Biomechanical testing in experimental bone interventions—May the power be with you

Abstract

Total variation in any measured variable, in conjunction with expected treatment effect, defines the minimum sample size (minSS) required to detect the expected effect with statistical confidence should the effect truly exist. A comprehensive literature survey of 3472 original studies was carried out to identify studies with biomechanical testing of whole bones. Total variation in common biomechanical traits and expected treatment effects in typical interventions were statistically determined. According to this survey, total variation in biomechanical traits between different species of experimental animals was similar, justifying the use of rat femur as a model in further analyses. Due to poorer precision, stiffness and energy absorption assessment require substantially larger sample size than breaking load. Due to same reason, minSS for femoral neck compression test is considerably larger than for femoral shaft three-point bending test. For the bending test, minSS to show a 10% treatment effect in the breaking load with 80% statistical power is 11rats/group, while corresponding minSS is 23 for the stiffness, and 53 for the energy absorption. For the femoral neck compression test, minSSs are 16, 51, and 134rats/group, respectively. Among the reviewed studies, the mean sample size was 11animals/group. This underscores the need for considerably larger sample sizes in experimental bone interventions which employ mechanical traits as primary outcome variables. In particular, poor precision and generally small expected treatment effects compromise the utility of stiffness and energy absorption assessments in experimental bone interventions.