134 Cancer Research UK (CRUK) website users were recruited via a ‘pop-up’. Using a 2 × 2 factorial design, participants were randomly allocated to one of four conditions and asked to: imagine they had to take tamoxifen, estimate the risks of 4 side effects, and indicate a presentation mode preference.
Those presented with absolute frequencies demonstrated greater accuracy in estimating 2 of 4 side effects, and of any side effect occurring, than those presented with frequency bands. Those presented with combined descriptors were more accurate at estimating the risk of pulmonary embolism than those presented with numeric descriptors only.
Absolute frequencies outperform frequency bands when presenting side effect risk information. However, presenting such exact frequencies for every side effect may be much less digestible than all side effects listed under 5 frequency bands. Combined numerical and verbal descriptors may be better than numeric only descriptors when describing infrequent side effects.
Information about side effects should be presented in ways that patients prefer, and which result in most accurate risk estimates.