Reconsidering exercise selection with EMG: poor agreement between ranking hip exercises with EMG

Key points

1. 01EMG rankings and muscle force rankings showed weak, non-significant agreement for both gluteus maximus (Spearman's rho = 0.29) and gluteus medius (rho = 0.51).
2. 02Peak EMG amplitude alone explained only 5% of gluteus maximus peak muscle force and 19% of gluteus medius peak muscle force.
3. 03When exercise type and individual participant variation were both accounted for, EMG explained 80-85% of muscle force, showing EMG is informative within the right context.
4. 04An illustrative example showed that hip hike and split squat produced identical gluteus maximus EMG (30% MVIC) but the split squat generated more than twice the muscle force (569 N vs 252 N).
5. 05EMG comparisons are only valid within the same individual, with the same electrode placement, and between biomechanically similar exercises.

How it was conducted

Design

Cross-sectional laboratory study with secondary EMG analysis of previously published biomechanics data

Participants

14 healthy female footballers, median age 24.1 years, no current lower limb injuries

Exercises

Eight hip-focused exercises: single-leg squat, split squat, single-leg Romanian deadlift, single-leg hip thrust, banded side-step, hip hike, side-lying leg raise, and side plank

Loads

Body weight and 12-repetition maximum (12RM) loads for each exercise

Measurements

Full-body kinematics, ground reaction forces, surface EMG from 12 lower limb muscles, and neuromusculoskeletal model-estimated gluteal muscle forces

Primary analysis

Spearman's rank correlation between exercise rankings by peak normalized EMG and peak normalized muscle force; linear mixed effects models for the EMG-force relationship

What they found

• Spearman's rank correlation between EMG-ranked and force-ranked exercises: gluteus maximus rho = 0.29 (95% CI -0.33 to 0.81, P = 0.310); gluteus medius rho = 0.51 (95% CI -0.05 to 0.87, P = 0.064) - both non-significant.
• EMG overestimated 7/14 (50%) and underestimated 5/14 (36%) exercise rankings for gluteus maximus; overestimated 7/14 (50%) and underestimated 6/14 (43%) for gluteus medius.
• Peak EMG amplitude alone (marginal R2): explained 5% of gluteus maximus peak muscle force (R2 = 0.05) and 19% of gluteus medius peak muscle force (R2 = 0.19).
• Adding exercise type random effects (conditional R2): 49% for gluteus maximus and 24% for gluteus medius.
• Adding participant random effects (conditional R2): 39% for gluteus maximus and 71% for gluteus medius.
• Adding both exercise type and participant random effects (conditional R2): 80% for gluteus maximus and 85% for gluteus medius.
• Representative example: hip hike and split squat both had gluteus maximus peak normalized EMG of 30% MVIC, but split squat produced 569 N versus 252 N for hip hike - more than twice the muscle force.
• Sensitivity analysis removing trials with greater than 20%, 10%, and 5% EMG model adjustment showed similar marginal R2 values (gluteus maximus 0.02-0.05, gluteus medius 0.13-0.25).

Limitations

• Small sample of only 14 female footballers limits generalizability to other populations, sex, or activity levels.
• Muscle forces are model estimates, not direct in vivo measurements; musculoskeletal model parameters were based on generic data and not fully personalized via medical imaging.
• No a-priori power analysis was performed as this was a secondary analysis of existing data.
• The neuromusculoskeletal model makes adjustments to EMG signals to match joint torques, which itself contributes to some of the observed disparity between raw EMG and estimated force.

Why it matters

For patients

Patients undergoing gluteal rehabilitation should not assume that exercises their therapist identifies as high-activation on EMG are necessarily producing the highest muscle forces or driving the most adaptation.

For clinicians

Clinicians prescribing hip exercises based on published EMG rankings may be selecting sub-optimal exercises for muscle loading; force-generating capacity depends heavily on muscle length, contraction velocity, and passive forces that EMG cannot capture.

For readers

This study challenges a widespread assumption in exercise science that higher EMG equals better exercise selection, and calls for future research linking exercise biomechanics to longitudinal muscle adaptation outcomes.