`
`At the Intersection of Health, Health Care and Policy
`
`Cite this article as:
`W. David Bradford, Andrew N. Kleit, Paul J. Nietert, Terrence Steyer, Thomas
`Mcllwain and Steven Ornstein
`How Direct-To-ConsumerTelevision Advertising For Osteoarthritis Drugs Affects
`Physicians’ Prescribing Behavior
`Health Affairs 25, no.5 (2006):1371-1377
`doi: 10.1377/hlthaff.25.5.1371
`
`The online version of this article, along with updated information and services, is
`available at:
`http://content.healthaffairs.org/content/25/5/137 1
`
`For Reprints, Links &
`Permissions:
`
`http://content.healthaffairs.org/1340_reprints.php
`
`Email Alertings : http://content.healthaffairs.org/subscriptions/etoc.dtl
`
`To Subscribe:
`
`_https://fulfillment.healthaffairs.org
`
`without prior written permission from the Publisher. All rights reserved.
`
`Health Affairs is published monthly by Project HOPE at 7500 Old Georgetown Road,
`Suite 600, Bethesda, MD 20814-6133. Copyright ©
`by Project HOPE - The People-to-People Health Foundation. As provided by United
`States copyright law (Title 17, U.S. Code), no part of
`may be reproduced, displayed, or transmitted in any form or by any means,electronic or
`mechanical, including photocopying or by information storage orretrieval systems,
`
`
`
`wealMHAqZL0z‘¢YoueyuosuIeyyYearAq/Bu0'suleyeuljeey}ue}U00//:d}]yWo)pepeojumMog
`
`Not for commercial use or unauthorized distribution
`
`1
`
`ALL 2069
`MYLAN PHARMACEUTICALS V. ALLERGAN
`IPR2016-01130
`
`1
`
`ALL 2069
`MYLAN PHARMACEUTICALS V. ALLERGAN
`IPR2016-01130
`
`
`
`MARKETWATCH
`
`
`MARKETWATCH
`
`How Direct-To-ConsumerTelevision Advertising
`For Osteoarthritis Drugs Affects Physicians’
`Prescribing Behavior
`
`DTC advertising of COX-2 inhibitors appears to have increased the
`number of prescriptions written for these products.
`
`by W. David Bradford, Andrew N. Kleit, Paul J. Nietert, Terrence Steyer,
`Thomas Mcllwain, and Steven Ornstein
`
`ABSTRACT:Concern aboutthe potential pernicious effect of direct-to-consumer (DTC) drug
`advertising on physicians’ prescribing patterns was heightened with the 2004 withdrawal
`of Vioxx, a heavily advertised treatmentfor osteoarthritis. We examine how DTC advertising
`has affected physicians’ prescribing behavior for osteoarthritis patients. We analyzed
`monthly clinical
`information on fifty-seven primary care practices during 2000-2002,
`matched to monthly brand-specific advertising data for local and network television. DTC
`advertising of Vioxx and Celebrex increased the number of osteoarthritis patients seen by
`physicians each month. DTC advertising of Vioxx increased the likelihood that patients re-
`ceived both Vioxx and Celebrex, but Celebrex ads only affected Vioxx use. [Health Affairs
`25, no. 5 (2006): 1371-1377; 10.137 7/hIthaff.25.5.13 71]
`
`Te U.S. FOOD AND DRUG Administra-
`
`tion (FDA) issued newregulations in
`August 1997 governing television ad-
`vertisingof prescription drugs. Shortly there-
`after, spending on direct-to-consumer (DTC)
`advertising for prescription drugs soared—
`from $596 million in 1995 to approximately
`$1.2 billion in 1997 and an estimated $3.8 bil-
`lion by 2004.' The effects of this spending
`have been the subject of much debate, al-
`though little is known aboutits actual im-
`
`pacts.Westudiedthisissuebyexaminingthe
`
`use of a popular (and controversial) class of
`prescription drugs: the anti-inflammatory
`and pain-relieving cyclooxygenase-2 (COX-
`2) inhibitors. The most popular of these were
`Vioxx (Merck) and Celebrex (Pfizer). We
`studied how prescriptions for these two
`drugs for patients with osteoarthritis re-
`sponded to changes in television advertising
`during 2000-2002.
`@ The Vioxx effect. One of the most
`
`David Bradford (bradfowd@musc.edu) is a professor at the Centerfor Health Economic and Policy Studies,
`DepartmentofHealth Administration andPolicy, at the Medical University ofSouth Carolina (MUSC) in
`Charleston. Andrew Kleit is a professorin the Department ofMeteorology, Centerfor Health Care and Policy
`Research, at the Pennsylvania State University in University Park. At MUSC, Paul Nietert is an assistant
`professorin the DepartmentofBiostatistics, Bioinformatics, and Epidemiology, Terrence Steyeris an assistant
`professor in the DepartmentofFamily Medicine; Thomas Mcllwainis an assistantprofessorin the Departmentof
`Health Administration andPolicy; and Steven Ornsteinis an associateprofessor in the DepartmentofFamily
`Medicine.
`
`EEE
`1371
`HFALTH AFFAIRS - Volume 25, Number 5
`
`DOL 10.1377#hIthaff.25.5.1371 ©2006 Project HOPE—The People-to-People Health Foundation, Inc.
`
`
`
`wee!MHAqZL0z‘€YoueWUOsueyyyeaHAq/Bio’ssreyeuyeey'jue}u0o//:di1yWo)pepeojumog
`
`2
`
`
`
`HEALTH TRACKING
`a
`
`heavily advertised products in recent years
`was Vioxx (rofecoxib). In calendar year 2000,
`Merck spent more than $160 million on adver-
`tising this product to consumers.? Merck
`withdrew Vioxx from the market in Septem-
`ber 2004 becauseof evidenceof increased risk
`of myocardial infarction and stroke associated
`withits use.’Its side effects have sparked much
`criticism of Merck's advertising strategy.
`M@ Previous studies. Published studies on
`the impact of television advertising for pre-
`scription drugs have yielded conflicting re-
`sults. Advancing an argument madeby Alison
`Masson and Paul Rubin, Alison Keith found
`that patients’ suggestions regarding drugs (as-
`pirin for cardiovascular disease) were impor-
`tant determinants in prescription decisions
`and that advertising tendedto lead to more ap-
`propriate care.
`Other studies found that drug companies’
`TV promotional activities preserved market
`share for existing products and also caused pa-
`tients to be less responsive to price.® The post-
`1997 era presented an opportunity for exami-
`nation of the new policy regime for DTC
`advertising. As Woodie Zachry and Diane
`Ginsburg pointed out, however, few studies
`examinethe actual effects of DTC advertising.’
`In one of these studies, John Calfee and col-
`leagues examined whether the 1997 FDA
`policy change increased the demandforstatins
`but were unable to find anysignificant direct
`effect.8 There are also studies that examined
`DTC advertising, using survey data. These
`studies tended to find that patients were posi-
`tively disposed toward DTC advertising and
`that advertising had someeffect on patient-
`physician interactions.°
`
`Study Data And Methods
`To assess the impact of DTC advertising, we
`examined separate models for (1) the average
`patient flow into each physician practice, (2)
`prescribing for Vioxx, and (3) prescribing for
`Celebrex, using the practice as the unit of anal-
`ysis. We examined data onvisits to fifty-seven
`primary care practices each month over three
`years (2000-2002). After exclusions (discussed
`below), this resulted in 1,589 observations in
`
`the form ofvisits per practice per month.
`@ Patient data. Data were obtained from
`the Practice Partner Research Network
`(PPRNet), located at the Medical University of
`South Carolina (MUSC). PPRNetis a prac-
`tice-based learning and research organization
`among primary care practices across the
`United States that use a common electronic
`medical record (EMR) (Practice Partner by
`Physician Micro Systems Inc. in Seattle).
`PPRNetpools longitudinal data on diagnoses,
`laboratory studies, medications, vital signs,
`and other nonidentifiable information quar-
`terly for research and quality improvementac-
`tivities. PPRNet has access to all EMR ex-
`tracts of ninety-one community-based
`primarycare practices in thirty-twostates. To
`create our practice-level data set, we extracted
`data on all patients who had a diagnosis for
`asteoarthritis from practices active during
`2000-2002. Because many unobservable fac-
`tors might drive prescribing, werestricted the
`original data to osteoarthritis patients.
`These patient data were then aggregated to
`the practice level by month. Also, we elimi-
`nated practices located more than 100 miles
`from the geographic center of the nearest me-
`dia market. Wealso eliminated practices that
`were in the database for fewer than twenty
`months or that wrote fewer than ten COX-2
`inhibitor prescriptions cumulatively over the
`three-year study period. Fifty-seven practices
`from forty-four markets remained after these
`restrictions were implemented.
`Advertising data. We obtained national
`and local advertising information from Com-
`petitive Media Reporting Inc. (CMR), which
`collects data on media advertisingforall prod-
`ucts, including pharmaceuticals, at the market
`(for example,city) level. The data are specific
`to the product's brand name and show which
`products were advertised and how many times
`they were advertised on both national andlo-
`cal television each month. We used counts of
`ads broadcast by month as our measure of
`DTC advertising. Patients and practices were
`assigned to the nearest local media market. (In
`the advertising data we received, media mar-
`kets were identified by metropolitan area,
`
`1372
`
`
`
`September/October 2006
`
`
`
`
`
`weelAAHAqZL0z‘€YoueWUOsvleyYYyeayAq/Bio'sureyeuyjeey'}Ue}U09//:d1]yLUOJ|papeojuMOGg
`
`3
`
`
`
`MARKETWATCH
`|
`
`such as Philadelphia or Denver.)
`@ Dependent variables. We examined
`the impact of DTC ads on three dependent
`variables: (1) the numberofvisits (for any rea-
`son) to the practice each month by osteo-
`arthritis patients; (2) the fraction of these vis-
`its associated with a prescription (new orre-
`newed) for Celebrex; and (3) the fraction of
`these visits associated with a prescription
`(new or renewed) for Vioxx (Exhibit 1). We
`could not distinguish reliably between the
`numberof new and renewedprescriptions, al-
`thoughthis is not a limitation with respect to
`our study’s goal.
`@ Explanatory variables. Advertising expo-
`sure. We measured advertising exposureas the
`numberof ads broadcastfor each brand adver-
`tised. We included separate measures for na-
`tional and local advertising, because national
`and local ads tend to be shown duringdifferent
`times of the day and during different programs.
`Retained information from ads. Since informa-
`tion presented in ads will not be immediately
`forgotten, we needed to accountfor retained
`information. To do this, we constructed two
`versions of each of our three models. First, we
`estimated models for the three dependentvari-
`ables that included the numberof ads placed
`in the current month. Second, we estimated
`
`models for the three dependentvariables that
`included the numberofads placed in both the
`current month and the previous month."
`Clinical and income variables. We also included
`variables we expected to affect either the de-
`mand. for treatment by patients or the supply
`of office visits (and therefore treatment) by
`practices. Factors expected to affect patients’
`demand for treatment were patients’ clinical
`comorbidities, average age, and sex. These
`clinical variables were calculated using the
`PPRNetdata onosteoarthritis patients in the
`practice. The income variable was taken from
`the Area Resource File (ARF) and was mea-
`sured at the county rather than the patient or
`practice level (and is thus a crude measure of
`average patient incomeforthe practice).
`Factors affecting general demandfor health care.
`Factors expected to affect the general demand
`for health care were county population;
`county per capita income; the average price of
`a physician office visit; and the percentage of
`county population that is over age sixty-five,
`employed, and either Caucasian or African
`American. We imputed theprice of an inter-
`mediate-lengthoffice visit with an established
`patient from the Council for Community and
`Economic Research’s (ACCRA’s) quarterly
`Cost of Living Index (using the average price
`
`EXHIBIT 1
`Means And Standard Deviations Of Dependent And Key Explanatory Variables, Study
`Of Direct-To-ConsumerTelevision Ads And Prescribing For Osteoarthritis (OA), 2000-
`2002
`
`Variable
`
`Mean
`Standard deviation
`
`
`420.814
`341.952
`Numberof monthly office visits by OA patients
`0.035
`Number of monthly Celebrex prescriptions per OA officevisit
`0.014
`
`0.028
`Number of monthly Vioxx prescriptions per OA office visit
`0.012
`0.761
`Numberof local Vioxx ads
`0.259
`37.742
`Numberof national Vioxx ads
`105.848
`Numberof local Celebrex ads
`12.296
`22,980
`102.262
`54.018
`Numberof national Celebrex ads
`
`
`Numberoflocal Vioxx ads, previous month
`Numberof national Vioxx ads, previous month
`Numberof local Celebrex ads, previous month
`Numberof national Celebrex ads, previous month
`
`0.243
`104.274
`11.755
`97.820
`
`0.727
`38.237
`23.057
`55.620
`
`SOURCE:Authors’ calculations, using data from PPRNet and the Centers for Medicare and Medicaid Services.
`NOTE: N = 1,589.
`
`HEALTH AFFAIRS ~ Volume 25, Number 5
`
`1373
`
`
`
`
`
`wee!MHAq/L0z‘€YuueWUosvleyyyyeayAq/Bio-ssujeyeuyeey'jue}u0o//:dy]YyLol)papeojumoq
`
`4
`
`
`
`
`
`wee!MHAqZL0Z‘€YouepyuosueyyYyeayAq/Bio'ssreyeuyjeey'jueuoss/:dy1ywos)papeojuMoG
`
`HEALTH TRACKING
`
`in the metropolitan area nearest the primary
`care practice site). Finally, we captured the
`market-level supply of services by including
`the numberof full-time physicians (bothpri-
`mary and specialty care) actively engaged in
`patient care in the county per thousand
`population, taken from the ARF.
`@ Clinical publications. Besides DTC ad-
`vertising and demand/supply-side economic
`factors, articles in medical journals might af-
`fect physicians’ prescribing habits. We con-
`trolled for confounding effects of clinical pub-
`lications in two ways. First, we conducted a
`Medline search using the keywords rofecoxib,
`celecoxib, Vioxx, Celebrex, and osteoarthritis
`(restricting the range to English-language
`journals and the years 2000-2002). We in-
`cluded measures of the numberof articles each
`month that focused on Celebrex or Vioxx, or
`both. We did not attempt to characterize
`whetherthe articles were favorable or unfavor-
`able. Wealso includeda variable that equals 0
`when the monthofpublication is before an im-
`portant August 2001 publication by Debabrata
`Mukherjee and colleagues, and 1 otherwise."
`This article was the first in a majorclinical
`journal to document concerns about Vioxx’s
`cardiovascular side effects.
`@ Statistical analysis. Two of the three
`models had dependent variables that were
`numbers of prescriptions written to osteo-
`
`arthritis patients divided by the numberof vis-
`its by such patients. By measuring these de-
`pendentvariables as percentages of the poten-
`tially “treated” population, we normalized the
`results for differences in practice size. We also
`included practice-level fixed effects in each of
`the three models (which controlled for any
`time-invariant practice characteristics). We
`estimated these models as linear regressions
`and adjusted the standard errorsfor clustering
`at the practice level (to controlfor the fact that
`we had repeated monthly observations on
`each practice) using the Stata “xtreg” proce-
`dure. We adjusted for clustering at the prac-
`tice level rather than at the media-marketlevel
`because we generally have only one practice
`per media market.
`
`Study Results
`Although Merck (Vioxx) investedin nearly
`twice as many ads as Pfizer (Celebrex) for the
`first nine months of 2000, advertising expo-
`sure for the two brands was roughly compara-
`ble over the remaining 2000-2002 time period
`(Exhibit 2). However, a different picture
`emerged atthe local level throughout this pe-
`riod: Celebrex local ads aired slightly more
`than twelve times a month on average, while
`Vioxx local ads aired fewer than once a month
`on average (data not shown). There was also
`evidencein the raw data of a more concentrated
`
`EXHIBIT 2
`
`Monthly Counts Of Direct-To-ConsumerTelevision Advertising Of Drugs For
`Osteoarthritis, Nationally And In Local Markets, By Brand Name, 2000-2002
`
`Total Vioxx ads
`Number of monthly ads
`
`200
`———--—== Total Celebrex ads
`
`
`
`
`100
`
`
`
`150
`
`1
`
`7
`
`19
`13
`Months since January 2000
`
`25
`
`31
`
`SOURCE:Authors’ analysis of data from Competitive Media Reporting Inc.
`
`a 1
`
`374
`
`September/October 2006
`
`5
`
`
`
`MARKETWATCH
`|
`
`strategy for Vioxx—with ads appearing often
`in some local markets and neverin others,”
`@ Ads’ impact on physician visits. Here
`wediscuss results from the fixed-effects re-
`gressions on the number of monthlyvisits to a
`practice by osteoarthritis patients. Because we
`estimatedlinear models, the parameters have a
`straightforward interpretation, implying how
`much of a change in the dependent variable
`would be induced by one additionaltelevision
`ad per month(in either a local or national con-
`text). We found consistent evidence in favor of
`the hypothesis that DTC advertising attracted
`patients to practices. Current-month local
`Vioxx advertising had a positive and signifi-
`cant (p =.01or better) effect in both specifica-
`tions. National Vioxx advertising also had a
`positive and significant effect in the first speci-
`fication. In terms of magnitudes,a 100 percent
`increase in local Vioxx ads would, on average,
`have increased the number of osteoarthritis
`visits per month by approximately 0.8 percent,
`while a 100 percentincreasein national Vioxx
`ads would haveled to a 7.4 percent increase in
`physician visits per month.In the secondspec-
`ification, a similar national Vioxx effect was
`measured for the lagged monthly advertising,
`although notfor the current-month DTC ad-
`vertising.
`Celebrex TVadvertising was effective only
`at inducing changesin patient flowsat the na-
`tional level (and only in our second specifica-
`tion), although the effect was somewhat com-
`plex. We found that current-month national
`DIC ads increased osteoarthritis patient flow,
`while lagged-month national DTC ads reduced
`patient flow. However,the neteffect was posi-
`tive, such that a 100 percent increase in DTC
`advertising for both time periods would have
`led to a 2 percent increase in monthly visits
`from osteoarthritis patients. Thus, on net, we
`did find support for the hypothesis that DTC
`advertising for COX-2inhibitors increased pa-
`tient flow into physician practices.”
`@ Ads’ impact on prescribing. In general,
`we foundlittle effect from DTC ads onthe rate
`of prescribing for Celebrex. Local DTC adver-
`tising for Vioxx had a marginally significant(p
`= .06) and a positive effect on the numberof
`
`Celebrexprescriptionsper patientvisit. How-
`ever, none of the other measures of the DTC
`levels was significantly associated with
`changes in Celebrex prescribing. Thus, al-
`though there could be a cross-brand effect for
`Celebrex, it appeared only weakly in our data,“
`Our results were somewhat stronger for
`Vioxx prescribing. Current-month local Vioxx
`DTCads had a positive and significanteffect (p
`= .04 or better) in both specifications. The
`magnitude of the effect was more modest,
`however; our results suggest that a tenfold in-
`crease in local Vioxx DTC spots (0.25-2.5, on
`average) would have induced about a 0.5 per-
`cent increase in the rate of Vioxx prescribing
`each month. Interestingly, national Celebrex
`ads had a positive and significant (p = .O1 or
`better) effect on the rate of Vioxx prescribing.
`The effect was larger than the impactof local
`Vioxx DTC advertising; a 50 percent increase
`in monthly national Celebrex advertising (102-
`153, on average) would have led to an increase of
`about0.5 percent in Vioxx prescribing,
`@ Impact of other factors. There was no
`evidence that journal articles on Celebrex and
`Vioxx affected prescribing? In particular,
`while there has been much discussion of
`Mukherjee and colleagues’ paper, we could not
`find a consistent effect of the publication in
`our linear model of prescribing.'*
`
`Discussion
`The issue of what impact,if any, DTC adver-
`tising has on the behaviorofpatients and phy-
`sician practices is of major policy importance.
`Pharmaceutical spending represents a large
`percentageof total spendingin state Medicaid
`programs. Even moderate changesin prescrib-
`ing for a small number ofproducts can lead to
`dramatic changes in Medicaid spending. The
`FDA has recently held public hearings to
`gather input for a reevaluation of its policy
`toward DIC advertising.
`This paper shedslight on the impact of ad-
`vertising of the two major COX-2 inhibitors
`available from 2000 to 2002: Vioxx and Cele-
`brex. Following Merck’s withdrawal of Vioxx,
`the New England Journal ofMedicine published an
`editorial that expressed skepticism about the
`
`Ee
`HEALTH AFFAIRS - Volume 25, Number 5
`1375
`
`
`
`
`
`wee!MHAq2102‘€YoueWUOsuleyYYyeayAq/Bio'sureyeuyeey'jua}uoo//:dy]yLUOW,papeojuUMOGg
`
`6
`
`
`
`HEALTH TRACKING
`SS.ee
`
`benefits to society from DTC advertising of
`COX-2 inhibitors.”
`B Patient flow. The results presented here
`bear directly on these issues. In general, we
`found that DTC advertising for Vioxx and
`Celebrex had positive effects on the flowof
`osteoarthritis patients into physician prac-
`tices. Advertising of both drugs generated an
`effect similar to that hypothesized by Keith
`and by Masson and Rubin, where increases in
`the potential ad exposure were associated
`with increases in patient flow." This effect,
`however, was somewhat attenuated for Cele-
`brex advertising.
`Once patients arrived at the physician's of-
`fice, it was clear that DTC advertising tended
`to change the rate at which COX-2inhibitors
`were prescribed. Theeffect of Vioxx DTC ads
`was consistently positive, increasing the pro-
`portion of visits at which a prescription was
`written for Vioxx. Vioxx DTC ads mightalso
`have had a positive effect on Celebrex pre-
`scribing rates, although the effect was only
`marginally significant for current-month local
`DTC ads. More national Celebrex DTC adver-
`tising was associated with higher rates of Vioxx
`prescribing but was notassociated withsignifi-
`cant changes in prescribing for Celebrex.
`lB “Detailing” versus advertising. There
`are a numberof possible explanations for the
`lack of an own-effect from Celebrex DTC ad-
`vertising. For example, data from IMS Health
`indicate that Pfizer devotedrelatively more ef-
`fort in 2000 to direct-to-physician marketing
`for Celebrex in the form of visits by pharma-
`ceutical representatives, who provided sam-
`ples (known in the industry as detailing) than
`Merck did on behalf of Vioxx. It is possible
`that this relative emphasis on detailing is re-
`flected in the relative ineffectiveness of DTC
`advertising at generating prescriptions. It is
`also possible that since Merck advertised rela-
`tively more heavily on behalf of Vioxx early in
`thelife cycle of the drugs, the class-level effects
`were exhaustedearly on,leavinglittle room for
`Celebrex advertising to increase the number of
`prescriptions. In the end, we did not have suf-
`ficient data to untangle why Vioxx advertising
`had a broaderclass effect and an own-effect,
`
`while Celebrex DTC advertising seemsto only
`have exhibited class effects for prescribing.
`@ Study limitations. Despite the impor-
`tance of our study, it has limitations. First,
`without monthly data for pharmaceutical de-
`tailing, we were unable to accountdirectly for
`the impact and interaction ofdetailing. Such
`physician-based marketing remains a larger
`component of pharmaceutical marketing ef-
`forts than DTC advertising.”It is possible that
`DTC advertising and detailing were coordi-
`nated; if so, the DTC advertising effects mea-
`sured here might have included somedetailing
`effect. Personal communication with drug
`company representatives, however, suggested
`that since both Vioxx and Celebrex were im-
`portant products to their manufacturers, the
`reps discussed these products at every oppor-
`tunity. This implies that there would belittle
`correlation betweenthelevels of detailing and
`local or national variation in advertising,
`which in turn would minimize the potential
`for bias in the results, Second, practices also
`generally have a supply of drug samples on
`handto give to patients when they write a pre-
`scription, Theavailability of samples might in-
`fluence which product is prescribed. Again,
`personal communication with physicians and
`drug reps suggested that most physicians
`would typically have had a stock of samples of
`both drugs on hand, which implies that the
`potential for omitted variables biasis limited,
`Also, the practice-level fixed effects included
`in the models capture any general tendency to
`favor one drug over another.
`
`I: SUMMARY, DTC ADVERTISING for
`
`COX-2inhibitors appears to have affected
`physician practices and patients. Pfizer's
`and Merck's DTC advertising had positive net
`effects in moving people toward more contact
`with their physicians. Once patients had con-
`tact with physicians, Merck’s DIC advertis-
`ing for Vioxx tended to stimulate Vioxx pre-
`scriptions and might have had a marginally
`important effect on Celebrex prescribing:
`however, Pfizer’s DTC advertising for Cele-
`brex seemsto only haveaffected (in a positive
`way) prescribing for Vioxx,
`
`
`
`wes]AAHAq2102‘€Yue]uOsveYYYearAq/Buo'sureyeuyeey'jueu0o//:diyWOJ)peapeojuMOoGg
`
`a1
`
`376
`
`September/October 2006
`
`7
`
`
`
`MARKETWATCH
`
`This paper wasfunded by grantsfrom the Agencyfor
`Healthcare Research and Quality (1 RO] HS011326-
`OIA2) andfrom the National Heart, Lung, and Blood
`Institute (1 RO] HLO77841-01). The authors thank
`RuthJenkinsfor assistance in extracting clinical data
`from the PPRNetsystem, and Olena Verbenko and
`Michael Kunzfor assistance in constructing the
`analysis datasets.
`
`NOTES
`1,
`T. Nordenberg, “TV Drug Ads That MakeSense,”
`Consumers’ Research Magazine 81,no. 3 (1998): 28-31;
`andS.Elliot, “With or Without Vioxx, Drug Ads
`Proliferate,” New York Times, 6 December 2004.
`M.Schumann, “Top 100 Megabrands: Chevrolet
`Leads the Race along Megabrand Road, but Mar-
`keters Hit 2001 Potholes,” Advertising Age 72, no,1
`(2001): 1.
`G. Kolata, “A Widely Used Arthritis Drug Is
`Withdrawn,” New York Times, 1 October 2004.
`EJ. Topol, “Failing the Public Health—Rofe-
`coxib, Merck, and the FDA,” New EnglandJowmalof
`Medicine 351, no. 17 (2004): 1707-1709; and D.
`Mukherjee, S.E. Nissen, and EJ. Topol, “Risk of
`Cardiovascular Events Associated with Selective
`COX-2 Inhibitors,” Journal of the American Medical
`Association 286, no. 8 (2001): 954-959.
`A. Masson and PH. Rubin, “Matching Prescrip-
`tion Drugs and Consumers: The Benefits of Di-
`rect Advertising,” New England Journal of Medicine
`313, no. 8 (1985): 513-515; and A. Keith, “Regulat-
`ing Information about Aspirin and the Preven-
`tion of Heart Attack,” American Economic Review
`85, no. 2 (1995): 96-99,
`M.A. Hurwitz and R.E. Caves, “Persuasion or In-
`formation? Promotion and the Shares of Brand
`Name and Generic Pharmaceuticals,” Journal of
`Law and Economics 31, no. 2 (1988): 299-320; C.
`King, “Marketing, Product Differentiation, and
`Competition in the Pharmaceutical Industry”
`(Working Paper, Massachusetts Institute of
`Technology, Department of Economics, 1996);
`and J.A. Rizzo, “Advertising and Competition in
`the Ethical Pharmaceutical Industry: The Case of
`Hypertensive Drugs,” Journal of Law and Economics
`42, no. 1 (1999): 89-116.
`WM.ZachryIII and D.B. Ginsburg, “Patient Au-
`tonomy and the Regulation of Direct-to-Con-
`sumer Advertising,” Clinical Therapeutics 23, no. 12
`(2001): 2024-2037.
`JE. Calfee, C. Winston, and R. Stempski, “Di-
`tect-to-Consumer Advertising and the Demand
`for Cholesterol-Reducing Drugs,” Journal of Law
`and Economics 45, no. 2, part 2 (2002): 673-690.
`FF. Gonul, F. Carter, andJ. Wind, “What Kind of
`
`Patients and Physicians Value Direct-to-
`Consumer Advertising of Prescription Drugs?”
`Health Care ManagementScience 3, no. 3 (2000): 215-
`226; N. Sumpradit, S.W. Fors, and L.
`McCormick, “Consumers’ Attitudes and Behav-
`ior toward Prescription Drug Advertising,” Amer-
`ican Journal of Health Behavior 26, no. 1 (2002): 68-
`75; WM.ZachryIII, J-E. Dalen, and T.R. Jackson,
`“Clinicians’ Responses to Direct-to-Consumer
`Advertising of Prescription Medications,” Ar
`chives of Internal Medicine 163, no. 15 (2003): 1808-
`1812; and J.S. Weissman et al., “Consumers’ Re-
`ports on the Health Effects of Direct-to-Con-
`sumer Drug Advertising,” Health Affairs 22 (2003).
`w82-w95 (published online 26 February 2003;
`10.1377/hithaf£w3.82).
`Weexplored alternative measures of DTC adver-
`tising. In one, we included current-month and
`six-month cumulative lagged ad measures. The
`six-month lagged measures were notsignificant,
`andthe current-month effects were unchanged.
`In a second, we included current-month and
`one- and two-month lagged measures. The pa-
`rameters on the current-month DIC ads were
`largely unchanged, but thesignificance level on
`all measures was loweras a result of collinearity
`between the three monthly measuresincluded in
`the same model.
`
`Mulkzherjeeetal., “Risk of Cardiovascular Events.”
`Theregression results are available as online ex-
`hibits. See Online Supplemental Exhibit 1 at
`http://content.healthaffairs.org/egi/eontent/full/
`25/4/1371/DC1.
`
`See Online Supplemental Exhibit2,ibid.
`See Online Supplemental Exhibit3, ibid.
`See Notes 13 and 14.
`
`Mukherjee et al,, “Risk of Cardiovascular
`Events.” This article had a significant effect in
`models where we estimated the DIC effect
`nonlinearly, but it had an insignificant effect in
`the linear models wepresenthere. Since this was
`not a majorfocus of the paper, we presented the
`morestraightforward linear models here.
`Topol, “Failing the Public Health.”
`Keith, “Regulating Information’; Masson and
`Rubin,“Matching Prescription Drugs”; P.H. Ru-
`bin, “Economics of Prescription Drug Advertis-
`ing,”Journal ofPharmaceutical Finance, Economics, and
`Policy 3, no. 4 (1991): 29-39; and P.H. Rubin and
`J.L. Schrag, “Mitigating Agency Problems by Ad-
`vertising, with Special Reference to Managed
`Health Care,” Southern Economic Journal 66, no. 1
`(1999): 39-60.
`US. Government Accountability Office, Prescrip-
`tion Drugs: FDA Oversight of Direct-to-Consumer Ad-
`vertising Has Limitations (Washington: GAO,
`2002).
`
`10.
`
`ll.
`12.
`
`13.
`14.
`15.
`16.
`
`17.
`18.
`
`19.
`
`HEALTH AFFAIRS - Volume 25, Number 5
`
`1377
`
`
`
`
`
`wes!AAHAqZLOZ‘€YEWUOsuIeyYyYyeoHAq/Buo'suseyeuyeay'jua}u0o//:dyyWO)pspeo|UMog
`
`8
`
`