Female medical students answered medical questions correctly more often than their male counterparts, but expressed less certainty about their answers, researchers reported.
Recognizing the association -- or lack thereof -- between accuracy and confidence in medical trainees may help reduce diagnostic errors caused by overconfidence, wrote three male medical students in a letter published in .
The findings "are a reminder that less confidence might not indicate a lack of knowledge and that confidence should not be mistaken for correctness," wrote , an MD/MBA student at New York University School of Medicine in New York City, and colleagues.
"There will always be a balance in medicine between running additional diagnostic tests and being reasonably confident about the diagnosis and plan given the (possibly limited) data at hand in a clinical setting," co-author , an MD candidate at Johns Hopkins School of Medicine, told 鶹ý in an email.
"As a physician, if you lean too far in either direction you're either incurring additional costs and additional stress for the patient, or you're missing potentially important diagnostic criteria," he added.
Theobald's group obtained data from a mobile Web platform called , developed by Haynes and classmate , an MD/MBA candidate.
Osmosis consists of user-generated question banks from 14,000 participants. After answering a question, a user selects one of three confidence ratings: "I'm sure," "Feeling lucky," or "No clue."
Researchers analyzed the responses of 1,021 users (617 men and 404 women) who answered at least 50 questions. Gender was determined based on name, and ambiguous names were omitted.
Women's answers were accurate 61.4% of the time, compared with 60.3% of men's (P=0.040).
But women selected "I'm sure" significantly less often than men did (39.5% versus 44.4% of responses).
Women averaged a higher accuracy in responses rated both "I'm sure" and "Feeling lucky" (80.5% versus 78.3% and 53.5% versus 49.8%, respectively).
Results aligned with previous work in different settings, Haynes said. "The decision to split groups by gender ... followed after reading several papers that found similar gender differences with respect to confidence in different settings such as mock clinical scenarios," he added.
Potentially inaccurate name-sex pairings were a limitation, authors wrote.
An additional limitation was that researchers could not verify that respondents were medical students. Gaglani noted in an email that respondents had “.edu†email addresses and that the Osmosis team aimed all their marketing at medical students. The authors are “fairly confident the vast majority, if not all, of our user subjects were medical students,†he wrote.