Reframing diagnostic reasoning: the Bayesian imperative in shoulder examination

Area of Science:

• Orthopedics
• Diagnostic Reasoning
• Bayesian Statistics

Background:

• Clinical shoulder tests like Jobe and Neer are widely used but often lead to binary diagnostic conclusions.
• Reported diagnostic accuracy for these tests varies, lacking quantitative rigor and uncertainty assessment.
• Existing summaries of sensitivity, specificity, and likelihood ratios do not adequately reflect study weighting or uncertainty.

Purpose of the Study:

• To formalize diagnostic reasoning in shoulder testing using a Bayesian framework.
• To demonstrate a transparent and reproducible method for calculating post-test probability.
• To highlight the limitations of individual test diagnostic value and advocate for sequential application.

Main Methods:

• Application of a Bayesian framework to integrate pre-test probability with pooled likelihood ratios.
• Utilizing published data from Hegedus et al. (2012) for likelihood ratios.
• Illustrating the method with a clinical example involving the Drop Arm Test and sequential test application.

Main Results:

• A positive Drop Arm Test moderately increased a 30% pre-test probability to approximately 51%.
• Sequential application of multiple tests substantially increased post-test probability to around 63%.
• The order of test application does not alter the final post-test probability but affects the intermediate diagnostic trajectory.

Conclusions:

• Individual clinical shoulder tests have limited standalone diagnostic value.
• Structured, sequential application of tests is crucial for accurate musculoskeletal diagnosis.
• Bayesian modeling offers a robust framework for interpreting clinical test results, accounting for diagnostic uncertainty.