Computation methods affect the reported values of in vivo human tendon stiffness.

PURPOSE Scientific validity is questionable when findings from studies cannot be used to make sense of physiological and/or biomechanical data. In particular, is the case of in vivo determination of tendon stiffness (K). Here, approaches range from taking the gradient (a) throughout the data range of resting to Maximal Voluntary Contraction (MVC), (b) tangents at individual data points, (c) linear regressions at discrete force levels ((b) and (c) being 'reference standard' as they utilise a number of distinct regions of the Force-Elongation Relationship (FER)). STUDY DESIGN A mathematical model approach is used to develop simple curvilinear FERs as seen when determining tendon mechanical properties, to allow variable calculations of K. OBJECTIVES To compare variability in K estimates using the various approaches currently seen in the literature. METHODS Three FER models were developed, representing low, medium and high K. Values of K were determined and compared using the approaches reported in the literature to estimate the magnitude of the difference between values attained of K. RESULTS Through mathematical modelling, we demonstrate that the impact on the recorded value of K is substantial: relative to the reference standard methods, computation methods published range from underestimating K by 26% to overestimating it by 51%. CONCLUSION This modelling helps by providing a 'scaling factor' through which the between studies variability associated with computational methods differences is minimised. This is especially important where researchers or clinicians require values which are consistent in the context of establishing the 'true' tendon mechanical properties to inform models or materials based on the biological properties of the human tendon.