How excellent can centres of excellence be? The impact of prevalence on service quality.

Centres of Excellence (CEs) are thought to provide better quality services for their speciality than Generic Services (GS). However, clinical test theory suggests this may arise from differences in the prevalence of these specialities' conditions in their referral populations, which affects the services' ability to detect diagnoses accurately, even with similar diagnostic sensitivities and specificities. Furthermore, GS' insensitivity to rarer diagnoses is necessary to avoid serious overdiagnosis despite using skills equivalent to CEs. Good GS can perform as well as CEs for disorders of 15% to 20% or greater prevalence in their referral populations, depending on the Minimal Clinically Important Difference (MCID) decided for their diagnoses' positive predictive values or degree of bias. CEs are necessary for rare disorders and have a role in determining MCIDs and the sensitivity and specificity of new measures. Sensitivity, specificity, positive & negative predictive values, and true diagnostic prevalence should be routine outcome measures.