REF-1260 · Visentini et al., 2022 · Journal of Science and Medicine in Sport · Paper · PubMed

Did my bike set-up cause my pain?

Not shown: in 18 studies no bike, body or load measure was strongly linked to overuse pain; load had moderate support, saddle height conflicting. Check what changed in training too.

Main caveat: From the abstract only. The full text (Elsevier) is not open, and OpenAlex lists no repository copy.
moderate evidenceMeasured on recreational to world-class athletes

Visentini et al. (La Trobe University; two authors also list a physiotherapy practice, Physiosports Brighton and Kiama Physiotherapy) asked which measures of the bike, the cyclist's body and training load go with overuse pain or injury in cycling.

They searched Medline, CINAHL, EMBASE, SPORTDiscus and the Cochrane Library, checked references, tracked citations, searched grey literature and reviewed key publications by known experts in the industry. Of 3,596 records, 18 studies met the criteria; 3 were judged at low risk of bias.

The best-evidence synthesis found moderate evidence of a relationship between load and symptoms, moderate evidence of no relationship between overuse injury or pain and many measures traditionally used in bike fitting, and conflicting evidence for seat height.

The authors conclude that no measure of the bike, the body or load is strongly linked to cycling overuse pain or injury, and that high-quality risk-factor studies are scarce.

Our reading: this is the base's first systematic statement that measured bike geometry has not been shown to predict pain while training load has moderate support; it fits the order the base already uses (find the load change first, SYNTH-73) and leaves saddle height as a candidate rather than a cause.

From the abstract, so which load measures, which fit measures and which pain sites were studied, and how large any association was, could not be read.

Protocol

Design: systematic review with a best-evidence synthesis (qualitative grading of evidence across studies, no pooled effect sizes).
Sources: Medline, CINAHL, EMBASE, SPORTDiscus and the Cochrane Library; reference checking and citation tracking of included articles; grey literature; a review of key publications by known experts in the industry. Search dates not in the abstract.
Eligibility: studies evaluating the association between specific measures of the bike, of the cyclist's body and of load characteristics, and overuse pain or injury in cycling.
Included: n = 18 studies from 3,596 records; 3 at low risk of bias. Designs, participants, disciplines and pain sites not given in the abstract.
Methods named in the reference list, not in the abstract (inference from the Crossref reference list): a prognostic-factor risk-of-bias tool (Hayden 2013), the Cochrane Back Review Group evidence-level method (van Tulder 2003) and best-evidence synthesis (Slavin 1995); that these were the methods used is our reading.
Funding and interests: Crossref and Europe PMC list the University of Melbourne as the funder (none of the affiliations printed is Melbourne). Printed affiliations: first author La Trobe University and Physiosports Brighton (electronic address at physiosports.com.au), second author Kiama Physiotherapy, third author La Trobe University; the records say nothing about what services the practices offer. No conflict-of-interest statement was available (full text not read).

Key Results

Factor groupBest-evidence synthesis (abstract)
Load characteristicsModerate evidence of a relationship with symptoms
Many measures traditionally used in bike fittingModerate evidence of no relationship with overuse pain or injury
Seat (saddle) heightConflicting evidence
Any measure of the bike, the body or loadNo strong evidence of a relationship
Search yield: 3,596 records, 18 included, 3 at low risk of bias.
Not in the abstract: which load measures (volume, hours, distance, intensity, recent change), which bike-fit measures were found unrelated, effect sizes, pain sites, and how many studies stood behind each grade.
The reference list (Crossref) cites, among cycling studies, the professional-rider survey (REF-921), the hand-nerve studies (REF-890, REF-1039), a club-rider prospective study, mountain-bike and recreational-tour surveys, a bike-fit comfort crossover and saddle and handlebar studies in riders; it does not cite the international bike-fit survey published in December 2018 (REF-1263). Which were included is not stated in the abstract.

Mechanisms

Measured: none; the review grades associations reported by observational studies.
Authors' interpretation: many theories exist about the cause of overuse symptoms in cycling, but there are few high-quality studies of risk factors, so none can be called established.
Inference: "moderate evidence of no relationship" for fit measures comes from riders on bikes they already ride, mostly set within usual ranges; a narrow spread of positions makes a relationship hard to find, so it does not show that a set-up far outside the usual range is harmless.
Inference: load and symptoms measured at the same time can run both ways — more riding exposes the tissue, and pain can change how much a rider rides — so "moderate evidence of a relationship" is not evidence that cutting load prevents pain.

Practical Application

Did my bike set-up cause my pain? Not shown: in 18 studies no bike, body or load measure was strongly linked to overuse pain; load had moderate support, saddle height conflicting. Check what changed in training too.
Ask about load before blaming the bike (B+, from the abstract, observational): load was the factor with moderate support, measured bike-fit factors were not.Not from this paper; the base's own practice: find the load change that preceded the pain and take it out, not all training (SYNTH-73 step 4); the 2-6-week window and the checklist (volume, long rides, climbing, low-cadence work, a new indoor block, a race, then equipment) are new practice proposed for the bike-fit guide, not yet in SYNTH-73 (see also SYNTH-29 — for cyclists every ramp number is practice).
Saddle height stays a candidate, not a cause: the evidence on it conflicts. The mechanics are in REF-882 and its update REF-1261: set it from the knee angle while pedalling, and expect changes under about 4% of leg length to change joint loads little; neither shows that changing it relieves pain.
A fit is not shown to prevent injury: this review found no measure strongly linked to injury, and the one trial of a professional fit measured comfort and self-rated pain for 15 days, not injury (REF-1262); in a survey, a self-fit and a professional fit went with the same odds of pain-free riding (REF-1263).
What it does not show: that changing the bike, or the load, relieves or prevents pain — no trial is in the review's abstract. Red flags and the two-week rule stay as in SYNTH-73.

Limitations

From the abstract only. The full text (Elsevier) is not open, and OpenAlex lists no repository copy; the designs, populations, measures, effect sizes, risk-of-bias tool and the evidence-grading rules could not be read.
Weak primary evidence. 15 of 18 studies were not at low risk of bias; cross-sectional and retrospective designs (as far as the reference list suggests) cannot separate cause from consequence; a best-evidence synthesis counts concordant studies, so "moderate" can rest on a few small studies.
Undefined exposures. "Load" and "many measures traditionally used in bike fitting" are not specified in the abstract, so the finding cannot be mapped onto a particular adjustment.
Industry input to the evidence pool. The search included grey literature and key publications by known experts in the industry; how these were weighted is not in the abstract.
Search date unknown. The reference list does not cite the 2018 international bike-fit survey (REF-1263); the search may have closed before it, or it was excluded (our reading).
Interests. Funding is listed in Crossref and Europe PMC as the University of Melbourne; two of three authors list a physiotherapy practice as an affiliation (Physiosports Brighton, Kiama Physiotherapy); no competing-interests statement could be read.
How this entry was checked and changed
Why this level

from the abstract; a systematic search of five databases with reference checking, citation tracking and grey literature, and a risk-of-bias rating, but 15 of the 18 studies were not at low risk of bias, the synthesis counts studies rather than pooling effects, and the designs, sample sizes and the measures tested are not in the abstract. The rubric's A- is for reviews of trials; the base grades reviews of weak observational evidence B+ (as REF-919)