J O I N T C E N T E R
`
`AEI-BROOKINGS JOINT CENTER FOR REGULATORY STUDIES
`
`Direct-to-Consumer Advertising and the Demand for
`Cholesterol-Reducing Drugs
`
`John E. Calfee, Clifford Winston, and Randolph Stempski
`
`Related Publication 03-12
`
`June 2003
`
`The authors are John E. Calfee, American Enterprise Institute; Clifford Winston, AEI-
`Brookings Joint Center and the Brookings Institution; and Randolph Stempski, American
`Enterprise Institute. The authors gratefully acknowledge IMS Health and Scott-Levin for
`providing the data described in this paper.
`
`1
`
`ALL 2070
`MYLAN PHARMACEUTICALS V. ALLERGAN
`IPR2016-01128
`
`

`

`DIRECT-TO-CONSUMER ADVERTISING AND THE
`DEMAND FOR CHOLESTEROL-REDUCING DRUGS*
`
`JOHN E. CALFEE,
`American Enterprise Institute
`
`CLIFFORD WINSTON,
`Brookings Institution
`
`and
`
`RANDOLPH STEMPSKI
`American Enterprise Institute
`
`Abstract
`In August 1997, the Food and Drug Administration (FDA) reinterpreted its ad-
`vertising regulations to ease limits on the use of broadcast media when advertising
`prescription drugs directly to consumers. We estimate the effect of direct-to-consumer
`advertising on demand, using 1995–2000 data from the market for the statin class
`of cholesterol-reducing drugs. We find no statistically significant effect from any
`form of advertising and promotion on new statin prescriptions or renewals and no
`evidence of adverse market effects from advertising or the FDA policy change. We
`did find evidence, however, that television advertising increased the proportion of
`cholesterol patients who had been successfully treated, which suggests that advertising
`reinforces compliance with drug therapy.
`
`The subject of advertising is marked by diverse and conflicting perspec-
`
`I.
`
`Introduction
`
`tives. Popular writers and social critics, for example, often portray advertising
`as wasteful and manipulative, while some academic economists argue that
`advertising can provide useful information for consumers and lower prices.1
`
`* The authors gratefully acknowledge IMS Health and Scott-Levin for providing the data
`described in this paper.
`1 A publication by the Media Foundation, Adbusters, seeks to identify the manipulative
`aspects of advertising. Studies by economists that found that restrictions on advertising raised
`prices include Lee Benham, The Effect of Advertising on the Price of Eyeglasses, 15 J. Law
`& Econ. 337 (1972); and John E. Kwoka, Jr., Advertising and Price and Quality of Optometric
`Services, 74 Am. Econ. Rev. 211 (1984). Pauline M. Ippolito & Alan D. Mathios, Information,
`Advertising and Health Choices: A Study of the Cereal Market, 21 Rand J. Econ. 459 (1990),
`and Pauline M. Ippolito & Alan D. Mathios, Information, Policy, and the Sources of Fat and
`Cholesterol in the U.S. Diet, 13 J. Pub. Pol’y & Marketing 200 (1994), found that the con-
`sumption of high-fiber cereals increased and the consumption of fat and saturated fat decreased
`when manufacturers were allowed to advertise the health content of their products. An overview
`of the academic and popular debate over advertising is contained in John E. Calfee & James
`K. Glassman, Fear of Persuasion: A New Perspective on Advertising and Regulation (1997).
`
`[Journal of Law and Economics, vol. XLV (October 2002)]
`䉷 2002 by The University of Chicago. All rights reserved. 0022-2186/2002/4502-0015$01.50
`
`673
`
`2
`
`

`

`674
`
`the journal of law and economics
`
`As the nation’s health care costs continue to rise, it is not surprising that the
`pharmaceutical industry’s multibillion-dollar direct-to-consumer (DTC) ad-
`vertising expenditures are attracting their share of critics and defenders.
`Some physicians, such as Matthew Holland, complain that DTC advertising
`encourages patients to ask physicians to write inappropriate prescriptions.2
`Health care providers, such as managed care organizations, and health care
`payers, such as employers, charge that advertising is increasing health care
`costs. In fact, according to Reuters, several state legislatures are considering
`curbs on DTC advertising.3
`The nation’s guardian against false or deceptive advertising, the Federal
`Trade Commission, has taken issue with the critics, arguing in 1996 and
`2002 comments to the FDA that DTC advertising can be valuable for con-
`sumers. In 2002, the National Health Council, an organization of health
`associations (including the American Medical Association) and businesses,
`issued a consensus statement that concluded that “DTC advertising is an
`effective tool for educating consumers and patients about health conditions
`and possible treatments.”4
`Recent changes in federal policy toward prescription drug advertising have
`intensified the debate. In 1997, the Food and Drug Administration (FDA)
`sharply reduced restrictions on drug advertising originally imposed in 1985.
`The original DTC advertising regulations required print ads to include a
`detailed “brief summary” of risk and other information and required broadcast
`ads to include a “major statement” of risks, while also making “adequate
`provision” for viewers to obtain full FDA-approved prescribing information.
`Although it was not feasible for broadcast ads to meet these requirements,
`the FDA allowed ads of two kinds. The first could discuss an illness or
`condition, suggesting that consumers see a physician for treatment without
`mentioning a brand. The second could emphasize a pharmaceutical brand
`without stating what condition the drug could treat.
`For some years, FDA staff and others had expressed dissatisfaction with
`the disclosure requirements, partly because their research had strongly sug-
`gested that the requirements were of little benefit to patients.5 The FDA had
`also accelerated the pace of switching prescription drugs to over-the-counter
`(OTC) status, recognizing the greater role that consumers were taking in
`
`2 Matthew F. Hollan, Direct-to-Consumer Marketing of Prescription Drugs: Creating Con-
`sumer Demand, 281 JAMA 382 (1999).
`3 Karen Pallarito, States Target Direct-to-Consumer Drug Ads, Reuters Health, June 28, 2001.
`4 National Health Council, Direct-to-Consumer Prescription Drug Advertising (January 2002)
`(http://www.nationalhealthcouncil.org/advocacy/dtc.htm, accessed September 29, 2002); and
`National Health Council, Direct-to-Consumer Prescription Drug Advertising: Overview and
`Recommendations (January 2002) (http://www.nationalhealthcouncil.org/advocacy/DTC_paper
`.pdf, accessed June 30, 2002).
`5 Louis A. Morris & Lloyd G. Millstein, Drug Advertising to Consumers: Effects of Formats
`for Magazine and Television Advertisements, 39 Food Drug & Cosmetic L. J. 497 (1984).
`
`3
`
`

`

`direct-to-consumer advertising
`
`675
`
`their health care decisions.6 In August 1997, 6 months after the departure of
`FDA Commissioner David Kessler, an opponent of DTC advertising, the
`FDA greatly eased the burden for broadcast ads, allowing them to achieve
`“adequate balance” by including a concise summary of risks and related
`information (often via voiceover), specifying sources for more complete in-
`formation (for example, a toll-free number, an Internet Web site address, and
`either concurrent print ads or specified locations such as pharmacies), and
`stating that information is available from all physicians and pharmacists.7
`According to studies by Wayne Pines and Chris Adams, DTC advertising,
`especially on television, accelerated rapidly from $579 million in 1996 to
`$2.6 billion in 2000.8
`The FDA’s 1997 action that eased limits on the most powerful form of
`consumer advertising might be expected to increase the effect of DTC ad-
`vertising (and more generally promotion) on prescription drug demand, all
`else equal. We investigate this fundamental hypothesis as a step toward
`assessing the welfare effects of DTC advertising. We use market data to
`assess the effect of industry promotional activity, including DTC advertising,
`on the demand for an important class of drugs, the so-called statin drugs for
`reducing serum cholesterol. Surprisingly, we are unable to find any evidence
`that advertising has affected demand in the short run. Consumer behavior in
`this market appears to be influenced primarily by patients’ interactions with
`their doctors, the sequence of visits that must be made before a prescription
`is filled, and the growing dissemination of objective evidence that prescription
`drugs are effective in reducing cholesterol and preventing heart attacks. We
`do provide preliminary evidence that advertising reinforces these factors,
`while strengthening patient compliance with statin drug therapy. It is also
`possible that advertising may affect demand in the long run, but we have
`not been able to capture that phenomenon.
`
`II. The Statin Class of Cholesterol-Reducing Drugs
`Epidemiological evidence, such as data from the Framingham Heart Study,
`led researchers in the 1950s to hypothesize that higher levels of serum cho-
`lesterol increased the risk of coronary heart disease. Influential segments of
`the public health community, including the U.S. Department of Agriculture
`and some agencies of the U.S. Department of Health and Human Services,
`
`6 Food and Drug Administration, Center for Drug Evaluation and Research: From Test Tube
`to Patient: Improving Health through Human Drugs (1999).
`7 In August 1999, the FDA issued a final guidance for DTC advertising (Food and Drug
`Administration, Center for Drug Evaluation and Research, Guidance for Industry: Consumer-
`Directed Broadcast Advertisements (August 1999), 64 Fed. Reg. 43197 (1999), leaving re-
`quirements essentially unchanged from the August 1997 version.
`8 Wayne Pines, A History and Perspective on Direct-to-Consumer Promotion, 54 Food &
`Drug L. J. 495 (1999); and Chris Adams, FDA Plans to Review Policy Allowing Direct-to-
`Consumer Drug Ads for TV, Wall St. J., March 28, 2001, at B1.
`
`4
`
`

`

`676
`
`the journal of law and economics
`
`reached a consensus during the 1970s that reducing cholesterol, primarily
`through dietary changes, could substantially reduce the risk of heart disease.
`This limited consensus formed the basis for a number of federal policies,
`culminating in the 1985 National Cholesterol Education Program that urged
`physicians to counsel their patients to reduce dietary cholesterol or use
`cholesterol-reducing drugs to reduce the risk of heart disease.9
`As with many public health claims, the hypothesized link between cho-
`lesterol and heart disease had its detractors. Many observers such as Gary
`Taubes noted the lack of evidence from controlled clinical trials that dietary
`changes could substantially affect serum cholesterol or that changes in serum
`cholesterol would change the risk of heart attacks or death from coronary
`heart disease.10 And although several drugs were available by the 1970s to
`reduce serum cholesterol, all had serious side effects, and none had an es-
`tablished ability to reduce the risk of heart disease.11
`With the introduction of Mevacor in 1987, a new statin class of cholesterol
`drugs that were largely free of serious side-effects transformed the treatment
`of high cholesterol.12 Competing brands began to appear in 1991. By 1997,
`five manufacturers were producing six brands of statin drugs. Substantial
`clinical testing, required of all new drugs to win FDA approval but often
`continued thereafter, has found that statin drugs typically reduce the incidence
`of fatal and nonfatal heart attacks by 20–30 percent.13 Research continues
`on these drugs, on new statin drugs still under development, and on heart
`disease and its treatment.
`
`9 James Cleeman & Claude Lenfant, The National Cholesterol Education Program: Progress
`and Prospects, 280 JAMA 2099 (1998); and Gary Taubes, The Soft Science of Dietary Fat,
`291 Science 2536 (2001).
`10 Taubes, supra note 9.
`11 Martijn Katan, Review of D. John Betteridge, ed., Lipids: Current Perspectives, 336 New
`Eng. J. Med. 1394 (1997).
`12 Technically, the statin drugs are 3-hydroxy-3-methylglutarul coenzyme A (HMG-CoA)
`reductase inhibitors (id.). Cholesterol-reducing drugs are often referred to as lipid-lowering
`drugs. The most important side-effects are liver abnormalities and muscle disease. Elevated
`liver enzymes occur in approximately 1–2 percent of patients but return to normal following
`discontinuation of therapy (Richard S. Safeer & Cynthia L. LaCivita, Choosing Drug Therapy
`for Patients with Hyperlipidemia, 61 Am. Fam. Physician 3371 (2000)). Myopathy occurs in
`approximately .01 percent of patients (William C. Roberts, Twenty Questions on Atheroscle-
`rosis, 13 Baylor U. Med. Ctr. Proc. 139 (2000)). In August 2001, Baycol (introduced in 1997)
`was withdrawn because of an abnormal number of deaths from rhabdomyolysis, a rare muscle
`disease that is sometimes fatal.
`13 Susan D. Ross et al., Clinical Outcomes in Statin Treatment Trials, 159 Archives Internal
`Med. 11793 (1999). Statin drugs have also been found effective in treating heart attacks (Gergg
`C. Fonarow et al., Use of Lipid-Lowering Medications at Discharge in Patients with Acute
`Myocardial Infarction: Data from the National Registry of Myocardial Infarction, 103 Cir-
`culation 38 (2001)). In addition, a small but growing body of research has found that cholesterol
`reduction lowers the risk of strokes (Harvey D. White et al., 2000 Pravastatin Therapy and
`the Risk of Stroke, 343 New Eng. J. Med. 317 (2000)) and, possibly, the risk of neurological
`diseases such as Alzheimer’s (Susan J. Landers, Beyond Cholesterol: New Uses for Statins,
`Am. Med. News, June 18, 2001).
`
`5
`
`

`

`direct-to-consumer advertising
`
`677
`
`That research has yielded three findings that are particularly relevant here.
`First, the relationship between reducing cholesterol and preventing heart
`attacks has finally become firmly established and widely accepted by the
`medical community.14 Second, an expert panel’s findings reported in the
`Journal of the American Medical Association showed that the criteria for
`using cholesterol-reducing drugs have steadily expanded beyond treatment
`to the prevention of cardiovascular disease and, in the process, demonstrated
`that larger proportions of adults can benefit from statin drugs.15 Finally, the
`expert panel also showed that a large share of people who would benefit
`from cholesterol reduction, by drug therapy or other means, have failed to
`get appropriate diagnosis or treatment.16 The role for statin drug therapy has
`also expanded because it has proved difficult to reduce cholesterol substan-
`tially by traditional means such as dietary changes.17
`
`Statin Drug Competition and Marketing
`III.
`Developing drugs that reduce cholesterol and the risk of heart disease was
`a major medical breakthrough. But drug companies still had to compete for
`consumer demand. One way is to reduce side-effects. Another is to showcase
`results from clinical trials. Lipitor, for instance, is marketed as achieving the
`most rapid reductions in cholesterol levels (and increases in high-density
`lipoproteins (HDL), the “good” cholesterol); Zocor’s marketing focuses on
`reduced mortality from heart disease; Pravachol is touted as being shown to
`prevent first and second heart attacks.
`Price competition may also be important, but it is difficult to identify.
`Three of the four dominant brands (Lipitor, Zocor, and Pravachol) had com-
`parable average retail prices, while Baycol’s price was substantially lower.18
`Retail prices for the entire statin drug market have been very stable, increasing
`only 7 percent in real terms between 1995 and 2000.19 Because these data
`do not reflect private discounts and rebates from manufacturers to pharma-
`ceutical benefit managers or managed care organizations, prices are probably
`falling despite the surge of interest in these drugs. Stable statin drug prices
`
`14 D. Monkman, Treating Dyslipidaemia in Primary Care: The Gap between Policy and
`Reality Is Large in the U.K., 321 Brit. Med. J. 1299 (2000).
`15 Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults,
`Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP)
`Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
`(Adult Treatment Panel III), 285 JAMA 2486 (2001).
`16 Id.
`17 J. L. Tang et al., Systematic Review of Dietary Intervention Trials to Lower Blood Total
`Cholesterol in Free-living Subjects, 316 Brit. Med. J. 1213 (1998); and Taubes, supra note 9.
`18 Baycol has been withdrawn from the market, but this did not occur until after the period
`covered by our analysis.
`19 Data are from IMS Health. See http://www.imshealth.com.
`
`6
`
`

`

`678
`
`the journal of law and economics
`
`may actually be disguising a quality-adjusted decline in price because on-
`going research is revealing new and improved uses for these drugs.
`Product promotion, or marketing, is the most transparent and potentially
`most effective way for pharmaceutical firms to compete. In one prominent
`marketing strategy, called physician detailing, pharmaceutical representatives
`visit physicians, managed care organizations, health insurance companies, or
`affiliated organizations such as pharmaceutical benefit management firms to
`stress how their drugs can reduce health care or labor costs. Since the 1997
`FDA policy change, media advertising directed toward consumers through
`television, newspapers, and magazines has become especially important.
`Institutional features of the statin drug market appear to influence adver-
`tising’s content and potential effects. The Food and Drug Administration
`rules require patients to get a physician’s prescription to obtain a statin drug.20
`Given the large population of potential users who have no regular doctor
`and are unable to use reliable home tests or self-diagnoses to determine their
`cholesterol levels, DTC advertising could provide a strong stimulus for people
`to see a physician about the benefits from drug treatment.21 On the other
`hand, pharmaceutical firms have limited experience in using broadcast ad-
`vertising of prescription drug brands to motivate consumers to see their
`physicians about possible medical conditions. In any case, because roughly
`75 percent of prescription drug costs are covered by insurance and other
`third parties, DTC advertising invariably focuses on therapeutic benefits in-
`stead of price.22
`
`Some Empirics of Market Conditions
`IV.
`Market data from 1995 through 2000 reveal the salient characteristics of
`the statin drug market and motivate systematic analyses of possible causal
`relations. As expected in a market for an effective new drug, demand has
`been growing steadily (Figure 1).23 Although statin use has increased greatly
`
`20 Merck and Bristol-Myers-Squibb have asked the FDA to permit OTC sales of Mevacor
`and Pravachol, respectively. The Association of Black Cardiologists supported their petition,
`which the FDA declined to grant (Victoria Stagg Elliott, FDA Advisory Committee Vetoes
`OTC status for Low-Dose Anti-cholesterol Drugs, Am. Med. News, August 7, 2000). Should
`the manufacturers succeed in obtaining OTC status, the role of marketing in the statin drug
`market could change substantially.
`21 Home cholesterol measurement devices measure only total cholesterol, not high-density
`lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol, which are clinically impor-
`tant.
`22 IMS Health data, cited in Pharmaceutical Research and Manufacturers of America, Phar-
`maceutical Industry Profile (1999).
`23 Id. Sales data are collected by IMS Health on a weekly reporting basis; data were initially
`supplied for this study time aggregated such that the first two 4-week periods in a given quarter
`were reported as “months,” followed by a 5-week period reported as a third “month.” Obser-
`vations were adjusted by the authors such that a given observation represents the extrapolation
`of estimated daily sales over 30 days.
`
`1 3
`
`7
`
`

`

`direct-to-consumer advertising
`
`679
`
`Figure 1.—Total pills dispensed, statin. Data are from IMS Health.
`
`since the first quarter of 1995, it appears to have increased slightly less since
`the FDA rule change on DTC advertising. The growth in prescriptions de-
`composed into new and renewals (with brand switches recorded as new
`prescriptions) was very similar to the growth in total pills dispensed.24
`Although the trend in demand has been steadily upward, the trend in the
`promotion of statin drugs has been irregular. Total promotional spending has
`fluctuated considerably but has not increased much since the FDA rule change
`except for two brief periods including the end of 2000 (Figure 2). The mix
`of promotional expenditures, however, has changed dramatically. Figure 3
`decomposes these into broadcasting (mainly television), popular print, med-
`ical journals, and physician detailing and shows that television advertising
`has become far more important since the rule change, while popular print
`advertising appears somewhat less important. The variation in the data is
`consistent with “pulse advertising”—a series of brief periods of intensive
`advertising followed by a sharp slowdown or cessation—but it is not apparent
`how this variation has contributed to the growth in consumer demand.
`To explore this issue, we estimated a monthly time-series regression of
`total statin drug prescriptions on advertising expenditures and other poten-
`tially important influences including a price index supplied by IMS Health
`(deflated by the Consumer Price Index for Urban Consumers) and a variable
`to control for potentially persuasive medical research findings about the ben-
`
`24 As noted previously, real statin drug prices have remained constant; thus, the trend for
`statin drug expenditures is very similar to the trend exhibited by pills and prescriptions.
`
`8
`
`

`

`680
`
`the journal of law and economics
`
`Figure 2.—Total statin promotional dollars. Data are from IMS Health.
`
`efits of using statin drugs.25 Specifically, we included a dummy variable to
`represent the 1998 presentation of preliminary results of a clinical trial that
`attracted considerable attention because it demonstrated that for many heart
`disease patients, statin drug therapy was at least as effective as angioplasty.26
`We also used several alternative information variables based on the number
`of substantive news stories about cholesterol in the New York Times, Los
`Angeles Times, Washington Post, and USA Today.
`Because a Dickey-Fuller test revealed that the dependent variable was
`nonstationary, these models were estimated using first differences (in logs).
`Advertising expenditures were specified in several ways including simple
`lags, 3-month and 6-month stocks to capture the notion of pulse advertising,
`and as total expenditures, total DTC advertising, and individual components
`(television, print, journal, and detailing). (We also tried instrumenting ad-
`vertising expenditures for statin drugs using advertising expenditures for all
`drug prescriptions as an instrument.)
`Our consistent finding was that advertising expenditures had a statistically
`insignificant effect on prescription demand. Using a dummy variable to in-
`dicate the effect of promotions before and after the 1997 FDA rule change
`and employing a simple version of a switching regression did not change
`this finding. We also estimated models using alternative measures of demand
`
`25 Pierre Azoulay, Do Pharmaceutical Sales Respond to Scientific Evidence? Evidence from
`Anti-ulcer Drugs, 11 J. Econ. Mgmt. Strategy 551 (2002), found that both medical research
`findings and promotion increased the sales of anti-ulcer drugs.
`26 The study was Atorvastatin versus Revascularization Treatments (AVERT); preliminary
`results were presented at the November 1998 American Heart Association meeting. The results
`were published in Bertram Pitt et al., Aggressive Lipid-Lowering Therapy Compared with
`Angioplasty in Stable Coronary Artery Disease, 341 New Eng. J. Med. 70 (1999).
`
`9
`
`

`

`direct-to-consumer advertising
`
`681
`
`Figure 3.—Monthly statin promotional dollars by type. Data are from IMS Health.
`
`such as number of pills and revenues. Because prescriptions and pills pre-
`scribed may not necessarily correspond to consumption patterns (consumers
`can time shift by purchasing a larger prescription), we also constructed an-
`other demand measure by calculating the mix of 30-day and 90-day pre-
`scriptions and treating the 90-day prescription as the equivalent of one 30-
`day prescription for each of the subsequent 3 months (we termed this variable
`PATIENTS as a measure of the number of people undergoing statin treatment
`in a given month). None of these regressions yielded significant coefficients
`for any of the advertising variables. Finally, we examined the impact of
`advertising on pharmaceutical firms’ market shares and, again, found no
`statistically significant effect from advertising despite dramatic shifts in these
`shares during the sample period.27
`It is important to stress that several steps may occur between the time an
`advertisement motivates a consumer to explore drug therapy for cholesterol
`and when the drug is purchased, which makes it difficult to rule out the
`possibility that DTC advertising does influence demand. These steps typically
`include a visit to a doctor, a second visit to review cholesterol test results,
`perhaps advice to try dietary changes and exercise, perhaps another choles-
`
`27 Mevacor’s share of total prescriptions went from 37 percent in December 1995 to 2 percent
`in December 2000; Lipitor’s market share went from 14 percent in December 1997 to over
`50 percent in December 2000. Marta Wosinska, Effects of Direct-to-Consumer Drug Adver-
`tising on Prescription Choice (Working paper, Univ. California, Berkeley 2001), found that
`DTC advertising had a small but statistically significant effect on brand shares, conditioned
`on a physician deciding to prescribe a statin drug.
`
`10
`
`

`

`682
`
`the journal of law and economics
`
`Figure 4.—New high-cholesterol diagnoses. Data are from Scott-Levin, Physician Drug and
`Diagnosis Audit.
`
`terol test, and, finally, the first prescription. This may take a a few weeks to
`6 months or more, with considerable variance among patients and perhaps
`among physicians. Thus it may be possible to detect the effect of DTC
`advertising on consumer demand only with disaggregated data that link a
`patient’s cholesterol treatment history with the timing of DTC advertising
`expenditures.28
`
`V.
`
`Investigating Other Possible Effects of
`Direct-to-Consumer Advertising
`Because consumers may not respond to advertising in a way that is easy
`to detect empirically, we explored alternative ways in which advertising could
`affect this market. For example, advertising could affect demand by influ-
`encing consumers to visit their doctors, get checkups and diagnoses, and
`ultimately use a cholesterol-reducing prescription drug. Indeed, Figure 4
`documents a steady rise in physician office visits involving a new diagnosis
`of elevated cholesterol, on the basis of monthly data from Scott-Levin’s
`ongoing panel of physicians. New diagnoses can signal a treatable condition
`and lead to additional office visits. Figure 5 shows that total office visits for
`patients under treatment for high cholesterol have risen from roughly 1 mil-
`lion in 1996 to more than 2.5 million in 2000. And a large fraction of these
`
`28 Meredith B. Rosenthal et al., Demand Effects of Recent Changes in Prescription Drug
`Promotion (Working paper, Nat’l Bur. Econ. Res., June 28, 2002), aggregated five therapeutic
`drug classes to estimate the effects of drug advertising. However, they are likely to have
`overstated these effects because it appears that they did not detrend their sales and advertising
`data.
`
`11
`
`

`

`direct-to-consumer advertising
`
`683
`
`Figure 5.—Doctor visits, patients currently under treatment for high cholesterol. Data are
`from Scott-Levin, Physician Drug and Diagnosis Audit.
`
`people are being treated with statin drugs. Data from Scott-Levin indicate
`that the proportion of elevated cholesterol diagnoses resulting in a statin drug
`prescription increased from 75 to 95 percent during 1996–2000.29
`It is therefore possible to characterize a process that has been growing
`during the past 4 years in which consumers visit the doctor, get a diagnosis,
`and, when diagnosed with high cholesterol, usually get a statin prescription.
`Advertising may therefore have an indirect effect on statin drug demand by
`influencing people to visit their doctor for a checkup and diagnosis. (This
`potential relationship is not contemporaneous in our (Scott-Levin) data be-
`cause office visits are not coded as cholesterol related until the patient is
`diagnosed, which usually does not happen until a new patient visits the doctor
`a second time to review the results of a serum cholesterol test.)
`We explored this possible effect of advertising by estimating monthly time-
`series regressions of office visits for all reasons, visits related to cholesterol,
`and so on, as a function of advertising expenditures plus alternative infor-
`mation variables. A Dickey-Fuller test revealed that the visits variables were
`nonstationary, so we estimated models using first differences (in logs). We
`specified advertising expenditures in several alternative ways including sim-
`ple lags and 3-month and 6-month stocks. The consistent finding was that
`advertising expenditures had a statistically insignificant effect on office visits.
`In addition, if advertising were having a strong effect on office visits, one
`might expect that patients would be initiating requests for statin drugs. But
`the Scott-Levin physicians’ panel data indicate that the share of patients with
`
`29 Scott-Levin, Physician Drug and Diagnosis Audit (http://verispan.com/products/data_sheet
`.asp?cp2&pp16).
`
`12
`
`

`

`684
`
`the journal of law and economics
`
`elevated cholesterol who requested a statin drug prescription has been only
`2–5 percent since 1997.
`In the face of growing cholesterol-related patient visits and statin drug
`demand, it would strengthen our findings that advertising has had a statis-
`tically insignificant effect on this growth if we could point to what has caused
`it. We can only conjecture at this point. One relevant fact is that actual and
`potential patients can obtain objective evidence on the efficacy of statin drugs
`in reducing serum cholesterol. Could such emerging evidence affect demand?
`Using the Scott-Levin data, we constructed the variable SUCCESS as the
`percentage of existing patients with a high-cholesterol diagnosis whose total
`cholesterol fell below 200 and then regressed PATIENTS (as defined pre-
`viously) on lagged values of SUCCESS. The motivation is that information
`about successful treatment may influence demand; for example, people who
`are dramatically lowering their cholesterol may increase demand by spreading
`the word about the effectiveness of statin drugs and thus making patients
`more receptive to physicians’ recommendations for statin drug therapy. Es-
`timation results excluding a statistically insignificant information variable are
`as follows (t-statistics are in parentheses):
`
`%DPATIENTS p .016 ⫹
`t
`(8.6)
`
`.077%DSUCCESS ⫹ .031DSUCCESS ⫺
`t⫺1
`t⫺2
`(3.0)
`(1.1)
`
`.254MA(1);
`(⫺2.1)
`
`N p 57,
`
`R p .21.
`2
`
`The variable SUCCESS has a statistically significant effect with one lag;
`the coefficient falls by about half for a 2-month lag but still exceeds its
`standard error. This finding is consistent with the idea that information is
`playing a role in the growth of the statin drug market. That information is
`presumably not from advertising, but instead from successful treatments that
`we speculate are being discussed among friends, coworkers, and the public
`at large as well as among physicians.30 Given that the Scott-Levin data
`indicate that the average total cholesterol levels of existing cholesterol pa-
`tients declined from about 240 to about 215 between 1996 and the end of
`2000, it would not be surprising if patients were sharing their positive ex-
`periences with statin drug therapy.
`It seems likely that the doctor, acting as a gatekeeper with both professional
`advice and the power to prescribe, has, in combination with objective evi-
`dence from the effects of cholesterol-reducing drugs, “crowded out” adver-
`tising from playing a significant short-run role in the growth of the cholesterol
`drug market. But advertising may be having a subtle effect by reinforcing
`
`30 Fred Mannering & Clifford Winston, Automobile Air Bags in the 1990s: Market Failure
`or Market Efficiency? 38 J. Law & Econ. 265 (1995), found that consumers’ adoption of
`airbags was spurred by information disseminated among friends, family members, and the
`media about actual experiences with airbags.
`
`13
`
`

`

`direct-to-consumer advertising
`
`685
`
`doctors’ recommendations to try a statin drug and continue using it. To
`explore this possibility, we estimated the effect of television advertising
`expenditures on SUCCESS (the number of existing patients with a h