Background
For people taking medicines, information on side effects is a common concern¹. Current mandatory medicines information (i.e. the leaflet inside the pack) lists side effects without frequency data. A recent EC guideline² proposed that leaflets should, where possible, use five verbal descriptors relating to the following incidence rates:

very common >10%
common 1 � 10%
not common 0.1 � 1%
rare 0.01 � 0.1%
very rare <0.01%

There is no evidence that these rates are similar to lay people's understanding of the verbal terms. To test this, we compared people's estimate of side effect frequency when given incidence information, in either a verbal or numerical form.

Method
We undertook 3 linked studies with an experimental, between-groups design with counter- balancing. The sample comprised 312 adults, recruited in public places. Recruits were given written information describing a scenario of visiting a doctor and receiving a hypothetical medicine (Epidoxin). Recruits read that it is associated with two named side effects of a given frequency. Half were given verbal frequency information (e.g. this side effect is common) and half numerical information (it occurs in 5% people). Recruits were asked to estimate side effect frequency for people taking Epidoxin.

In Study 1, people were told about either mild or severe side effects, and asked to rate either the percentage of people who would have the side effect or the number out of 10,000. In Study 2 people were asked about either mild or severe effects and asked to make a frequency estimate between 0.001% and 100%. Study 3 examined only very common side effects (and their numerical equivalent). As well as estimating side effect frequency, recruits rated their attitudes to the information and medicine-taking.

Results
People given verbal descriptors significantly over-estimated the side effect frequency by, on average, factors of 5 to 400, with greatest over-estimation for rare and very rare side effects. In Study 3, people given verbal information were less satisfied with the information, thought the effects more severe, estimated a greater risk to health of the medicine and thought they and others would be less likely to take the medicine.

Conclusion
Understanding the risks associated with medicines is an important influence on people deciding whether or not to take them. The EC's recommended verbal frequency descriptors do not correspond with people's interpretation of them. Research is needed to see whether people taking real medicines show similar frequency over-estimation.

1. Raynor DK (1998). The influence of written information on patient knowledge and adherence to treatment. In L Myers & K Midence (Eds.) Adherence to treatment in medical conditions. Harwood.
2. European Commission, D-G III. A guideline on the packaging information of medical products for human use authorised by the community. Brussels: EC, 1997.