PS3-49 SELECTION OF KEY PROMIS DOMAINS FOR A PREFERENCE-BASED SCORING SYSTEM

Tuesday, October 20, 2015
Grand Ballroom EH (Hyatt Regency St. Louis at the Arch)
Poster Board # PS3-49

Janel Hanmer, MD, PhD1, David Feeny, PhD2, Baruch Fischhoff, PhD3, Ron Hays, PhD4, Rachel Hess, MD, MS5, Paul Pilkonis, PhD3, Dennis Revicki, PhD6, Mark S. Roberts, MD, MPH7, Joel Tsevat, MD8 and Lan Yu, PhD9, (1)The University of Pittsburgh, Pittsburgh, PA, (2)Hamilton, ON, Canada, (3)Pittsburgh, PA, (4)Los Angeles, CA, (5)University of Utah, Salt Lake City, UT, (6)Bethesda, MD, (7)Department of Industrial Engineering, University of Pittsburgh, Pittsburgh, PA, (8)Cincinnati, OH, (9)University of Pittsburgh, Pittsburgh, PA
Purpose: The set of health domains used in a preference-based measure should have face validity, be structurally independent, and be parsimonious. PROMIS currently has 44 available item banks assessing health domains. We sought to select a subset for preference-based valuation.


Methods: We used a modified Delphi procedure with 8 PROMIS experts to select the most appropriate domains for a general health measure. After 3 rounds of surveys, the experts agreed on a set of 10 potential domains. The 10 domains were presented to a sample of 50 adult community members to assess structural independence. Each subject was presented with 20 randomly-selected pairs of domains and asked to consider 1 pair at a time. For a given pair, the subject was asked if it were possible to simultaneously have high functioning in domain 1 but low functioning in domain 2, and vice versa, to provide input on independence.


Results: The 50 community respondents, ranged in age ranged from 22 to 70, 60% were female, 52% were white, and 24% had a high school education. The proportion of independent pairs within an individual ranged from 0.3 to 1.0 (mean = 0.77). The domain pair with the lowest rate of perceived independence was Anxiety and Sleep (with only 50% of respondents finding that pair to have structural independence). The pairs with the highest independence rates (with ≥95% reporting structural independence) were Physical Function and Depression, Physical Function and Pain Intensity, and Satisfaction with Social Roles and Sexual Function.


On the basis of legacy preference-based measures, Physical Function, Pain, and Depression were retained for face validity. Other domains were added to this set if they were at least as independent as the minimum set (independence reported by ≥67% of respondents). The final subset of domains was: Applied Cognition - General Concerns, Emotional Distress - Depression, Fatigue, Pain Interference, Physical Function, Satisfaction with Social Roles and Activities, and Sleep Disturbance.


Conclusion: With input from quality of life experts and community members, we developed a parsimonious set of 8 PROMIS domains with face validity and structural independence suitable for a general preference-based health-related quality of life measure.

See more of: POSTER SESSION III & CONTINENTAL BREAKFAST

See more of: 37th Annual Meeting of the Society for Medical Decision Making