The reliability of plain radiography in experimental fracture healing.

To investigate the reliability of radiographs in the evaluation of healing of closed fractures. A closed midshaft tibial fracture was created in 40 goats and stabilized with an external fixator. The animals were assigned to four groups: no injection, injection of 1 mg osteogenic protein-1 (OP-1), 1 mg OP-1 with collagenous carrier, or carrier alone. Radiographs were performed weekly until the animals were killed after 2 and 4 weeks. Healing was evaluated using radiographs, biomechanical testing, and histological examination. All radiographs were examined by two independent observers. Interobserver agreement was calculated and radiographic scores were compared with mechanical and histological scores using regression analysis. Regression analysis showed poor correlation between radiographic scores and biomechanical and histological data. Correlation coefficients varied between 0.39 and 0.63. Good agreement between the observers was seen in only three parameters: visibility of the fracture line, weightbearing ability, and a combined healing parameter. Plain radiography provides poor parameters for monitoring the fracture healing process.