Big versus small data and the generalizability of the rate of computed tomography overuse in minor head injury.