Academic journal article Psychological Test and Assessment Modeling

Measurement Equivalence of the Patient Reported Outcomes Measurement Information System® (PROMIS®) Anxiety Short Forms in Ethnically Diverse Groups

Academic journal article Psychological Test and Assessment Modeling

Measurement Equivalence of the Patient Reported Outcomes Measurement Information System® (PROMIS®) Anxiety Short Forms in Ethnically Diverse Groups

Article excerpt

Introduction

Item banks developed using item response theory (IRT) are being promoted for efficient assessment of health-related constructs, particularly as applied to physically frail populations. Some of these banks have focused on anxiety. For example, Walter et al. (2007) established an anxiety item bank with 50 items, calibrated with the generalized partial credit model. Precise estimates were obtained with as few as six to eight items administered. Somatic anxiety symptoms such as dizziness, dyspnea and palpitations were excluded. Another well-known collection of item banks is from the Patient Reported Outcomes Measurement Information System® (PROMIS®), developed as a part of the U.S. National Institutes of Health (NIH) roadmap initiative (see www.nihpromis.org) to measure self-reported health for clinical research and practice. As a subdomain to measure emotional distress, the PROMIS Anxiety item bank consists of 29 items and several short forms (Cella et al., 2010; Pilkonis, Choi, Reise, Stover, Riley, & Cella, 2011). Originally developed in English, the PROMIS Anxiety item bank has been translated into several languages including: Spanish, German, Mandarin (short form only) and Dutch (short form only). According to the NIH PROMIS webpage (http://www. nihpromis.org/measures/translations), translation of the PROMIS Anxiety item bank into several other languages (e.g., Portuguese, Hebrew) is currently in progress.

Differential Item Functioning (DIF) analyses of the PROMIS Anxiety item bank

Given that systematic measurement bias in measures used for research and practice could lead to misleading group comparisons and inaccurate prevalence rates, a critical first step for the PROMIS Anxiety item bank is to establish measurement equivalence across diverse groups. Despite the importance of measurement equivalence, differential item functioning analyses have not been performed widely in studies using PROMIS measures. Only a few studies examined DIF for PROMIS measures and even fewer studies of DIF are available for the PROMIS anxiety measure (e.g., Choi, Gibbons, & Crane, 2011). Choi et al. evaluated the 29 item PROMIS Anxiety bank for age DIF using a sample of 766 adults. Five of 29 items evidenced modest levels of DIF: "I felt fearful"; "I was anxious if my normal routine was disturbed"; "I was easily startled"; "I worried about other people's reactions to me"; and "Many situations made me worry." Magnitude of DIF was small. Aggregate DIF impact was very small; however, based on examination of the standard error of measurement, salient score changes for some subjects were observed such that there was some individual level impact. Given that the sample used in the analyses by Choi et al. did not permit analyses by race or ethnicity, there is a need for DIF analyses of the PROMIS Anxiety item bank in patient populations from diverse cultural backgrounds.

DIF Analyses of general anxiety measures

Several recent studies have examined DIF in the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983) in different populations, such as primary care (Cameron, Scott, Adler, & Reid, 2014); Parkinson's disease (Forjaz, Rodrigues-Blázquez, & Martinez-Martin, 2009); spinal cord injury (Müller, Cieza, & Geyh, 2012); motor neurone disease (Gibbons et al. 2011); chronic obstructive pulmonary disease (Tang, Wong, Shiu, Lum, & Ungvari, 2008); musculoskeletal rehabilitation (Pallant & Tennant, 2007); breast cancer (Osborne et al., 2004), and caregivers to cancer patients (Lambert, Pallant, & Girgis, 2011). Nearly all investigators used the Rasch model for analyses, and most concluded that little DIF was observed. In one study (Cameron et al., 2014) of the HADS, DIF was observed for gender or age for three items. Only one item with gender DIF was identified (Guillén-Riqueime & Buela-Casal, 2011) in the State-Trait Anxiety Inventory (Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.