Testimony to FDA
Testimony to FDA Open Hearings (DTC Part 15 Hearing)
November 1, 2005
My submission today is a response, in the spirit of linguistics and communication science, to the FDA’s repeated calls for hard research on the communicative workings of televised drug advertising. Among the public’s many concerns about televised drug advertising, the provision of risk information stands out. Just 1 consumer in 3 interviewed for Prevention Magazine’s 1998 survey thought that the risk information was satisfactory. How to get people to act on risk information is, of course, a communications equivalent of the Holy Grail. But when people don’t even believe they’re being adequately warned, you have a problem of an entirely different order.
The research of Louis Morris and others in the 80s on the formatting of ads – where to position the risk information, whether it should be visual or audio, how much of it there should be – was used by FDA (rightly or wrongly) to justify the type of ads we watch today.  However, the fair balance between benefit and risk information enshrined in FDA guidance has neither been rigorously operationalized nor empirically demonstrated, i.e. no practical procedure exists for judging fair balance, or for proving it. As a result, FDA has found itself reacting to events, not guiding them.  
            My submission today draws in part on collaboration with Jon Schommer, Professor of pharmaceutical care and health systems at the University of Minnesota, who cannot be here today. We are engaged in re-visiting the key issue of the formatting of TV ads and its communicative effect, which leads us to question some of the working assumptions in current drug advertising regulation.
In a recent paper, ‘Advertisement Format and the Provision of Risk Information about Prescription Drug Products (Research in Social and Administrative Pharmacy 1(2), pp. 185-210, 2005), we examined ways to improve the required risk messages in TV prescription drug advertisements.
We took two TV ads, one for a lower-risk and one for a higher-risk medication, and using male and female voice-over artists we produced two alternative ‘male’ and ‘female’ versions of the ads, with the risk information shifted to the end and supplemented by a risk message in captions.
        For the higher-risk medication, we found that placing the risk information at the end of the advertisement with added captions had the following effects:
  1. (1) It significantly improved the students’ short-term memory of general and specific side effect information, but without making the medication appear more risky;
  2. (2) It created a perception that the advertisement was more informative;
  3. (3) It reduced feelings of distraction;
  4. (4) It reduced feelings of information overload, of bewilderment or finding it hard to follow.
However, little effect was found with the advertisement for the lower risk medication – a medication being taken by many of the students – perhaps because they simply didn’t care to pay much attention to the risk messages. On the other hand, the female voice-over had considerably more effect than the male voice-over – intriguing, but again we cannot be sure why.
While our experiment was restricted to a group of 1st-year pharmacy students, most of them female, we believe that it points to ways of improving risk information in drug advertising to the general consumer. We hope to extend our work to a broader population. In addition, and of great importance in our view, there is an urgent need to study the response of the elderly, the semi-literate and other vulnerable groups. The elderly, for example, have been shown to have difficulty with rapid or disorganized texts and elaborate inferencing. Even at the best of times, TV ads, like any spoken message, cannot offer the viewer the same opportunity for scrutiny as printed advertisements.

            Empirical studies of communication, be they surveys or focus groups or experimental manipulations, are notoriously difficult and expensive to construct. Often, the setting is artificial. Nor do they easily yield robust or general conclusions.  If we are to conduct the kind of case-by-case and experimental studies that Morris & Pahl have pinpointed as a chief desideratum (in the major review they published in 1998), one of the most promising and practical avenues is linguistics and discourse analysis.
Linguistics studies linguistic structures – sounds, words, syntax and meaning. Discourse analysis opens the lens further, to look for patterns in whole stretches of text, and to see how they relate to the situation – to who the speakers and addressees are, how they relate socially, what they are doing culturally, and what other systems of communication are being used – visuals, body language and so on. Linguistics and discourse analysis are not – in and of themselves – a way of predicting quite how readers or listeners will understand a text. But they can call upon massive databases of verbal behavior, and on the analyst’s own well-honed intuitions, to anticipate what a text will ‘mean’.
Equally important, however, discourse analysis alerts us to a complex and often daunting web of meaning. Behind the disembodied semantics of a sentence is the ‘pragmatics’, the use we make of it in context, the associations in a phrase like ‘Ask your doctor’ ('ask' can mean 'inquire' but is there a cunning hint here of its other meaning, 'request'?)), tone of voice, suggestivities and strategic ambiguities, including all kind of innuendos for which we wouldn’t wish to be held strictly accountable. All of this is part of language, and when a trial judge assesses what something means to the ordinary man or woman something of this web of meaning will generally be taken into account. But not enough: it takes a trained analyst to disentangle such meaning -- and then to point the way for empirical investigation of key phenomena.
The two most challenging and sophisticated types of text for analysis of meaning are the poem and the ad. Theorists often portray advertising as being either informational or persuasive. However, discourse analysis sees the advertising data differently: Consumers should not expect to find any logic in an ad. Guy Cook (‘The Discourse of Advertising’, Oxford) has argued that an intrinsic playfulness and creativity in advertising is inherently ‘word play’ and extraordinary creativity, constantly seeking to surprise and inveigle and to reinvent the rules of the game. We now face this in a particularly potent form: the post-modern, ironic or retro ad, where the words are less and less important and the image says it all. The consumer plays along – to varying degrees.
What then of literalism, the popular belief that a sentence has a literal meaning quite independent from its implied meaning? This belief is inconsonant with the way ordinary people actually understand language – with far-reaching implication for the study of advertising. In the words of Dwight Bolinger, President of the Language Society of America: “The most insidious of all concepts of truth is that of literalness. Advertising capitalizes on the legal protection it affords.”
As an example of what linguistic and discourse analysis can bring to the evaluation of drug advertisements, I micro-analyzed the text and visuals of several televised Rx drug ads, using linguistic and discourse analytic procedures. (Glinert, L H.  TV commercials for prescription drugs:
A discourse analytic perspective. Research in Social and Administrative Pharmacy 1(2), pp. 185-210, 2005)
My research questions were: what the ad was seeking to do, how the wording and visual patterns related, and what messages (overt or implied) a viewer was likely to derive.
Here are the main findings:
(1) Functionally, the advertisements were a bewildering blend of promotion, information and entertainment – with many touches of postmodernist irony.
Thus, one celebrated campaign made a point of mixing science with science fiction in its ads. Or again, what does the viewer make of a finale, delivering a final punch but distorting some of the key medical information delivered elsewhere in the ad? Is the outcome a cognitive dissonance; or will viewers perhaps draw upon the entire message in reaching their conclusions and take this finale as merely a rhetorical flourish? And is this kind of finale is in breach of FDA Guidance on fair balance?
(2) Risk messages frequently competed with up-beat music and visuals. Having to process the warning caption “Your results may vary” together with an image of a beaming woman and a quick-fire string of superlatives, the average viewer may well not consider the possibility that, for many people, the effect may be less than complete cure.
            (3) There was an intense switching and fusing of styles – here, the cognitive and persuasive effect on those of non-American cultural backgrounds should be investigated.
(4) Strategic linguistic ambivalence was frequently used. Thus, running through testimonials for an asthma medication was a tension between absolute and relative claims of efficacy. Three of the testimonials were cautiously couched: 'gets out more', 'with fewer symptoms', ' doesn’t have to use his reserve inhaler as often ', 'more nights restful sleep', 'fewer asthma symptoms' -- but the fourth testimonial and the epilogue had an absolute ring: 'for day and for night control of asthma symptoms' can be construed as a very strong claim, maybe toned down by 'helps control' in the epilogue; but here again, ‘helps control asthma symptoms all day and all night’ has equally sweeping connotations.
            (5) There was much variation and ambivalence as to the source of authority and identity of addressee.  For example, the traditional advertising distinction between personal and impersonal – ‘company’ and ‘consumer’ – was sometimes ostentatiously flouted: characters in the story line sometimes unexpectedly took on the mantle of spokesperson for the product.
These results point to textual features that deserve experimental or other empirical study. I would stress that one of the central issues in analyzing a text in communications terms is the relative impact of its component elements. Among the variables to consider is the nature of text reception: Is it read or heard in toto? Linearly?  How often?
It is the nature of this kind of qualitative micro-analysis to focus on a limited group of texts and study them as organic wholes, while seeking points of comparison and contrast with advertising and linguistic practice in order to ground one’s categories. Undoubtedly, a large-scale study is a desideratum, using a team of analysts and a system of crosschecking, and also deploying corpus analytical techniques. The results could in turn be compared with survey and focus group studies of viewers’ responses to the same ads.
Complex though it is, discourse analysis holds out promise of providing a useful rapid evaluation of advertising copy requiring prior or post-facto approval. More generally, it can contribute to our general understanding of how TV advertisements convey meaning with respect to drug benefits and risks, with implications for advertisers, regulators and patient education.