Adding Disease-Specific Concerns to Patient-Reported Outcome Measures [Internet]

Excerpt

Background: Patient-reported outcomes, which are generic or condition-specific, are used for several reasons, including clinical care, clinical research and trials, and in national efforts to monitor the quality of health care delivery. Creating patient-reported outcome measures (PROMs) that meet different purposes without overburdening patients, health care systems, providers, and data systems is paramount.

Objectives: The overall aim of this study was to create a methodology to incorporate condition-specific concerns in generic health assessments. We used PROMIS® as our resource for generic measures and started with 2 prevalent and high-impact conditions: heart failure (HF) and osteoarthritis of the knee (OA-K). The 3 primary objectives to reach this aim were the following:

1. Objective 1: Obtain patient and clinician expert input on the condition-specific relevance of existing and new items to develop improved brief assessments (PROMIS-PLUS) of these 2 high-impact conditions.

2. Objective 2: Validate PROMIS-PLUS in these 2 conditions to document key measurement properties of reliability, validity, patient-perceived relevance, and clinical usability.

3. Objective 3: Produce crosswalks from PROMIS to PROMIS-PLUS and selected legacy instruments for these 2 conditions and disseminate the new instruments and crosswalks.

Methods: We used a longitudinal transformation mixed-methods study design at 13 clinical sites to develop and test the instruments. The design built on PROMIS methods and the International Society for Pharmacoeconomics and Outcomes Research's Task Force recommendations for developing PROM items. Specific methods included patient focus groups, clinician interviews, item gap analysis, new item development, cognitive interviews, draft instruments development, draft instruments testing, data analysis, and crosswalks between legacy measures and the new measures.

Results: For HF, we identified 64 existing PROMIS items and developed 22 new items, across 18 domains. For OA-K, we identified 52 existing PROMIS items and developed 24 new items, across 14 domains. Our psychometric analyses evaluated measure reliability (internal consistency and test-retest estimates) and measure validity (convergent, divergent, and concurrent correlational analyses; and known-groups analyses, based on high vs low Global Health Physical and Mental status); these all supported measure usefulness. Our assessments of change in domain status across time (paired t tests of baseline vs follow-up status and change score comparisons of meaningful vs no/not meaningful Global Rating of Change Score status) supported the instruments' sensitivity to measuring change.

Conclusions: We developed a methodology for creating condition-specific instruments in 2 high-impact conditions that bridge gaps in existing generic measures. The methodology creates instruments that gather the patient's perspective while allowing health systems, researchers, and other interested parties to monitor and compare outcomes over time, conditions, and populations.

Limitations: We noted 4 primary limitations: (1) Focus group participants were not representative of the diversity of education, racial, or ethnic backgrounds in the United States; we were unable to account for nonrespondents in the cross-sectional panel; (3) the final instruments may be too long for ongoing use in clinical settings; and (4) we retained less than half of respondents with HF in the longitudinal study owing to their disease burden and our limited access to patients.