Reliability of lumbar spine radiograph reading by chiropractors

An intraobserver and interobserver study on the reproducibility of data was performed.

This study investigates the variability in the interpretation of lumbar spine radiographs by chiropractors working in private practice.

In chiropractic practice radiographs are used often, but this use is currently under debate. Therefore, there is a need for further study of the value of diagnoses made by radiographs by chiropractors. An acceptable intra- and interobserver agreement in radiograph reading is a prerequisite for a useful application of radiographs as a diagnostic tool in daily practice and in research.

Four chiropractors read 100 blinded sets of standard, erect anteroposterior and lateral lumbar radiographs independently. The same set was read in two separate sessions with a 2-month interval. The first session revealed the interobserver agreement. The comparison of the ratings by the same assessor in the two sessions indicated the intraobserver agreement. The assessors used a specially developed criteria list with emphasis on "nonspecific" radiographic findings. The prevalence of some important categories was increased artificially. Agreement was expressed in percentage agreement and generalized kappa, combining the results of all four assessors.

Most kappas ranged from 0.40 to 0.75, representing fair to good agreement. In general, intraobserver agreement was better than interobserver agreement. The low kappas that were found may be explained partially by the high-agreement-low-kappa paradox as a result of a low prevalence.

The kappas and percentage agreement were acceptable, although not excellent. These results will be beneficial for future research on the value of radiograph diagnosis of nonspecific findings for delivery of safe and effective chiropractic therapy.