Objective: To examine factors associated with musculoskeletal referral and determine whether referral influences clinical outcomes.
Methods: Patients included in the study presented with knee or shoulder pain to primary care physicians affiliated with an academic teaching hospital. The primary clinical outcome was change in pain or function measured up to 12 months after initial presentation. Covariates included baseline pain and function, duration of complaint, initial diagnosis, insurance status, and several demographic factors.
Results: Forty-one percent (65 of 160 patients) were referred for knee or shoulder conditions, 47 (29%) patients were referred only to an orthopedic surgeon, and 12 (8%) only to a rheumatologist; 6 (4%) patients saw both specialists. For patients with knee pain, the only variable correlated with referral was an initial diagnosis of internal derangement (p = 0.02). No variable was significantly associated with referral for shoulder pain. Baseline pain, baseline function, duration of complaint, age, and insurance status were not associated with referral for either knee or shoulder pain (all p values > 0.05). The variables most associated with improvement in pain and function were more severe pain and function at baseline (all p values < 0.0001). In multivariate analyses controlling for clinical and demographic factors among patients with shoulder complaints, referral was associated with significantly less improvement in clinical outcomes than non-referral (p = 0.02). Referral was not associated with clinical outcomes for patients with knee pain.
Conclusions: Referral was common for patients with knee or shoulder conditions. The only baseline variable correlated with referral was a diagnosis of knee internal derangement. Referral was not associated with improvement in pain or function and may actually be correlated with worse outcomes among patients with shoulder pain, although this is likely due to unmeasured factors contributing to the referral decision.