Evaluation of Clinician and Machine Performance in the Assessment of Low Back Pain.
Gracovetsky SA, Newman N, Richards M, Asselin S, Lanzo V and Marriott A.
Spine
23(5) : 568-575, 1998.
Abstract
Study design
A prospective blind study to test the performance of
clinicians (Evaluators) compared to that of an automated physical examination by the
Machine (the Spinex International), presenting randomly designated Simulators/Dissimulators and
Honest subjects for the assessment of acute benign low back pain (LBP).
Objectives
To test the impact of reported pain and history on the
clinical examination, and to compare the ability of clinicians and the Machine to
recognize normality in a controlled group of subjects with and without benign LBP.
Background
The literature raises serious questions regarding the
efficacy of the clinical examination for LBP subjects.
Method
A "gold standard" (clinical examination by LBP
experts) was established against which the clinical examination by the Evaluators, and the
Machine assessment, in Honest subjects and Simulators/Dissimulators, were compared using
the ROC technique. The selection of subjects was performed according to strict
inclusion/exclusion criteria.
Results
The Evaluators were more accurate with the Honest subjects,
the Machine more accurate with the Simulators/Dissimulators and, for the entire population
tested, they were equivalent (71% vs 72% concordance). Results from the Machine's expert
system and from clinician readers of the Machine data compared favorably. The Machine's
concordance with the gold standard increased with increasing loads lifted by the subject.
Conclusion
By relying primarily on the subject's self-presentation,
often to the exclusion of objective findings, the clinician may err in evaluating low back
function when the patient does not report his or her true condition. The additional
functional analysis provided by the Machine offers the clinician objective pertinent
information to complement the clinical examination.
