EXPLAINING RISK REDUCTIONS TO PATIENTS: DO PHYSICIANS USE NUMBERS?

Purpose: Traditional measures of benefit from risk-reducing drug therapies are absolute risk reduction (ARR), relative risk reduction (RRR) and number needed to treat (NNT).  We explored whether physicians use such numbers in their encounters with patients and how they perceive the benefit of statin therapy for high-risk patients.

Method: General practitioners (n=450) and internists (n=450) were mailed a questionnaire and asked whether, in clinical practice, they explain the benefits of risk-reducing drug therapies in qualitative terms, such as “great risk reduction” versus “small risk reduction”, numerical terms, such as ARR, RRR or NNT, or both. The physicians were asked to indicate their perception of benefit from statin therapy on a zero to ten Likert scale anchored at “everybody benefits” (zero) and “many patients must be treated for each patient that benefits” (ten), i.e. like a lottery. 

Result: We obtained responses from 428 physicians (48%). The proportion using qualitative terms only was 58%, compared to 3% for numerical terms only, and 34% for using both qualitative and numerical terms. The proportions using ARR, RRR and NNT were 31%, 27% and 22%, respectively. Multivariate logistic regression analysis indicated that male physicians were more likely to use numbers (OR 1.65, 95% CI 1.02-2.66).  Male sex was a predictor for using ARR (OR 1.94, CI 1.15-3.26), being an internist predicted using ARR (OR 1.86, CI 1.18-2.91) and RRR (OR 1.74, CI 1.09-2.77) whereas age was negatively associated with using NNT (OR 0.97 per year, CI 0.94-0.99). On the Likert scale the mean and median scores were 4.2 (SD 2.7) and 3.0, respectively, i.e. the majority tended to believe that most patients benefit from statin therapy. However, 28% scored 6 or higher, suggesting that they tended to perceive statin therapy as a lottery. Scores on the Likert scale were not associated with age, sex or specialty. Using NNT was associated with a high score on the Likert scale (OR per level 1.12, CI 1.03-1.22).   

Conclusion: Male physicians, younger physicians, internists and physicians who perceived statin therapy as a lottery were more likely to use numbers when explaining risk reductions to patients.  The associations were modest, however, and the majority of physicians seem to avoid using numbers in their encounters with patients.