The Debate on Influencing Doctors' Decisions: Are Drug Characteristics the Missing Link?
`Author(s): Sriram Venkataraman and Stefan Stremersch
`
`Source: Management Science, Vol. 53, No. 11 (Nov., 2007), pp. 1688-1701
`
`Published by:
`INFORMS
`
`Stable URL:
`http://www.jstor.org/stable/20122327
`Accessed: 04-02-2016 21:01 UTC
`
`
`
`
`
`Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
`info/about/policies/terms.jsp
`
`
`JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
`in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
`For more information about JSTOR, please contact support@jstor.org.
`
` INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Management Science.
`
`
`
`
`http://www.jstor.org
`
`This content downloaded from 208.85.77.1 on Thu, 04 Feb 2016 21:01:17 UTC
`All use subject to JSTOR Terms and Conditions
`
`SENJU EXHIBIT 2297
`LUPIN v. SENJU
`IPR2015-01097
`
`Page 1 of 15
`
`

`
`MANAGEMENT SCIENCE M3H.
`
`11, November
`Vol. 53, No.
`1688-1701
`2007, pp.
`DOI i0.l287/mnsc.l070.0718
`ISSN 0025-19091 EissN 1526-55011071531111688 ? 2007 INFORMS
`
`on Influencing Doctors' Decisions:
`The Debate
`Link?
`Are Drug Characteristics
`the Missing
`
`Goizueta
`
`Business
`
`Fuqua
`
`School
`
`of Business,
`
`Sriram Venkataraman
`School, Emory University,
`Atlanta, Georgia
`Stefan Stremersch
`North Carolina
`Duke University,
`Durham,
`
`30322,
`
`svenka2@emory.edu
`
`27708,
`
`stefan.stremersch@duke.edu
`
`has
`
`come
`
`In fact,
`
`there
`
`patient
`
`requests
`
`policy
`public
`effectiveness
`
`on patients'
`treatment
`under
`increased
`by physicians
`scrutiny.
`public
`making
`is a
`on
`fair amount
`actions
`firms
`of pharmaceutical
`of debate
`the
`effects
`of marketing
`toward
`Decision
`physicians
`find a strong and positive
`and their impact on physician prescription behavior. While
`some scholars
`influence
`even
`some
`find negative
`is also
`find
`and
`only moderate
`of marketing
`others
`effects.
`Debate
`effects,
`actions,
`(such as patient
`on the role of other
`in physician decision making, both on
`influencers
`requests)
`mounting
`and sample dispensing. The authors argue that one factor that may
`in this debate
`tip the balance
`prescriptions
`a
`as a
`is the
`role of drug
`such
`effectiveness
`and
`side
`effects.
`characteristics,
`drug's
`drug's
`a unique
`as
`data
`that marketing
`show
`and
`set,
`they
`Using
`efforts?operationalized
`symposium
`detailing
`patient requests do affect physician decision making differentially across
`of firms to physicians?and
`meetings
`to marketing
`find
`that
`the
`of physicians'
`brands. Moreover,
`decision
`efforts
`and
`responsiveness
`they
`making
`effectiveness
`the drug's
`and
`side
`clear
`paper
`upon
`presents
`guidelines
`depends
`envisions
`and
`of drug
`and managerial
`the
`of
`the
`role
`such
`characteristics,
`practice
`study
`in this surging public debate.
`lead to valuable
`and side effects, may
`insights
`Key words: physician decision making; marketing
`effort; patient request; drug effectiveness;
`prescription;
`sample dispensing; detailing; pharmaceuticals;
`public policy
`sampling;
`9, 2005. This paper was
`History: Accepted by Linda V. Green, public sector applications;
`received November
`in Articles
`for 3 revisions. Published online
`the authors 6 months
`in Advance September 14, 2007.
`with
`
`effects.
`
`This
`
`that
`
`for
`as
`
`side effect; drug
`
`the drugs
`regarding
`come under
`increased
`in the United
`rise
`
`sharply
`and
`governments
`to
`factors
`that may
`decision making.
`drug
`are market
`attention
`firms
`
`1.
`Introduction
`Decision making
`by physicians
`treat patients with
`has
`they
`scrutiny As pharmaceutical
`expenses
`countries
`and other
`States
`developed
`with
`of
`the population,
`aging
`turn
`their attention
`regulators
`affect physician
`(adversely)
`that draw
`Factors
`particular
`ing actions
`of pharmaceutical
`targeted directly
`for a
`at physicians
`and patient
`requests
`specific
`a public
`in
`"There has been
`outcry
`especially
`drug.
`over
`doctors
`between
`the cozy
`America,
`relationship
`are illegal, others
`Some practices
`and drug companies.
`are simply part of
`flat
`trio of
`the customary
`food,
`2005, p. 9). The
`(The Economist
`tery, and
`friendship"
`for the
`of Merck
`for its marketing
`actions
`prosecution
`case
`is a very recent, heavily publicized,
`in
`drug Vioxx
`(The Wall Street Jour
`that regulators
`take notice
`point,
`nal 2006).
`Pharmaceutical
`
`a huge
`ever
`and
`firms
`spend
`on detailing
`calls by
`visits
`(sales
`increasing
`budget
`The
`and meetings.
`representatives)
`pharmaceutical
`in the pharmaceutical
`number
`of sales representatives
`a six-fold
`in the
`increase
`industry has undergone
`
`last
`
`at
`
`to approximately
`20 years
`today, and 77% of
`100,000
`are planning
`to further
`the companies
`their
`expand
`in 2005
`sales
`force
`2004). Detailing
`(30.6%)
`(Hradecky
`to 81% of
`amount
`to physicians
`and sampling
`(50.6%)
`in 2000
`firms
`promotion
`spending
`by pharmaceutical
`et al. 2003).
`In addition,
`increas
`(Rosenthal
`patients
`ingly request a certain brand of drug
`from
`the physi
`one
`in three patients
`In the United
`cian.
`States,
`some point has asked about a drug by name
`(Calabro
`It is a commonly
`that such patient
`held belief
`2003).
`are often
`by direct-to-consumer
`requests
`triggered
`at an all-time high of $4
`(DTC) advertising,
`presently
`in the United
`States
`billion
`(Edwards
`2005).
`of a physician,
`The most
`decision
`important
`it concerns
`if
`practice
`general
`physicians,
`cially
`to use
`in treatment
`of patients. The deci
`which
`drug
`on drug
`can be wit
`treatment
`sions physicians make
`behavior. They
`nessed
`prescription
`through observing
`can also be observed
`as samples
`in sampling
`behavior,
`a prescription
`are provided
`(as a finan
`together with
`or instead of a prescription
`to the patient),
`cial subsidy
`(as a trial, e.g., when
`about drug-patient
`uncertainty
`interaction
`is high). Sample dispensing
`by physicians
`is an
`is rarely
`studied.
`important
`physician
`Sampling
`as well,
`to pre
`lead
`decision
`because
`sampling may
`
`espe
`
`is
`
`1688
`
`This content downloaded from 208.85.77.1 on Thu, 04 Feb 2016 21:01:17 UTC
`All use subject to JSTOR Terms and Conditions
`
`Page 2 of 15
`
`

`
`Venkataraman
`
`Management
`
`The Debate
`and Stremersch:
`Science 53(11), pp. 1688-1701, ?2007
`
`on Influencing Doctors' Decisions
`INFORMS 1689
`
`and Koenigs
`treatment
`scribed
`(Morelli
`long-term
`for
`thus have
`berg 1992), and
`consequences
`significant
`firms and public health.
`pharmaceutical
`turned
`have
`and regulators
`Academic
`scholars
`actions of pharmaceu
`how both marketing
`assessing
`firms and patient
`influence
`tical
`requests
`physician
`on drug
`decision making
`treatment, both prescription
`this point, most
`behavior. At
`and
`research
`sampling
`on how marketing
`tar
`has been
`conducted
`efforts
`to physicians
`affect physicians'
`prescription
`geted
`as a
`Patient
`behavior.
`factor
`requests
`influencing
`as a physi
`decision making
`and
`sampling
`physician
`so far.
`have
`less attention
`cian decision
`received
`stream
`in the relatively
`Even
`research
`developed
`con
`on marketing
`efforts and prescription
`behavior,
`some
`recently. While
`studies
`troversy has been
`raised
`has a pos
`et al. 2001)
`find
`that detailing
`(e.g., G?n?l
`effect on prescriptions
`itive and
`written,
`significant
`find either a very modest
`other
`studies
`effect
`(Mizik
`and Jacobson 2004) or no effect at all (Rosenthal
`et al.
`on brand
`or
`sales.
`of detailing
`2003)
`prescriptions
`the rea
`et al.
`that
`(2004) posited
`Recently,
`Leeflang
`son for these
`is that prior models
`results
`incongruent
`in that
`the effect of
`be misspecified,
`may
`they pool
`across brands, while
`brands
`expenditures
`marketing
`in fact differ
`in the extent
`to which
`may
`physicians
`are responsive
`a firm
`to the marketing
`expenditures
`to promote
`makes
`them
`through detailing, meetings
`or other promotional
`the
`instruments.
`This
`is also
`stance we
`take
`in the present
`study.
`such as
`This
`that drug
`characteristics,
`study posits
`are a potential
`source
`and effectiveness,
`side effects
`in the respon
`if any,
`for brand-specific
`differences,
`of physicians'
`brand prescription
`siveness
`behavior
`to marketing
`firms. Our
`efforts by pharmaceutical
`to resolving
`the controversy
`contribute
`insight may
`on how marketing
`of pharmaceutical
`efforts
`firms
`affect prescription
`behavior. We also examine
`the role
`in
`of
`these drug
`the effect
`of other
`characteristics
`as patient
`and other
`such
`"influencers,"
`requests,
`such as
`A
`decisions,
`sample
`physician
`dispensing.
`from our empirical
`coherent
`arises
`picture
`analysis.
`the influ
`We
`find that drug characteristics
`affect both
`ence patients
`(in this study
`through patient
`requests)
`as
`as well
`the pharmaceutical
`(in this
`firms
`study
`to physi
`their marketing
`efforts
`through
`targeted
`exert on physician
`decision making,
`both
`cians)
`a physician's
`and a physician's
`prescription
`decisions. Thus, we underscore
`dispensing
`tance of
`characteristics
`drug
`including
`of
`influence
`firms and/or
`patients
`by
`a physician makes.
`treatment decision
`this study
`is the first attempt
`between
`influencers
`(e.g., detailing
`firm) and drug
`characteristics
`pharmaceutical
`on physician
`behavior.
`efficacy)
`
`edge,
`actions
`
`sample
`the impor
`in any
`study
`on any drug
`By our knowl
`to test
`inter
`for
`the
`
`by
`
`(e.g.,
`
`to
`
`in
`
`a unique
`this study, we have
`data
`For
`composed
`sources. The
`set that matches
`first contains
`three data
`on manufacturers'
`detailed
`information
`vis
`detailing
`at manufactur
`its to physicians,
`attendance
`physician
`ers' meetings,
`of patients
`and drug
`for 2,774
`requests
`as the num
`States, as well
`in the United
`physicians
`ber of prescriptions
`and
`written
`samples
`on a monthly
`by each of
`these physicians
`sets we
`and
`third data
`second
`composed
`contain data on
`These
`(1) effectiveness,
`in our database.
`effects of each drug
`The next
`back
`theoretical
`section
`the
`discusses
`our data
`set and
`3 describes
`Section
`the
`ground.
`we use. Section
`our
`4 presents
`analysis methodology
`our findings,
`results. Section 5 discusses
`their
`implica
`and management
`for public policy
`and
`tions
`practice,
`the study's
`limitations.
`
`dispensed
`basis. The
`ourselves.
`and
`(2) side
`
`2. Background
`on
`This
`first discusses
`the
`research
`section
`prior
`efforts on
`effects of pharmaceutical
`firms' marketing
`on
`and
`their effects
`physician
`prescribing
`explores
`until
`behavior
`the physician,
`which
`sampling
`by
`Second, we discuss
`the
`remained
`unstudied.
`today
`limited prior research on the effects of patient
`requests
`on physicians'
`and
`prescription
`sample-dispensing
`behavior. Third, we
`the role
`char
`that drug
`explore
`acteristics may play on physician
`and
`their
`decisions
`interactions with
`efforts and patient
`firms' marketing
`requests. Fourth, we discuss
`vari
`any other
`relevant
`that may
`ables
`and
`affect physicians'
`prescription
`behavior.
`sample-dispensing
`
`2.1. Effects
`Efforts
`
`of Pharmaceutical
`Firms' Marketing
`on Physician
`and
`Prescription
`Behavior
`Sample-Dispensing
`can divide
`One
`the
`the prior
`literature
`regarding
`on
`effect of pharmaceutical
`efforts
`firms' marketing
`individual
`into two
`behavior
`physicians'
`prescription
`one
`effects and one
`streams, namely,
`finding positive
`effects, at best. We discuss
`each stream
`finding mixed
`in
`turn.
`G?n?l
`
`gunta
`ceutical
`
`et al.
`(2001) and Manchanda
`and Chinta
`find
`that marketing
`efforts by pharma
`(2004)
`to the physician
`affect
`companies
`positively
`by a physician,
`there are
`issued
`but
`prescriptions
`returns
`to detailing. Manchanda
`et al.
`diminishing
`find
`that detailing
`affects prescrip
`(2004)
`positively
`tion behavior,
`but
`that high-volume
`physicians, while
`are
`to detailing,
`being detailed more,
`less responsive
`as compared
`to
`low-volume
`physicians.
`Narayanan
`and Manchanda
`(2004) find
`that while
`influ
`detailing
`in an overwhelming
`num
`enced physicians
`positively
`there was
`of
`ber
`cross-sectional
`cases,
`significant
`in physician
`and
`responsive
`ness
`et al.
`that
`find
`(2005)
`
`temporal heterogeneity
`to detailing.
`Janakiraman
`
`This content downloaded from 208.85.77.1 on Thu, 04 Feb 2016 21:01:17 UTC
`All use subject to JSTOR Terms and Conditions
`
`Page 3 of 15
`
`

`
`Venkataraman
`
`and Stremersch:
`1690 Management
`
`on Influencing Doctors' Decisions
`The Debate
`Science 53(11), pp. 1688-1701, ?2007
`INFORMS
`
`are
`
`to both
`
`responsive
`nonpersistent
`physicians
`while
`and
`persistent
`symposium meetings,
`detailing
`are only
`to symposium meet
`responsive
`physicians
`(sales or
`that use aggregate
`studies
`ings. Also, many
`a positive
`effect of detailing
`find
`data
`prescription)
`on drug
`and Desiraju
`sales
`2005;
`(e.g., Chintagunta
`et al. 2004, 2005; Neslin
`2001; Rizzo
`1999).
`Narayanan
`to
`firms' market
`the prior
`literature,
`According
`a positive
`effect on prescription
`ing efforts may have
`visits or symposium meet
`because
`behavior
`detailing
`on efficacy
`to the physician
`information
`ings provide
`(G?n?l et al. 2001).
`In line
`and side effects of the drug
`a long
`in economics
`and
`(e.g., Becker
`with
`tradition
`and Shapiro
`1993, Grossman
`1984, Leffler
`Murphy
`et al. (2005) have argued
`that firms'
`1981), Narayanan
`actually have both an informa
`efforts may
`marketing
`and a
`tive role
`uncertainty)
`(e.g., reducing
`cognitive
`role
`affect).
`inducing positive
`(e.g.,
`persuasive
`find
`that marketing
`and
`Mizik
`Jacobson
`(2004)
`to
`the physi
`efforts by pharmaceutical
`companies
`affect new
`issued
`cian positively
`prescriptions
`by
`are very modest.
`a physician,
`the effect
`sizes
`but
`about a strong and positive
`cast doubt
`Their
`findings
`on physician
`effect of marketing
`efforts
`prescription
`as evidenced
`in studies using
`and
`behavior
`aggregate
`and Vanden Abeele
`data. Parsons
`individual-level
`behavior
`find
`that physician
`(1981)
`prescription
`to marketing
`efforts by pharma
`quite unresponsive
`to the physician,
`and
`sales
`calls may
`firms
`ceutical
`even have a negative
`et al. (2003) did
`effect. Rosenthal
`at
`not
`for detailing
`find robust and significant
`effects
`brand
`level.
`the
`individual
`there has been no
`the best of our knowledge,
`To
`that examines
`the effect of marketing
`research
`prior
`efforts on sample-dispensing
`the physi
`behavior
`by
`for our purposes
`cian. The most
`useful
`research
`that exam
`in medicine
`literature
`the sparse
`probably
`have when
`the motives
`ines
`dispensing
`physicians
`to their patients. Motives
`that have been
`free samples
`(2) conve
`cited are:
`(1) financial
`for patients;
`savings
`initiate
`(4) demon
`nience;
`(3)
`therapy
`immediately;
`use
`to patients;
`strate
`the appropriate
`(5) adjust
`is pur
`full prescription
`doses before
`the
`prescribed
`or adverse
`early effectiveness
`chased; and
`(6) evaluate
`et al. 2003).
`(Chew et al. 2000, Duffy
`effects
`
`is
`
`is
`
`2.2. Effects
`
`Most
`
`on Physician
`of Patient Requests
`Behavior
`and Sample-Dispensing
`Prescription
`the effects of patient
`that studies
`of the research
`on physician
`is driven by
`decision making
`requests
`in the
`of DTC
`the growing
`advertising
`importance
`the FDA's
`1997 Draft
`after
`United
`States, mostly
`on DTC
`advertisements.
`DTC
`broadcast
`Guidance
`is an
`driver of patient
`requests
`important
`studied
`et al. 2003), and
`scholars have only
`triggered by DTC advertising,
`requests when
`reason.
`than
`other
`any
`
`advertising
`(Mintzes
`patient
`rather
`
`In a study using
`that por
`standardized
`patients
`27% of all patients
`request
`trayed major depression,
`a prescription
`ing Paxil
`for
`also
`received
`it, 26%
`an
`and
`47%
`received
`alternative
`antidepressant,
`received no antidepressant,
`while
`only 3% of patients
`the same condition were prescribed
`if they
`with
`Paxil
`did not explicitly
`et al. 2005).
`request Paxil
`(Kravitz
`found a positive
`in other
`rela
`scholars
`Also,
`settings,
`between
`and prescription
`requests
`patient
`tionship
`et al. 2003)
`et al. 2003, Lyles
`2002, Mintzes
`(Kravitz
`et al. 2003). This pos
`referral
`and physician
`(Kravitz
`itive relationship
`is driven
`and
`pressure,
`by patient
`do not com
`research has shown
`that when
`physicians
`are
`ply with
`less
`satisfied
`requests,
`patient
`patients
`(Kravitz et al. 2003).
`with
`visit
`their physician
`in this area
`studies
`is the notion
`typical
`Underlying
`that patient
`if triggered
`by DTC
`requests,
`especially
`are often
`or
`for mild
`trivial
`ailments
`advertising,
`et al. 2004, Wilkes
`et al. 2000). Kravitz
`et al.
`(Weissman
`found
`that subjective
`health distress
`(2003)
`predicted
`for physician
`services
`(referrals and prescrip
`requests
`than did an objective
`tions) more
`count
`powerfully
`to conclude
`of chronic
`them
`that
`conditions,
`leading
`be driven more
`than dis
`"requests may
`by anxiety
`ease burden"
`(p. 1680). To the best of our knowledge,
`no research
`that examines
`exists
`the effect of patient
`on sample dispensing
`the physician.
`
`requests
`
`by
`
`2.3. Moderating
`of Drug Characteristics
`Role
`though prior research has
`Even
`that drug char
`stated
`their role
`acteristics may moderate
`the above
`effects,
`in the effect of firms' marketing
`efforts and patients'
`remains unex
`on physician
`decision making
`requests
`can
`a drug
`et al. 2004). While
`plored
`(Leeflang
`such as
`be characterized
`dimensions,
`among many
`its
`its dosage,
`its potency,
`its approved
`indications,
`its
`interac
`method
`and
`administration
`frequency,
`its toxicity, and
`its
`tions with
`food and other drugs,
`focus on
`study we will
`in this first exploratory
`price,
`two very
`the
`salient product
`characteristics,
`namely,
`and
`side effects.
`the drug's
`effectiveness
`drug's
`to which
`the
`is the extent
`A drug's
`effectiveness
`end
`clinical
`the
`of negative
`reduces
`likelihood
`drug
`side effects are secondary,
`and usu
`points. A drug's
`effects of a drug. For
`for statins,
`instance,
`ally adverse,
`a drug's
`to which
`is the extent
`it reduces
`effectiveness
`such as
`of negative
`the
`clinical
`likelihood
`endpoints,
`or coronary
`(fatal or nonfatal) myocardial
`infarction
`are
`statins may
`show
`heart disease. The
`side effects
`such as gastro-intestinal
`effects
`reactions,
`headaches,
`nausea.
`and
`Above, we
`that
`literature
`found
`referenced
`prior
`firms'
`of
`informative
`effects
`and persuasive
`positive
`efforts on physician
`decision making. Now
`marketing
`we
`to which
`the extent
`firms'
`the effects
`of
`explore
`efforts on physician
`decision making may
`marketing
`
`This content downloaded from 208.85.77.1 on Thu, 04 Feb 2016 21:01:17 UTC
`All use subject to JSTOR Terms and Conditions
`
`Page 4 of 15
`
`

`
`Venkataraman
`
`Management
`
`The Debate
`and Stremersch:
`Science 53(11), pp. 1688-1701, ?2007
`
`on Influencing Doctors' Decisions
`INFORMS 1691
`
`uncertainty
`affect
`toward
`scientific
`
`also
`
`and side effects
`effectiveness
`the drug's
`upon
`depend
`a more
`firm promotes
`effective
`the
`profile. When
`to a
`as compared
`its abil
`less effective
`drug,
`drug,
`ity to lower physician
`the drug and
`about
`the drug
`is higher,
`increase physicians'
`as
`to back
`there will be
`evidence
`stronger
`2002). The effect of
`the marketing
`effort
`up
`(Azoulay
`of side effects on the relationship
`the number
`between
`a firm's marketing
`effort and a physician's
`decision
`a drug
`is more
`the one hand,
`speculative. On
`making
`creates a high
`with many
`level of physi
`side effects
`(e.g., on
`cian uncertainty
`all
`between
`the
`interaction
`can be effectively
`these
`side effects), which
`reduced
`a drug with
`efforts, while
`few
`firms' marketing
`by
`a
`uncer
`creates
`low
`level of physician
`side effects
`the return
`thus reducing
`the need
`for?and
`tainty,
`reduction
`firms' marketing
`on?uncertainty
`through
`et al. 2005). On
`the other hand,
`efforts
`(Narayanan
`it will be harder
`to persuade
`to
`for firms
`physicians
`that has a high number
`a drug
`treat patients with
`of
`a low num
`to a drug with
`side effects as compared
`ber of side effects. Hence,
`the total
`effect of
`interaction
`side effects and a firm's marketing
`is difficult
`efforts
`ex ante, and hence
`to predict
`is worthy
`of empirical
`investigation.
`requests, we
`to prior
`to patient
`referred
`As
`to occur more
`literature
`that
`found patient
`requests
`conditions. Thus, we
`that patient
`often
`for mild
`expect
`side effects are honored
`for drugs with many
`requests
`in fewer cases
`than patient
`the physician
`requests
`by
`reason
`for drugs with
`is that
`side effects. The
`few
`easily do more
`drugs with many
`side
`effects may
`to the patient
`than
`the damage
`the
`ini
`from
`damage
`et al. 2005). We
`tial mild
`condition
`(Kravitz
`expect
`that patient
`for drugs with higher
`effective
`requests
`ness are honored
`in more
`cases
`the physician
`than
`by
`for drugs with
`lower
`effectiveness.
`requests
`a
`react more
`one
`hand,
`may
`posi
`physician
`to an effective
`as she or he has
`request
`tively
`drug
`value.
`less uncertainty
`the drug's
`about
`therapeutic
`the other hand, a physician
`that reacts
`to
`On
`favorably
`a patient
`for an effective drug
`is more
`to
`request
`likely
`receive
`afterwards
`than when
`he
`favorable
`feedback
`to a patient
`for an
`reacts
`ineffective
`request
`favorably
`drug. Given
`increase
`this feedback,
`the physician will
`it con
`to patient
`favorable
`reaction
`requests, when
`his
`cerns
`the effective
`and will decrease
`his
`favor
`drug,
`it concerns
`to patient
`the
`able reaction
`requests, when
`ineffective
`drug.
`a priori,
`the following:
`we,
`expect
`Summarizing,
`the effects
`effectiveness
`may
`Drug
`strengthen
`on prescription
`of marketing
`efforts
`and
`the physician.
`behavior
`by
`effectiveness may
`Drug
`requests on prescription
`patient
`ior by
`the physician.
`
`patient
`On
`the
`
`sampling
`
`the effects of
`strengthen
`behav
`and sampling
`
`Side effects of a drug may weaken
`or strengthen
`on prescription
`the effects
`of marketing
`and
`efforts
`behavior
`information?
`upon
`(depending
`
`sampling
`trade-off).
`persuasion
`Side effects of a drug may weaken
`the effects of
`on prescription
`and sampling
`behav
`patient
`requests
`the physician.
`ior by
`
`2.4. Other Variables
`as well,
`We
`control
`for other variables,
`af
`that may
`First, we
`and
`fect prescription
`behavior.
`sampling
`sam
`control
`for
`the number
`and
`of prescriptions
`in the prescription model,
`for competing
`brands
`ples
`while we
`in the sam
`control
`for competitive
`samples
`on Mizik
`pling model.
`Based
`and Jacobson
`(2004),
`we
`or neg
`that
`these effects may be positive
`expect
`a clear ex ante expectation.
`ative, without
`They may
`as prescriptions
`be negative
`and samples
`for compet
`share of the focal brand
`take away
`ing brands
`(brand
`as
`also be positive,
`They may
`switching).
`increasing
`can
`and
`of competing
`brands
`prescriptions
`samples
`in the drug
`be
`indicative
`of growth
`of
`the
`category
`focal brand
`(category
`growth).
`Second, we
`control
`for the effect of sample dispens
`ing of the own brand on prescriptions.
`This effect may
`or negative,
`the reason
`be positive
`upon
`dependent
`a sample
`the physician
`(see above).
`why
`dispenses
`that a physi
`and Manchanda
`(2006) argue
`Narayanan
`a sample,
`is uncer
`as she or he
`cian may
`dispense
`tain about a patient's
`to the focal drug. This
`response
`a negative
`would
`effect of
`contemporaneous
`imply
`own
`on own prescriptions,
`as
`the
`samples
`sample
`of a prescription. On
`comes
`at the expense
`the other
`and Manchanda
`(2006) also argue
`hand, Narayanan
`that a physician may
`subsidize
`low-income
`financially
`or
`sample dispensing,
`patients
`low-coverage
`through
`case a drug prescription
`comes with
`in which
`usually
`a free sample. This would
`imply a positive
`contem
`effect.
`poraneous
`Third, we
`for carry-over
`control
`effects,
`allowing
`to
`interact with
`and
`these effects
`effectiveness
`drug
`is an often observed
`side effects. Physician
`persistence
`by habit persistence
`and
`feed
`driven
`phenomenon,
`et al. 2005). We
`back of patients
`(Janakiraman
`expect
`to be more
`the more
`physician
`persistence
`positive
`is, as this will
`effective
`the drug
`increase positive
`feed
`to the physician. On
`back of patients
`the other hand,
`the more
`the more
`the drug has,
`side effects
`nega
`tive feedback
`the physician will
`receive
`from patients,
`in turn will
`which
`lower physician
`persistence.
`
`3. Data
`
`and Analysis
`
`3.1. Data
`sets used
`The data
`for the empirical
`analysis
`include
`panel data,
`(a) physician-level
`(b) drug
`and
`(c) clinical
`trial reports. The
`database,
`
`study
`approval
`
`in this
`
`This content downloaded from 208.85.77.1 on Thu, 04 Feb 2016 21:01:17 UTC
`All use subject to JSTOR Terms and Conditions
`
`Page 5 of 15
`
`

`
`1692
`
`Venkataraman
`
`and Stremersch:
`
`Management
`
`on Influencing Doctors' Decisions
`The Debate
`Science 53(11), pp. 1688-1701, ?2007
`INFORMS
`
`Table 1
`
`Descriptive Statistics and Correlation Table
`
`Descriptive
`
`statistics
`
`Variable
`
`Mean
`
`Std. dev. Min.
`
`Max.
`
`Patient Competitive
`Prescriptions Meeting Detailing request prescriptions Samples
`
`Competitive
`samples
`
`Correlation
`
`table
`
`51 1
`
`1.23
`Prescriptions
`0.02
`Meeting
`0.73
`Detailing
`0.07
`Patient request
`Competitive prescription 3.69
`0.15
`Samples
`0.46
`
`Competitive samples
`
`1.86
`0.17
`0.97
`0.48
`3.94
`0.50
`0.99
`
`0
`0
`0
`0
`0
`0
`0
`
`18
`13
`41
`80
`19
`24
`
`1
`0.05
`1
`0.04
`-0.00
`0.05 0.00 -0.02
`0.36
`0.01
`0.20
`0.03
`
`0.02 0.01
`0.03
`
`-0.03
`-0.04
`
`1
`-0.03
`-0.03
`0.06
`
`1
`0.02
`0.17
`
`1
`0.24
`
`We
`
`provides
`
`a meta
`from
`effectiveness
`obtained
`drug
`trial reports
`of clinical
`Insti
`(source: National
`analysis
`tute
`and Clinical
`This meta
`for Health
`Excellence).
`a
`measurement
`standardized
`analysis
`a
`mea
`Z-score
`standardized
`effectiveness,
`namely,
`of a brand
`sure of the overall
`to
`relative
`effectiveness
`a placebo.
`these are standardized,
`the rela
`Because
`can be compared
`tive effectiveness
`of brands
`directly.
`are explained
`in the
`in full detail
`The measurements
`in the e-companion).2
`online
`appendix
`(provided
`Table 1 provides
`and Pear
`the descriptive
`statistics
`son correlations
`interest. Table
`for the variables
`of
`reflects variance
`in both
`the dependent
`variables
`written
`(RX) and
`interest,
`i.e., prescriptions
`samples
`at
`The database
`the monthly
`includes,
`dispensed.
`level, all prescriptions
`cat
`within
`the examined
`drug
`egories by a panel of 2,774 physicians.
`In all, we have
`1, we
`From Table
`also observe
`39,880 observations.3
`that the correlations
`variables
`the independent
`among
`are small, hence
`attenuating multicollinearity
`in the analysis. No physician
`lems
`prescribes
`to all his or her patients.
`brand
`
`of
`
`1
`of
`
`prob
`the same
`
`two years
`panel data1
`span
`monthly
`physician-level
`a
`come
`from
`and
`2002-December
`2003)
`(January
`mar
`in pharmaceutical
`firm
`that
`specializes
`large
`we
`can
`to confidentiality
`keting. Due
`agreements,
`source. The data
`sets
`not
`the data
`contain
`reveal
`on
`three
`information
`categories,
`namely,
`therapeutic
`and coagulation
`(1) statins,
`(2) gastrointestinal
`drugs,
`is a
`and
`(ED). The panel
`(3) erectile
`dysfunction
`across
`of physicians
`balanced
`representative
`sample
`vol
`and prescription
`regions,
`specialties,
`geographic
`umes. Monthly
`vari
`brand-specific
`physician-level
`total samples
`ables
`include
`total prescriptions
`written,
`total number
`total number
`of details,
`dispensed,
`and
`total number
`of patient
`of meetings
`attended,
`data are collected
`from
`the
`requests. These
`directly
`an electronic
`database
`that
`office
`physician
`through
`and detailing-call
`collects
`information.
`prescription
`our database
`Unlike
`researched
`databases,
`previously
`on samples
`the physi
`information
`has
`dispensed
`by
`a more
`cian,
`complete
`understanding
`facilitating
`across
`two key
`variables?
`of physician
`behavior
`and samples dispensed. We
`written
`cali
`prescriptions
`on the four most prescribed
`brate our empirical model
`in each category. The shares of the focal brands
`brands
`are 85%
`in Category
`in Category
`2, and 88%
`1, 78%
`in Category
`3.
`Our measures
`for drug characteristics,
`effects were
`constructed
`and
`side
`of side effects
`obtained
`the number
`the FDA
`database
`from
`approval
`a history
`not only
`of drug-application
`and drug-innovation
`dates,
`approval
`a
`that
`list of
`side
`effects
`but
`also
`updated when new
`indications
`and/or
`announced.
`
`effectiveness,
`as
`follows. We
`from
`the drug
`that
`includes
`filing dates,
`classifications,
`is periodically
`side effects are
`
`1
`Note
`
`keting
`Due
`
`that our physician-level
`and prescription
`efforts
`to
`institutional
`factors
`retail distribution,
`coverage,
`not accurately
`reflect
`might
`to direct measures
`access
`have
`get a more
`accurate
`behavior.
`physician
`
`picture
`
`of mar
`includes measures
`database
`at
`level.
`data directly
`the physician
`like availability
`of generics,
`insurance
`at
`etc., data
`collected
`the pharmacy
`Because we
`actual physician
`behavior.
`variables, we
`can
`of physician-level
`on
`activities
`of effects
`of marketing
`
`3.2. Analysis
`section describes
`This
`the empirical model. We begin
`this sec
`the econometric model
`and end
`by specifying
`on
`a discussion
`tion with
`the estimation
`
`procedure.
`of market
`variables?
`
`the effects
`decision
`
`3.2.1. Model.
`To estimate
`on
`two physician
`ing activities
`and
`(b)
`(a) prescriptions
`dispensed?we
`samples
`our
`model
`below.
`estimated
`econometric
`describe
`that the model we
`intricacies
`Note
`the
`specify, given
`is a descriptive model
`of the available
`that does
`data,
`not allow normative
`claims
`(Franses 2005).
`
`3.2.1.1. Dependent
`the
`include
`These
`Variables.
`(to new and previously
`total number
`of prescriptions
`and
`the
`total number
`patients) written
`
`diagnosed
`
`of
`
`2
`An
`online
`
`electronic
`version
`
`companion
`that can be
`
`as part of
`to this paper
`is available
`at http://mansci.journal.informs.
`found
`
`the
`
`org/.
`3
`Our
`
`panel
`physicians
`24 months.
`
`is an unbalanced
`in
`the panel
`for
`
`as we
`do not
`panel
`data window,
`the complete
`
`observe
`which
`
`all
`is
`
`This content downloaded from 208.85.77.1 on Thu, 04 Feb 2016 21:01:17 UTC
`All use subject to JSTOR Terms and Conditions
`
`Page 6 of 15
`
`

`
`Venkataraman
`
`Management
`
`and Stremersch:
`The Debate
`Science 53(11), pp. 1688-1701, ?2007
`
`on Influencing Doctors' Decisions
`INFORMS 1693
`
`a category, making
`side effects between
`drugs within
`a count of the number
`of side effects a valid measure.
`that newer drugs
`Also, while
`readers may
`think
`ically have
`fewer
`side effects
`than older drugs
`ing a possible
`confound
`between
`the number
`and
`the passing
`effects
`this
`is not
`of
`time),
`in our data.
`In fact, regressions
`of the number
`on
`time
`effects
`show
`that
`the coefficient
`for
`
`typ
`(creat
`of side
`the case
`of side
`time
`is
`
`insignificant.
`The measures
`for effectiveness
`and side effects
`in more
`in the online
`also provided
`detail
`appendix.
`the main
`We
`include
`effects of the product
`character
`istics and
`interactions with
`their
`all other variables
`above.
`
`are
`
`posited
`
`;
`
`t
`
`samples dispensed
`These are denoted
`
`of brand
`as
`
`RXjpt
`
`prescriptions
`Because
`
`j at time
`t by physician
`p.4
`and
`Samplesjpt, respectively
`3.2.1.2.
`As
`Variables.
`Independent
`stipulated
`above, we
`the effect
`of drug manufacturers'
`study
`and meetings)
`efforts
`marketing
`detailing
`(through
`on physician
`and patient
`and
`requests
`prescription
`decisions.
`effort by
`the
`sample-dispensing
`Detailing
`; at
`t to physician
`manufacturer
`for brand
`time
`p,
`as
`is measured
`denoted
`the
`total number
`by Detjpt,
`of detailing
`calls made
`the sales
`force
`for brand
`by
`spirit, we define
`to physician
`p dit time
`t. In similar