Background & Aims

Brief Pain Inventory (BPI) is a commonly used tool to assess patient-reported pain outcomes and has been translated into multiple languages. Most research shows that the BPI has two factors (pain severity, pain interference) in most populations, and can be valid with three factors (pain severity, pain activity interference, pain affective interference) in some populations such as people with HIV/AIDS and cancer. This study examined the psychometric performance of the BPI, particularly the factor structure using confirmatory factor analysis (CFA) and item difficulty and discrimination using Item Response Theory (IRT) analyses in U.S. Veterans who experience knee pain prior to Total Knee Arthroplasty (TKA).

Methods

A total of 660 U.S. Veterans were consented and screened for enrollment in an on-going, double-blind, two-arm, randomized controlled trial, evaluating the efficacy of an Acceptance and Commitment Therapy (ACT) intervention in veterans at risk for Persistent post-surgical pain (PPSP) following total knee arthroplasty (TKA). All Veterans completed the Brief Pain Inventory (BPI) during post-consent screening to assess individuals’ eligibility with regard to pain level for the trial. Based on factor structures of the BPI reported in prior literature, confirmatory factor analysis was conducted to examine the fit of the data to a 2-factor structure (pain severity and pain interference) and a 3-factor structure (pain severity, pain activity interference, and pain affective interference). Appropriate factor structures that confirm unidimensionality were used for IRT analysis, where items within each factor were estimated for difficulty and discrimination.

Results

All 660 Veterans (age mean =66.7 years, SD=9.1 ; 88% male; 72.2% white, 61% married) had knee pain with varied duration, and 61.5% had other ongoing, chronic or persistent pain for at least the last six months. The 2-factor structure showed acceptable mixed model fit and appropriate discrimination with strong correlation factors (r=.776, p<.001). The 3-factor structure showed better model fit than the 2-factor structure and had appropriate discrimination with strong correlation of pain severity with both pain activity interference (r= .738, p<.001) and pain affective interference (r= .766, p<.001), and very strong correlation between pain activity and affective interference (r= .856, p<.001). BPI showed acceptable mixed model fit as a 2-factor tool, and better model fit as a 3-factor model to IRT. BPI items in both factor structures showed acceptable item discriminations and evenly spread item difficulty estimates. Item and test information curves were unimodal and symmetric.

Conclusions

While both 2-factor and 3-factor structures of the BPI may be appropriate and the 3-factor structure had better model fit, it is possible that the 3-factor structure may be too parsimonious. Findings provide preliminary information supporting the use of the 2-factor structure or the 3-factor structure based on study aims and concepts of interest. For example, if pain interference in general is of interest (versus pain activity and affective interference), and/or sample size is relatively small, a 2- factor structure vs. 3-factor structure, can be pursued.

References

•Atkinson, T.M., et al., Using confirmatory factor analysis to evaluate construct validity of the Brief Pain Inventory (BPI). Journal of pain and symptom management, 2011. 41(3): p. 558-565.
•Byrne, B. M. (2012). Structural equation modeling with Mplus: Basic concepts, applications, and programming. Routledge.
•de Ayala, R. J. (2009). The theory and practice of item response theory. The Guilford Press.
•DeVellis, R. F. (2016). Scale development: Theory and applications. Sage publications.
•Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1-55.
•Kline, R. B. (2011). Principles and practice of structural equation modeling. Guilford press.
•Muthén, L. K., & Muthén, B. O. (2010). Mplus user’s guide Author. Los Angeles, CA.
•Rasch, G. (1960). Probabilistic models for some intelligence and attainment tests. University of Chicago Press.
•Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York, NY: Guilford Press.
•Garson, D. (2010). Statnotes: Topics in multivariate analysis: Factor analysis. Retrieved from https:// faculty.chass.ncsu.edu/garson/PA765/statnote.htm
•Marsh, H. W., Hau, K.-T., & Wen, Z. (2004). In search of golden rules: Comment on hypothesis approaches to setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler’s (1999) findings. Structural Equation Modeling, 11(3), 320–341.
•Schmitt, T. A. (2011). Current methodological considerations in exploratory and confirmatory factor analysis. Journal of Psychoeducational Assessment, 29(4), 304–321
•Yu, C.-Y. (2002). Evaluating cutoff criteria of model fit indices for latent variable models with binary and continuous outcomes (Unpublished doctoral dissertation). University of California, Los Angeles.

Presenting Author

Barbara Rakel

Poster Authors

Wen Liu

PhD, MSN, BSN

University of Iowa

Lead Author

E-mail

Kyung Soo Kim

University of Iowa

Lead Author

Jennie Embree

University of Iowa

Lead Author

E-mail

M. Bridget Zimmerman

University of Iowa

Lead Author

E-mail

Katherine Hadlandsmyth

University of Iowa

Lead Author

E-mail

Tracey Smith

Baylor College of Medicine

Lead Author

E-mail

Joseph Buckwalter

MD

University of Iowa

Lead Author

E-mail

David Green

MD

Baylor College of Medicine

Lead Author

E-mail

Lilian Dindo

Baylor College of Medicine

Lead Author

E-mail

Barbara Rakel

University of Iowa

Lead Author

E-mail

Topics

  • Assessment and Diagnosis