Mapping to EQ-5D Utility in Knee Osteoarthritis: A Comparative Evaluation of WOMAC and SF-36 Algorithms in the STEP 9 Trial
Author(s)
Vagia Daki, MSc1, Karthik Ramakrishnan, MPH2, Konstantinos Soulanis, MSc1, Ines Guerra, MSc3, Theodora Oikonomidi, PhD1, Inger Smith, Msc4.
1IQVIA, Athens, Greece, 2Novo Nordisk A/S, Bengaluru, India, 3IQVIA, London, United Kingdom, 4White Box Health Economics Ltd, Brighton & Hove, United Kingdom.
1IQVIA, Athens, Greece, 2Novo Nordisk A/S, Bengaluru, India, 3IQVIA, London, United Kingdom, 4White Box Health Economics Ltd, Brighton & Hove, United Kingdom.
OBJECTIVES: To critically review published EQ-5D mapping algorithms using Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) symptom scoring in knee osteoarthritis (KneeOA) and apply the mapping to STEP-9 trial data. Furthermore, to compare the results with a widely used EQ-5D mapping algorithm using SF-36 data (Rowen 2009).
METHODS: The HERC database was utilized to identify WOMAC to EQ-5D mapping algorithms. Further, a desk search of the NICE website was conducted to understand HTA submissions in knee disorders that utilized such mapping algorithms.
RESULTS: Six mapping algorithms were identified: Barton 2008, Xie 2010, Wailoo 2014, Price 2019, Bilbao 2020, and Ayala 2021. Two algorithms (Barton 2008, Xie 2010) exclusively used linear regression, which is considered not appropriate for EQ-5D mapping. Three studies (Ayala 2021, Bilbao 2020, Price 2019) did not externally validate the suggested algorithm. Most identified algorithms performed poorly in predicting EQ-5D utilities in severe health states based on WOMAC scores (Ayala 2021, Bilbao 2020, Xie 2010, Barton 2008). Furthermore, a weak association between the WOMAC stiffness subscale and EQ-5D was found in three studies (Ayala 2021, Bilbao 2020, Wailoo 2014). Wailoo 2014 was judged as most appropriate for STEP9 mapping considering its methodology and similarity to the STEP9 population, however, its implementation resulted in low utility values which were misaligned with expected observed utility values in KneeOA. Further results varied greatly in comparison to SF-36 to EQ-5D mapping using STEP-9 data. Review of NICE submissions showed no evidence of WOMAC to EQ-5D mapping, whilst SF-36 to EQ-5D mapping was implemented in one instance.
CONCLUSIONS: Mapping from WOMAC to EQ-5D may be limited due to weak associations—particularly with the stiffness subscale—and the absence of mental health domains. Mapping from SF-36 to EQ-5D could be appropriate in Knee OA, and requires further research.
METHODS: The HERC database was utilized to identify WOMAC to EQ-5D mapping algorithms. Further, a desk search of the NICE website was conducted to understand HTA submissions in knee disorders that utilized such mapping algorithms.
RESULTS: Six mapping algorithms were identified: Barton 2008, Xie 2010, Wailoo 2014, Price 2019, Bilbao 2020, and Ayala 2021. Two algorithms (Barton 2008, Xie 2010) exclusively used linear regression, which is considered not appropriate for EQ-5D mapping. Three studies (Ayala 2021, Bilbao 2020, Price 2019) did not externally validate the suggested algorithm. Most identified algorithms performed poorly in predicting EQ-5D utilities in severe health states based on WOMAC scores (Ayala 2021, Bilbao 2020, Xie 2010, Barton 2008). Furthermore, a weak association between the WOMAC stiffness subscale and EQ-5D was found in three studies (Ayala 2021, Bilbao 2020, Wailoo 2014). Wailoo 2014 was judged as most appropriate for STEP9 mapping considering its methodology and similarity to the STEP9 population, however, its implementation resulted in low utility values which were misaligned with expected observed utility values in KneeOA. Further results varied greatly in comparison to SF-36 to EQ-5D mapping using STEP-9 data. Review of NICE submissions showed no evidence of WOMAC to EQ-5D mapping, whilst SF-36 to EQ-5D mapping was implemented in one instance.
CONCLUSIONS: Mapping from WOMAC to EQ-5D may be limited due to weak associations—particularly with the stiffness subscale—and the absence of mental health domains. Mapping from SF-36 to EQ-5D could be appropriate in Knee OA, and requires further research.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
PCR6
Topic
Patient-Centered Research
Topic Subcategory
Health State Utilities, Patient-reported Outcomes & Quality of Life Outcomes
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity), Musculoskeletal Disorders (Arthritis, Bone Disorders, Osteoporosis, Other Musculoskeletal)