RSNA 2016: Do radiologists fall victim to gambler's fallacy?

All radiologists are familiar with the justification process for individual scans, weighing the risk of the dose versus the benefits of the imaging. However, modern-day radiologists may be placing too much weight on dose history, according to an RSNA 2016 presentation by Colin Walsh, MSc, a medical physicist at St James’s Hospital in Dublin.

There’s no doubt that EMR have streamlined patient care; the ability for clinicians to instantly see all patient data is invaluable for both diagnostic and therapeutic care. Including past imaging in the EMR may reduce the need for additional scans in the future, if they provide actionable information. At the same time, having the entire dose history of a patient can cloud a radiologist’s thinking with the gambler’s fallacy, according to Walsh.

The classic example of the gambler’s fallacy is repeating a chance event, such as flipping a coin. While three results of heads in a row may lead an observer to bet on tails as the next result, the true probability for each coin flip is 50/50.

“In a similar way, we have that tendency to think that patients who’ve had previous x-rays are at more risk from the next x-ray,” said Walsh. “This can create a cognitive bias.”

Exposure to radiation either causes cancer or it doesn’t, according to Walsh. If an exam has caused cancer, refraining from preforming future studies is not going to undo the harmful cell division. On the other hand, if a scan has not caused cancer—by far the most likely outcome—than the previous exposure has no bearing on risk with the current exposure.

The fallacy can affect decisions with limiting economic factors, for example, if a patient is receiving a CT exam, then for the next exam they could be given priority for MRI to reduce ionizing radiation dose. Measures of these may be unnecessary, according to Walsh.

“Dose history can be useful to check previous imaging, but we need to be wary of the strong potential for dose history to create confusion over risk estimates,” said Walsh.

The idea that a plethora of past exams results in stored risk in a patient runs counter to probability theory and the biological mechanisms of irradiation.

“We’re skewing the justification proves with cognitive errors,” said Walsh. “All risk is the same, we should check if the previous image is giving us what we need. If not, then we can justify the new scan.”