How Risk Statistics Are Presented May Mislead

How statistical results are presented may mislead patients and clinicians, a new study has found. (Image: Alex Slobodkin/iStockphoto.com)

The way in which many clinical studies present risk may mislead clinicians and patients about the magnitude of a treatment’s benefit and hamper their ability to make informed decisions, according to a Cochrane systematic review published today.

The review from the Cochrane Collaboration, a group that analyzes and synthesizes findings from medical research, examined the results of 35 studies that assessed how health consumers, health care professionals, or both perceived the benefits of a treatment depending on how the statistical results were presented.

Consider, for example, a hypothetical study in which 5 hip fractures occurred among 100 study participants who took an osteoporosis drug for 3 years and 10 hip fractures occurred among 100 study participants who did not take the drug. In this study, one could say that this drug cut the risk of developing a hip fracture by 50%; such an assumption would be based on the description of the relative risk of experiencing a fracture while taking the drug compared with the risk of having a fracture without treatment. But when the same data are presented in the form of absolute risk, the absolute reduction of risk during the 3 years of the study would be 5% (the difference between 10% of the untreated group having a hip fracture and 5% of the group taking the osteoporosis drug and still having a hip fracture). Or the results could be presented as the number needed to treat, which in this example would translate to needing to treat 20 patients with this drug for 3 years to prevent 1 hip fracture.

The researchers found that both health professionals and consumers were more likely to view the benefits of a treatment favorably if the results are presented as a relative risk reduction rather than as an absolute risk reduction or as the number needed to treat. The Cochrane review found that such a misperception of risk may alter decision making among these groups; however, the researchers did not examine whether study participants acted on this type of misperception.

Study author Holger Schünemann, MD, PhD, chair of the department of clinical epidemiology and biostatistics at McMaster University in Hamilton, Ontario, Canada, and his colleagues concluded that “there are strong logical arguments for not reporting relative values alone, as they do not allow a fair comparison of benefits and harms as absolute values do.”



Categories: Evidence-Based Medicine, Journalology/Peer Review/Authorship, Patient Education/Health Literacy, Statistics and Research Methods