Evidence for overdiagnosis in non-cancer conditions was assessed: a meta-epidemiological study using the 'Fair Umpire' framework

OBJECTIVE: To evaluate the strength of the evidence for, and the extent of, overdiagnosis in non-cancer conditions.

STUDY DESIGN AND SETTING: We systematically searched for studies investigating overdiagnosis in non-cancer conditions. Using the 'Fair Umpire' framework to assess the evidence that cases diagnosed by one diagnostic strategy but not by another may be overdiagnosed, two reviewers independently identified whether a Fair Umpire - a disease-specific clinical outcome, a test result or risk factor that can determine whether an additional case does or does not have disease - was present. Disease-specific clinical outcomes provide the strongest evidence for overdiagnosis, follow-up or concurrent tests provide weaker evidence and risk factors provide only weak evidence. Studies without a Fair Umpire provide the weakest evidence of overdiagnosis.

RESULTS: Of 132 studies, 47 (36%) did not include a Fair Umpire to adjudicate additional diagnoses. When present, the most common Umpire was a single test or risk factor (32% of studies), with disease-specific clinical outcome Umpires used in only 21% of studies. Estimates of overdiagnosis included 43%-45% of screen-detected acute abdominal aneurysms, 54% of cases of acute kidney injury and 77% of cases of oligohydramnios in pregnancy.

CONCLUSION: Much of the current evidence for overdiagnosis in non-cancer conditions is weak. Application of the framework can guide development of robust studies to detect and estimate overdiagnosis in non-cancer conditions, ultimately informing evidence-based policies to reduce it.