How do validated measures of functional outcome compare with commonly used outcomes in administrative database research for lumbar spinal surgery?