A new study finds that physicians believe that preventing diagnostic errors can be aided using technology like decision support and artificial intelligence, but they will never replace the physician’s role in diagnosis.
The 6,400-clinician study was conducted by QuantiaMD, which touts itself as the largest mobile and online physician community.
Key findings of the study include:
- 47 percent of the respondents encounter diagnostic errors (e.g., missed, late, or wrong diagnoses) at their practice at least monthly.
- 64 percent said that up to 10 percent of misdiagnoses they have experienced have directly resulted in patient harm.
- More than 96 percent of clinicians believe diagnostic errors are preventable at least some of the time.
Study respondents identified atypical patient presentation, failure to consider other diagnoses and inadequate patient history as the top contributors to diagnostic errors. Respondents also noted external factors such as a over-testing to avoid malpractice risk.
Clinicians indicated the top five diagnoses at greatest risk for misdiagnosis as:
- Pulmonary embolism
- Bipolar disorder
- Breast cancer
- Myocardial infarction
An estimated 40,000 to 80,000 US hospital deaths result from misdiagnosis annually, according to a 2002 JAMA article. However, these errors have received little attention due to a combination of under-reporting and under-developed measurement techniques. QuantiaMD's Do No Harm Special Interest Group initiated this study to shed light on this problem and to engage the nation's physicians in a mobile and online interactive educational program aimed at helping to prevent these errors.
"Diagnostic errors have been a long-neglected aspect of the patient safety movement, not getting the attention they deserve," said Robert M. Wachter, MD, professor and associate chairman, Department of Medicine, University of California, San Francisco and the moderator of the new series. "This QuantiaMD program, which includes presentations by many of the world's leading experts on diagnostic mistakes, lays the foundation for physicians to come together to begin discussing the challenge of improving our diagnostic acumen, and learning about the utility of new tools, including new kinds of cognitive training, systems-based decision support tools, and even artificial intelligence."
Looking to the future, study respondents feel that efforts to prevent diagnostic errors must strike a balance between the science and the art of diagnosis. Three quarters of clinicians were confident that decision support tools and artificial intelligence aids will help to prevent diagnostic errors in the future. However, 56 percent believe that artificial intelligence aids, such as a medical version of Watson - IBM's artificial intelligence computer system - will never be able to replace the physician's role in diagnosis.
"Unlike many patient safety topics, which often focus on management systems and are executed by various members of the care team, diagnostic errors are 100 percent physician territory," said Michael Paskavitz, editor in chief of QuantiaMD. "What excites us is that QuantiaMD members initiated this study and are proactively collaborating to deliver solutions. The clinical and financial impact of diagnostic errors are profound. Equally profound, however, is the opportunity that comes with tens of thousands of physicians focusing their attention on improving a vital element of a patient's care experience."
QuantiaMD's 8-part series, "Preventing Diagnostic Errors", is available for viewing here. In it, Wachter and his faculty of experts from across the nation discuss the latest thinking around misdiagnosis, including current and future approaches to overcome many of the challenges identified in this study and elsewhere. Each interactive segment takes on a different aspect of the issues, such as exploring systems and cognitive errors, focusing on improving diagnostic reasoning, and discussing policies and future approaches to prevention.
Click here to view the study “Physician Perspectives on Preventing Diagnostic Errors.”