Assessment of the clinically relevant change in pain for patients with sciatica

Authors:
Open Access
Publisher website
Google Scholar
Abstract
We performed an ancillary study to a clinical trial in order to estimate the minimum clinically relevant change in pain for patients with sciatica. The main study was a randomised double blind trial conducted to assess the effectiveness of epidural corticosteroid injections in patients with sciatica, presumably due to a herniated nucleus pulposus, for 15–180 days.6 We thus administered three epidural injections at 48 hour intervals with 2 ml prednisolone acetate (50 mg) or 2 ml isotonic saline. Self evaluation was the primary outcome, measured at day 20 on a four item scale (recovery, important improvement, poor improvement, or worse). The first two items were then pooled, thus defining a treatment success, while the last two items defined a treatment failure. As a secondary outcome we assessed the severity of pain on a VAS both at day 0 and day 20. Eighty five patients were included in the main study. However, in the present work, analyses were performed on a subsample of 75 patients for whom we had both the self evaluation at day 20 and the two pain assessments. The subsample comprised 45 men and 30 women with a mean (SD) age of 41 (11) years. Those patients had had pain for a mean (SD) duration of 49 (41) days and their mean baseline pain was 58 (16). For 39 of those 75 patients, treatment was a success. We performed analyses without taking into account the randomisation arm: there is indeed no foundation for a different relationship between self evaluation and pain assessment according to a patient’s treatment. For each patient, we calculated both the crude and relative changes in pain assessment and plotted receiver operating characteristic curves (fig 1 ⇓ ). Table 1 ⇓ shows the performance indexes associated with several cut off points. Optimal cut off points, which thus offer the best compromise between sensitivity and specificity, are estimated at −30 mm for the crude change and −50% for the relative change. These values are higher than those estimated by Todd et al1 and Farrar et al.2, 3 Such discrepancies may be due to a difference in the “gold standard”. They may also be related to the length of the period of observation because we focused on a 3 week period.
Keywords