Focusing Research Efforts in Priority Areas

The NAM outlined a broad palette of research topics in need of further study for the field to make robust progress on this complex topic, but the report also clearly pointed the field towards early wins through an initial focus on “identifying the most common diagnostic errors, 'don’t miss' health conditions that may result in patient harm, [and] diagnostic errors that are relatively easy to address.” While the most frequent diagnostic errors are likely with common conditions such as asthma or migraine, malpractice and autopsy studies consistently find that roughly 50-80% of the serious harms resulting from missed or delayed diagnoses are linked to one of three key disease categories:

1. vascular events (e.g., stroke, heart attack, pulmonary embolus)
2. infections (e.g., sepsis, meningitis, appendicitis), and
3. cancers (e.g., lung cancer, colon cancer, breast cancer).

We refer to these major disease categories responsible for the lion’s share of misdiagnosis-related harms as ‘The Big Three.’ The ‘Big Three’ are not evenly distributed across practice settings and patient populations—missed vascular events dominate in emergency care, missed infections are most common among children, and missed cancer diagnoses lead the way in primary care. Nevertheless, ‘Big Three’ diseases account for at least three (and up to five) of the ‘top five’ diseases across practice settings.

Table: Proportion of serious harms attributed to “The Big Three” in frontline care settings

Clinical Settings	Vascular Events	Infections	Cancers	Total
Emergency Medicine	~30%	~20%	~9%	~59%
Pediatrics	-	~40%	~13%	~53%
Adult Primary Care	~12%	~8%	~60%	~80%

 

It is noteworthy that causes (and therefore solutions) likely differ substantially across the ‘Big Three.’ The Table below identifies one key disease example for each of the ‘Big Three’ categories.

 

Table: Exemplars from each “Big Three” category with diagnostic error causes and possible solutions

Misdiagnosis	Principal Cause	Solution(s) in Need of Further Research
Stroke	No specialty expertise	Telemedicine & device-based decision support
Sepsis	Overwhelmed by data	Big data visual analytics & machine learning algorithms
Lung Cancer	Results not communicated	Direct-to-patient reports & EHR triggers to close loops