Presentation (CTI)

OBJECTIVES: Routinely collected healthcare databases are widely used for epidemiological, health economics and outcomes research (HEOR), and regulatory evidence generation. Their suitability depends on the representativeness of the underlying population and the completeness of routine healthcare information. This study evaluated the representativeness of IQVIA Medical Research Data (IMRD) using independent national demographic, mortality and disease benchmarks for England.
METHODS: A repeated annual cross-sectional analysis used IMRD, a longitudinal UK primary care electronic healthcare database. Representativeness was assessed using national benchmarks for England across three domains: demographic structure (2024), mortality (annual standardised mortality ratios [SMRs], 2014-2024), and age- and sex-standardised prevalence of common chronic diseases (2024). Disease definitions were based on NHS England QOF SNOMED CT codelists and prevalence estimates were directly standardised to the 2024 England population.
RESULTS: IMRD included over 2.1 million actively registered patients in 2024. The age and sex distribution was broadly similar to that of the England population, with modest differences in the age structure. Annual standardised mortality ratios (SMRs) ranged from 0.88 to 1.00 between 2014 and 2024. Age- and sex-standardised prevalence estimates were comparable with published national benchmarks for England, including diabetes (7.52% vs 7.7-7.9%), heart failure (1.39% vs 1.33%) and chronic kidney disease stage 3-5 (4.39% vs 4.6%). Lastly, the prevalence ratio of type 2 to type 1 diabetes was approximately 14:1, consistent with national estimates.
CONCLUSIONS: IMRD demonstrated good agreement with independent national demographic, mortality and disease prevalence benchmarks for England. Modest differences in age structure support the use of age- and sex-standardisation when comparing disease prevalence with national estimates. The consistently high SMRs provide reassurance regarding mortality ascertainment and maintenance of an accurate registered patient denominator. Multi-domain benchmarking provides a practical framework for assessing the suitability of routinely collected healthcare databases for epidemiological and HEOR research.