Amidst justified concerns about the potential for overdiagnosis it is easy to forget the great challenges associated with establishing a timely and accurate diagnosis in the first place. Many thousands of patients each year are believed to die or suffer serious harm because of missed diagnoses. In the US, in the last 25 years diagnostic errors were responsible for more than one in four malpractice claims, leading to an estimated average of 1.5 billion dollars paid out in damages every year. The International Diagnostic Error in Medicine conference is dedicated to research on improving the accuracy and timeliness of diagnosis in every field of medicine. It is organised by the Society for Improving Diagnosis in Medicine and the sixth such annual conference was held recently in Chicago.
The event started with a preconference “crash course” on diagnostic error theory. Diagnostic errors can be generated by both system (such as over-stretched capacity) and psychological (cognitive and emotional) factors. The latter can result from “cognitive biases.” Emergency physician and educationalist, Pat Croskerry described several types of such biases including the notorious “anchoring bias”—the inability to “move on” from an initially suspected diagnosis, in spite of a changing clinical picture and/or discordant subsequent test results. However, many diagnostic errors result from both system and psychological factors, as commented several times during the conference.
I met many doctors passionately involved in medical training. There were several workshops on teaching clinical reasoning and “cognitive de-biasing.” The organisers have also made an extensive clinical reasoning training resource available online.
One approach to improving diagnostic performance is the mining of already collected electronic data (i.e. in relation to abnormal laboratory tests) to identify patient groups at higher risk of a missed diagnosis. For example, recent research by Hardeep Singh’s Houston-based group indicates that such “electronic triggers” can help to spot missed diagnoses of cancer. In general, diagnostic tests can both resolve diagnostic uncertainty and contribute to the generation of diagnostic errors, an intriguing antithesis explored in a workshop dedicated to diagnostic testing and further discussed in this paper by the conference host Paul Epner and colleagues.
Many substantive contributions were made by patients directly affected by diagnostic errors. Patients can be a valuable resource in helping to “co-produce” their diagnosis—and if so, interventions aiming to empower patients can help to reduce diagnostic errors in the future. This of course assumes that patients are happy and able to take up a more active role during consultation; and that doctors in general do improve upon their inter-personal care and communication skills.
Diagnostic errors represent a special type of healthcare related adverse events. Unsurprisingly, relevant research has many methodological and practical similarities with other patient safety research. But many speakers lamented that diagnostic errors have been “overlooked” by quality initiatives in recent years (chiefly focusing on errors of commission, such as prescribing errors, post-operative mortality and healthcare associated infections). This is in spite of the fact that diagnostic errors seem to “outperform” any other type of medico-legal complaints in ambulatory care settings, as shown in a recent paper by Gordy Schiff and colleagues, and by Larry Casalino’s group previously.
A potential reason why quality improvement initiatives have so far tended to exclude diagnostic safety is that diagnostic errors are difficult to define and measure. Acknowledging these challenges, I have argued that in the context of non-acute care, the number of times a patient had seen a doctor before the correct diagnosis was suspected has good potential as a measure of diagnostic quality.
There were discussions about interventions that aim to help improve diagnosis during the consultation, such as “checklists” and clinical decision support tools (manual or electronic). UK based primary care and cancer researchers will be familiar with the use of risk assessment tools to help diagnose cancer promptly. We are at the start of a research journey aiming to evaluate both the benefits and also the potential harms that may result from such interventions. Bob Wachter highlighted the potential adverse impact on communication with patients that may result from the use of electronic decision support tools. He used a captivating picture drawn by a seven year old patient (previously published by Elizabeth Toll): during a medical consultation going clearly wrong, it portrays a doctor “examining” his own computer screen (instead of the patient) during the consultation.
The conference coincided with the publication of a great collection of new papers dedicated to the subject of diagnostic error included in a special issue of BMJ Quality and Safety (all articles are open access). There is also a new journal dedicated to publishing research to improve medical diagnosis. It is called Diagnosis and its editors are Mark Graber and Mario Plebani.
Georgios Lyratzopoulos (@GLyratzopoulos) is a public health and health services researcher working at the Cambridge Centre for Health Services Research. He focuses on the earlier diagnosis of cancer, and other aspects of cancer healthcare quality and cancer epidemiology. He is a member of BMJ’s Board of Fellows.
Funding declaration: GL is supported by a post doctoral fellowship award by the National Institute for Health Research (NIHR PDF-2011-04-047). The views expressed in this publication are those of the author and not necessarily those of the NHS, the National Institute for Health Research, or the Department of Health.
Conflict of interest statement: I participated and presented my work in the conference related to this blog and have met with several of the organisers during the event.