Facebook
Twitter
LinkedIn
Email
 |
Checking for direct PDF access through Ovid | |
Excerpt
Re: A Study of the Diagnostic Accuracy and Reliability of the Scoliometer and Adam’s Forward Bend Test (Spine 1999;23:796–802).
In response to William P. Bunnell’s letter, we would like to make the following points. Before a clinical tool is used to screen or diagnose, it must be evaluated with a sound study design and “state of the art” clinical epidemiology methods. The first step in selecting study design is to clearly state a research question. The purpose of our study was to determine the reliability and validity of the Scoliometer as a diagnostic test in a sample of subjects who present to a tertiary care clinic. Dr. Brunnell would agree that any attempts made identifying alternative ways to accurately measure spinal curves magnitude, while decreasing the amount of radiation to an adolescent must be encouraged. This goal was clearly stated in the paper.
Our inclusion and exclusion criteria were clearly described and the tests were administered in a consistent manner. Furthermore, we used sophisticated analytic techniques to answer our research question. One of these methods, the ROC curve analysis is very efficient at illustrating the trade-offs between sensitivity and specificity. We would encourage Dr. Brunnell to refer to Figures 1 and 2 of our paper.
Another interesting comment pertains to precision. We believe that a 5° measurement error is a sign of poor precision when attempting to identify a curve that measures at least 5° on the Scoliometer. The danger of using a 5° cut-point in clinical practice is that it would significantly misclassify the curve status of young adolescents, and thus limit its use as a diagnostic test.
Longitudinal data is not necessary to conclude that the Scoliometer lacks reliability and validity, and therefore, that its usefulness is questionable as an outcome measure or as an instrument used to monitor curve progression. If important errors are associated with one measurement, even larger errors would cumulate from repeated measurements.
Finally, we find the suggestion of chiropractic bias amusing. We would prefer to think that our bias is rather the result of our graduate training in epidemiology.
