This presentation is part of: Oral Concurrent Session A - Patient and Physician Behavior/Preferences 2
COMMUNICATING THE BENEFITS OF RISK REDUCING INTERVENTIONS – NUMBER NEEDED TO TREAT (NNT) OR POSTPONEMENT OF ADVERSE EVENTS? RANDOMIZED TRIAL OF LAY PEOPLES’ OPINIONS
Peder A. Halvorsen, MD, University of Tromsø, Norway, Institute of Community Medicine, 9516 ALTA, Norway, Ivar Sønbø Kristiansen, PhD, University of Southern Denmark, Odense, Institute of Public Health, The Research Unit of General Practice, N-0853 Oslo, Norway, and Randi Selmer, PhD, The Norwegian Institute of Public Health, Division of Epidemiology, Department of Chronic Diseases, N-0403 Oslo, Norway.
Purpose: To explore whether people respond differently to NNT than to postponement of adverse events when considering risk reducing drug therapies. Methods: A sample of attendees (n=1,754) to a general population health survey was mailed a questionnaire and asked about preferences for a hypothetical drug therapy to prevent heart attacks. Based on the 4S study, treatment effect after 5 years was presented in 3 formats: “For every heart attack that is avoided, 13 patients must take the therapy” (NNT) or “for all patients taking therapy, heart attack is postponed by 2 months” or “for one out of four patients taking therapy, heart attack is postponed by 8 months, while the others don’t benefit”. Respondents were randomly allocated to one of these formats. Another 1,000 attendees were asked about preferences for a drug therapy against hip fractures. Based on the FIT study, benefit from 5 years of drug therapy was presented either as ”for every hip fracture that is avoided, 57 patients must take the therapy” (NNT) or “for all patients taking therapy, hip fracture is postponed by 16 days” or “for three out of 100 patients who take therapy, hip fracture is postponed by 16 months, while the others don’t benefit”. Again, allocation to one of the effect formats was random. Results: The overall response rate was 81%. Presented with the heart attack scenario, 93% consented to the drug therapy in response to the NNT-format, 69% consented in response to “smaller postponement for everybody”, while 82% consented in response to “greater postponement for a fraction” (χ2=89.6, p<0.001). Corresponding figures for the hip fracture scenario were 74%, 34% and 56% respectively (χ2=91.5, p<0.001). In multivariate logistic regression models, additional predictors for consent to drug therapy were poor education, desire to know one’s risk status and perception of the risk reduction format as easy to understand. However, 40% of the respondents reported difficulties in understanding the effect format. Conclusion: Treatment effects presented in terms of NNTs yield higher consent rates than postponement of adverse events. The results may reflect preferences for substantial but uncertain benefits over certain, but small benefits, or simply a response to prospects of “complete prevention”, “great postponement” and “small postponement”, disregarding probabilities. Finally, the results may reflect difficulties in understanding the effect measures.
