is observational — people who arrived healthier did better, which is not proof that getting healthier changes your result.
Quoted on 1 page
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
No formal consensus statement on cartilage repair rehabilitation exists.
The 2024 ESSKA-AOSSM-AASPT formal consensus covers the meniscus, not cartilage. Re-checked in August 2026: still none. Two structured Delphi documents exist — the twelve-surgeon industry-sponsored survey several pages cite, and a 2024 international Delphi whose panel was surgeons only and in which five of its eleven rehabilitation statements failed to reach consensus at all. Neither is a society-commissioned guideline. If a cartilage equivalent is published it would become the single highest-value source for this project.
Nothing cites this entry directly — it is the reason the sources that do get cited are as weak as they are.
No study ever compared an earlier return — to running, work, or sport — against a later one.
Every return timeline on this site is surveyed practice or one institution's convention. No trial randomized the timing, and no study followed what happened to the people who went back sooner. The two attributed running timelines disagree, and the site prints both rather than averaging them into a third nobody used.
Quoted on 3 pages
Program · Stage V (tibiofemoral) · Library · Returning to work · Library · Running progression
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
No human study has ever tested NSAID use in a chondrocyte implantation population.
The review the field cites for avoiding NSAIDs after cartilage repair contains eighteen studies: four in vitro, thirteen animal, and one human — and the human one is an ACL reconstruction registry where NSAIDs looked neutral-to-favorable. The avoidance recommendation for this population rests entirely on laboratory and animal work showing NSAIDs suppress the cell differentiation a maturing graft depends on. Whether that translates to a worse knee in an actual cartilage-repair patient has never been measured, in either direction. Human evidence about NSAIDs and cartilage does exist outside this population — including randomized evidence that one drug, indomethacin, accelerated osteoarthritis progression — and it is drug-specific and contested. None of it involves a knee carrying a graft.
Nothing cites this entry yet — the medication content it conditions is still being written. It is recorded now because the search that established it has been performed.
No study of any design examines a corticosteroid injection into a knee that carries a cartilage graft.
Searched directly: no clinical study, registry analysis, guideline, or consensus statement exists, and the 2026 four-society injection guideline does not mention cartilage-repair patients. What exists is triangulation — repeated steroid injections thinned osteoarthritic cartilage in a randomized trial, steroids are dose-dependently toxic to chondrocytes in the laboratory, and injections within about a month of knee arthroscopy carried several-fold higher infection risk. None of that is a MACI knee, and the absence of evidence cuts both ways.
Nothing cites this entry yet — the risks content it belongs to is still being written. Recorded now because the search has been performed.
There is no data on driving after cartilage repair — and the borrowed numbers probably run optimistic here.
Zero studies of braking or driving in any cartilage repair population, and the MACI surgeon consensus mentions driving once, as a transportation aside. The 4–6 week figures published for other knee surgery assume early weight-bearing; a tibiofemoral MACI patient is not fully weight-bearing until 7–9 weeks and wears a range-limiting brace that measurably impairs braking in healthy people at every setting tested. Every published number is a brake-response surrogate — no study links any of them to actual crashes.
Nothing cites this entry yet — the practical content it conditions is still being written. Recorded now because the search has been performed.
Cryotherapy has never been isolated as a variable after any cartilage repair procedure.
It sits inside the trial protocols as standard comfort care, so its use is documented practice — but no study of any design has tested whether it changes anything after cartilage repair. The adjacent knee-surgery evidence shows small, short-lived, low-certainty benefits for pain and opioid use; and surface icing genuinely cools the joint interior, which means a maturing graft is exposed to the temperature change with no evidence about what, if anything, that does.
Nothing cites this entry yet — cryotherapy appears on the site only inside practical content. Recorded now because the search has been performed.
No rehabilitation protocol of any kind addresses a knee with grafts in both compartments at once.
The weight-bearing trials enrolled femoral condyle lesions; the patellofemoral guidance is consensus and surveyed practice; no study, protocol, or consensus statement describes a patient whose grafts sit in both compartments simultaneously — even though multifocal disease is common enough that the long patellofemoral series are full of it. Where the two location programs disagree, nothing published says which restriction governs. Every combined instruction on this site is authored inference, labelled as such, and the disagreements it has not yet resolved are marked rather than averaged.
Quoted on 1 page
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
Nobody has compared how fast to progress *within* a phase — only how fast to reach full weight-bearing.
The randomized trials in this field varied one thing: the week full weight arrived. Between those points nothing has been compared — not the order movements are added in, not how long a knee should hold at one level before the next, not whether a phase should be divided at all. So the divisions this site draws inside a stage are its own reading of where the published change-points fall, and where a source falls silent the division is a judgement rather than a finding. It is stated on every one of them rather than left to be inferred.
Referenced by every beat in content/beats.ts, which the reach walk does not yet cover — it indexes gates, claims and benchmarks only. Teaching it to walk beats is owed, and this note comes out when it does (rule 26 enforces both directions).
The research this site rests on names the post-operative warning signs and attaches a number to none of them; the one number on the red-flags page, a temperature, comes from clinics' patient instructions and a physiotherapy guideline instead.
Infection, clot, wound breakdown and graft-related locking appear across this literature as complications to watch for, and the risk windows are documented — but no study in this site's research states the temperature, the swelling, or the range loss at which a reader should pick up the phone. The temperature on the red-flags page comes from the post-operative instructions two clinics give for this kind of surgery and from a national physiotherapy guideline for joint replacement, and the two do not give the same number. This is not a claim that medicine at large lacks thresholds; it is a statement about what this site can and cannot cite. On that page the cost of calling unnecessarily is a phone call and the cost of not calling can be a pulmonary embolism, so every line there is written to err toward calling, and no threshold on it should be read as a permission to wait.
Reached by the red-flags page, whose warning signs carry no number from a study — RedFlag carries no citations, which is the reason this entry exists.
The swelling scale this site asks you to grade yourself on has never been searched for in this project's literature.
The stroke test — sweep the fluid up out of the hollow, watch whether and how fast it comes back — is a standard knee examination with a published grading scale behind it, and it is the single measurement this site leans on hardest: it sets the step-back rule, it colours the ledger, and it is half of the pair that tells you whether yesterday's dose was right. None of that rests on anything in this site's research. The scale does not appear in any file in docs/research/, its reliability after this operation has not been looked up, and no source here establishes that a grade on any given morning means what the site treats it as meaning. That is a gap in the work rather than a proven gap in the evidence, and it is recorded as the former. Until it is closed, read the trend rather than the grade: a knee that is fuller three mornings running is telling you something whatever number you put on it.
Reached from /record and the ledger through the tracker registry, which the reach walk does not cover — it indexes gates, claims and benchmarks only.
No trial has compared the high-density technique against standard MACI in people, in either direction.
The argument for the higher cell dose is one sheep study, which found five million chondrocytes per cm² produced less variable, more normal-looking cartilage than one million. The human evidence is single-arm case series from the centre that developed the technique — good enough to show the operation works in the knees it has been done in, and silent on whether it works better than the alternative. Nobody has run the comparison, so neither direction has been shown: this is an untested question, not a null result. Read any claim of superiority, in either direction, as a claim about mechanism rather than about outcomes.
Quoted on 1 page
Each row is the claim’s own words and links to it in place. This list is generated by walking the citation graph at publish — it cannot drift from the pages, and the count in the register is its length, not a typed number.
Entries are never renumbered and never deleted. When one is resolved it stays in place, marked, with what changed linked from it.