In the USA the Kefauver-Harris Amendment of 1962 required that prescription drug advertisements contained a fair balance of both negative and positive information. This resulted in the inclusion of a ‘brief summary’ of side-effects, warnings and contraindications. For a good history of the regulation of DTCA in the US, see Wilkes et al.207. There is a surprisingly large amount of research on the possible effect of the ‘fair balance’ requirement, of which the high quality work was done by Louis Morris from the FDA208. Some of the research on this topic looks at advertisements for consumers, some at advertisements for doctors.
In three articles, Morris and different co-authors look at ways of conveying risk information to consumers in DTCA. This type of analysis is important for working out how best to regulate DTCA, what kind of risk information should be presented and how. These papers describe aspects of a study that experimented with different forms of presenting risk information in mock DTCA advertisements for two fictitious prescription drugs. Morris et al.141 focused on TV advertisements in which the amount of information, the content (either general or specific), the format (video, audio) and the placement of information about risks were varied. Six hundred and seventy six people, (of whom 50% had one or more of the conditions that the drugs treated), viewed the advertisements. They found that more risk messages were recalled if more were included in the advertisement, if these were more specific, if they were both written and spoken, and if they were spread out throughout the advertisement. When people remembered high levels of risk information they tended to have less knowledge and awareness of the benefits of the medicines. Morris et al.142 outlined results from both the TV advertisements and magazine advertisements. Presenting the entire patient package insert in magazine advertisements led to lower knowledge of risks than advertisements that included the information in other forms. Morris et al.208 also described the results related to both TV and magazine advertisements, and looked at differences among the sample. For example, older people were more positive about both the medicine and the advertisement.
Similarly, but in a much smaller and less rigorous study, Larson et al.209 tested a small number of doctors’ assessments of the believability of mock advertisements which included or did not include brief summaries. They also looked at whether the country in which the advertisement was found (USA or Mexico) affected believability, and found that it did not. Although the inclusion of brief summaries led to statistically significant increases in believability, this was relatively unimportant in explaining the variance in believability. They use attribution theory to explain their results. Larson and Smith209 appear to describe the same study. Hurd211 described a small study where pharmacy students were asked to rate the credibility and honesty of a sales representative, and whether they would purchase the product, in a series of hypothetical situations. Including negative information about the product did not affect their preference for the product, but they rated the sales representative’s honesty more highly.
Cady and Larson212 attempted to find out whether the inclusion of a brief summary made ads more believable. They used advertisements for two mock products and asked students to rate their believability with and without the brief summary. The major claim in one advertisement was more believable in the version with the brief summary, but the results were somewhat equivocal overall. Schommer et al.213 assessed what patients had learnt after viewing an DTCA advertisement which included both positive and negative information about an antihistamine. The study used patients waiting for appointments at a clinic, and only 34% of those approached agreed to participate. The participants were confused by the claim that the antihistamine gave ‘non-drowsy’ relief, and the statement that about one in a hundred people experience drowsiness. The authors suggest that it may be confusing to have both risk and benefit information in the same advertisement. (However these claims do seem to be contradictory and the viewers’ confusion seems entirely appropriate). Tucker and Smith214 asked 192 people in a shopping mall to view four different versions of an advertisement for a fictional influenza vaccine. Those which included risk information were rated as more ‘appealing’ than those without, but those with no warning or a general warning (all medicines have side-effects, please consult your doctor), were rated as providing more ‘security’.
It is hard to interpret the practical significance of these studies. While Morris et al. usefully explore different formats for conveying risk information, many of the other studies simply compare advertisements with and without risk information.
CONCLUSION: Communicating risk information to achieve “fair balance” requires great care and testing.