MEASUREMENT PRECISION MATTERS: PROPR DETECTS TREATMENT EFFECTS MISSED BY EQ-5D-5L IN THE CONVINCE RANDOMIZED TRIAL
Author(s)
Christoph P. Klapproth, MD1, Anna G. Schappert, M.D.2, Leili Riazy, M.Sc.2, Felix Fischer, PhD2, Giovanni F.M. Strippoli, PhD3, Krister Cromm, MA,MBA4, Peter J. Blankestijn, PhD5, Michiel L. Bots, PhD5, Kathrin Fischer, BA, MSc, PhD6, Jörgen Hegbrant, PhD7, Bernard Canaud, PhD8, Andrew Davenport, PhD9, Mark Woodward, PhD10, Claudia Barth, PhD11, Matthias Rose, PhD2.
1Internist, Research Fellow, Charité University Medicine Berlin, Berlin, Germany, 2Charité University Medicine Berlin, Berlin, Germany, 3Polytechnic University of Bari, Bari, Italy, 4Fresenius Medical Care, Bad Homburg, Germany, 5UMC Utrecht, Utrecht, Netherlands, 6Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany, 7Lund University, Lund, Sweden, 8Montpellier University, Montpellier, France, 9University College London (UCL), London, United Kingdom, 10Imperial College London, London, United Kingdom, 11B. Braun Avitum, Melsungen, Germany.
1Internist, Research Fellow, Charité University Medicine Berlin, Berlin, Germany, 2Charité University Medicine Berlin, Berlin, Germany, 3Polytechnic University of Bari, Bari, Italy, 4Fresenius Medical Care, Bad Homburg, Germany, 5UMC Utrecht, Utrecht, Netherlands, 6Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany, 7Lund University, Lund, Sweden, 8Montpellier University, Montpellier, France, 9University College London (UCL), London, United Kingdom, 10Imperial College London, London, United Kingdom, 11B. Braun Avitum, Melsungen, Germany.
OBJECTIVES: Preference-based measures (PBMs) are central to estimating quality-adjusted life years in cost-effectiveness analyses, yet their psychometric performance depends on that of their underlying descriptive systems. The PROMIS Preference Score (PROPr) incorporates item-response theory-based PROMIS domains and may provide greater measurement precision and sensitivity to small treatment effects. This may be particularly relevant in long-term randomized controlled trials (RCT) characterized by high attrition and modest differences in symptom burden. We compared the PROPr to the EQ-5D-5L in the CONVINCE RCT of high-dose hemodiafiltration (HDF) versus conventional hemodialysis (HD).
METHODS: CONVINCE randomized 810 patients from 61 centers across 8 countries to high-dose HDF (n=413) or HD (n=397), with follow-up through 36 months. EQ-5D-5L and PROPr were administered at baseline and every 6 months. Longitudinal changes in PBM scores were analyzed using linear mixed-effects models including fixed effects for time, treatment group, and treatment-by-time interaction, under a missing-at-random assumption. Secondary analyses evaluated EQ-5D-5L dimensions and PROMIS domains.
RESULTS: Baseline PROPr scores were identical between groups (mean 0.42±0.23), whereas EQ-5D-5L index scores were 0.73±0.30 for HD and 0.75±0.28 for HDF. Over follow-up, PROPr detected a significant longitudinal treatment effect favoring HDF versus HD (interaction β=0.002, 95% CI 0.000 to 0.003; p=0.032), whereas EQ-5D-5L did not (interaction β=0.002, 95% CI -0.000 to 0.003; p=0.079). Both instruments demonstrated similar overall time-related decline (β=-0.004 per assessment interval). Domain-level analyses identified significant treatment-by-time interactions favoring HDF for PROMIS Physical Function (β=0.067, 95% CI 0.014 to 0.120; p=0.014), PROMIS Social Participation (β=0.089, 95% CI 0.011 to 0.167; p=0.026), and EQ-5D-5L Self-Care (β=-0.008, 95% CI -0.014 to -0.002; p=0.009). Attrition reached 87% by 36 months.
CONCLUSIONS: In a large RCT with substantial attrition, PROPr identified treatment differences that were not detected by EQ-5D-5L. These findings suggest that greater measurement precision at the domain level of PBMs may improve measurement precision in cost-effectiveness analyses.
METHODS: CONVINCE randomized 810 patients from 61 centers across 8 countries to high-dose HDF (n=413) or HD (n=397), with follow-up through 36 months. EQ-5D-5L and PROPr were administered at baseline and every 6 months. Longitudinal changes in PBM scores were analyzed using linear mixed-effects models including fixed effects for time, treatment group, and treatment-by-time interaction, under a missing-at-random assumption. Secondary analyses evaluated EQ-5D-5L dimensions and PROMIS domains.
RESULTS: Baseline PROPr scores were identical between groups (mean 0.42±0.23), whereas EQ-5D-5L index scores were 0.73±0.30 for HD and 0.75±0.28 for HDF. Over follow-up, PROPr detected a significant longitudinal treatment effect favoring HDF versus HD (interaction β=0.002, 95% CI 0.000 to 0.003; p=0.032), whereas EQ-5D-5L did not (interaction β=0.002, 95% CI -0.000 to 0.003; p=0.079). Both instruments demonstrated similar overall time-related decline (β=-0.004 per assessment interval). Domain-level analyses identified significant treatment-by-time interactions favoring HDF for PROMIS Physical Function (β=0.067, 95% CI 0.014 to 0.120; p=0.014), PROMIS Social Participation (β=0.089, 95% CI 0.011 to 0.167; p=0.026), and EQ-5D-5L Self-Care (β=-0.008, 95% CI -0.014 to -0.002; p=0.009). Attrition reached 87% by 36 months.
CONCLUSIONS: In a large RCT with substantial attrition, PROPr identified treatment differences that were not detected by EQ-5D-5L. These findings suggest that greater measurement precision at the domain level of PBMs may improve measurement precision in cost-effectiveness analyses.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
PCR147
Topic
Clinical Outcomes, Methodological & Statistical Research, Patient-Centered Research
Topic Subcategory
Health State Utilities, Patient-reported Outcomes & Quality of Life Outcomes
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Urinary/Kidney Disorders