D r u g C o s t G r o w t h
`
`The U.S. Pharmaceutical
`Industry: Why Major
`Growth In Times Of
`Cost Containment?
`
`Four factors affecting drug use have driven costs upward since
`1994, but their future role is uncertain.
`
`by Ernst R. Berndt
`
`100
`
`DRUG
`INDUSTRY
`
`ABSTRACT: Growth in utilization rather than price, particularly since 1994, has
`been the primary driver of increased pharmaceutical spending. In this paper I
`focus on four factors that have increased utilization, even as cost containment
`efforts have flourished: (1) “the importance of being unimportant”; (2) in-
`creased third-party prescription drug coverage; (3) the introduction of success-
`ful new products; and (4) aggressive technology transfer and marketing efforts
`by pharmaceutical firms. I also consider the roles that these four factors are
`likely to play in the future.
`
`For mo st med ica l ca re indu st ries in the United States,
`
`the 1990s were turbulent, as managed care and other cost
`containment efforts flourished, rooting out overutilization, al-
`tering incentives, and affecting health care quality in ways not yet
`well understood. Yet during this same decade the U.S. pharmaceuti-
`cal industry experienced relatively high rates of domestic sales
`growth. Why such significant growth in times of cost containment?
`n Recent spending growth patterns. In terms of average an-
`nual growth rates in pharmaceutical sales, while the rate of 12.8
`percent for the more recent 1994–1999 time period is only slightly
`larger than the rate of 11.9 percent for 1987–1994, the composition of
`this spending growth has changed dramatically (Exhibit 1).
`Using price index formulae analogous to those used by the U.S.
`Bureau of Labor Statistics, IMS Health regularly decomposes pre-
`scription drug expenditures into those attributable to price (the
`change in spending if last year’s mix of drugs were purchased today),
`those attributable to spending on new products (defined as less than
`
`Ernst R. Berndt is a professor of applied economics at the Massachusetts Institute of Technol-
`ogy’s Sloan School of Management and director of the National Bureau of Economic Re-
`search (NBER) Program on Technological Change and Productivity Management.
`
`H E A L T H A F F A I R S
`
`~ V o l u m e
`
`2 0 , N u m b e r
`
`2
`
`©2001 Project HOPE–ThePeople-to-PeopleHealth Foundation, Inc.
`
`IMMUNOGEN 2276, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D R U G I N D U S T R Y
`
`EX H IB I T 1
`The U.S. Prescription Pharmaceutical Market: Total Annual Sales Growth And Its
`Sources, 1987–1999
`
`Percent growth rate
`
`Other
`Price
`
`20
`
`16
`
`12
`
`48
`
`0
`
`1990
`1993
`1987
`1996
`SOURCE: IMS Health, “Retail Provider Perspective, 2000,” reproduced in Pharmaceutical Industry Profile 2000:
`Research for the Millennium (Washington: Pharmaceutical Research and Manufacturers of America, 2000), Figure 4-11.
`NOTE: Annual averages were as follows. Sales growth: 1987–99, 12.6 percent; 1987–94, 11.9 percent; 1994–99, 12.8
`percent. Price growth: 1987–99, 4.8 percent; 1987–94, 6.4 percent; 1994–99, 2.5 percent. Residual growth: 1987–99,
`7.8 percent; 1987–94, 5.8 percent; 1994–99, 10.3 percent.
`
`1999
`
`DRUG COST
`GROWTH
`
`101
`
`a year old), and the residual (those attributable to volume and mix
`on incumbent products). Hereafter I refer to the latter two nonprice
`factors as “utilization” components. From 1987 through 1994, of the
`11.9 percent average annual rate of spending growth, about half re-
`flected the direct effects of increased prices, while the remaining half
`is attributed to utilization growth. In contrast, from 1994 through
`1999 the growth rate remained in double digits, but only about
`one-fifth was directly attributable to price changes; nearly 80 percent
`of increased drug spending was related to growth in utilization.1
`In this paper I offer four hypotheses to help explain why use of
`pharmaceuticals has continued to grow even as managed care and
`other cost containment efforts have flourished. The four factors on
`which I focus, not necessarily in order of importance, are (1) “the
`importance of being unimportant”—pharmaceuticals’ modest share
`of total U.S. health care costs; (2) the dramatic growth of third-party
`prescription drug coverage; (3) the successful new product innova-
`tion emerging from the pharmaceutical industry; and (4) pharma-
`ceutical firms’ aggressive technology transfer and marketing efforts.
`Factor 1: ‘The Importance Of Being Unimportant’
`Alfred Marshall, a famous nineteenth-century economist, reasoned
`that certain characteristics of goods and services made their demand
`more or less price-responsive, or more or less immune to cost-
`cutting efforts. Among the four laws of demand that Marshall enun-
`ciated, one has been dubbed “the importance of being unimportant.”
`
`H E A L T H A F F A I R S
`
`~ M a r c h / A p r i
`
`l
`
`2 0 0 1
`
`IMMUNOGEN 2276, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D r u g C o s t G r o w t h
`
`To Marshall, if spending on some good or service is perceived to be
`only a small portion of total costs, that good or service will not be as
`likely to be on cost cutters’ radar screens; instead, they will tend to
`focus more on big-ticket items. Although Marshall provided no ana-
`lytic basis for this argument, it is plausible to argue that, other
`things being equal, it may be rational for budget managers to focus
`most of their attention on the largest budget items.
`Hospital spending (outpatient plus inpatient) continues to be the
`single largest component of health care costs (Exhibit 2). Despite
`the shift from inpatient to outpatient settings, total hospital costs
`are still the largest single health care component. The second-largest
`spending item has consistently been physician services, whose share
`of total health care spending has remained relatively constant over
`the past four decades at about 20 percent.
`In third or fourth place is spending for outpatient prescription
`drugs. Even at their current 8 percent share, prescription drug costs
`are still relatively unimportant. However, this 8 percent represents
`an average, and the variance across subpopulations is considerable.
`For example, data from the 1995 Medicare Current Beneficiary Sur-
`vey (MCBS) indicate that while Medicare beneficiaries’ average to-
`tal spending on prescription drugs was $536, the variance was $741.2
`Also, it is likely that the prescription drug share is larger for payers
`that cover the nonelderly working population, a subgroup with rela-
`tively low rates of hospitalization.
`Within the past decade, as the prescription drug cost share has
`grown, pharmacy benefit management (PBM) tools have been devel-
`oped and have flourished. These tools include drug utilization re-
`view, generic substitution, prior authorization, step-care protocols,
`therapeutic interchange, increasingly restrictive formularies, three-
`tier copayment structures, academic detailing, and various physi-
`
`102
`
`DRUG
`INDUSTRY
`
`EXHIBIT 2
`Health Care Expenditure Cost Shares, By Category, 1960–1998
`
`Hospital care
`Physician services
`Prescription drugs
`Nursing home care
`All other
`
`Total health care
`expenditures (billions)
`
`34.6%
`19.7
`10.0
`3.0
`32.7
`
`38.3%
`18.6
`7.5
`5.7
`29.9
`
`41.5%
`18.3
`4.9
`7.1
`28.2
`
`36.7%
`20.9
`5.4
`7.3
`29.7
`
`34.9%
`20.3
`6.1
`7.6
`31.1
`
`34.6%
`20.1
`6.6
`7.7
`31.0
`
`34.0%
`20.0
`7.2
`7.8
`31.0
`
`33.3%
`20.0
`7.9
`7.6
`31.2
`
`$26.9
`
`$73.2
`
`$247.3
`
`$699.4
`
`$993.3
`
`$1,039.4
`
`$1,088.2
`
`$1,149.1
`
`SOURCES: K. Levit et al., “National Health Spending Trends in 1996,” Health Affairs (Jan/Feb 1998): 35–51 (for 1960–1990
`data); and K. Levit et al., “Health Spending in 1998: Signals of Changes,” Health Affairs (Jan/Feb 2000): 124–132 (for
`1995–1998 data).
`NOTE: “All other” includes dental and other professional services, home health care, nonprescription drugs and medical
`durables, vision products, net cost of private health insurance, government public health activities, and research/construction.
`
`H E A L T H A F F A I R S
`
`~ V o l u m e
`
`2 0 , N u m b e r
`
`2
`
`IMMUNOGEN 2276, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D R U G I N D U S T R Y
`
`“The information technology revolutions have contributed to the
`diffusion of drug coverage into benefit plans.”
`
`DRUG COST
`GROWTH
`
`103
`
`cian capitation schemes. While use of these PBM tools has undoubt-
`edly constrained drug spending growth, a detailed analysis of their
`impacts is beyond the scope of this paper.
`It is worth noting, however, that formulary compliance by physi-
`cians involves information gathering and monitoring costs. Such
`costs are likely to be higher the larger the number of payers with
`which a physician contracts. Relatively few physicians today have
`only one managed care contract. Based on data from the 1996–97
`Community Tracking Survey of Physicians, Nancy Beaulieu reports
`that 61 percent of primary care physicians and 64 percent of special-
`ists surveyed had six or more managed care contracts.3 The ability of
`any one payer to greatly affect prescribing decisions is constrained
`when physicians simultaneously interact with so many different
`payers and their formularies.
`Thus, until recently prescription drug costs have not on average
`been as important as the health care cost shares of hospital and
`physician services. In the context of nonpharmaceutical expendi-
`tures, there is some evidence suggesting that managed care has had
`a much larger impact on prices paid for health care services than on
`their use.4 This may be particularly true for drugs, whose average
`cost share in 1998 was still relatively unimportant at 8 percent.
`Factor 2: Growth In Third-Party Drug Coverage
`Prescriptions dispensed at retail pharmacies have been paid for in a
`variety of ways. Historically, for consumers with private third-party
`drug coverage, the drug recipient initially made a full cash payment
`to the pharmacy and then was reimbursed in whole or in part by the
`insurer. Until the 1990s this somewhat cumbersome procedure was
`the norm. The transaction costs—first saving and storing prescrip-
`tion receipts in shoe boxes, then gathering them together, and fi-
`nally filling out forms and sending them off to claims proces-
`sors—were considerable, for both beneficiaries and insurers.
`n Impact of information technology. Recent technological
`progress, particularly involving information technology and tele-
`communications equipment, has dramatically changed the way in
`which third-party drug claims are processed at pharmacies, making
`covered insurance transactions much more convenient and less
`costly than they were a decade ago. Today, for example, the privately
`insured beneficiary usually pays a copayment or coinsurance to the
`
`H E A L T H A F F A I R S
`
`~ M a r c h / A p r i
`
`l
`
`2 0 0 1
`
`IMMUNOGEN 2276, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D r u g C o s t G r o w t h
`
`pharmacy upon receipt of the prescription. After monitoring the
`pharmacy claim request to ensure compliance with formulary provi-
`sions, the third-party insurer then seamlessly reimburses the phar-
`macy electronically for the remainder, based on their contractual
`arrangement. For publicly provided drug insurance such as Medic-
`aid, even when there is a copayment, the entire transaction is typi-
`cally processed instantaneously and electronically.
`Technological developments involving electronic transactions
`have also facilitated inexpensive, instantaneous monitoring for
`safety and formulary compliance by PBMs. Indeed, it could well be
`argued that the very existence of PBM techniques owes much to the
`revolutions in information technology and telecommunications.5
`But what do these technological revolutions have to do with
`increased drug use? Undoubtedly the tight U.S. labor market in the
`past decade has contributed enormously to enhanced employee
`compensation in the form of more generous prescription drug cover-
`age. However, because they have reduced pharmacies’ and insurers’
`costs; offered consumers increased convenience and less bookkeep-
`ing; and enabled PBMs to monitor transactions, enforce formulary
`provisions, and perform drug utilization reviews at very low cost,
`the information technology revolutions have contributed as well to
`the diffusion of drug coverage into benefit plans.
`n Changing role of third-party insurance. The Health Care
`Financing Administration (HCFA) has produced data that docu-
`ment the changing role of third-party coverage in paying for pre-
`scription drugs (Exhibit 3). As seen in this exhibit, in 1965 (prior to
`the 1967 precedent-setting agreement between Ford Motor Com-
`pany and the United Auto Workers enshrining drug insurance
`benefits as part of employees’ benefit package), private insurance
`
`104
`
`DRUG
`INDUSTRY
`
`EXHIBIT 3
`Share Of Prescription Drug Spending, By Source Of Payment, Selected Years
`1965–1998
`
`1965
`1970
`1975
`1980
`1985
`
`1990
`1995
`1996
`1997
`1998
`
`3.5%
`8.8
`12.2
`20.1
`29.9
`
`34.4
`46.8
`48.8
`50.8
`52.7
`
`92.6%
`82.4
`75.4
`66.0
`55.4
`
`48.3
`33.9
`31.6
`29.1
`26.6
`
`0.0%
`7.6
`10.8
`11.7
`11.8
`
`13.5
`15.8
`16.1
`16.5
`17.1
`
`3.9%
`1.2
`1.6
`2.2
`2.9
`
`3.8
`3.4
`3.5
`3.6
`3.6
`
`SOURCES: Health Care Financing Administration (HCFA) National Health Accounts; and Report to the President: Prescription
`Drug Coverage, Spending, Utilization, and Prices (Washington: DHHS, April 2000), Table 2-30.
`
`H E A L T H A F F A I R S
`
`~ V o l u m e
`
`2 0 , N u m b e r
`
`2
`
`IMMUNOGEN 2276, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D R U G I N D U S T R Y
`
`covered only about 3.5 percent of prescription drug spending. More-
`over, in 1965 Medicaid did not yet exist. By 1980 the nonreimbursed
`cash share had fallen to 66.0 percent, the private third-party portion
`rose to 20.1 percent, and Medicaid grew from nothing to 11.7 percent.
`By 1995 the private third-party portion surpassed that of out-of-
`pocket cash payments. In 1998, the last year for which HCFA data
`were available, the out-of-pocket share had fallen to about a quarter,
`while the portion covered by private and public insurance was al-
`most 70 percent. Although the rate of increase has fallen recently, it
`is clear that the predominant form of payment for prescription drugs
`has changed dramatically from uninsured cash to electronically
`processed third-party insurance coverage transactions.
`n Relationship of use and coverage. But why is this change
`significant for understanding increased use of prescription drugs?
`At least since the RAND Health Insurance Experiment in the 1980s,
`it has been widely understood that per capita drug use is strongly
`associated with the extent of drug coverage.6 For example, in the
`RAND experiment, when patient coinsurance rates for prescription
`drugs fell from 95 percent to 25 percent, per capita prescription drug
`spending increased 33 percent (per capita prescriptions increased
`22 percent). Furthermore, when the patient coinsurance rate fell to
`zero (free to the patient), per capita drug spending relative to the 25
`percent coinsurance rose another 32 percent (per capita prescrip-
`tions increased another 22 percent).
`A caveat is worth noting. The RAND experimental data are now
`several decades old, and since the experiment involved simultane-
`ously changing coinsurance rates for drugs and nondrug health care
`services, rather than adding drug coverage to existing health bene-
`fits, their quantitative values may not provide reliable guidance in
`the current policy environment. Moreover, because of selection
`problems into insurance plans, in nonexperimental settings it is
`difficult to quantify the extent to which increased drug coverage has
`contributed to increased drug use. Thus, for example, evidence that
`in the 1990s elderly Americans with drug coverage used drugs con-
`siderably more intensively than did those without such coverage
`does not help us to quantify reliably what drug use would be if drug
`benefits were added to Medicare.7
`n Decreasing copayments. Increased drug use has also been
`associated with decreased copayments as a share of costs. In 1996,
`1997, and 1998, for example, as private health insurers’ payments for
`prescription drugs increased by 17.5 percent, 18.7 percent, and 19.7
`percent, respectively, beneficiaries’ out-of-pocket payments in-
`creased by only 5.3 percent, 4.9 percent, and 5.4 percent, respec-
`tively.8 In such an insurance-protected environment, it is not at all
`
`H E A L T H A F F A I R S
`
`~ M a r c h / A p r i
`
`l
`
`2 0 0 1
`
`DRUG COST
`GROWTH
`
`105
`
`IMMUNOGEN 2276, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D r u g C o s t G r o w t h
`
`106
`
`DRUG
`INDUSTRY
`
`surprising that employees’ prescription drug use has surged.
`n Deceleration ahead. Growth in the insured drug coverage
`share has decelerated considerably in the past few years, however
`(Exhibit 3). Whether the 70 percent insurance coverage portion will
`continue to increase in the future depends in large part on public
`policy decisions such as those involving Medicare universal drug
`coverage. Absent such developments, there does not appear to be
`any compelling reason to expect further growth in the impact of
`drug coverage on use, particularly if the U.S. labor market softens.
`Factor 3: Introduction Of Successful New Products
`While the first two factors identified above operate essentially by
`affecting demand, the third factor functions primarily through the
`supply side—namely, the successful introduction of new pharma-
`ceutical products. These new products are the fruits of substantial
`research and development (R&D) efforts. Some of the recent inno-
`vations have provided effective therapies where previously very few
`had existed (for example, the protease inhibitors for acquired im-
`munodeficiency syndrome [AIDS], the human growth hormone
`Protropin, and Viagra for erectile dysfunction). Others have been
`new entrants in already crowded therapeutic classes, but their
`manufacturers have claimed lower prices and greater cost-
`effectiveness (for example, Lipitor, a statin lipid-lowering agent,
`and Protonix, a proton pump inhibitor).9 Yet other sets of innovative
`pharmaceutical products have offset nonpharmaceutical costs, such
`as third- and fourth-generation antidepressants that have facilitated
`reductions in the intensity of costly psychotherapy, and the beta-
`blockers and blood pressure medications that have reduced costs of
`cardiovascular-related hospitalizations and surgeries.10 Finally, in-
`creases in the variety of products within therapeutic areas facilitate a
`better matching between idiosyncratic patients and their medications.
`n Approval time falling. In recent years the number of new
`molecular entities (NMEs) approved by the U.S. Food and Drug
`Administration (FDA) has increased considerably. From 1987
`through 1993 the FDA approved on average twenty-four NMEs per
`year. In the past six years, however (1994–1999), this approval rate
`increased by almost 50 percent to 35.5 NMEs per year. Part of the
`reason more new drugs are coming onto the U.S. market each year is
`that the mean approval time at the FDA has fallen. From 1987
`through 1993 the mean approval time was 26.3 months, but by 1999
`it had fallen to 12.6 months.11
`n Clinical testing time rising. At the same time, however, the
`mean time in the clinical testing phase of drug development in-
`creased from 5.5 years in the 1980s to 6.7 years in 1990–1996; the
`
`H E A L T H A F F A I R S
`
`~ V o l u m e
`
`2 0 , N u m b e r
`
`2
`
`IMMUNOGEN 2276, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D R U G I N D U S T R Y
`
`number of clinical trials per new drug application (NDA) increased
`from sixty in 1989–1992 to sixty-eight in 1994–1995; and the mean
`number of patients involved in each NDA increased about 20 per-
`cent, from 3,567 in 1989–1992 to 4,237 in 1994–1995. Clinical trials
`take longer and are more costly. This is partly the result of the FDA’s
`and providers’ demands for more information on patient subpopula-
`tions and on interactions with other drugs, necessitating larger and
`more complex multisite trials.12
`n Impact of new products on spending. How important are
`new pharmaceutical products as drivers of increased drug spending?
`The answer to this question depends in part on what one means by
`a “new” pharmaceutical product, and on this there is inherent ambi-
`guity. IMS Health defines a “new product” as any product having a
`new National Drug Classification (NDC) code and launched during
`the twelve months ending with the last calendar quarter. Although
`a one-year window is rather narrow, new NDCs include all new
`products, such as new generics, new brand-name products, and new
`line extensions of existing molecules. While it would be preferable
`to have a new-product definition that excluded new generic drugs
`having comparable brand-name presentations and strengths, spend-
`ing on new generic drugs is typically much less than that for new
`brands, and thus the IMS definition is a reasonable first-cut esti-
`mate of the impact of new products on drug spending.
`Since 1997 about 46 percent, on average, of drug spending growth
`can be attributed to growth in new elements, about 32 percent to
`volume and mix changes involving older drugs, and 22 percent to the
`price growth of older drugs (Exhibit 4). The role of successful new
`products in increasing drug spending is therefore a significant one.
`n Role of patent protection. Pharmaceutical products differ
`from many other consumer products in the critical role of patent
`protection on sales. For pharmaceuticals, once patent protection
`and market exclusivity expire, generic entry typically cuts sharply
`into the pioneer’s revenues. A year after patent expiration in mid-
`1997, for example, pharmacy costs for brand-name Zantac (generic
`name ranitidine) fell to about 15 percent of their prepatent expira-
`tion level, and a year later to only about 10 percent of that level. The
`generic share sold increased from 0 percent to about 80 percent (one
`year) and 90 percent (two years). For Tagamet (generic name
`cimetidine), during the first year following patent expiration, ex-
`penditures also fell about 80 percent, and at twelve months after
`expiration the generic share already reached slightly more than 80
`percent.13 While not all brand-name drugs experience such rapid
`drops in revenues following patent expiration (for some, reputation
`and brand-name loyalty effects persist), for others such as Capoten
`
`H E A L T H A F F A I R S
`
`~ M a r c h / A p r i
`
`l
`
`2 0 0 1
`
`DRUG COST
`GROWTH
`
`107
`
`IMMUNOGEN 2276, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D r u g C o s t G r o w t h
`
`EX H IB I T 4
`Trends In Growth Rates In The U.S. Prescription Pharmaceutical Market,
`By Component And Quarter, 1996–2000
`
`Percent growth
`
`Volume and mix
`Price
`New elementsa
`
`21
`
`18
`
`15
`
`12
`
`9
`
`36
`
`0
`
`Q3
`1996
`
`Q1
`1997
`
`Q3
`1998
`
`Q1
`Q3
`Q1
`1998
`1997
`1996
`SOURCES: IMS Health, "Retail Provider Perspective, 2000."
`a Includes products and line extensions launched during the twelve months ending with the latest calendar quarter.
`
`Q1
`1999
`
`Q3
`1999
`
`Q1
`2000
`
`(generic name captopril), the revenue reduction following patent
`expiration was apparently even more dramatic.14
`In this context of patent expiration and generic penetration, two
`points are worth noting. First, in the ten years following passage of
`the Drug Price Competition and Patent Term Restoration Act (the
`Hatch-Waxman Act) in 1984, the generic share of dispensed pre-
`scription drug units in the United States more than doubled, from
`18.6 percent to 41.6 percent. Since 1994 this share has increased at a
`more modest rate, reaching 47.1 percent in 1999.15 Hence, today
`roughly half of all prescription drug units dispensed in the United
`States are off-patent generic drugs.
`Second, a substantial number of brand-name drugs are expected
`to lose patent protection and market exclusivity in the next few
`years and are likely to experience sharp losses in revenues as gener-
`ics enter and capture market share. Among these are blockbusters
`such as Vasotec, Prozac, Pepcid, Augmentin, Mevacor, Claritin, and
`Prilosec. According to one set of industry analysts, based on their
`total 1998 annual U.S. sales, drugs expected to lose patent protec-
`tion in 2000 will result in annual revenue reductions to brand-name
`firms of about $6.5 billion, slightly less than the $6.7 billion for those
`drugs whose patents are due to expire in 2001.16 The cumulative
`effect of these revenue reductions over time is of course even larger.
`
`H E A L T H A F F A I R S
`
`~ V o l u m e
`
`2 0 , N u m b e r
`
`2
`
`IMMUNOGEN 2276, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D R U G I N D U S T R Y
`
`“Post-launch research can greatly alter treatment guidelines,
`medical practice, and the demand for pharmaceuticals.”
`
`DRUG COST
`GROWTH
`
`109
`
`Factor 4: Aggressive Technology Transfer And
`Marketing Efforts
`Even though prescription drugs are becoming increasingly “user-
`friendly” (having fewer side effects, fewer adverse interactions with
`other drugs, more convenient dosing, and enhanced tolerability),
`these drugs are still complex “high-tech” products, and their appro-
`priate and wise use requires a great amount of knowledge-base
`development and technology-transfer information. Marketing plays
`a critical role in this information-sharing process.
`n Post-launch research. After new products are launched,
`pharmaceutical firms often undertake or help to support long-term
`studies to compare alternative therapies, to assess whether preemp-
`tive treatment is efficacious, or to compare outcomes among sub-
`populations. As does basic research, post-launch research incorpo-
`rates multilateral information flows among researchers in the drug
`industry, academe, and government. The findings from such studies,
`typically involving interactions with large numbers of physicians
`and patients, can greatly alter treatment guidelines, medical prac-
`tice, and the demand for pharmaceuticals. It is useful to consider
`several examples and their impacts on medical practice.
`Cholesterol-reducing drugs. Based on a number of widely publicized
`studies documenting the benefits of more aggressive treatment, in
`the mid-1990s the National Cholesterol Education Program adopted
`new treatment guidelines for patients with coronary heart disease,
`lowering the target for treatment from a low-density lipoprotein
`(LDL) level of less than 130 to less than or equal to 100.17 Treating
`high cholesterol more aggressively has benefited patients and has
`undoubtedly increased the use of cholesterol-lowering medications.
`Treatment after heart attacks. A second example involves post–heart
`attack treatment with beta-blockers, which have been shown to be
`effective in reducing morbidity and mortality associated with heart
`disease, as well as in decreasing the probability of having a second
`heart attack. In 1992, as part of the Cooperative Cardiovascular
`Project, HCFA initiated a study in which the use of beta-blockers
`was monitored following a myocardial infarction. In 1992, data indi-
`cated that only 31.8 percent of eligible patients received this treat-
`ment. In 1996 the National Committee for Quality Assurance
`(NCQA) implemented a measure of beta-blocker use among health
`plans it audited, and later it set a 90 percent post–heart attack
`
`H E A L T H A F F A I R S
`
`~ M a r c h / A p r i
`
`l
`
`2 0 0 1
`
`IMMUNOGEN 2276, pg. 10
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D r u g C o s t G r o w t h
`
`110
`
`DRUG
`INDUSTRY
`
`beta-blocker treatment rate as a benchmark requirement for health
`plans to obtain the NCQA “excellent” quality rating accreditation.
`According to the NCQA, the average treatment rate among its
`audited health plans rose from 62.2 percent in 1996 to 79.9 percent
`in 1998.18
`Diabetes diagnosis. A third example concerns altering the criterion
`for a diabetes diagnosis. Based on pivotal clinical studies, in 1997 the
`American Diabetes Association recommended that a person be con-
`sidered to have diabetes if his or her fasting blood glucose level was
`above 126, a decrease of 10 percent from the level of 140 that had been
`the previous diagnostic criterion.19 This added to the number of
`persons considered to have diabetes and surely has increased use of
`medications to control diabetes.
`In each of these examples, post-launch studies have changed the
`knowledge base concerning the appropriate and effective use of new
`drugs. Although a somewhat nontraditional form of marketing, the
`fostering and facilitation of research that promulgates new diagnos-
`tic criteria and treatment guideline standards both benefits patients
`and on balance increases the use of pharmaceuticals.
`n More traditional marketing. Other technology transfer and
`marketing efforts involving pharmaceuticals are much more tradi-
`tional. In this context, it is informative to examine a widely used
`measure of advertising intensity—advertising-to-sales ratios, meas-
`ured in current dollars—for common over-the-counter (OTC)
`medications. Advertising for OTC drugs entails spending on various
`print media, radio, television, and billboards. For pain medications,
`1993 advertising-to-sales ratios were 17 percent for Tylenol, 26 per-
`cent for Advil and Bayer, and 21 percent for Excedrin. For antacids,
`the ratios were 33 percent for Alka-Seltzer, 24 percent for Mylanta,
`and 20 percent for Tums.20 Are these ratios low or high?
`Search goods versus experience goods. Marketing science researchers
`have long noted that the magnitude of advertising-to-sales ratios
`varies systematically across goods, depending in part on how one
`obtains information about the likely impact of the good or service on
`a given person.21 This research distinguishes “search goods” as polar
`opposites of “experience goods.” The characteristics and effective-
`ness of search goods can be determined simply by searching their
`specifications, which often can be quantified. A pure search good
`does not require one’s own consumption experience to gain infor-
`mation on its characteristics. By contrast, the characteristics and
`effectiveness of experience goods are to a great extent idiosyncratic
`and unpredictable for a given person. An experience good, therefore,
`requires a person to consume it directly to obtain reliable informa-
`tion on its performance. Not surprisingly, other things being equal,
`
`H E A L T H A F F A I R S
`
`~ V o l u m e
`
`2 0 , N u m b e r
`
`2
`
`IMMUNOGEN 2276, pg. 11
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`D R U G I N D U S T R Y
`
`advertising-to-sales ratios tend to be higher for experience goods
`than for search goods, since firms need to continuously entice a
`consumer to engage in a trial with the experience good and to re-
`mind him or her of previous successful consumption experiences so
`that he or she won’t defect to another product. Brand loyalty tends
`to be stronger for experience goods than for search goods.
`Although precise demarcation between search and experience
`goods and their attributes is not possible, the distinction is useful
`and informative, and it helps us to interpret advertising-to-sales
`ratios for pharmaceuticals. In 1998, for example, ratios for primarily
`“search good” companies were Home Depot, 1.6 percent; Phillips
`Electronics, 2.4 percent; American Express, 3.2 percent; Circuit
`City, 4.5 percent; Sony Corporation, 4.7 percent; and Intel, 5.8 per-
`cent. In contrast, 1998 ratios for “experience good” companies were
`Ralston Purina, 14.9 percent; Wendy’s International, 16.7 percent;
`Coors, 18.8 percent; McDonald’s, 21.1 percent; Estée Lauder Cosmet-
`ics, 22.2 percent; Revlon, 24.0 percent; and L’Oréal, 26.5 percent.22
`Clearly, prescription drugs are predominantly experience goods,
`and thus one would reasonably expect that advertising-to-sales ra-
`tios for them would be relatively high. Moreover, since physicians
`primarily make prescribing decisions, much pharmaceutical mar-
`keting is focused on them, with detailers providing information and
`free samples to physicians to encourage them to experiment with
`their product. How one measures total marketing expenditures and
`ratios for prescription drugs is inherently ambiguous and problem-
`atic, particularly when marketing involves traditional and less tradi-
`tional promotions. If one includes as components of marketing jour-
`nal advertising, detailing to physicians, product samples, and
`direct-to-consumer (DTC) marketing, ratios of 10–20 percent are
`not unreasonable estimates. Using this definition of marketing, IMS
`Health estimates that in 1999 drug companies spent $13.9 billion on
`marketing, which when combined with IMS Health final sales esti-
`mates of $113.0 billion, yields a marketing-to-sales ratio of 12.3 per-
`cent. This is a higher ratio than for, say, Sony at 4.7 percent, but is
`lower than that for McDonald’s at 21.1 percent.
`In the past decade, U.S. drug companies have marketed their
`experience goods aggressively. Marketing provides technology-
`transfer information to patients and providers on efficacy in the
`treatment