Fundamental Measurement and Quality-Adjusted Life Years

We write in response to a recent discussion Value Health over the preference weights for EQ-5D-5L.1Norman R. Olsen J. Competing views on EQ-5D-5L valuation set.Value Health. 2020; 23: 574-575Abstract Full Text PDF PubMed Scopus (4) Google Scholar It would appear that there is disagreement and options open move from EQ-5D-3L value framework support reference case lifetime models NICE assessment. In ongoing lengthy debate of merits multiattribute instruments, protagonists have overlooked significant point: EQ-5D assessment framework, common with other utility fails meet required axioms fundamental measurement. The implication this failure, which has been recognized 30 years, scores are nothing more than ordinal or raw score measures.2Merbitz C. Morris Grip Ordinal scales foundations misinference.Arch Phys Med Rehabil. 1989; 70: 312Google They cannot claims therapy. For measure be used quality-adjusted life year (QALY) analyses, it require ratio properties. instruments available at level. This means they addition, subtraction, multiplication, division scores. To calculate QALYs, an outcome instrument must level (ie, true zero). Then possible multiply time spent disease state by its utility. easily demonstrated generates ranging –0.59 (a worse death) 1.0 (apparently perfect health). Even here not zero because produced scoring system. system also leads negative utilities QALYs. Attempting rescale otherwise lose no answer; will still yield A developing measurement requirements does exist. Ratio challenge. Rasch Measurement Model (RMM). rests, using simultaneous conjoint measurement, assumption probability affirming patient-reported item depends 2 factors: ability respondent difficulty item. RMM compatible either classical test theory (CTT) theory. are, as Bond Fox point out, competing paradigms.3Bond T. Applying Model: Fundamental Human Sciences.3rd ed. Routledge, New York, NY2015Crossref takes perspective if standards, then we should adopt data-to-model paradigm. If concerned with, happy ignore, questions can follow model-to-data key distinction uses procedures physical sciences point. aim standards by, Stevens pointed out 1940s, allocating numbers events according certain rules.4Stevens S. On measurement.Science. 1946; 103: 677-680Crossref (2359) these rules comprise RMM. therapy taken seriously then, determined within paradigm, scale unidimensional. reflective single latent construct. follows apply sciences. Multiattribute clearly avoid adding together several differing attributes into score. Every contribute underlying attribute. By measures such create even do determine Furthermore, concept QALY impossible mathematical employed education last 60 years needs-based quality-of-life past 20 years. Health, ISPOR's flagship journal, address issue inform membership may rethink modeled cost-per-QALY claims. Author Contributions: Concept design: Langley, McKenna Drafting manuscript: Critical revision paper important intellectual content: Conflict Interest Disclosures: authors reported conflicts interest. Funding/Support: received financial research.