REF-1267 · Viau et al., 2025 · Physiotherapy Theory and Practice · Paper · PubMed

Will a physio-led bike fit fix my knee pain on the bike?

Maybe: in 76 riders with knee pain, worst pain fell 2.5 of 10 points by 12 weeks after a fit plus cadence and load advice and exercises. No control group.

Main caveat: No control group. Natural recovery and regression to the mean are expected in riders who seek care at the peak of a pain episode.
limited evidenceMeasured on recreational to trained athletes

Viau and colleagues followed recreational road cyclists who came to a physiotherapy clinic specialised in cycling with knee pain while riding.

Every rider received the Physiotherapy-led Bike Fitting (PBF) intervention: a comprehensive bike fit "focusing on key measurements such as knee flexion and knee alignment relative to the pedal axis while cycling", tailored education on cycling cadence and training progression, and an exercise prescription.

Knee pain during cycling was rated on the numerical pain rating scale (NPRS, 0-10) at baseline, 4 and 12 weeks. In 76 riders, worst pain fell by 2.52 points (95% CI −3.04 to −2), average pain by 1.81 (−2.27 to −1.36) and least pain by 0.7 (−1.02 to −0.38) at 12 weeks (all p < .001).

The authors conclude that riders exposed to the PBF intervention reported less cycling-related knee pain and disability.

Our reading: the only cohort the base holds of symptomatic cyclists treated with a fit, and the pain fell; but there is no comparison group, so the fall includes whatever would have happened anyway after a rider seeks care at the worst point of a pain episode, and the advice on training progression may have done as much as the fit.

It does not show that a fit, or any saddle change, causes the improvement (SYNTH-73 still holds that changing saddle height has not been shown to resolve knee pain).

Protocol

Design: single-group prospective observational longitudinal cohort; all included riders received the intervention; outcomes at baseline, 4 and 12 weeks; within-group change over time by linear models (abstract). Ethics approval, registration and follow-up completeness are not given in the abstract.
Participants: n = 76 recreational road cyclists with knee pain while cycling, who consulted for cycling-related knee pain at a physiotherapy clinic specialised in cycling (abstract). Age, sex, symptom duration, the site of knee pain, clinical diagnoses and training volume are not given in the abstract; the abstract does not name the clinic.
Intervention (PBF): a comprehensive bike fit focusing on knee flexion and knee alignment relative to the pedal axis while cycling; tailored education on cycling cadence and training progression; exercise prescriptions (abstract). Target ranges, the measurement system (2D or 3D, static or dynamic beyond "while cycling"), the number of sessions, which components were changed for each rider, and the exercises are not given in the abstract.
Primary outcome: knee pain during cycling on the NPRS (0-10), as worst, least and average pain. Disability is named in the conclusion; its instrument and results are not given in the abstract.
Analysis: linear models of within-group change at 4 and 12 weeks; 12-week changes reported with 95% CIs.
Affiliations, funding and interests: Viau, Tremblay and Coutu — Physiovélo clinic, Longueuil, Québec; Desmeules — School of Rehabilitation, Université de Montréal, and Hopital Maisonneuve-Rosemont Research Center; Lafrance — Department of Social and Preventive Medicine, Université Laval, and Centre de recherche du CHU de Québec (PubMed record).Funder in the Crossref and Europe PMC records: Ordre Professionnel de la Physiothérapie du Québec (the provincial professional order of physiotherapists; no award number). The competing-interest statement is not available (abstract only); the abstract does not address any interest of the authors' clinic in the result.

Key Results

Knee pain during cycling (NPRS, 0-10)Change from baseline at 12 weeks (95% CI)p
Worst pain−2.52 (−3.04 to −2)< .001
Average pain−1.81 (−2.27 to −1.36)< .001
Least pain−0.7 (−1.02 to −0.38)< .001
Not in the abstract: baseline pain values, the 4-week changes, the number of riders followed to 12 weeks, the proportion who improved by any threshold, adverse events and the disability results.
Abstract tension: the conclusion reports "a reduction in cycling-related knee pain and disability", but no disability measure or result appears in the abstract.
Clinical importance: the abstract does not say whether the changes exceed a minimal clinically important difference on the NPRS.

Mechanisms

Measured: knee pain during cycling fell within the group after the package, most for worst pain (−2.52) and least for least pain (−0.7).
Authors' interpretation: a physiotherapy-led bike fit that includes a fit, education and exercise prescription "can be helpful among cyclists with knee pain".
Inference: with no control group the fall includes natural recovery, regression to the mean (riders consult when pain is at its worst) and any change in training after the advice; the study cannot say which component helped, or whether the fit added anything. Load is the one factor with moderate evidence of a link to cycling overuse symptoms in the systematic review the base now holds (REF-1260), and advice on training progression was part of this package.
Inference: the fit set knee flexion "while cycling", which matches the review evidence for setting saddle height from the dynamic knee angle rather than a static one (REF-1261); the abstract gives no target range, so the study adds no number for a saddle-height rule.

Practical Application

Will a physio-led bike fit fix my knee pain on the bike? Maybe: in 76 riders with knee pain, worst pain fell 2.5 of 10 points by 12 weeks after a fit plus cadence and load advice and exercises. No control group.
What a clinician-led fit for knee pain looked like here: knee flexion and knee alignment over the pedal axis checked while pedalling, plus advice on cadence and on how fast to build training, plus exercises (B, from the abstract, uncontrolled). Expect a package, not a single saddle tweak.
Ask what changed in the training first. Not from this paper; the base's own practice: find the load change that preceded the pain and take it out (SYNTH-73 step 4; knee pain during low-cadence work: SYNTH-55), change one thing at a time, and see a clinician if it is not improving within about two weeks (SYNTH-73 step 6).
Pain that does not settle, or red flags (swelling, locking or giving way, night pain, pain that changes the pedal stroke) go to a clinician (SYNTH-73, practice).
What it does not show: that saddle height alone, or any single fit change, resolves knee pain; that a clinician-led fit does better than a careful self-adjustment or another fitter (REF-1263 is the survey on self-fit vs professional fit; REF-1262 the randomised trial of a 3D fit vs a handout in 162 amateur cyclists, with comfort and pain rated at 15 days).

Limitations

No control group. Natural recovery and regression to the mean are expected in riders who seek care at the peak of a pain episode; the size of the change cannot be credited to the intervention.
A package. Fit, education on cadence and training progression, and exercises were delivered together; the contribution of the fit itself is unknown.
Who delivered it. The intervention was delivered at a physiotherapy clinic specialised in cycling; three of the five authors list Physiovélo clinic (Longueuil) as their affiliation. The abstract does not name the study clinic or address the clinic's interest in the result; blinding is not described, and pain was self-rated by riders who all received the intervention.
Who was studied. Recreational road cyclists who chose to consult a specialised clinic; not riders who never seek care, not other disciplines.
Self-reported pain on an 11-point scale; no performance, training-volume or return-to-riding outcome in the abstract.
Abstract only: baseline values, number of sessions, adherence, dropouts, the disability instrument and results, the 4-week values, adverse events and the fit targets could not be checked.
Funding and interests: funded by the Ordre Professionnel de la Physiothérapie du Québec (Crossref record); competing interests not available.
How this entry was checked and changed
Why this level

(from the abstract) — the rubric's B row for an uncontrolled pre-post design (n = 76). Not lowered further because the outcome is the rider's own pain and the sample is not small; not higher because nothing separates the package from natural recovery or regression to the mean, the three components cannot be separated, the intervention was delivered at a physiotherapy clinic specialised in cycling while three of the five authors list Physiovélo clinic, Longueuil, QC as their affiliation (PubMed record; the abstract does not say whether it was the study clinic), and baseline values, follow-up completeness and the disability results are not in the abstract. The full text is not open (Taylor & Francis; no repository copy listed by OpenAlex)