Estimating minimally important differences for the PROMIS pain interference scales: results from 3 randomized clinical trials. Issue 4 (April 2018)

• Record Type: Journal Article
• Authors: Chen, Chen X.; Kroenke, Kurt; Stump, Timothy E.; Kean, Jacob; Carpenter, Janet S.; Krebs, Erin E.; Bair, Matthew J.; Damush, Teresa M.; Monahan, Patrick O.
• Is Part Of: Pain. Volume 159:Issue 4(2018)
• Publication Date: 2018-04
• Subjects: PROMIS -- Pain measurement -- Pain interference -- Minimally important difference -- Psychometrics -- Patient-reported outcome measures; Pain -- Periodicals; Douleur -- Périodiques; Anesthésie -- Périodiques; Pain; Electronic journals; Periodicals; Electronic journals; 616.0472
• DOI: 10.1097/j.pain.0000000000001121
• Languages: English
• ISSNs: 0304-3959

Abstract

Abstract : Abstract: Minimally important difference (MID) refers to the smallest meaningful difference that carries implications for patient care. Minimally important differences are necessary to help interpret patient-reported pain outcomes in research and clinical practice. The PROMIS pain interference scales were validated across diverse samples; however, more information about their MIDs could improve their interpretability. The purpose of this study was to estimate MIDs for 4 fixed-length PROMIS pain interference scales, including the 6-item Pain Short Form and the 4-, 6-, and 8-item pain interference scales used in the PROMIS profile instruments. Data were analyzed from 3 randomized controlled trials (N = 759). The 3 samples, respectively, consisted of patients with chronic low back pain (n = 261), chronic back pain or hip/knee osteoarthritis pain (n = 240), and a history of stroke (n = 258). For each sample, anchor- and distribution-based approaches were used to estimate MIDs. Standard error of measurement and effect sizes were used as distribution-based MID estimates. Anchor-based MID estimates were established by mapping PROMIS pain interference scores onto established anchor measures, including the Brief Pain Inventory, and retrospective and prospective global ratings of change. The distribution- and anchor-based MID estimates showed convergence. For the pain samples, MID estimates ranged from 2 to 3 T-score points. For the nonpain sample, MID estimates ranged from 3.5 to 4.5 T-score points. The MID estimates were comparable across the 4 fixed-length scales. These MIDs can be used to evaluate treatment effects in research and clinical care and to calculate estimates for powering clinical trials. Abstract : Supplemental Digital Content is Available in the Text.For PROMIS Pain Interference (PROMIS-PI) scales, the minimally important difference estimates were sample dependent; however, the estimates were comparable across the PROMIS-PI scales of different lengths.