OP0075 Ultrasound-detected shoulder pathologies cluster into groups with different clinical associations: data from a prospective study of 500 community referrals for shoulder pain

Background Shoulder pain is common and its management remains challenging, often resulting in poor outcomes: 50% of people continue to have shoulder pain at 18 months. This may be in part due to inaccurate clinical diagnosis. Ultrasound offers accurate detection of pathology and its use is increasing. However, the relationship between ultrasound findings and clinical phenotype is unclear. Objectives A prospective study was undertaken to explore latent class groupings and explore the association between patient reported outcome measures and the different groups. Methods 500 primary care patients attending for shoulder ultrasound were prospectively recruited. Radiologists and sonographers underwent training to ensure standardised reporting. Baseline data was collected via self-reported questionnaires. Outcome measures collected included pain, function, quality of life, treatments received, activity, self-efficacy and levels of acceptable symptom states. These measures underwent Rasch analysis. Latent class analysis was undertaken to identify groups. Results Mean age was 53.6% and 52% were female. Latent class analysis confirmed the existence of 4 groups: limited bursitis; extensive inflammation; RC tears; limited pathology. The oldest age group were those with RC tears, and the youngest was those with limited bursitis. The rotator cuff tear group had the highest levels of pain and disability, and the lowest levels of acceptable symptom states. Those with limited pathology had the highest levels of acceptable symptom states. The extensive inflammation group had the lowest activity scores. Summary statistics for classes obtained prior to covariate adjustment (complete data only) Conclusions This study confirms that ultrasound pathologies cluster into groups. These groups appear to differ in symptom associations at baseline; expanding the LCA to include covariates will allow us to formally explore these associations. A longitudinal study will provide understanding of the relevance of these groups to long-term patient outcomes. Disclosure of Interest None declared