Treatment variability for shoulder pain between physician and non-physician clinicians based on initial setting and specific shoulder diagnosis: a health system analysis

Shoulder pain is common, can arise from various causes and has a highly variable prognosis. Treatment may differ based on the clinician delivering the care and initial care settings (primary, specialty, or emergency care). The purpose of this study was to investigate how the management of shoulder pain differs depending on clinical care settings and clinician type. This was an observational cohort study. Using routinely collected health information from Military Health System electronic medical records and claims data, we assessed initial care setting and provider types for common shoulder disorders that occurred between July 1, 2013, and March 31, 2019. We identified shoulder-related care using current procedural terminology (CPT) and ICD-10 diagnosis codes marked in encounters within three months of initial diagnosis. Care was categorized into pharmacological treatment, non-pharmacological treatment, and imaging procedures, and compared across initial care settings and provider types. There were 246,041 unique individuals in the cohort with a mean(SD) age of 37.9(12.3) years, 21.9% female, 63.1% on active duty, and 76.3% enlisted. Most patients were initially seen in a primary care setting (80.2%), followed by specialty care (16.7%) and emergency care(3.1%), and 44.4% of the patients were seen by physicians. Across all settings and clinician types, non-specific shoulder diagnosis (i.e. non-specific or unspecified shoulder pain) was the most common (73.9%), followed by rotator cuff-related pain disorders (15.9%), multiple specific diagnoses (3.7%), glenohumeral osteoarthritis (2.0%), and hypomobility disorders (1.8%). Patients who saw a specialty care non-physician were more likely to receive exercise or physical therapy than other clinician-location types, which was fairly consistent across all diagnostic groups. Diagnostic subgroups of shoulder pain and initial care settings influence practice variability between physician and non-physician management of shoulder pain. Additionally, three out of every 4 patients received a non-specific shoulder diagnosis. These findings highlight the lack of coding using specific diagnostic labels, which prevents a deeper assessment of care care variability across specific shoulder diagnostic subgroups. Initial care settings should also be considered when exploring variability in care pathways for shoulder disorders.