Differential item functioning of the patient-reported outcomes information system (PROMIS®) pain interference item bank by language (Spanish versus English)

Abstract

Background: About 70% of Latinos, 5 years old or older, in the United States speak Spanish at home. Measurement equivalence of the PROMIS^® pain interference (PI) item bank by language of administration (English versus Spanish) has not been evaluated.

Methods: A sample of 527 adult Spanish-speaking Latinos completed the Spanish version of the 41-item PROMIS^® pain interference item bank. We evaluate dimensionality, monotonicity and local independence of the Spanish-language items. Then we evaluate differential item functioning (DIF) using ordinal logistic regression with item response theory scores estimated from DIF-free "anchor" items.

Results: One of the 41 items in the Spanish version of the PROMIS^® PI item bank was identified as having significant uniform DIF.

Conclusions: English- and Spanish-speaking subjects with the same level of pain interference responded differently to 1 of the 41 items in the PROMIS^® PI item bank. This item was not retained due to proprietary issues. The original English language item parameters can be used when estimating PROMIS^® PI scores.

Keywords: Item response theory; Language DIF; PROMIS®; Patient-reported outcomes; Psychometrics.

Figure 1

Comparison of test characteristic curves (TCC) for English and Spanish with all items (left) and only DIF items (right).

Figure 2

Assessment of Individual-Level DIF impact.