Article Text

Download PDFPDF
101 Can we detect overdiagnosis? exploring indicators of possible overdiagnosis outside cancer screening contexts
  1. Katy Bell1,
  2. Jenny Doust2,
  3. Leon Smith1,
  4. Ian Harris3,
  5. Rachelle Buchbinder4,
  6. Louise Cullen5,
  7. Alexandra Barratt1
  1. 1University of Sydney, Sydney, Australia
  2. 2Bond University, Gold Coast, Australia
  3. 3University of New South Wales, Sydney, Australia
  4. 4Monash University, Melbourne, Australia
  5. 5University of Queensland, Brisbane, Australia


Objectives Although the risk of overdiagnosis in cancer screening was identified almost 50 years ago, evidence of cancer overdiagnosis was slow to emerge, and has required development of new approaches and study designs. As a result, responses to mitigate overdiagnosis, and to inform the public and professionals, have been delayed. This meant cancer screening activities became well established in clinical practice and policy before the problem overdiagnosis was recognised, making reversal difficult.

It would be beneficial to avoid repeating the same delays outside the cancer setting – whether in the context of screening programs (e.g. gestational diabetes screening among pregnant women) or in managing already symptomatic patients (eg children with ADHD symptoms, adults with joint pain) – so that timely policy responses to overdiagnosis in these settings can be developed. We therefore aimed to identify early ‘indicators’ of potential overdiagnosis.

Method We used concepts described in key papers about methods for detecting and quantifying cancer overdiagnosis, the clinical utility of diagnostic tests, and for modifying the definition of diseases, to identify early indicators of potential overdiagnosis. We identified four indicators: 1. Expanded detection of condition/disease; 2. Subclinical forms of the condition/disease; 3. Increasing diagnosis and treatment of condition/disease; and 4. Balance of benefit vs harm of diagnosis/treatment likely to be unfavourable based on existing evidence.

We provide an overview of the proposed approach, and illustrate it using two examples that are common in everyday, clinical investigation of non cancer conditions: MRI in the investigation of adult knee pain, and high sensitivity troponin testing in emergency department workup of patients with suspected myocardial infarction. We evaluate and discuss the application of the indicators to these examples, using analysis of routinely collected, local data, and current research evidence.

Results Example One: Is there potential overdiagnosis of knee pathology in older Australians? Knee MRI is positive on indicators 1, 2 and 4 for possible overdiagnosis (expanded detection of condition (disease); subclinical forms of the condition (disease); balance of benefit vs harm of diagnosis/treatment likely to be unfavourable). Indicator 3 (increasing diagnosis and treatment of the condition) is inconclusive. Example Two: Is there potential overdiagnosis of Myocardial Infarction in Australians? Non ST Elevation Myocardial Infarction (NSTEMI) meets indicators 1 and 3 (expanded detection of condition (disease); increasing diagnosis and treatment of condition (disease)). Indicators 2 (subclinical forms of the condition (disease)) and 4 (benefit vs harm of diagnosis/treatment likely to be unfavourable) are inconclusive.

Conclusions The proposed indicators may have utility in helping to identify possible overdiagnosis sooner than would normally come about through standard processes of surveillance of changes in clinical testing and its impact. We found that in our two examples, more indicators were clearly positive for knee MRI than for hs Tn/NSTEMI, suggesting more evidence of potential overdiagnosis for the former than for the latter. The approach is necessarily different from methods that have been developed for cancer screening because, in non-cancer conditions, patients may be asymptomatic (as in the context of early detection of disease) OR symptomatic yet still at risk of overdiagnosis. These four indicators may help identify tests that warrant the investment of research resources to assess the impact of the new testing patterns and disease definitions on long term, patient-relevant outcomes to avoid harm through overdiagnosis and overtreatment.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.