Health Numeracy and Relative Risk Comprehension in Perioperative Patients and Physicians

Abstract

Helping patients to understand relative risks is challenging. In discussions with patients, physicians often use numbers to describe hazards, make comparisons, and establish relevance. Patients with a poor understanding of numbers-poor "health numeracy"-also have difficulty making decisions and coping with chronic conditions. Although the importance of "health literacy" in perioperative populations is recognized, health numeracy has not been well studied. Our aim was to compare understanding of numbers, risk, and risk modification between a patient population awaiting surgery under general anesthesia and attending physicians at the same center. We performed a single-center cross-sectional survey study to compare patients' and physicians' health numeracy. The study instrument was based on the Schwartz-Lipkus survey and included 3 simple health numeracy questions and 2 risk reduction questions in the anesthesiology domain. The survey was mailed to patients over the age of 18 scheduled for elective surgery under general anesthesia between June and September 2019, as well as attending physicians at the study center. Two hundred thirteen of 502 (42%) patient surveys sent and 268 of 506 (53%) physician surveys sent were returned. Median patient score was 4 of 5, but 32% had a score of ≤3. Patients significantly overestimated their total scores by an average of 0.5 points (estimated [mean ± standard deviation (SD)] = 4.3 ± 1.2 vs actual 3.8 ± 1.3; P < .001). Health numeracy was significantly associated with higher educational level (gamma = 0.351; P < .001) and higher-income level (gamma = 0.397; P < .001). Physicians' health numeracy was significantly higher than the patients' (median [interquartile range {IQR}] = 5 [4-5] vs 4 [3-5]; P < .001). There was no significant difference between physicians' self-estimated and actual total numeracy score (mean ± SD = 4.8 ± 0.6 vs 4.7 ± 0.6; P = .372). Simple health numeracy (questions 1-3) was predictive of correct risk reduction responses (questions 4, 5) for both patients (gamma = 0.586; P < .001) and physicians (gamma = 0.558; P = .006). Patients had poor health numeracy compared to physicians and tended to overrate their abilities. A small proportion of physicians also had poor numeracy. Poor health numeracy was associated with incomprehension of risk modification, suggesting that some patients may not understand treatment efficacy. These disparities suggest a need for further inquiry into how to improve patient comprehension of risk modification.