A Study of Physicians' Perception of Advertising Judged Deceptive By the Fda

Richard L. Oliver, Washington University
R. Hoyt Walbridge, Institute for Survey research, Temple University
Peter H. Rheinstein, Office of Drugs, Food and Drug Administration
ABSTRACT - A fictitious medical journal was used as a medium to determine if a prescription drug advertisement judged deceptive by the FDA achieved that alleged effect. One hundred fifty physicians were exposed to the misleading ad and an FDA approved non-misleading version in a controlled field experiment. Interview results showed that physicians were inclined to perceive more efficacious drug claims in the misleading treatment and that this effect increased for the more inaccurate claims. Implications for the advertising and medical communities and for FDA's regulatory activities were discussed.
[ to cite ]:
Richard L. Oliver, R. Hoyt Walbridge, and Peter H. Rheinstein (1984) ,"A Study of Physicians' Perception of Advertising Judged Deceptive By the Fda", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 224-228.

Advances in Consumer Research Volume 11, 1984      Pages 224-228

A STUDY OF PHYSICIANS' PERCEPTION OF ADVERTISING JUDGED DECEPTIVE BY THE FDA

Richard L. Oliver, Washington University

R. Hoyt Walbridge, Institute for Survey research, Temple University

Peter H. Rheinstein, Office of Drugs, Food and Drug Administration

[This study was supported by Food and Drug Administration Contract #223-75-3017 and was conducted by the Institute for Survey Research, Temple University. The authors wish to thank Paul T. Hugstad, Leonard A. LoSciuto, Michael B. Mazis, and Richard B. Vanderveer for their conceptual and methodological inputs.]

ABSTRACT -

A fictitious medical journal was used as a medium to determine if a prescription drug advertisement judged deceptive by the FDA achieved that alleged effect. One hundred fifty physicians were exposed to the misleading ad and an FDA approved non-misleading version in a controlled field experiment. Interview results showed that physicians were inclined to perceive more efficacious drug claims in the misleading treatment and that this effect increased for the more inaccurate claims. Implications for the advertising and medical communities and for FDA's regulatory activities were discussed.

Background

The essence of deception in advertising continues to be an unresolved point of contention among government and industry spokesmen. Until recently, writers and other interested parties sought to define deception as that which exists when consumers (or some "n%" of consumers) respond with impressions or beliefs which are not true, not justified, or potentially misleading (e.g., Gardner 1975; Jacoby and Small 1975). The higher level issue of whether inaccurate impressions have a detrimental impact on buying behavior was frequently regarded as axiomatic by the FTC and other governmental agencies (Preston 1977), despite the fact that industry prefers to view misdirected purchasing behavior as an ultimate test of deception (Dillon 1973). In any prevailing political environment, however, definitions used by the regulatory agencies are those which the advertising to unity is obligated to observe.

In a more current interpretation of deceptive advertising, Russo, Metcalf, and Stephens (1981) critiqued three criteria of misleadingness. The first of these, the widely accepted 'In% rule" noted above, was deemed insufficient because it is unable to discriminate between false belief caused by advertising exposure and that caused by other communications and preexisting misconceptions. The authors did not consider, however, the possibility that this criterion may be acceptable for first exposure to a product (e.R.s an innovation).

A second criterion, increased false belief after exposure, was judged to be a sufficient but a not necessary condition because it fails to detect "non-incremental" forms of misleadingness. This latter case could exist if an ad exploited preexisting beliefs (e.g., Brand X granola is a natural food) which would be expected to strengthen a related false belief (e.g., Brand X granola is a nutritious food). In this example, the latter belief is secondary to the implied claim and not directly addressed.

The third criterion and the one preferred by Russo et al. (1981) is the degree to which fa se belief is decreased after the offending material in the ad has been purged or corrected. This solution, originally proposed by Jacoby and Small (1975), demonstrates only the impact of the misleading ad component by controlling for all other preexisting and extraneous influences. The present study illustrates an application of this approach.

The FDA's Interpretation of Deception

The FDA, in its regulatory capacity, continues to monitor advertising and other promotion to a select subpopulation, that of physicians. Unlike consumer advertising, pharmaceutical promotions of ethical drugs must contain technical information (labeling) about he drug's approved uses (indications), contraindications, adverse reactions, warnings, and precautions. Promotional copy must be consistent with the technical labeling and cannot contain unapproved claims of effectiveness. The mere suggestion o' a claim at variance With FDA's labeling is deemed false and misleading in violation of the Federal Food, Drug, and Cosmetic Act as amended in 1962. Note that this legal definition is more subjective and restrictive than Russo et al. (1981) would prefer.

Nonetheless, industry copywriters and art directors agree that pharmaceutical advertising is no less creative or sophisticated as a result of the narrow limits placed on promotional claims by the FDA (O'Connor 1978). From time to time, however, this creativity may extend beyond the limits of what the FDA has approved as an accurate statement of a drug's benefits and risks. The present study was designed to determine physician response to advertising claims which the FDA had judged to be deceptive.

In a beginning effort to augment the decision process with objective ,inputs, a strategy being used with greater frequency by the FTC (Brandt and Preston 1977; Preston 1976), the FDA selected a violative promotional piece for testing. The advertising in question promoted a large pharmaceutical company's minor tranquilizer product as partial treatment for cardiovascular problems, gastrointestinal disorders, and hyperthyroidism. None of these indications had been approved by the FDA. This drug was approved only for anxiety and anxiety accompanying chronic medical illnesses such as those referred to in the ad copy.

METHOD

A sample of physicians, stratified by listed specialty (general/family practice, internal medicine), was selected from the telephone directories of five large SMSAs (Atlanta, Boston, Chicago, Los Angeles, and Philadelphia) using a systematic random sampling plan designed for the individual sites. Each subject was sent an advance letter requesting participation and later was contacted by an interviewer for an appointment. A $30 fee was offered as incentive. Two hundred thirty physician contacts were required before the desired sample size of 150 was attained, a 65% participation rate. The average respondent was 52 years of age and had been in practice for 25 years.

Procedure

Personal interviews with participating physicians were conducted on a "double blind" basis by a professional research institute so that neither the interviewer nor the physician was aware of the study's sponsor. The subjects were told only that the study had been undertaken to determine how doctors evaluate drug information as part of an effort to make drug advertising maximally informative. During the course of the interview, each physician was asked to read through the contents of a mock journal containing five articles and four prescription drug advertisements selected by the FDA. The respondent was then asked a number of questions pertaining to the journal and general readership behavior. The content of one of the ads was the treatment as described in the following discussion.

Treatment

The promotional message in question was a four color, four page advertisement with bold copy on the first page, graphics and descriptive copy on the next two, and technical labeling on the last. The information judged to convey false impressions consisted of the bold copy and the first of the two pages of graphics. The message in hold copy was a quotation from the scientific literature suggesting that anxiety may precipitate or exacerbate chronic medical illness. This statement, when taken alone. was not necessarily false or misleading.

The second page contained printed and graphic material selectively extracted from writings on coronary heart disease. colitis, and thyrotoxicosis. This material was presented in such a way that, in the FDA's opinion, it capitalized on the non-deceptive bold copy to convey the impression that drug-induced changes in anxiety levels would partially treat or reduce the severity of each of the three medical conditions. Moreover, the FDA believed that the copy further suggested that data existed to show that a reduction in anxiety through the use of the drug in question would reduce the incidence (morbidity) of these illnesses with a resulting decrease in mortality rates. In short, the FDA objected to the implication that, if A is thought to be one cause of B, then the absence of A will result in a prior or subsequent absence of 8.

To determine if physicians also perceived this same fallacious implication, the FDA constructed a nonmisleading version of the ad for comparison purposes by eliminating the entire second page. In addition, the first page of bold copy was changed to read simply:

"Anxiety may accompany many chronic medical illnesses."

One of the two versions of the ad was presented to each physician on a random basis, balanced across the two physician specialties in a 2 x 2 design. Subjects were allowed to read the mock Journal at their own pace, after which tine they were asked a number of questions pertaining to content recall. Of particular concern to the FDA were four items measuring uses (indications) for the product:

- Is useful for treating anxiety.

- Is useful in the treatment of anxiety accompanying. chronic medical illness.

- Is useful in treating chronic medical illnesses, in addition to accompanying anxiety.

- Will decrease morbidity and/or mortality of chronic medical illnesses.

The first two of the four claims were approved by the FDA, while the regaining two, if believed by physicians, would represent false impressions on the part of the subjects. The FDA believed that the last claim was the more deceptive of the two unapproved statements.

The measure of deception used here reflected historical precedent in the federal regulatory agencies rather than true beliefs, per se. Early judgmental agency approaches were based on a concept the FTC called the "doctrine of tendency to deceive" (Kinter 1971). Under this doctrine, actual deception need not be found if the "capacity" or "tendency" to deceive can be demonstrated. The FDA later 0 adopted a working definition based on the "n%" rule 2 whereby deception was deemed to exist when a representative group of physicians had a "common impression or belief" which is incorrect or not justified (Jacoby and Small 1975). Both definitions allow subjective interpretation in lieu of more objective evidence.

Thus, perceptions of deception as opposed to internalized inaccurate beliefs were used as criteria in a manner similar to that used by Shimp (1978) and others because it was felt that readers' judgments of the implied content in an ad were sufficient to demonstrate deceptive potential, particularly in the case of advertising for an established product. In addition, the higher level criterion of impression reduction as a result of corrective copy was available (the third criterion of Russo et al. 1981), thus satisfying the control conditions for causality with comprehension as the criterion. To this end, subjects were asked to specify in 5-point Likert scale (agree-disagree) format the degree to which each of the indications was implied by the misleading or the non-misleading version of the ad.

Analysis

Responses to each of the four statements were analyzed separately in a 2 X 2 ANOVA design as a function of the copy treatment (accurate, misleading) and physician specialty (general practice, internal medicine), and then as a single multivariate response using MANOVA. To control for the possibility that prior familiarity with the drug in question affected the responses, subjects were asked to estimate the number of prescriptions they had written for the specific product in the prior month. This measure was entered as a covariate in the univariate and multivariate ANOVAs.

FINDINGS

Response frequencies for the four claims in the accurate and misleading versions of the test ad are shown in Table 1. As would be expected, the more misleading the claim, the greater the degree to which subjects disagreed with the suggestion of implied deception. However, it is also apparent that a substantially greater proportion (z = 4.01, p < .01) of physicians agreed or strongly agreed that a decrease in morbidity was implied by the deceptive version of the ad. The results were in the expected direction for the lesser of the two deceptive claims but this difference did not attain significance. An analysis of the total distribution of responses did show significant effects, however, as will be discussed. It is also worthy of note that, with the exception of the most general claim, the misleading ad generated greater agreement with product claims generally, an observation which will be tested in the multivariate analysis

Table 2 shows the cell mean scores for the treatment by physician specialty design over the four claims. The data support the observation from Table 1 that physicians are less likely to perceive the more misleading of the two implied deceptions, but the data also show that the ad deemed misleading resulted in higher agreement in seven of eight comparisons. There was also a slight tendency for general practitioners to display greater agreement with the suggestion of implied deception, a finding which may be explained by more favorable attitudes resulting from prior experience or a general tendency to use journal advertising as an information source. This latter interpretation will be explored in the discussion which follows.

TABLE 1

LIKERT-SCALE RESULTS FOR ACCURATE AND MISLEADING AD TREATMENTS

TABLE 2

TREATMENT BY SPECIALTY CELL MEAN FINDINGS

TABLE 3

UNIVARIATE AND MULTIVARIATE ANOVA AND ANCOVA RESULTS

Main effect and interaction univariate F ratios for the four claims taken separately are shown in Table 3 with and without previous prescribing behavior as a covariate. A number of observations are apparent. First, the treatment effect was significant for the two claims deemed misleading; the more misleading the claim, the greater the effect. Second, differences in specialty were evident only for the more misleading of the two implied deceptions. Third, only one claim produced a barely significant interaction effect. Fourth, a MANOVA of all claims taken simultaneously using an exact F test on Wilks' Lamda confirmed the univariate findings. Finally, the addition of previous prescribing behavior as a covariate in all analyses produced only modest changes in the findings. Generally, the covariate reduced the impact of the specialty effect as general practitioners tend to see more patients than do internists.

The most interesting finding, that the ad deemed deceptive was perceived as conveying such an impression to a greater degree than FDA's version, is suggested by the percent of agreement figures in Table 1 and by the univariate and multivariate results of Table 3. It appears that the violative ad generated greater agreement with both unapproved claims and with all claims taken simultaneously. While agreement declined as the nature of the claim became more severe, a greater proportion of the physician sample perceived misleading representations in the violative as opposed to the non-violative ad.

DISCUSSION

A number of policy implications of significance to FDA in the regulation of drug promotion emerged from this study. First, it was demonstrated that subjective judgment can be verified and augmented with empirical research. Prior to this approach, industry and government traditionally "negotiated" the meaning and intent of an advertisement. When industry acceded to the FDA's interpretation, the outcome more often resulted from "jawboning" than from a realization on a company's part that an advertisement may be potentially misleading.

Second, an initial estimate of the "n% rule" proposed by Jacoby and Small (1975) was suggested by the results. Because of the cautions noted by Russo et al. (1981), we are inclined to view this benchmark not as an absolute value but rather as a simple estimate of the difference between the percentage agreement in the deceptive treatment and that in the approved version of the ad. In the present case, the more deceptive claim resulted in a difference of 30 percentage points, significant beyond the .01 level. The FDA views this as significant in its own right as this proportion suggests that a conservative estimate of 50,000 practicing physicians on a national basis may be mislead by promotional copy of the nature tested here. It should also be noted that the lesser of the two claims deemed misleading resulted in a 13% difference. Although this latter figure did not attain significance, it does support the direction of effect observed with the morbidity claim.

Third, the criterion of less misleading false beliefs for corrected than for original advertisements proposed by Russo et al. (1981) was "validated" using a professional population and specialty advertising. Moreover, their notion of exploiting secondary beliefs (exploitive misleadingness) was also supported in that the MANOVA suggested that the misleading version of the ad may have a "halo" effect on both approved and unapproved claims. With the exception of the most general claim, mean scale scores for all claims were higher in the case of the misleading treatment. This was true even for the second approved claim where the difference attained significance in the covariance analysis.

Fourth, the data suggest that some subpopulation segments may be in greater agreement with unapproved implied claims in drug advertising. In the present study, general practitioners displayed this tendency when compared to internists for the more severe of the two unapproved claims. It Is generally accepted that general practitioners have heavier caseloads and see a broader cross-section of complaints than internists. For this reason, it is thought that they are unable to devote as much time to the reading of technical articles and may, as a result, rely more heavily on advertising as an information source (Herman and Rodowskas 1976). Unfortunately, this interpretation is somewhat clouded by the results obtained with the first of the unapproved claims. In this instance, internists were in greater agreement with the FDA's interpretation of deception in the misleading ad, thus resulting in the barely significant interaction effect observed in Table 3. These observations hopefully will alert readers to the possibility of differing magnitudes of effect across population subsegments.

It should be noted in stressing the limitations of the study that only the potential for deception was measured here. It was assumed by the FDA that recognition of an implied claim was sufficient to establish this potential. Physicians were not asked if the implied claims were deceptive nor were they asked the degree to which they believed these claims. To this extent, the FDA did not establish deception or the ability to recognize deception. In its capacity as a regulatory agency, the FDA has chosen to approach this sensitive issue in stages, allowing industry to respond to the methodology and conclusions. It is expected that future research efforts will employ more direct measures of the misleading potential of advertising and its effects on prescribing behavior.

In conclusion, a methodology was proposed and used to establish the degree to which the FDA's interpretation of a labeling violation was justified. The findings showed that a substantial proportion of a practicing physician sample perceived implied claims in the selected test ad which were not approved. Moreover, the differences between the agreement scores over the approved and unapproved ad versions were significant when viewed univariately and multivariately. Thus, the FDA's determination that the misleading advertisement was communicating information which had been designated a priori as deceptive was clearly supported. Because of this outcome, the FDA was satisfied that a standardized protocol for testing future cases of misleadingness had been successfully constructed. Future modifications would most certainly involve the use of actual beliefs as a criterion and a "no ad" control group for assessing the second criterion of Russo et al. (1981). In any event, FDA perceives this effort as merely an initial step in determining the deceptive potential of advertising practice in the prescription drug industry and urges individual industry members to pursue their own interpretation of misleadingness so that alternative methodologies can be compared.

REFERENCES

Brandt, Michael T. and Ivan L. Preston (1977), "The Federal Trade Commission's Use of Evidence to Determine Deception," Journal of Marketing, 41 (January), 54 62.

Dillon, Tom (1973), "What is Deceptive Advertising?" Journal of Advertising Research, 13 (October), 9-12.

Gardner, David M. (1975), "Deception in Advertising: A Conceptual Approach," Journal of Marketing, 39 (January), 40-46.

Herman, Colman M. and Christopher A. Rodowskas, Jr. (1976), "Communicating Drug Information to Physicians," Journal of Medical Education, 51 (March), 189-196.

Jacoby, Jacob and Constance Small (1975), "The FDA Approach to Defining Misleading Advertising," Journal of Marketing, 39 (October), 65-73.

Kinter, Earl W. (1971), A Primer on the law of Deceptive Practices, New York: Macmillan.

O'Connor, John J. (1978), "Admen Find Bold Art, Plain Talk Keys to Effective Drug Ads," Advertising Age, (February 13), 27-28.

Preston, Ivan L. (1976), "A Comment on 'Defining Misleading Advertising' and 'Deception in Advertising'," Journal of Marketing, 40 (July), 54-60.

Preston, Ivan L. (1977), "The FTC's Handling of Puffery and Other Selling Claims Made 'By Implication'," Journal of Business Research, 5 (June), 155-181.

Russo, J. Edward, Barbara L. Metcalf, and Debra Stephens (1981), "Identifying Misleading Advertising," Journal of Consumer Research, 8 (September), 119-131.

Shimp, Terence A. (1978), "Do Incomplete Comparisons Mislead?" Journal of Advertising Research, 18 (December), 21-27.

----------------------------------------