Validity of Torque-Data Collection at Multiple Sites: A Framework for Collaboration on Clinical-Outcomes Research in Sports Medicine

in Journal of Sport Rehabilitation
Restricted access

Purchase article

USD  $24.95

Student 1 year online subscription

USD  $76.00

1 year online subscription

USD  $101.00

Student 2 year online subscription

USD  $144.00

2 year online subscription

USD  $192.00


Collecting torque data using a multimode dynamometer is common in sports-medicine research. The error in torque measurements across multiple sites and dynamometers has not been established.


To assess the validity of 2 calibration protocols across 3 dynamometers and the error associated with torque measurement for each system.


Observational study.


3 university laboratories at separate institutions.


2 Biodex System 3 dynamometers and 1 Biodex System 4 dynamometer.


System calibration was completed using the manufacturer-recommended single-weight method and an experimental calibration method using a series of progressive weights. Both calibration methods were compared with a manually calculated theoretical torque across a range of applied weights.

Main Outcome Measures:

Relative error, absolute error, and percent error were calculated at each weight. Each outcome variable was compared between systems using 95% confidence intervals across low (0–65 Nm), moderate (66–110 Nm), and high (111–165 Nm) torque categorizations.


Calibration coefficients were established for each system using both calibration protocols. However, within each system the calibration coefficients generated using the single-weight (System 4 = 2.42 [0.90], System 3a = 1.37 [1.11], System 3b = –0.96 [1.45]) and experimental calibration protocols (System 4 = 3.95 [1.08], System 3a = –0.79 [1.23], System 3b = 2.31 [1.66]) were similar and displayed acceptable mean relative error compared with calculated theoretical torque values. Overall, percent error was greatest for all 3 systems in low-torque conditions (System 4 = 11.66% [6.39], System 3a = 6.82% [11.98], System 3b = 4.35% [9.49]). The System 4 significantly overestimated torque across all 3 weight increments, and the System 3b overestimated torque over the moderate-torque increment.


Conversion of raw voltage to torque values using the single-calibration-weight method is valid and comparable to a more complex multiweight calibration process; however, it is clear that calibration must be done for each individual system to ensure accurate data collection.

Kuenze is with the Dept of Kinesiology, Michigan State University, East Lansing, MI. Eltouhky is with the Dept of Kinesiology and Sport Sciences, University of Miami, Coral Gables, FL. Thomas is with the Dept of Kinesiology, University of North Carolina Charlotte, Charlotte, NC. Sutherlin and Hart are with the Dept of Kinesiology, University of Virginia, Charlottesville, VA.

Address author correspondence to Christopher Kuenze at