Using beat frequency in music to adjust running cadence in recreational runners: a randomized multiple baseline design

Key points

1. 01Eight running sessions with faster-beat music increased cadence by an average of 8.5% (+14 steps per minute) in all five participants.
2. 02The higher cadence was maintained 3-5 weeks after the music program ended, with a group-level increase of 7.9% (+13 steps per minute, p = .001).
3. 03Heart rate and running speed did not significantly change, suggesting the mechanical load reduction comes without added metabolic cost.
4. 04All participants rated running with music as enjoyable (average 8/10) and found the higher cadence easier and more pleasant by the end of the study.
5. 05One participant dropped out with a calf injury, consistent with known increased calf loading when cadence rises.

How it was conducted

Design

Randomized concurrent multiple-baseline design (MBD); each participant served as their own control

Participants

5 completers (7 enrolled; 2 dropped out); recreational runners aged 18-60 with baseline cadence below 170 steps per minute; mean age 42.4 years

Intervention

4 weeks (8 sessions) of running with music set 7.5-10% above individual baseline cadence, played via Bluetooth earbuds through a Garmin Forerunner 645

Study duration

12 weeks total: variable baseline phase (2-5 weeks), 4-week intervention, variable post-intervention phase (3-6 weeks); runs twice per week

Primary outcome

Running cadence (steps per minute) measured by Garmin watch during 5+ km outdoor runs

Analysis

Randomization tests for multiple-baseline single-case design; visual inspection using 2-SD bands; design-comparable effect size estimated via Pustejovsky model MB2

What they found

• Cadence increased by 8.5% (+14 steps per minute) during the music intervention phase versus baseline (p = .003).
• Cadence remained elevated by 7.9% (+13 steps per minute) during the post-intervention phase versus baseline (p = .001).
• Mean cadence: baseline 165 +/- 3.8 spm, intervention 179 +/- 3.9 spm, post-intervention 178 +/- 3.9 spm.
• Design-comparable between-subject effect size (post-intervention vs. baseline, adjusted for small-sample bias): d = 3.24 (95% CI 1.8-4.7).
• Heart rate did not significantly change during intervention+post-intervention combined (p = .37) or post-intervention alone (p = .42); mean values were 152, 155, and 151 bpm across the three phases.
• Speed did not significantly change during intervention+post-intervention combined (p = .64); a statistically significant but not practically relevant speed reduction of 4 seconds per kilometer was observed in the post-intervention phase alone (p = .042).
• All five participants individually showed a practically relevant increase in cadence (>7%) during and after the intervention based on 2-SD visual inspection.

Limitations

• Very small sample (n = 5 completers) with no control group, limiting generalizability.
• Selection bias: participants self-enrolled specifically because of low cadence, potentially inflating motivation and effect size.
• Music genre was not personalized and the intervention was stopped abruptly rather than gradually faded, which may not reflect the optimal motor-learning approach.
• Participants were instructed to maintain the higher cadence after the music stopped, making it unclear whether cadence would remain elevated without this explicit reminder.

Why it matters

For patients

Recreational runners with a low stride rate can use a simple music-based app for eight sessions to raise their cadence durably, which may reduce ground impact forces and injury risk without making runs feel harder.

For clinicians

A low-cost, field-based music cadence protocol appears feasible and comparably effective to lab-based gait retraining; cadence increases of roughly 8% persisted weeks after the intervention, though calf loading warrants monitoring.

For readers

This small but internally rigorous pilot provides a useful proof-of-concept and a sample-size estimate (8 per group) for a full RCT; the effect size is large but must be interpreted cautiously given the 5-participant sample.