`in the Proton Pump Inhibitor Market:
`A Case Study
`
`Yite John Lu
`Joel F. Farley
`Richard A. Hansen
`
`ABSTRACT. This case study of the proton pump inhibitor market
`examines prescription volume, promotional spending, and their interre-
`latedness between the years 2000 and 2004. Our results show that share
`of voice and share of market are strongly correlated (r2 values from 0.79
`to 0.91) and that the temporal relationship is clear: share of market
`follows share of voice. Exogenous market factors such as generic entry
`and over-the-counter entry disrupt the relationship between share of
`market and share of voice and influence the decision to advertise. For
`example, generic entry increased pantoprazole (Protonix®) advertis-
`ing (p < 0.05) while over-the-counter availability decreased rabepra-
`zole (Aciphex®) and pantoprazole advertising (p < 0.01 andp < 0.05,
`respectively). In comparing the different promotional media, direct-to-
`
`Yite John Lu is a PharmD candidate and Joel F. Farley, PhD, and Richard A. Hansen,
`PhD, are assistant professors, all in the Division of Pharmaceutical Outcomes and
`Policy, School of Pharmacy, University of North Carolina at Chapel Hill.
`Address correspondence to: Richard A. Hansen, PhD, Division of Pharmaceutical
`Outcomes and Policy, School of Pharmacy, University of North Carolina at Chapel
`Hill, Campus Box 7360, Chapel Hill, NC 27599 (E-mail: rahansen@unc.edu).
`This work was supported by the American Foundation for Pharmaceutical Educa-
`tion New Investigators Program and the Pharmacy Foundation of North Carolina.
`Dr. Hansen is supported by grant K12 RR023248.
`
`Journal of Pharmaceutical Marketing & Management, Vol. 17(3/4) 2006
`Available online at http://jpmm.haworthpress.com
`© 2006 by The Haworth Press, Inc. All rights reserved.
`doi:10.1300/J058v17n03_04
`
`39
`
`Exhibit 1095
`IPR2017-00807
`ARGENTUM
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000001
`
`
`
`40
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`consumer advertising did not statistically significantly increase pre-
`scription volume, but physician-directed advertising was related to an
`additional 43,662 prescriptions for every 1% increase in share of voice
`(p < 0.001). For marketing managers, our study demonstrates the rela-
`tionship between share of voice and share of market and sheds light on
`the relative effectiveness of advertising strategies in the proton pump
`inhibitor market. doi:10.1300/J058v17n03_04 [Article copies available for
`a fee from The Haworth Document Delivery Service: 1-800-HAWORTH.
`E-mail address: <docdelivery@haworthpress.com> Website: <http://www.Haworth
`Press.com> © 2006 by The Haworth Press, Inc. All rights reserved.]
`
`KEYWORDS. Advertising, promotional effectiveness, promotional
`spending, generic entry
`
`INTRODUCTION
`
`Pharmaceutical promotion has undergone considerable scrutiny as
`the cost of prescription medications continues to rise. The pharmaceuti-
`cal industry spent $30 billion on advertising and promotions in 2003,
`and that number remains on an upward trend (1). These promotional
`efforts have been shown to influence which medications physician pre-
`scribe and the way patients use medications (2, 3). Pharmaceutical pro-
`motion is especially important in oligopolistic markets, where only a
`few brands compete. When the difference in therapeutic efficacy be-
`tween competing drugs is small in such markets, companies are able to
`use promotional efforts to push their brand. When a newcomer drug
`enters such an established drug market, it may be difficult for the mar-
`keting team to overcome the pioneer brand’s hold of market share. To
`overcome this barrier to entry, companies often spend large amounts
`of money on promotion (4). However, it is difficult to know how to bud-
`get advertising expenditures and which promotional efforts are most
`effective.
`Pharmaceutical promotion comes in several forms, such as physician
`visits (detailing), direct-to-consumer (DTC) advertising, drug sampling,
`and medical journal advertising. Detailing historically has received
`the largest share of any promotional budget. In a survey of 2,700 phy-
`sicians, the Kaiser Family Foundation showed that three quarters of
`physicians rate information from pharmaceutical representatives as
`either “very” or “somewhat” useful (5). A literature review by Wazana
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000002
`
`
`
`Lu, Farley, and Hansen
`
`41
`
`showed a number of interesting relationships between detailing and
`physician prescribing (6). Physician detailing was shown to cause
`greater prescribing of newer single-source medications and fewer ge-
`nerics, leading to greater medication costs. Furthermore, although de-
`tailing increased physicians’ awareness of medications, it tended to
`cause physicians to prescribe nonrationally. A frequent component of
`detailing involves the provision of free prescription samples to physi-
`cians. In a study performed by Mizik and Jacobson, detailing and sam-
`pling were both shown to have a modest effect in increasing physician
`prescribing (7).
`Despite evidence of effectiveness with physician-directed promo-
`tions, a recent study suggests that managers who continue to increase
`their sales force are experiencing diminished effectiveness with this ap-
`proach (8). As an alternate venue of marketing efforts, DTC advertising
`has witnessed significant expansion during the past decade. To illus-
`trate, spending on DTC advertisements increased from approximately
`$800 million in 1996 to $2.7 billion in 2001 (9). Although spending
`on DTC advertisements has grown over the past decade, DTC adver-
`tisements tend to be focused in a handful of disease areas. Studies which
`have investigated the effect of DTC advertising within these therapeutic
`markets have shown them effective at increasing prescription sales. In
`the nonsedating prescription antihistamine market, for example, DTC
`advertising has been shown to increase both brand share and category
`sales (10). The same study also concluded that DTC advertising had
`a positive synergistic effect with detailing, although the return on
`investment for DTC advertising was not as high as for detailing (10).
`Another medium frequently used by pharmaceutical firms for adver-
`tising is professional journals. A study by PERQ/HCI Research con-
`cluded that journal ads provide positive return on investment, especially
`in conjunction with pharmaceutical detailing (11). Similarly, in the ROI
`Analysis of Pharmaceutical Promotion study, 391 different drugs’ sales
`and marketing data were analyzed monthly from 1995 to 1999 (12). Re-
`gression analysis determined that the median (or average) return on in-
`vestment (ROI) was $1.72 for detailing, $0.19 for DTC advertising, and
`$5.00 for journal advertising. The authors concluded that DTC is over-
`used and journal advertising is underused (12). Conversely, other stud-
`ies have shown that journal advertising does not increase ROI, and
`when analyzed together with promotional efforts at meetings and events,
`journal advertising decreases the ROI of detailing and DTC (10). When
`looking at relative media value, which is defined as “the relative ability of
`the next exposure to communicate a message as part of a normal media
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000003
`
`
`
`42
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`mix,” journal advertising seems to be less effective than DTC advertising
`and detailing (13).
`In reviewing the literature, it seems apparent that each of the different
`types of promotional spending employed by pharmaceutical firms has
`a positive relationship with prescription sales. For most pharmaceutical
`markets, we believe that the amount of money spent on advertising by a
`drug manufacturer in relation to its competitors’ spending will influ-
`ence its prescription market share. In this article, we use this concept to
`evaluate promotional spending in relation to prescription volume in the
`pharmaceutical market using a single drug class as a case study. We ex-
`amine the impact of various forms of promotional spending using a
`share of voice (SOV) and share of market (SOM) approach. SOV is a
`term used to describe the fraction of promotional spending between
`competitors in the same therapeutic class. SOM, which is also known as
`share brand, describes the fraction of prescriptions written for one drug
`in relation to all prescriptions written in that therapeutic class.
`Evaluation of SOV and SOM is based on the premise that competing
`products can affect market share through their promotional voice. This
`is especially important in an oligopolistic, highly competitive market
`because competitors will keep track of each others’ advertising and
`counter by changing their advertising budget (14). For example, if a
`competing company’s brand is the sole DTC advertiser for a given ther-
`apeutic market, then consumers are less likely to become aware of other
`firms’ products. Similarly, if a company has five salesmen for every
`competitor’s salesman, then physicians will have more exposure to the
`first company’s brand. Past studies which have looked into the relation-
`ship between SOV and SOM have shown a high correlation between
`the two (15). Authors have also concluded that gaining SOM requires a
`very high SOV and that losing SOV causes a decrease in SOM (15). Using
`SOM and SOV allows for managers to plan their current and future ex-
`penditures based on the market environment, not inwardly on revenue,
`expense, and profit values.
`As a case study, we examine promotional spending and prescription
`volume data for the proton pump inhibitor (PPI) market (Figure 1),
`where the competing brands are believed to be similar in scientific effi-
`cacy (16). By looking at past trends (between the years 2000 and 2004),
`we test the correlation between SOV and SOM in the PPI market. Be-
`cause promotion is not the only factor likely to influence prescription
`sales, we also investigate the role of other market variables that may
`have influenced advertising expenditures and prescription volume. The
`specific goals of this study are to determine the relationship between
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000004
`
`
`
`Lu, Farley, and Hansen
`
`43
`
`FIGURE 1. Proton Pump Inhibitor Market Timeline.
`
`Aciphex
`
`Nexium
`
`Prevacid
`
`Prilosec
`
`Protonix
`
`Generic
`Prilosec
`
`Prilosec
`OTC
`
`1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
`
`SOM and SOV, to determine the influence of exogenous market factors
`on the decision to advertise, and to examine the relationship between
`advertising and prescription volume conditioned on the effect of these
`exogenous market factors.
`
`METHODS
`
`Data Sources
`
`We studied the PPI market (rabeprazole, Aciphex®, Eisai Inc.; esome-
`prazole, Nexium®, AstraZeneca; lansoprazole, Prevacid®, Tap Pharm;
`omeprazole, Prilosec®, AstraZeneca; Prilosec OTC®, AstraZeneca; panto-
`prazole, Protonix®, Wyeth) by combining data representing US prescrip-
`tion volume, promotional spending, and drug approval dates. Estimates
`of PPI prescription volume (Uniform System of Classification class
`23420) were obtained from IMS Health’s National Prescription Audit
`Plus™ (NPA Plus) for the time period spanning January 2000 thro-
`ugh October 2004. The NPA Plus estimates the number of prescriptions
`dispensed nationally using data collected from 34,000 independent,
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000005
`
`
`
`44
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`chain, mass merchandiser, food store, mail service, and long-term care
`pharmacies. Our analysis is based on a random selection of more than
`20,000 of these pharmacies. All stores contributed prescription drug
`data, but only half of the stores contributed information on over-the-
`counter (OTC) sales. We aggregated data across strengths and product
`formulations and created summary estimates for each molecular entity,
`keeping generic and OTC entities separate from brand-name, prescription-
`only entities.
`Promotional spending data for the PPI class also were obtained from
`IMS Health for a comparable time period. The IMS Health Integrated
`Promotional Services data reflect dollars spent on provider detailing,
`patient samples, journal advertising, and DTC advertising for each
`product in this class. We aggregated promotional spending at the mo-
`lecular level but still kept OTC and generic spending separate. Promo-
`tional data were matched to prescription volume data at this level.
`
`Prescription Volume and Promotional Spending
`
`We graphed trends in prescription volume and advertising expendi-
`tures in the PPI market between January 2000 and October 2004. These
`illustrations were used to visualize market dynamics with regard to pre-
`scription volume and promotional spending. To explore the relationship
`between promotional spending and market share, we calculated SOV
`and SOM for each product during each month for the time period being
`studied. We calculated SOV for specific promotional media (i.e., detail-
`ing, sampling, journal, DTC advertising) and for aggregate promotion
`by dividing product-specific promotional spending by the sum of all
`promotional spending in the market. Similarly, SOM was calculated by
`dividing product-specific prescription volume by the sum of all pre-
`scriptions in the market for corresponding time intervals. We plotted
`the SOV against the SOM by year from 2000 to 2004. For each year, we
`added a regression line forcing the intercept through zero. The r2 value
`of this regression line was used as a rough estimate of the strength of
`this relationship. We also plotted the relationship between SOV and
`SOM over time for each drug. By doing so, we were able to visualize
`not just the correlation between the two parameters but also the time
`effect SOV can have on SOM and vice versa. This time model also
`helped illustrate the potential impact of dynamics other than SOV in the
`market, such as generic and OTC entry.
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000006
`
`
`
`Lu, Farley, and Hansen
`
`45
`
`Predictive Models
`
`One of the goals of this paper was to examine the impact of various
`promotional strategies on prescription volume. A primary concern in
`the estimation of prescription volume resulting from advertising was
`the direction of causality. Previous analyses have attempted to control
`for this endogeneity issue by using an instrumental variables approach
`(17). Although effective in removing endogeneity from the advertising/
`volume relationship, we were unable to ascertain an effective instru-
`ment for this study. Instead, we attempted to investigate for the presence
`of endogeneity using interrupted time-series analyses (18, 19). These
`models examined the impact of two separate external market dynamics
`(generic and OTC PPI market penetration) on the level of advertising
`for both DTC and physician-directed promotions. The basic regression
`model for the interrupted time-series analyses was:
`
`4+
`
`b Month + b Gen Entry + b Gen Month
`y = b
`1
`2
`3
`0
`+ b OTC En
`try + b OTC Month
`5
`
`The dependent variable (y) was modeled separately for DTC and
`physician-directed advertising. Month is a continuous variable equal to
`the month of observation from month 1 (January 2000) to month 58
`(October 2004). Gen Entry is a dummy variable equal to 0 for months
`preceding entry of generic omeprazole and 1 for any month after. Gen
`Month is a continuous variable counting the number of months after ge-
`neric omeprazole market entry. OTC Entry is a dummy variable equal
`to 0 for months preceding Prilosec OTC entry and 1 for any month after.
`OTC Month is a continuous variable counting the number of months
`after Prilosec OTC entry. In the model, b2 and b4 capture a one-time
`change in prescription volume or promotional spending associated with
`OTC or generic entry into the PPI market, respectively. Change in the
`level of monthly promotional spending associated with generic and
`OTC entry are captured by coefficients b3 and b5, respectively. We spe-
`cified Prais-Winsten regression procedures to control for autocorre-
`lation and robust standard errors to control for heteroscedasticity in
`each time-series regression.
`We next used a fixed effects panel model to estimate the effect of in-
`dividual PPI advertising behavior on prescription volume. The bene-
`fit of using a fixed effects model for this study was that it allowed for
`estimation of the effect of advertising among each individual PPI
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000007
`
`
`
`46
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`medication on prescription volume while controlling for individual PPI
`heterogeneity patterns. The model was specified to allow assessment of
`multiple types of promotional voice on PPI sales. The basic model was
`summarized as:
`
`Y
`it
`
`=
`
`a
`+
`
`+
`Month
`i
`b
`SMPL
`it 5
`
`b
`+
`DTC
`it 1
`b
`+
`GEN
`t 6
`
`b
`+
`JOURN
`it 2
`b
`+
`+
`MKT
`it 7
`
`+
`
`DTAIL
`
`b
`it 4
`
`b
`it 3
`e
`
`it
`
`where Yit represents monthly medication sales for each PPI drug (i) in
`month t; ␣
`i is the intercept for each individual PPI drug; Monthit
`indicates the observation month ranging from 1 to 58; DTCit repre-
`sents the promotional voice for DTC advertising; JOURNit represents
`the promotional voice for journal advertising; DTAILit represents the
`promotional voice related to physician detailing; SMPLit represents
`the promotional voice related to the provision of samples to physicians;
`GENt is a 0-1 dummy variable representing months prior to and after
`generic entry of omeprazole, respectively; and MKTit indicates the num-
`ber of years the PPI has been on the market for each monthly observation.
`
`2 to 5 which represented PPI sales
`The coefficients of interest were 
`conditioned upon the effect of the other advertising covariates.
`We tested for the presence of multicollinearity among advertising
`variables a priori using Pearson’s correlations and found potential
`collinearity between detailing and sampling (r = 0.83). We therefore
`reorganized the dependent variables into physician-directed advertising
`(detailing, sampling, and journal) and consumer-directed advertising.
`A separate regression using these two advertising variables was also
`specified using fixed effects panel estimation. All analyses were con-
`ducted using STATA-SE version 9.0.
`
`RESULTS
`
`Descriptive Analysis of Prescription Volume
`and Promotional Spending
`
`After aggregating annual prescription data, we looked for trends and
`prominent changes in the PPI market (Figure 2). Prescription volume
`for Prilosec–the market leader in 2000 with 32 million prescriptions
`written–steadily decreased until it was the least prescribed PPI with
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000008
`
`
`
`Lu, Farley, and Hansen
`
`47
`
`FIGURE 2. Annual Prescription Volume.
`
`2 million prescriptions written in 2004. Nexium entered the market in
`February of 2001, and its prescription volume rose to 24 million by 2004,
`giving it the second largest market share behind Prevacid. Protonix’s
`prescription volume also increased during this time period after be-
`ing introduced to the market in February of 2000. Aciphex and Prevacid’s
`prescription volume remained planar during the four-year period.
`We analyzed annual promotional expenditures in the PPI market,
`noting changes that might explain trends in prescription volume and
`changes which might be explained by other market variables, such as gen-
`eric and OTC entry (Figure 3). AstraZeneca, the maker of both Prilosec
`and Nexium, stopped promotional spending for Prilosec after its patent
`expiration in November of 2002. In contrast, Nexium came to market in
`February of 2001, and AstraZeneca launched an aggressive promo-
`tional campaign. Promotional expenditures for Prevacid, Protonix, and
`Aciphex steadily increased during this time period. However, in 2003
`when Prilosec OTC was introduced, promotional spending on Prevacid
`and Aciphex decreased slightly, while spending on Protonix continued
`to increase.
`The strength of correlation between SOV and SOM in the PPI market
`was assessed both graphically and through an r2 value. We observed a
`relatively unstable correlation between SOV and SOM for the entire
`PPI market during the years of 2001 through 2003. This period was
`likely one of transition within the PPI market with Nexium’s market
`entry in February of 2001 and Prilosec’s patent expiration in November
`of 2002. Given instability within the PPI market, we conducted our
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000009
`
`
`
`48
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`FIGURE 3. Annual Advertising Dollars.
`
`analysis excluding Prilosec and Nexium data between the years 2001
`and 2003. After exclusion of Prilosec and Nexium data in the years
`2001-2003, recalculated r2 values showed strong correlations between
`SOV and SOM in the PPI market for years 2000-2004. The r2 values
`were 0.9029, 0.9084, 0.8464, 0.7894, and 0.8382 for the years of 2000,
`2001, 2002, 2003, and 2004, respectively, with the intercept forced
`through zero. The relationship between SOV and SOM in the PPI
`market was plotted for years 2000 and 2004 (Figure 4). We did not graph
`the years 2001-2003 because of the transition being made with Prilosec
`coming off patent and Nexium being introduced to the market.
`Graphing the trends of SOV and SOM for each PPI drug over time
`allowed us to see the similarities between the two values (Figure 5). For
`Nexium, a sharp rise in SOV was followed by a rise in SOM. In con-
`trast, Prilosec experienced a sharp drop in SOV followed by decreasing
`SOM. Prilosec and Nexium exemplify the temporal phenomenon that
`a change in SOV is followed by a change in SOM of the same direction.
`Prevacid and Aciphex both had small variations of SOV and SOM over
`the five-year period. Protonix had an increasing trend toward a larger
`SOV and SOM.
`
`Predictive Models
`
`Figure 6 shows graphically the effect of generic omeprazole entry on
`DTC advertising. Visual inspection of this figure suggests no association
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000010
`
`
`
`Lu, Farley, and Hansen
`
`49
`
`FIGURE 4. Share of Voice versus Share of Market for the Years 2000
`and 2004.
`
`between generic or OTC entry and the level of spending for DTC ad-
`vertising among any of the proton pump inhibitors examined. This is
`confirmed in the interrupted time-series regressions seen in Table 1. Non-
`significance of coefficients from the interrupted time-series analysis of
`DTC advertising and prescription volume suggests no association be-
`tween either generic or OTC entry of omeprazole and spending on DTC
`advertising.
`Unlike the DTC advertising analysis, Figure 7 suggests an association
`between generic and OTC penetration and the level of physician-dire-
`cted promotion for certain PPI drugs. There was evidence of a potential
`drop in physician-directed promotions associated with generic PPI en-
`try for Nexium and Prevacid. Similarly, OTC penetration appears to
`have been associated with a decline in physician-directed promotion for
`Protonix and Aciphex.
`Visual observations from Figure 7 were tested for significance using
`the interrupted time-series analysis shown in Table 2. Significance of
`the variable “Month” suggests an increase in monthly physician-directed
`advertising prior to generic market penetration for Aciphex, Protonix,
`and Nexium. Nonsignificance of the “Generic Entry” and “OTC Entry”
`coefficients suggests no evidence of a disruption in the continuity of
`the time trend associated with generic or OTC market entry. Looking
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000011
`
`
`
`50
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`FIGURE 5. Timeline Plots of SOM and SOV for the PPIs.
`
`0.6
`0.5
`0.4
`0.3
`0.2
`0.1
`0
`0.6
`0.5
`0.4
`0.3
`0.2
`0.1
`0
`0.6
`0.5
`0.4
`0.3
`0.2
`0.1
`0
`0.6
`0.5
`0.4
`0.3
`0.2
`0.1
`0
`0.6
`0.5
`0.4
`0.3
`0.2
`0.1
`0
`
`Aciphex
`
`Prevacid
`
`Protonix
`
`Nexium
`
`Prilosec
`
`Jan-01
`Jan-00
`Jan-03
`Apr-01
`Jan-02
`Apr-00
`Apr-03
`Apr-02
`Jan-04
`Apr-04
`Jul-01
`Jul-00
`Jul-03
`Jul-02
`Jul-04
`Oct-01
`Oct-00
`Oct-03
`Oct-02
`Oct-04
`
`SOV
`
`SOM
`
`at the “Generic Month” coefficient, there was a significant increase
`in monthly physician advertising for Protonix after generic penetration
`of omeprazole. This contrasts to the “OTC Month” coefficient which
`showed a decline in physician advertising associated with OTC penetra-
`tion of Prilosec.
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000012
`
`
`
`Lu, Farley, and Hansen
`
`51
`
`FIGURE 6. Monthly Direct-to-Consumer Advertising.
`
`TABLE 1. Interrupted Time Series Analyses of PPI Level DTC Advertising.
`
`Intercept
`
`Month
`
`Generic entry
`
`Generic month
`
`OTC entry
`
`OTC month
`
`Prevacid
`
`796,703.62
`(1,326,347.64)
`
`176,521.86*
`(87,887.94)
`
`1,905,720.44
`(3,035,968.04)
`
`193,242.56
`(540,006.65)
`
`⫺1,347,067.42
`(5,252,351.18)
`
`⫺475,106.58
`(687,493.88)
`
`Aciphex
`
`⫺1,755.97
`(1,828.37)
`
`150.49
`(154.97)
`
`⫺3,298.95
`(3,471.38)
`
`⫺158.90
`(154.15)
`
`22.50
`(26.96)
`
`8.41
`(9.20)
`
`Protonix
`
`⫺61.97
`(43.07)
`
`4.04
`(2.76)
`
`⫺214.26
`(227.17)
`
`35.42
`(51.45)
`
`Nexium
`
`8,611,075.78
`(6,935,354.08)
`
`550,797.16
`(457,968.45)
`
`⫺4,186,514.36
`(4,168,089.10)
`
`⫺303,135.83
`(769,488.09)
`
`⫺3,368.90
`(2,231.66)
`
`731,431.22
`(5,106,663.78)
`
`620.20
`(352.26)
`
`⫺289,946.87
`(697,961.38)
`
`*p⬍ 0.05
`Standard error in parenthesis
`
`Given the significance of exogenous market factors on the decision
`to advertise, we next ran the fixed effects regression models of various
`promotional advertisements on predicted prescription volume. Table 3
`presents results of the fixed effects panel estimation of promotional
`voice on prescription volume. Because promotional voice is a ratio
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000013
`
`
`
`52
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`FIGURE 7. Monthly Physician-Directed Advertising.
`
`TABLE 2. Interrupted Time-Series Analyses of PPI Level Physician Advertising.†
`
`Intercept
`
`Month
`
`Generic entry
`
`Generic month
`
`OTC entry
`
`OTC month
`
`Prevacid
`28,967,975.17***
`(1,375,080.39)
`515,960.87
`(74,369.81)
`⫺11,355,962.91
`(11,059,916.23)
`1,483,058.70
`(1,809,318.16)
`⫺16,056,247.26
`(8,972,307.16)
`⫺1,933,667.71
`(1,916,154.80)
`
`Aciphex
`12,157,332.39***
`(1,169,781.03)
`359,610.34***
`(57,154.52)
`⫺6,431,390.79
`(4,354,811.20)
`735,699.24
`(610,063.24)
`3,940,401.73
`(3,383,566.29)
`⫺1,932,238.81**
`(716,718.80)
`
`Protonix
`10,566,037.78***
`(1,734,602.83)
`210,656.18*
`(92,737.69)
`1,604,713.93
`(2,191,391.24)
`978,979.31*
`(458,437.04)
`96,749.00
`(3,312,295.55)
`⫺1,165,453.39*
`(502,119.16)
`
`Nexium
`28,634,858.55***
`(2,148,702.90)
`1,284,942.39***
`(191,805.49)
`2,666,336.37
`(9,850,139.54)
`⫺2,307,075.93
`(1,370,515.46)
`⫺2,669,918.09
`(7,677,021.55)
`793,998.56
`(1,477,266.51)
`
`†Physician advertising is the sum of detailing, samples, and journal advertising.
`*p⬍ 0.05, **p⬍ 0.01, ***p⬍ 0.001
`Standard error in parenthesis
`
`variable whereby 1 equals 100% of PPI promotional voice and 0 repre-
`sents 0% of PPI promotional voice, for convenience we chose to inter-
`pret coefficients on the basis of a 1% increase in promotional voice for
`each variable. Furthermore, to aid in interpretation, we also ran the model
`using advertising dollars as the independent variable for each type of
`promotion to facilitate ease in interpretation of results (data not shown).
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000014
`
`
`
`Lu, Farley, and Hansen
`
`53
`
`TABLE 3. Fixed Effects Panel Estimation of Prescription Volume Conditioned
`on Promotional Voice.
`
`Full Model Estimation
`
`DTC/Physician Estimation
`
`95% Confidence
`Interval
`
`b
`
`Intercept
`
`205,338.84
`
`Time
`
`16,891.71***
`
`Generic entry
`
`⫺166,532.54
`
`OTC entry
`
`⫺216,689.57*
`
`DTC advertising
`
`⫺155,655.91
`
`Physician
`advertising†
`
`Detailing
`
`⫺1,520,198.40*
`
`Samples
`
`4,650,951.26***
`
`Journal
`
`654,688.59*
`
`(⫺140,865.87,
`551,543.55)
`
`(9,918.25,
`23,865.17)
`
`(⫺342,890.88,
`9,825.80)
`
`(⫺409,197.93,
`⫺24,181.20)
`
`(⫺422,290.45,
`110,978.64)
`
`b
`
`⫺16,607.85
`
`19,765.64***
`
`⫺204,673.18*
`
`⫺248,237.23*
`
`6,728.74
`
`4,366,237.60***
`
`95% Confidence
`Interval
`
`(⫺288,611.57,
`255,395.86)
`
`(13,412.44,
`26,118.84)
`
`(⫺373,638.33,
`⫺35,708.03)
`
`(⫺445,535.94,
`⫺50,938.52)
`
`(⫺310,358.64,
`323,816.12)
`
`(3,571,049.06,
`5,161,426.13)
`
`(⫺2,942,833.67,
`⫺97,563.12)
`
`(3,752,154.23,
`5,549,748.30)
`
`(150,370.53,
`1,159,006.65)
`
`⫺
`
`⫺
`
`⫺
`
`⫺
`
`⫺
`
`⫺
`
`†
`
`Physician advertising is the sum of detailing, samples, and journal advertising.
`*p ⬍ 0.05, **p ⬍ 0.01, ***p ⬍ 0.001
`
`The results from Table 3 suggest nonsignificance between DTC pro-
`motional voice and prescription volume among individual PPI medica-
`tions. This contrasts to significance of the coefficient for sample voice
`which suggests an increase of approximately 46,510 prescriptions for
`each 1% increase in promotional voice. Using advertising spending in
`dollars as the dependent variable, for each additional dollar spent on
`samples 0.035 additional prescriptions were dispensed. Interestingly,
`there was evidence of a decline in prescription volume attributable to
`physician detailing, furthering our concern of multicollinearity among
`physician-directed types of promotion.
`Given the potential for multicollinearity in the models described
`above, we also ran fixed effects regression models dividing the type
`
`Journal of Pharmaceutical Marketing & Management Downloaded from informahealthcare.com by Hochschule Osnabruck on 04/14/15
`
`For personal use only.
`
`000015
`
`
`
`54
`
`JOURNAL OF PHARMACEUTICAL MARKETING & MANAGEMENT
`
`of promotion into physician-directed media (journal, detailing, and
`sample media) and DTC media. This reduced model again showed
`nonsignificance between DTC promotional voice and prescription volume.
`However, the coefficient for physician-directed advertising was statis-
`tically significant. A 1% increase in physician-directed promotional
`voice was associated with an increase of 43,662 additional PPI prescrip-
`tions dispensed in the reduced model. Using advertising spending in
`dollars as the dependent variable, for each additional dollar spent on
`physician advertising 0.037 additional prescriptions were dispensed.
`Interpretation of the coefficient “Time” suggests an increase in the
`number of individual PPI medications dispensed over time. The coeffi-
`cients “Generic Entry” and “OTC Entry” suggest a decline in the num-
`ber of prescriptions dispensed during the time periods associated with
`generic and OTC entry into the PPI market, respectively.
`
`DISCUSSION AND CONCLUSION
`
`Examining the relationship between share of promotional spending
`and share of prescription volume in the PPI market revealed a relatively
`strong association (r2 ⱖ 0.8). This is consistent with analyses conducted
`by Jones and Schroer, where SOM was shown to routinely follow
`changes in SOV (15, 20). These authors contend that the two values
`cannot be drastically different or market share can no longer be main-
`tained. In other words, any attempt to diminish advertising once adver-
`tising has been initiated can significantly reduce a firm’s share of the
`market. Our analysis of the relationship between SOV and SOM in the
`PPI market is consistent with this, although the strength of the relation-
`ship over time was influenced by other market dynamics. By plotting
`the changes of SOV and SOM over time for each individual drug, we
`are able to visually observe these relationships.
`The influence of market dynamics (e.g., new market entrant, generic
`entrant, OTC entrant) on the relationship between SOV and SOM ap-
`pears to have a near-term effect and then stabilize over time. For ex-
`a