Accuracy of rheumatoid arthritis diagnosis coding in primary care: a validation study

Accurate identification of rheumatoid arthritis (RA) within routinely collected health data is essential for disease surveillance, service planning and research. This study aims to determine the proportion of recorded RA diagnoses in primary care records that met predefined validation criteria for RA ascertainment.

We conducted a retrospective diagnostic validation study of subjects with a SNOMED-coded RA diagnosis recorded in the Pinnacle Midlands Health Network primary care database in 2018-2025. RA status was assessed using a structured, hierarchical validation pathway incorporating rheumatology specialist involvement, disease-modifying antirheumatic drug (DMARD) exposure and RA-specific serology (rheumatoid factor [RF] and anti-cyclic citrullinated peptide [aCCP] antibodies), with independent review of equivocal cases by two senior rheumatologists. Positive predictive value (PPV) was calculated against this composite reference standard. Additional internal validation was performed in random samples of DMARD-exposed subjects with and without specialist involvement.

Among 3,831 subjects with a SNOMED-coded RA diagnosis, 3,306 were confirmed to have RA, corresponding to a PPV of 86.3% (95% confidence interval 85.2-87.4%). PPV was highest among subjects with documented rheumatology specialist involvement (94.4%). Misclassification was concentrated among subjects without specialist involvement, DMARD exposure or serological data, accounting for most false-positive diagnoses. Internal validation demonstrated modest residual misclassification when DMARD exposure alone was used, particularly in subjects without specialist involvement.

SNOMED-coded RA diagnoses in primary care demonstrate high PPV, but reliance on coding alone may lead to misclassification. Adoption of validated case definitions, alongside refinement of diagnostic coding practices, is essential to support accurate epidemiology research and health service planning.