Statistics for Communicating Risk

August 28, 2009 at 2:47 pm | In Prescribing Extra - Other |

PLoS Medicine has published the results of study that aimed to assess the effect of presenting statistics to patients in different ways when making decisions about taking statin medication to reduce cardiovascular risk.

The study recruited 2,978 eligible participants who were given access to online information that presented benefit data relating to statin medication in one of six different ways. The data were presented as relative risk reduction (RRR), absolute risk reduction (ARR), number needed to treat (NNT), event rate (ER), tablets needed to take (TNT) or natural frequency (NF).

The RRR data resulted in a greater number of participants indicating that they would accept treatment when compared to the five methods of absolute rate reporting. Of the five different statistical methods of presenting data in an absolute form, natural frequency appears to be preferable based on the participant preference, understanding, satisfaction and confidence in the data.

The authors note that this study "supports the contention that RRR is a more persuasive summary statistic. It is also noted that the study aimed to recruit 4,500 participants (750 to 800 participants for each intervention arm). The sample size requirement was therefore not met but the authors assert that the randomisation protocol worked well. In addition it is unclear how these results might apply in other situations.

Action: Clinicians should already be aware of the different methods of presenting data to patients to aid an informed decision. This study may help to identify which methods are likely to be best understood and be most trusted.