Background Osteoarthritis (OA) is recognised as the most prevalent joint disorder in the world. However, accurate estimates of true disease prevalence are affected by a number of factors which include the definition of OA applied. UK guidance recommends application of a working diagnosis of OA in primary care, allowing diagnosis to be made in older adults (45 years and over) without recourse to further investigation. However, evidence shows that older adults may receive either an OA or joint pain morbidity code suggesting different diagnostic thresholds are applied by different clinicians. This may have consequences for both patient outcomes and accuracy of prevalence estimate calculations, hence healthcare service provision.

Objectives By using morbidity codes as a proxy for diagnosis, this study aimed to explore whether certain patient characteristics which are known risk factors for OA alter the diagnostic threshold in older patients presenting with joint symptoms.

Methods The study was set in ten UK general practices (population 66,231) which contribute anonymised consultation information to the Consultations in Primary Care Archive (CiPCA). Patients aged 45 years or over who presented in a two year period with a new episode of a joint-related problem, either coded as OA or as joint symptom (JS), were identified. Cross-sectional analysis explored the association between risk factors for OA (gender, age, BMI, deprivation status) and code attribution (JS or OA). Longitudinal analysis explored associations with final code attribution.

Results Of 3,950 patients presenting with a joint-related problem, 2,711 (68.6%) received an initial JS code and 1,239 (31.4%) an OA code. Female gender [odds ratio (OR) 1.38 (95% CI 1.20-1.60)] and increasing age [60-69: OR 2.18 (95% CI 1.83-2.61); 70-79: OR 2.83 (95% CI 2.35-3.41); ≥80: OR 3.25 (95% CI 2.59-4.08)] were associated with an increase in the likelihood of being attributed an OA code as a result of initial consultation. No such association was found with BMI or deprivation status. Seven hundred (25.8%) of those receiving an initial JS code were attributed an OA code during the eight year follow-up period with the highest proportion (42.9%) occurring within the first year. The association between likelihood of OA code attribution and female gender [hazard ratio (HR) 1.24 (95% CI 1.06-1.45)] and increasing age [60-69: HR 1.83 (CI 1.52-2.20); 70-79: HR 2.03 (CI 1.67-2.47); ≥80: HR 2.03 (CI 1.56-2.65)] endured over time.

Conclusions Approximately two-thirds of patients were coded as JS at initial presentation, suggesting that they did not receive a diagnosis of OA. Approximately one quarter of these patients received an OA code over the next eight years. Given UK recommendations for primary care clinicians, in a sample of older patients presenting with joint-related problems a greater proportion would have been expected to receive an OA diagnosis hence OA code. Although patient age and gender are associated with the diagnosis made, further research should investigate other factors such as clinician characteristics which may be associated with OA diagnosis and whether the diagnosis made affects long-term patient outcomes.

Acknowledgements Time part supported by NIHR CLAHRC WM. Academic extension funded by West Midlands Deanery. CiPCA supported by North Staffordshire Primary Care Research Consortium and NCCRCD

Disclosure of Interest None declared