Development and Analysis of PROMIS Pain Intensity Scale.

TitleDevelopment and Analysis of PROMIS Pain Intensity Scale.
Publication TypeJournal Article
AuthorsChen W-H, Revicki D, Amtmann D, Jensen MP, Keefe FJ, Cella D
JournalQuality of life research
IssueSuppl 1

Aims: The primary objective of this study is to develop a PROMIS Pain Intensity Scale by evaluating the unidimensionality and completing item calibration of the pain intensity items developed in the PROMIS Wave I study. This document provides a summary of the item selection process based on the results of the confirmatory factor analysis (CFA) and item response theory (IRT) analysis. Methods: The PROMIS project is focused on developing item banks and assessment instruments for pain and other patient-reported outcome domains. The draft PROMIS pain related items were developed based on literature reviews, clinician interviews, and qualitative research with patients with pain. In addition to the three item banks related to pain (pain interference, pain quality, and pain behavior), six items were identified as pain intensity items. The data used in this study included: 1) PROMIS Wave I sample where internet survey data were collected from 838 participants who responded to all six pain intensity items and 5,059 participants who responded to at least one pain intensity item; 2) American Chronic Pain Association (ACPA) sample where 967 participants completed subset of the pain intensity items; and 3) Northwestern University sample where 532 participants completed another subset of the pain intensity items. Participants reporting no pain were excluded from the analysis. We evaluated inter-item correlations, confirmatory factor analysis (CFA), item response theory analysis, and correlations with other PROMIS global items of these six pain intensity items. Results: Inter-item correlation ranges from 0.33 to 0.93. CFA shows good fit of the six items to a unidimensional model: comparative fix index (CFI)=0.989, Tucker-Lewis index (TLI)=0.986, and root mean square error of approximation (RMSEA)=0.093. Based on results of IRT analysis results three items are removed owing to local dependency and model misfit. The IRT slope parameters of the three remaining items were 1.84, 3.15, and 4.42. The category threshold parameters ranged from -2.30 to 3.23. Correlation with global pain item is 0.68, and 0.61 with PROMIS global physical health score. Conclusions: The PROMIS pain intensity scale provides a measure of characteristic pain that could be useful in clinical and research settings.