A comparison of bone strain measurements at anatomically relevant sites using surface gauges versus strain gauged...

Instrumented bone staples were first introduced as an alternative to surface-mounted strain gauges for use in human in vivo bone strain measurements because their fixation to bone is secure and requires not only minimally invasive surgery. Bench-top bone bending models have shown that the output from strain gauged bone staples compares favorably to that of traditional mounted gauges. However their within- and across-subject performance at sites typically instrumented in vivo has never been examined. This study used seven human cadaver lower extremities with an age range of 23-81 years old and a dynamic gait simulator to examine and compare axial strains in the mid tibial diaphysis and on the dorsal surface of the second metatarsal as measured simultaneously with strain gauged bone staples and with traditional surface-mounted gauges. Rosette configurations were used at the tibial site for deriving principal compression and tension, and shear strains. Axial outputs from the two gauge types demonstrated strong linear relationships for the tibia (r(2)=0.78-0.94) and the second metatarsal (r(2)=0.96-0.99), but coefficients (slopes) for the relationship were variable (range 7-20), across subjects and across sites. The apparent low reliability of strain gauged staples may be explained by the fact that both strain gauged staples and surface strain gauges are inexact to some degree, do not measure strains from exactly the same areas and strain gauged staples reflect surface strains as well as deformations within the cortex. There were no relationships for the principal tibia compression, tension or shear strain measurements derived from the two rosette gauge types, reflecting the very different anatomical areas measured by each of the constructs in this study. Strain gauged bone staples may be most useful in comparing relative axial intra-subject differences between activities, but inter-subject variability may require larger sample sizes to detect differences between populations.