Relative responsiveness of condition-specific and generic health status measures in degenerative lumbar spinal stenosis

Abstract

The objective of this study was to compare the relative responsiveness of a condition-specific spinal stenosis measure and two generic health status measures for outcome assessment of surgery for degenerative lumbar spinal stenosis, and to examine whether responsiveness statistics and measures of the ability to distinguish clinically important improvement rank the instruments consistently. Physical function and symptom severity scales of the spinal stenosis measure were compared to the Sickness Impact Profile (SIP) and the Roland scale, which is derived from the SIP. Responsiveness was calculated with the standardized response mean, the effect size, and Guyatt's responsiveness statistic. The discriminative ability of the instruments to distinguish patients who improved from those who did not was assessed using satisfaction with surgery as an external criterion. Minimal clinically relevant improvement was estimated using patient satisfaction as the external criterion. All responsiveness statistics revealed the same order of responsiveness; the physical function scale (SRM = 1.07) and symptom severity scales (SRM = 0.96) were more responsive than the Roland scale (SRM = 0.77) which was only slightly more responsive than the SIP (SRM = 0.69). Strikingly, the physical dimension of the SIP (SRM = 0.62) was even less responsive than the global SIP. The shape of and the area under the ROC curves showed that the physical function and symptom severity scales discriminate better between satisfied and unsatisfied patients than the Roland scale and SIP. The sensitivity to detect clinically important changes was somewhat lower at the ends of the scales, especially for the SIP and the Roland scale. Statistical approaches that assess the ability to distinguish clinically important changes and overall responsiveness statistics ranked the measures consistently. On the basis of these findings, we suggest that a condition-specific spinal stenosis measure is preferable as the primary end point in evaluative studies of degenerative lumbar spinal stenosis.