OECD Health Working Paper: The economics of diagnostic safety (28 March 2025)

Summary

Diagnosis is complex and iterative, therefore liable to error in accurately and timely identifying underlying health problems, and communicating these to patients. Up to 15% of diagnoses are estimated to be inaccurate, delayed or wrong. Diagnostic errors negatively impact patient outcomes and increase use of healthcare resources.

This Health Working Paper from the Organisation for Economic Co-operation and Development (OECD) defines the scope of diagnostic error and illustrates the burden of diagnostic error in commonly diagnosed conditions. It also estimates the direct costs of diagnostic error and provides policy options to improve diagnostic safety.

Content

Key findings of this report included:

• Most people will experience at least one diagnostic error in their lifetime, sometimes resulting in severe patient harm, as it is estimated that 80% of all harm caused by delayed or misdiagnosis may be preventable.
• Tests, tools, diagnostic procedures and information systems are proliferating across healthcare settings to help patients and providers identify the exact nature of health problems. Despite these technological advances, health systems may still fail to identify and communicate health conditions correctly or in a timely way.
• Diagnostic errors negatively impact patient outcomes and increase the use of healthcare services, with associated increased costs. An estimated 2.6 million diagnostic errors occur in the United States each year, resulting in approximately 371,000 deaths and 424,000 permanent disabilities due to misdiagnosis.
• The report estimates that the direct consequences of diagnostic error on healthcare budgets account for 17.5% of total healthcare expenditure. In the United States this would amount to USD 870 billion each year.
• Deficits in health system design and governance, clinical environments, and individual provider competencies can drive poor diagnostic outcomes.
• Internationally, guidelines and standards on accurate and timely diagnosis for health conditions can be lacking and not systematically adopted.
• Even a relatively modest target of halving diagnostic error rates would not only reduce considerable patient suffering and distress but could free up as much as 8% of healthcare expenditure. Across OECD countries, this would equate to USD 676 billion a year.

Setting out what policymakers can do to improve diagnostic safety, the paper suggests the following set of actions:

• Clinical directors should foster changes in medical work culture and clinical environment for peer feedback and multidisciplinary approach to patient diagnosis and review.
• Patient perspectives and preferences should be taken into account when making and reviewing a diagnosis.
• Medical specialty associations should set national or international standards and guidelines for ordering diagnostic testing and interpreting results, to minimise diagnostic error, harms and wasteful healthcare expenditure.
• National patient safety agencies should routinely collect, report and publish quality assurance indicators for error and safety for diagnosis of common conditions such as cancer screening, mental health disorders and sepsis.
• Health financing should report on regional or institutional variations or anomalies in expenditure and reimbursement for diagnosis rates or diagnostic testing, indicative or poor quality care.
• Healthcare insurers and providers should review policies for financing and reimbursement of diagnostic practices that do not conform to best international practice or guidelines in order to enable healthcare expenditure savings.
• Healthcare systems should leverage digital health architecture to prioritise development of integrated health information flows between patients and different healthcare providers, to ensure timely and systematic follow-up and communication of diagnosis.
• The use of language learning models and AI to analyse multiple clinical, biomedical and radiological patient data sources to achieve a more accurate and timely diagnosis requires clinical validation and ongoing refinement, but may be of use in conditions where clinical diagnosis is currently challenging or reliable diagnostic testing is lacking.