Exploring the Test-Retest Reliability of Exercise-Induced Hypoalgesia: A Study of Single-Joint Dynamic and Isometric Resistance Exercises

Abstract

Objective To establish the test-retest reliability of exercise-induced hypoalgesia (EIH) after single-leg knee extension isometric (IRE) and dynamic resistance exercise (DRE), and to examine individual-level between-session heterogeneity, overall proportional bias and heteroscedasticity via Bland-Altman Limits of Agreement (LoA) analysis in healthy, pain-free adults.

Background EIH test-retest reliability is poor to fair across several exercise paradigms, undermining its research and clinical applications. No study has examined EIH reliability after DRE or single-joint IRE. Additionally, individual-level EIH measurement error between-sessions remains unexplored.

Study Design A single-group test-retest reliability design was used over two identical testing sessions with a target separation of seven days, in 32 participants. A combined familiarisation and maximal strength testing session preceded Session 1 by 72 hours, establishing individualised exercise doses (25% maximal voluntary isometric contraction for IRE; 10-repetition maximum for DRE).

Method Pressure pain thresholds (PPT) were assessed via handheld algometry at the index leg’s knee medial joint line and contralateral forearm, directly before and after each condition: quiet rest, DRE, and IRE. Test-retest reliability was determined using an ICC(3,1) single-rater, single-measure, absolute-agreement, two-way mixed-effects models with 95% confidence intervals, and standard error of measurement (SEM). EIH responder agreement was assessed via Cohen’s kappa. Bland-Altman LoA plots examined group-level proportional bias and heteroscedasticity and screened for individual-level outliers. Post-hoc sensitivity analyses examined reliability after adjusting for between-session exercise dose differences and outlier removal.

Results Thirty participants completed the study after two participants withdrew due to unforeseen circumstances. Most participants completed Sessions 1 and 2 seven days apart (83%). ANOVA testing confirmed local EIH occurring at the knee for both conditions; remote EIH at the forearm could not be confirmed. Within-session PPT reliability was excellent (ICC ≥ 0.82) across all sites and conditions; between-session baseline PPT reliability was excellent (ICC = 0.81) at the forearm but fair for the knee (ICC = 0.54). Between-session EIH reliability was poor (ICC ≤ 0.39) across all sites and conditions. EIH SEMs generally exceeded mean EIH magnitudes. Responder agreement was fair for IRE at the knee (κ = 0.345) and poor elsewhere. Adjusting for significant between-session differences in exercise parameters did not improve reliability. Bland-Altman analysis showed large mean biases and wide LoAs across all conditions, with no heteroscedasticity or proportional bias. Six participants (20%) were identified as condition-specific outliers. Outlier removal substantially improved knee absolute ICC values for both IRE (ICC = 0.60) and DRE (ICC = 0.58), with IRE knee SEM reducing to below the mean EIH magnitude. However, reliability remained below the COSMIN acceptability threshold (ICC ≥ 0.70) across all conditions.

Conclusion EIH after single-joint IRE and DRE demonstrated poor between-session reliability. Group-level estimates were substantially influenced by six condition-specific outliers. However, even after removal, reliability estimates remained insufficient for individual-level use. These findings extend EIH reliability evidence to previously untested resistance exercise modalities and provide the first evidence that individual-level heterogeneity can substantially suppress group-level EIH reliability. Future studies should incorporate Bland-Altman LoA analysis alongside ICC and SEM to detect individual-level measurement heterogeneity not identifiable by group-level statistics alone.