Patient and Surgeon Variables Associated With a Diagnosis of Shoulder Periprosthetic Joint Infection

There is no consensus reference standard for diagnosing prosthetic shoulder joint infection (PJI). There may be inadequate distinction between colonization and infection. It is not clear that culture of a common shoulder commensal organism can account for unexpected pain or stiffness after shoulder arthroplasty. In this survey-based experiment, we asked: 1) What patient and surgeon variables are associated with diagnosis of shoulder PJI? And 2) What is the surgeon interobserver agreement for diagnosis of shoulder PJI?

One hundred seven members of the Science of Variation Group reviewed ten hypothetical patient scenarios with five randomized variables: examination findings, serum markers, synovial fluid markers, histological assessment, and organism cultured from deep tissue specimens. Participants diagnosed the presence or absence of shoulder PJI. Mixed multilevel logistic regression sought variables associated with PJI diagnosis. Interobserver agreement was measured with Fleiss kappa.

Surgeon diagnosis of shoulder PJI was independently associated with deep tissue culture growth of C. acnes (OR=235 [95% CI 19 to 2933]; P < 0.01), S. epidermidis (OR=147 [95% CI 8.4 to 2564]; P < 0.01), and S. aureus (OR=110 [95% CI 6.9 to 1755]; P < 0.01) much more so than presence of a sinus tract on examination (OR=43 [95% CI 3.7 to 505]; P < 0.01), inflammatory histology (OR=15 [95% CI 4.0 to 58]; P < 0.01), inflammatory synovial fluid markers (OR=13 [95% CI 3.9 to 45]; P < 0.01), and serum inflammatory markers (OR=5.8 [95% CI 2.0 to 17]; P < 0.01). The reliability of surgeon diagnosis for shoulder PJI was poor (Fleiss kappa = 0.013 [95% CI -0.0039 to 0.031]).

The observation that surgeons may not adequately distinguish colonization and infection - considering any positive culture as an infection - combined with the low reliability of diagnosis observed, suggests possible overdiagnosis and overtreatment of shoulder PJI.