Method: A random sample of 5,000 general internists was purchased from the American Medical Association’s Masterfile and physicians were sent an invitation to participate in an online survey. The survey contained five hypothetical vignettes assessing physician decision-making in different clinical contexts. Each hypothetical vignette varied by one framing factor in a binary fashion: (1) risk of cervical cancer provided as frequency or percentage, (2) risks and benefits of cancer prophylactic medications provided as losses or gains, (3) prophylactic mastectomy presented as increase in survival or decrease in mortality, (4) intoxicated patient presented as male or female, and (5) cost information of lumbar screenings provided or not. Participants were asked to make recommendations (i.e., about frequency of screening, medication type, prophylactic mastectomy, hospital admission and discharge, and type of screening). 159 physicians completed the survey.
Result: 1) When the predicted probability of developing invasive cervical cancer was presented in frequency, a statistically significant higher proportion of physicians (79.3%) recommended a shorter period for the next Pap test (in 1-2 years), compared to scenario when the predicted probability was presented in percentage (62.5% physicians recommended the next Pap test in 1-2 years) (p=.019). (2) A higher proportion of physicians preferred Raloxifene to Tamoxifen when it was presented as a double-gain (41.7%) than when it was presented as a double loss (26.6%) (p=.017). (3) When an increase in mortality rate instead of a decrease in survival rate for the same magnitude was employed in the question, physicians were more likely to recommend contralateral prophylactic mastectomy (63.7% versus 45.8%, respectively) (p=.027). (4 & 5) We did not find any statistically significant difference in physicians’ recommendations or decisions when costs were provided and when gender-context was presented.
Conclusion: Physicians, including those who often keep up with training and education, are susceptible to framing effects in clinical judgment and decision-making. These findings have multiple implications for clinical decision-making, continuing training for physicians, physician-patient communication, ethical issues, societal cost reduction, and scientific reports. We argued that this bias susceptibility can be managed, or can be utilized to architect ethical-based or scientific-based preferences and choices for physicians.
See more of: The 35th Annual Meeting of the Society for Medical Decision Making