Evaluating the Influence of a Risk Calculator on Physician Risk Perception and Decision-Making in IPMN Surveillance: A Randomized Trial.

backgroundRisk calculators (RCs) support clinicians estimating the likelihood that a pancreatic intraductal papillary mucinous neoplasm (IPMN) would progress so that surveillance might be discontinued for low-risk lesions. We tested the effect of an RC on clinicians' judgment and decision-making and identified their cancer risk threshold for changing their decision. PATIENTS AND

methodsWe presented clinicians with three vignettes (V1, V2, and V3) of patients with low-risk IPMNs and asked them to assess the likelihood that the IPMN would progress to develop high-risk features and whether they would recommend continuing surveillance imaging. Clinicians were randomly assigned to the clinical vignettes alone (n = 35) or supplemented by data from the Dutch-American Risk Stratification Tool (DART-1 RC: n = 37). We compared clinicians' judgments and decisions between groups and assessed their cancer risk threshold (level of risk at which recommendation would change).

resultsAcross all vignettes, the RC resulted in no change in clinicians' judged likelihood of IPMN progression (V1 8.49 vs. 8.41%, p = 0.09; V2 4.39 vs. 6.75%, p = 0.99; V3 13.61 vs. 13.29%, p = 0.27) or recommendation to continue surveillance (V1 57 vs. 41%, p = 0.78; V2 41 vs. 59%, p = 0.55; V3 66 vs. 34%, p = 0.31). Clinicians varied in their reported risk threshold (V1 9%, interquartile range [IQR] 2, 13%; V2 9% [IQR 1, 15%], V3 8% [IQR 3, 20%]).

conclusionsAn RC did not significantly influence clinicians' risk perception or decision to continue surveillance, although the study was limited by low sample size. The cancer risk threshold at which clinicians would change their recommendation varies widely. Future work is needed to understand why RCs do not appear to alter decision-making.