How age,outcome severity,and scale influence general medicine clinic patients’ interpretations of verbal probability terms

Objective: To assess whether the type of scale used (scaling effects) and the severity of outcome (outcome severity) influence patients’ numerical interpretations of verbal probability expressions. Design: Cross-sectional survey of patients in a general medicine clinic. Setting: A university-based Department of Veterans Affairs Medical Center. Participants: 210 patients seen consecutively in a general medicine clinic. Measurements and results: The patients were randomized to scale and health outcome (complications of surgery). Two scales (a long form and a short form ) were used to expressly allow patients to choose probabilities less than 1%. The long form had a lower bound of “<1 out of 1,000,000”; the short form had a lower bound of “<1 out of 1,000.” Two complications were used: “death from anesthesia” and “severe pneumonia.” In the context of being told that their surgeon believed that the chance the complication would occur was “rare,” patients were asked to give the numerical estimate of that chance. The values elicited on both scales were significantly different for the two outcomes, with the “rare” risk of death from anesthesia being characterized as less likely than the “rare” risk of severe pneumonia (F=5.24, p=0.023). Linear regression and three-factor analysis of variance showed significant differences in the probabilities elicited for scale, outcome, and age, with older patients generally responding with higher probabilities than did younger patients. Conclusions: These findings suggest that the severity of the associated outcome and the scale used to elicit patients’ numerical estimates of verbal probability expressions influence patients’ quantitative interpretations of the verbal probability statement; and older patients respond with higher probabilities of negative outcomes than do younger patients. Future studies must continue to explore whether verbal probability expressions are adequate for communicating medical risk to patients or whether patients should be provided with numerical estimates of frequency. Supported in part by the National Science Foundation under contract SES-9020984 with Carnegie Mellon University.