Gait retraining for runners with patellofemoral (front-of-knee) pain, mainly a 5-10% higher step rate and soft-landing cues taught with feedback that is faded out, went with short-term gains in pain and function and lower knee-loading measures, in an umbrella review of 8 reviews (5 systematic), mostly against habitual running and with no pooled effect sizes
Main caveat: No pooled estimates and no certainty grading (the authors applied no GRADE), so the review gives direction and consistency but not the size of any benefit.
Petri and colleagues (Santa Catarina, Brazil, with Migliorini and Maffulli) brought together review-level evidence on gait retraining for patellofemoral pain in runners, with prevention as a secondary question.
They searched two databases to March 2026 and included five systematic reviews (Agresta 2015, Neal 2016, Alexander 2022, Anderson 2022, Figueiredo 2025) and three narrative or mixed-methods reviews (Barton 2016, Fyock 2020, Gaudette 2022).
AMSTAR-2 rated four of the systematic reviews high quality and Agresta 2015 low (no protocol, no risk-of-bias assessment, no list of excluded studies); SANRA scored the narrative reviews 10-11 of 12.
The strategies studied were mainly a 5-10% higher step rate, cues to land more softly, a change of foot strike, and auditory or visual feedback, usually introduced progressively with the feedback gradually withdrawn.
The most consistent findings across reviews were short-term gains in pain during running and in self-reported function, and lower impact-loading measures (vertical loading rates more consistently than hip or knee angles), especially with step-rate and soft-landing strategies.
Foot-strike change, proximal kinematics, longer-term outcomes and injury prevention were less consistent; the one prevention signal, a soft-landing cue in novice runners, came from healthy runners.
The authors conclude that gait retraining may be a feasible add-on, not a stand-alone treatment, for runners with patellofemoral pain, best delivered supervised with feedback that fades.
Participants
Eight reviews.By Table 1: Agresta 2015, 10 studies (healthy runners, patellofemoral pain, chronic exertional compartment syndrome); Neal 2016, 28 studies in runners with patellofemoral pain; Alexander 2022, 30 RCTs in novice and recreational runners; Anderson 2022, 37 studies in injured runners including patellofemoral pain; Figueiredo 2025, 18 studies in recreational and trained runners; Barton 2016, 46 studies, mostly in runners without symptoms (5 in injured runners, 4 with clinical outcomes); Fyock 2020, 3 studies in adult recreational runners with patellofemoral pain; Gaudette 2022, a narrative review of symptomatic and asymptomatic runners.
The number of runners, their sex split and ages are not given by the umbrella review.
Protocol
Question (PICO): runners with patellofemoral pain, recreational or competitive (other runners only where a review reported prevention or mechanism findings separately); gait retraining (step rate, foot strike, feedback, posture, or combinations); comparators of no intervention or other interventions; outcomes pain, function, biomechanics, safety, and incidence or recurrence of patellofemoral pain.
Search: PubMed/MEDLINE and Scopus, inception to March 2026; reference lists; medRxiv and SportRxiv.
Selection: 145 records, 69 duplicates removed, 76 screened, 23 full texts, 8 included; full texts excluded for no specific gait-retraining analysis (5), a focus on strengthening, footwear or orthoses (3), or no reproducible method (7). Screening, extraction and appraisal in pairs.
Appraisal: AMSTAR-2 for systematic reviews; SANRA for narrative reviews, which had to score at least 10 of 12 and were used for context, not for effectiveness conclusions.
Overlap: corrected covered area 0.18, which the authors read as low (below 25%); most overlap among the reviews of step-rate and visual-feedback trials (Neal 2016, Barton 2016, Figueiredo 2025).
Synthesis: narrative, by strategy and outcome domain; follow-up classed as immediate, short term (up to 3 months) or mid term (3-6 months). No meta-analysis; no GRADE.
Key Results
| Strategy or domain | What the reviews converged on | Consistency |
|---|---|---|
| Step rate +5-10% | lower knee extensor moment, patellofemoral joint loading measures, vertical ground reaction forces and loading rates; short-term gains in pain and function | most consistent |
| Soft-landing cues | lower impact-loading measures; one prevention trial in healthy novices (about 66% fewer knee injuries, as reported for Alexander 2022) | consistent for loading; prevention from one trial |
| Foot-strike change (rearfoot to mid or forefoot) | lower impact peaks and loading rates, more variable; may shift load to the ankle and Achilles | less consistent; often combined with other cues |
| Hip adduction, pelvic drop, knee valgus | mixed or null across studies | less consistent |
| Feedback (visual, auditory) | best read as a way to deliver retraining, faded over time | — |
| Pain and function | short-term gains, more consistent for pain during running and self-rated function than for pain at rest | short term (mostly ≤3 months); mid-term data sparse |
| Injury prevention | uncertain; no direct evidence that step-rate change lowers injury rates | inconsistent |
| Safety | reported infrequently; generally tolerated when introduced gradually; symptom flare or calf/Achilles discomfort when changes were made too fast | sparse |
No pooled effect sizes, confidence intervals or numbers needed to treat are given; the umbrella review reports direction and consistency only.
Partial maintenance of gains up to 6 months is reported from longitudinal studies.
The authors state that the biomechanical changes are plausible explanations, not shown to cause the symptom improvement.
Practical Application
For a runner with diagnosed front-of-knee pain, a modest step-rate increase (about 5-10%) or a cue to land more softly is a reasonable add-on to a clinician's plan, not a treatment on its own. Across 8 reviews it went with short-term gains in pain and function and lower knee-loading measures; no size of benefit was pooled and most follow-up ended by 3 months.
This is not treatment advice; a physiotherapist or physician diagnoses the knee pain and decides whether gait retraining belongs in the plan.
Introduce it gradually, with feedback that is withdrawn over time: changes made too fast went with symptom flares and calf or Achilles discomfort. A metronome, mirror, phone or wearable can give the feedback (the authors' implementation advice, not a tested comparison).
How it sits with [SYNTH-94] and [SYNTH-55]: the guides put a clinician's knee-targeted exercise and education first (REF-1381) and offer step-rate change as an option, not a treatment, because in the one runner trial gait retraining showed no evidence of added benefit over education alone (REF-917).This review does not overturn that: its gains were mostly measured against habitual running or no retraining, not against education or exercise, and the authors themselves call retraining an adjunct, not a substitute for strengthening, load management and education.
Limitations
No pooled estimates and no certainty grading (the authors applied no GRADE), so the review gives direction and consistency but not the size of any benefit.
Three of the eight reviews are narrative or mixed-methods (one a critically appraised topic of 3 studies); the authors say this may have added risk of bias and used them for context.
AMSTAR-2 rates the reviews, not the primary trials; the four systematic reviews rated high by AMSTAR-2 used risk-of-bias tools, but the umbrella review does not report what those found.
Primary trials heterogeneous in step-rate targets, feedback, duration, supervision and comparators; outcomes mostly up to 3 months.
Overlap: the authors report a corrected covered area of 0.18 and call it low (below their 25% cut-off); the greatest overlap was among Neal 2016, Barton 2016 and Figueiredo 2025 on cadence and visual-feedback trials, and the authors say they did not count repeated trials as independent support; which trials recur is not listed, so how far the convergence rests on shared trials cannot be checked (our reading).
Several included reviews mix healthy and injured runners; biomechanical results in healthy runners were treated as context.
Two small points: the "about 66%" fewer knee injuries the umbrella review gives for Alexander 2022 (Table 1, Discussion) is close to but not the same as the RR 0.32 in that review's own abstract (REF-1393), a difference with another source rather than inside this paper; and the safety statement cites Neal 2016 with Fyock 2020 in the Results but Neal 2016 with Barton 2016 in the Discussion.
Funded by Brazilian public research agencies (CAPES, FAPESC, CNPq); the authors declare no competing interests.
How this entry was checked and changed
Why this level
an umbrella review whose quality appraisal we read in the full text (the rubric's A+ row), held at A- because it pools no effect sizes, draws partly on narrative reviews, applies no certainty grading, and rests on primary trials that are heterogeneous, small and mostly no longer than 3 months