VALIDATING PRIMARY AND SECONDARY HEALTHCARE RESOURCE USE AND COSTS ESTIMATES IN OMOP-MAPPED VS SOURCE CPRD-HES DATA IN THE UK
Author(s)
Gianluca Fabiano, PhD1, Njoki Njuki, Bachelor of Science1, Antonella Delmestri, PhD1, Rafael Pinedo-Villanueva, BA, MSc, PhD2;
1University of Oxford, Oxford, United Kingdom, 2University of Oxford, Associate Professor, Oxford, United Kingdom
1University of Oxford, Oxford, United Kingdom, 2University of Oxford, Associate Professor, Oxford, United Kingdom
Presentation Documents
OBJECTIVES: To validate estimates of primary and secondary healthcare resource use (HCRU) and costs derived from OMOP-mapped CPRD Aurum data linked to Hospital Episode Statistics (HES) vis-á-vis those generated using the source data for a cohort of postmenopausal women with fragility fractures.
METHODS: We conducted a retrospective cohort study (2010-2018) of women >50 years with a first fragility fracture. HCRU and costs were estimated for up to two years post-index fracture using OMOP-mapped and source CPRD-HES data. Primary care costs were estimated using healthcare professional specialty unit costs; inpatient and outpatient costs using Healthcare Resource Groups and NHS unit costs. The OMOP-based analysis was built on variables reporting standardised OMOP concepts; supplementary analyses incorporated source variables retained within OMOP. HCRU and cost differences were summarised using means and bootstrap confidence intervals.
RESULTS: The OMOP-mapped cohort included 22,900 women (mean follow-up 578 days) and the source CPRD-HES cohort 23,106 (577 days), with ≥1 primary care encounters in 97.3% and 97.4%, respectively. Mean number of encounters was 39.0 in OMOP and 38.3 in source, with mean costs of £1,112.8 and £1,087.1, respectively. Hospitalisations were observed for 52.4% in OMOP and 52.3% in source, with a mean of 1.16 per woman in both datasets. Hospital costs using standardized OMOP variables were 35.1% lower than in source (£2,451.7 vs £3,777.3). Using retained source-specific variables, the difference reduced to 4.7% (£3,599.6 vs £3,777.3). Outpatient appointments were recorded for 89.6% of women with mean 7.28 visits (£1,183.9) in OMOP and 7.27 (£1,199.1) in source. Emergency care was recorded for 67% of women in both cohorts, with a mean 1.26 encounters (£351.8) in OMOP and 1.25 (£447.6) in source.
CONCLUSIONS: OMOP-mapped data produced HCRU and cost estimates largely equivalent to those derived from CPRD-HES. Leveraging OMOP-retained source variables and refining specialty vocabularies and mappings would improve the alignment.
METHODS: We conducted a retrospective cohort study (2010-2018) of women >50 years with a first fragility fracture. HCRU and costs were estimated for up to two years post-index fracture using OMOP-mapped and source CPRD-HES data. Primary care costs were estimated using healthcare professional specialty unit costs; inpatient and outpatient costs using Healthcare Resource Groups and NHS unit costs. The OMOP-based analysis was built on variables reporting standardised OMOP concepts; supplementary analyses incorporated source variables retained within OMOP. HCRU and cost differences were summarised using means and bootstrap confidence intervals.
RESULTS: The OMOP-mapped cohort included 22,900 women (mean follow-up 578 days) and the source CPRD-HES cohort 23,106 (577 days), with ≥1 primary care encounters in 97.3% and 97.4%, respectively. Mean number of encounters was 39.0 in OMOP and 38.3 in source, with mean costs of £1,112.8 and £1,087.1, respectively. Hospitalisations were observed for 52.4% in OMOP and 52.3% in source, with a mean of 1.16 per woman in both datasets. Hospital costs using standardized OMOP variables were 35.1% lower than in source (£2,451.7 vs £3,777.3). Using retained source-specific variables, the difference reduced to 4.7% (£3,599.6 vs £3,777.3). Outpatient appointments were recorded for 89.6% of women with mean 7.28 visits (£1,183.9) in OMOP and 7.27 (£1,199.1) in source. Emergency care was recorded for 67% of women in both cohorts, with a mean 1.26 encounters (£351.8) in OMOP and 1.25 (£447.6) in source.
CONCLUSIONS: OMOP-mapped data produced HCRU and cost estimates largely equivalent to those derived from CPRD-HES. Leveraging OMOP-retained source variables and refining specialty vocabularies and mappings would improve the alignment.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
RWD151
Topic
Real World Data & Information Systems
Topic Subcategory
Distributed Data & Research Networks
Disease
SDC: Musculoskeletal Disorders (Arthritis, Bone Disorders, Osteoporosis, Other Musculoskeletal)