Federal government-interest patent disclosures for recent top-selling drugs

Abstract

Objective: To provide updated evidence on government-interest patent disclosures in US patents for top-selling small-molecule drugs.

Methods: IQVIA National Sales Perspectives data identified 300 top-selling drugs, defined by peak 2013–2017 US sales. For the 197 approved through New Drug Applications (NDAs), data were collected from a recently-released dataset of all patents listed in 1985–2016 Annual Editions of the FDA Orange Book. Data on patent assignees and Government Interest Statements (if any) were collected from the US Patent and Trademark Office online database. The percentage of drugs with at least one government-interest patent disclosure was calculated, as was the percentage of patents with a disclosure, by type (drug product, drug substance, or method of use). Government-interest patent disclosures were defined as those for which: the patent application contained a Government Interest Statement; and/or any of the patent assignees was a US government agency.

Results: Few patents for the top-selling drugs had a government-interest patent disclosure, 2.6% on average. By patent type, figures ranged between 1.6% (for patents with drug product claims) and 3.6% (for patents with drug substance claims). Accounting for multiple patents per drug, 8.6% of top-selling drugs analyzed had at least one patent with a Government Interest Statement; 1.5% had at least one with a US government agency assignee; and 10.2% met either criterion (none met both).

Limitations: Analyses were limited to top-selling NDA-approved drugs (generally excluding biologics) and Orange Book-listed patents. Patents with government-interest patent disclosures could also have relied on non-government funding. Patents were not characterized by relative economic investments, importance in the discovery and development process, or contribution to clinical value.

Conclusion: Results were generally comparable to a prior analysis that found that 9.0% of new drugs approved between 1988 and 2005 had either a Government Interest Statement disclosure or a government agency first-listed patent assignee.

Keywords:

• Prescription drugs
• economics
• patents
• research and development

JEL CLASSIFICATION CODES:

• I11
• O31

Introduction

The Bayh-Dole Act and Federal Government-interest patent disclosure requirements

The University and Small Business Patent Procedures Act of 1980 (the “Bayh-Dole Act” or “Bayh-Dole”) governs the intellectual property rights associated with federally-funded research, such as basic research grants from the National Institutes of Health (NIH) to universities and other organizations (e.g. teaching hospitals and non-profit research institutes). It changed the research and development (R&D) landscape in the US in at least two fundamental ways. First, it created consistency across federal agencies by creating a uniform framework to encourage technology transfer from academic organizations to the private sector. Second, Bayh-Dole allowed universities to retain the title to patents arising directly from their sponsored research activities. With these clear patent rights in-hand, universities are then free to license the right to use these technologies to private sector partners, who then assume the multi-year risk and cost of development and commercialization¹. By allowing universities to retain patent rights arising from federal research grants, Bayh-Dole created a defined and viable route for patentable research insights from the university to make their way from “bench to bedside”². Before Bayh-Dole, the government had the option to retain title rights to all inventions resulting from federally-funded research and development. While grantees could request a waiver of the government’s right to title either in advance during contract negotiations, or on a case-by-case basis after they disclosed the invention to the federal agency sponsoring the research, the process was thought to be inconsistent and a disincentive to private sector investment, with one observer referencing the often-cited estimate that “by 1978, the government had licensed less than 5% of the 25,000 to 30,000 patents it owned”³.^*

Patent rights are viewed as essential incentives to develop new drugs, due to the lengthy, costly, and risky nature of the innovative (R&D) process, the lower cost and risk of “free-rider” imitation (in the form of generic copies, which can be marketed after patents and any statutory exclusivity provisions expire, entering the market on average 12–14 years after brand launch⁴), and the likelihood that early-stage, pre-revenue companies only have intellectual property assets with which to attract investment capital to fund continued development. With the right to retain and license the results of federally-funded research, universities and other recipients of these federal funds (such as teaching hospitals, and non-profit research institutes) must meet certain requirements, including filing for patent protection on behalf of individual inventors (for patentable inventions) and sharing a portion of license revenue with inventors. In addition, when an organization (such as a university) claims title to a subject invention for which it has obtained assignment, it must disclose the government’s interest in patent applications and notify the government before abandoning any patent application⁵.^† These disclosures of government support for research can be found in the form of “Government Interest Statements” in those patents (sometimes referred to as “Bayh-Dole disclosures”).

Private and public sector roles in drug discovery and development

Untangling the various aspects of the complementary and evolving roles played by the public and private sectors in the lengthy and complex process of scientific discovery and drug development is challenging. Not only does the nature of scientific activities change as technology advances, but collaboration may evolve and take different forms across the biopharmaceutical innovation ecosystem, from licensing to “industry-academic partnerships, venture capital, disease foundations, as well as public-private, pre-competitive consortia”⁶. Earlier, researchers explored interactions between public and private sector researchers across “upstream” research exploring the basic physiological and biochemical mechanisms underlying disease and “downstream” applied clinical research. They found evidence of valuable and complex information exchange across public and private sector researchers and the development process in a set of case studies, supplemented with quantitative analysis of publishing activity⁷. Later, researchers undertook case study reviews of subsets of drugs to explore these topics further^8–10.^‡

Others explored the public sector’s role in discovery and drug development by means of quantitative analysis of the information in Government Interest Statements in patents. These researchers calculated the percentage of drugs approved by the Food and Drug Administration (FDA) with patents listed in the so-called FDA “Orange Book” (formally the “Approved Drug Products with Therapeutic Equivalence Evaluations”) that acknowledged government support in those patents¹¹. This research¹¹ examined a sample of 379 New Drug Applications (NDAs) approved between 1988 and 2005 for drugs classified as New Molecular Entities (NMEs) and that had at least one patent reflected in the Orange Book.^§ The authors found that 9.0% of these 379 drugs either had: a Government Interest Statement disclosure of funding; or a US government agency as the first-named patent assignee (or both).

One of the barriers to additional systematic analysis has been the need for a complete list of patents linked to each drug on which to perform the analysis, to ensure that all relevant patents for a given set of drugs are included in the analysis. Due to the nature of the FDA Orange Book database (described further below), this has proved challenging. The objective of this study is to update the previous analysis of the rates of government-associated patent disclosures, focusing on a set of recent top-selling drugs, and using a novel comprehensive database of US Orange Book-listed patents recently made available by others that ensures that the list of patents associated with the drugs analyzed is complete.

Methods

Data sources

Data for the analysis were combined from the following sources: IQVIA sales data to identify the set of top-selling drugs to be analyzed; FDA Orange Book-listed patent data from the newly-available database; corresponding data from a current version of the FDA Electronic Orange Book (EOB) for drugs approved in 2016; and data from the US Patent and Trademark Office (USPTO) online database on patent assignee names and Government Interest Statement text. Each is described below.

IQVIA (formerly IMS Health) National Sales Perspectives data identified the top-300-selling brand drugs for the period 2013–2017, as defined by peak annual US invoice sales computed over the 5-year period at the brand-level (noting invoice sales exclude discounts and rebates). Peak annual sales were used in order to identify high-selling drugs in any of the years. Drugs were restricted further to those approved by the FDA for marketing through the NDA process, as patents for these drugs are listed in the Orange Book; patents for drugs approved by the FDA through the Biologics License Application (BLA) process are not listed in the Orange Book. After excluding such BLA-approved drugs, the final list of drugs to be analyzed totaled 197 top-selling NDA drugs from the years 2013–2017.

The FDA Orange Book contains a list of all prescription drug products that are approved by FDA for safety and effectiveness, together with their therapeutically equivalent generic drugs. Since the passage of the 1984 Hatch-Waxman Act, it also contains patent information. A version of this information is available on-line. This electronic version of the Orange Book (the “EOB”) contains the current version of information about drugs, their listed patents, and their generic equivalents. One of the challenges for researchers is that it is a “live” database, rather than a cumulative set of all information about drugs and their patents ever listed. It does not include expired patents, nor information for drugs that are discontinued or no longer marketed, for example for drugs that have experienced generic entry or been removed from the market.

To counteract these limitations, others recently have compiled a cumulative electronic version of the patent and exclusivity information from the Annual Editions of the Orange Book for the period 1985–2016.¹² The Annual Editions for each year are current with information as of the end of the previous year. Data were digitized by the researchers from PDF copies of Annual Editions of the Orange Book from 1985 through 2016, obtained in response to a Freedom of Information Act Request (FOIA) submitted to the FDA. This Orange Book patent database contains patent observations (i.e. patent number, patent expiration date, patent type) by Annual Edition year, application type (i.e. NDA; Abbreviated New Drug Application or ANDA, used for generic drug applications; or Not Available), FDA application number, active ingredient, trade name, and product number (corresponding to unique combinations of dosage form-route of administration-strength combination for a given NDA number). Edition years span 1985–2016 (omitting 1986, when no Orange Book was published). Patent type indicates whether the patent listed is associated with a drug substance (DS) claim code covering the active ingredient, a drug product (DP) claim code covering the formulation and composition, or a method of use (MOU) claim code covering a particular indication or use of the drug product.^∥ An individual patent may be associated with one or more types of codes, including one or more method of use codes, depending on the specific claims appearing in the patent.

Further information regarding the new Orange Book patent database, including details of the process used to convert the data from PDF copies to text files, and to perform cross-checks and random audits to confirm accuracy, is available on-line.¹³

The list of top-selling 2013–2017 drugs was merged with data from the new Orange Book patent database. Where there were naming mismatches between brand drug names appearing in the sales data and in the Orange Book patent data, they were resolved manually. To be conservative (i.e. to increase the probability a brand drug would be identified as having at least one government-associated patent disclosure), the more general brand name in the sales data was adopted.^¶ Orange Book patent-listing data on drugs launched in a given year is provided in the subsequent year’s edition (e.g. patent data for drugs launched in 2007 will appear for the first time in the 2008 edition). For drugs approved in 2016, comparable EOB information was gathered directly from the current online version.^** These data would have been provided in the 2017 and 2018 editions. Because each strength results in a separate product number in the Orange Book, patents are often listed multiple times for the same brand-name drug in the same edition year. All listed strengths of the top-selling drugs were considered in the calculations; if at least one strength had a patent listed meeting the criteria, it was characterized as having a government-interest patent disclosure. In addition, each Annual Edition includes data on all drugs, their product numbers, and their patents included in the Orange Book in that year, and thus much information is duplicated from year to year.

This process resulted in a total of 26,860 data entries for analysis for the 197 drugs over the period 2013–2017, and 1,151 unique patent numbers.^†† (Observations take the form, e.g. Annual Edition 2004, Brand Name, NDA Application Number, Product Number, Patent Number, Patent Claim Type Indicators).

As noted, patent information was added to the Orange Book after the passage of the Hatch-Waxman Act of 1984. The new Orange Book patent database reflects use codes (indicating an MOU patent) beginning with the 1988 edition, and DS and DP indicators beginning with the 2004 edition. DS and DP indicator variables are coded “N/A” for edition years 1985–2003; thereafter, they are coded either “0” or “1”. Patents appearing both in edition years before and after 2004 reflect multiple codes, by year – “N/A” for the prior years (1985–2003) and either “0” or “1” for years 2004 and later. As a result, many of the patents included in editions before 2004 are nonetheless characterized. Of the 1,151 unique patents in the dataset for analysis, only 46 (or 4%) have no patent type information entered in any edition year.

Information was collected from the USPTO on each of the 1,151 unique patent numbers listed for any of the top-selling drugs, by using the online search function, to determine which of them contained at least one government-interest patent disclosure.¹⁴ Data were collected on the names of all patent assignees to determine whether any is a US government agency, together with the text appearing in the Government Interest Statement (if any).

Definitions and calculations

Generally consistent with prior research¹¹, a government-interest patent disclosure was defined as one which met either or both of the following criteria: first, a “Bayh-Dole disclosure” appeared in the Government Interests section of the patent (e.g. “The invention described herein was made with Government support under grant no. xxx awarded by the National Institutes of Health. The Government has certain rights in this invention”); second, the patent was assigned to a US government entity (e.g. “The United States of America as represented by the Department of Health and Human Services” is listed as a patent assignee).

The Government-Interest statement appears in a separately-labeled section of the patent titled “Government Interests”, directly following the section titled “References Cited”. It is clearly and consistently labeled, always found in the same location, and non-ambiguous in interpretation. All listed patent assignees were included in the data analyzed.

As noted in the prior research by others, this definition relies on patent holders accurately reporting government support in their patent applications. Contractors are required to file a patent application on an elected subject invention within specified periods and to acknowledge government support (if relevant), and there are penalties for non-disclosure. Others have noted the potential for under-reporting government interests in patents, while also observing that determining whether a subject invention was in fact “conceived of or first actually reduced to practice in the performance of work under a funding agreement” (the relevant Bayh-Dole reporting standard) can be difficult and subject to complex patent litigation¹⁵.

With regard to Orange Book listings, manufacturers are required by law to list eligible patents in the Orange Book, even after FDA approval.^‡‡ Moreover, they have an incentive to do so because companies seeking to market a generic version of the drug must certify in their Abbreviated New Drug Application (ANDA), for each patent listed by the manufacturer in the Orange Book, the basis for that challenge. The basis may be that the patent either has expired, will expire before the generic drug comes to market, or is not subject to a patent; or that the patent is not valid or will not be infringed by generic entry, which triggers so-called “Paragraph IV” certification litigation between the parties. When manufacturers respond by timely suing over the infringement for a listed patent, an automatic 30-month stay of FDA generic drug approval is issued, in order to allow for the litigation to proceed and resolve the matter. However, not all patents relating to the drug must, or may, be listed. Patents related to methods of manufacturing or to formulations and methods of use that do not relate to the approved drug product and its approved indications may not be listed (they would be listed, if relevant and allowable, with future applications).^§§

As described, the percentage of drugs having at least one patent with a government-interest patent disclosure was calculated at the drug brand name-level, which may encompass multiple NDA numbers, FDA-listed product numbers, and patents (in the data, product numbers range from a minimum of one to a maximum of 13 for a given drug). For the 197 drugs analyzed, there were 26,860 total Orange Book Annual Edition patent observations in the dataset, and 1,151 unique patent numbers. If at least one of the dosage-strength observations for a given branded drug had an Orange Book-listed patent meeting the definition of having a government-interest patent disclosure, the drug was recorded as having met the definition.

All figures presented are simple averages, at either the drug- or patent-level (i.e. they are not weighted by sales or any other variable).

Results

Patents by claim type and percentage with a government-interest patent disclosure

Table 1 shows the total number of Orange Book-listed patents in the dataset for the drugs reviewed, together with the percentage of each type. Of the 1,151 unique patents associated with the set of top-selling drugs, drug product (DP) and method of use (MOU) claim patents were the most common, with somewhat over half of the patents having DP and MOU claims (59% and 58%, respectively). Drug substance (DS) claim patents were about half as common, with an average of 29% of patents having a claim of this type. Because individual patents may include multiple claim types, the percentages by patent claim types do not sum to 100%. Forty-six of the 1,151 patents (4%) did not contain patent claim type data (i.e. DS, DP, MOU) in the FDA data. They have not been excluded from the calculations, unless otherwise noted.

Table 1. Orange Book-listed patents for top-selling 2013–2017 drugs, with patent claim types.

	All Orange Book-listed patents	Percent of all patents with given claim type
Method of use patents	666	57.9%
Drug product patents	677	58.8%
Drug substance patents	335	29.1%
Total Orange Book patents	1,151	N/A*

*Patent claim types do not sum to total; individual patents may include multiple claim types.

Notes: Method of use (MOU), drug product (DP), and drug substance (DS) patent claim types are as recorded in FDA Orange Book.

Source: Author’s analysis of FDA Orange Book, US PTO data (see Notes 11, 15, and 17).

Table 2 shows the total number and percent of Orange Book-listed patents in the dataset for the drugs reviewed that have a government-interest patent disclosure (i.e. have either a Bayh-Dole mandated disclosure in the Government Interest section of the patent, or have a US government patent assignee among any of the assignees, or both).

Table 2. Orange Book-listed patents for top-selling 2013–2017 drugs, by patent claim type and government-interest patent disclosure.

	All Orange Book-listed patents	Patents with government-interest patent disclosure	Percent with government-interest patent disclosure
Method of use patents	666	21	3.2%
Drug product patents	677	11	1.6%
Drug substance patents	335	12	3.6%
Total Orange Book patents	1,151	30	2.6%

Notes: Method of use (MOU), drug product (DP), and drug substance (DS) patent claim types are as recorded in FDA Orange Book. Patent claim types do not sum to total; individual patents may include multiple claim types. Government-interest patent disclosure is either: government-interest statement patent disclosure; or US government patent assignee (or both).

Source: Author’s analysis of FDA Orange Book, US PTO data (see Notes 11, 15, and 17).

Few patents of any type for the top-selling drugs have a government-interest patent disclosure. On average, 2.6% of the patents had a government-interest patent disclosure; by patent type, figures ranged between 1.6% (for patents with DP claims) and 3.6% (for patents with DS claims).

Average numbers of patents per drug by patent type

Table 3 shows the average number of Orange Book-listed patents per drug, in total and segmented by patent claim type (i.e. drug substance claim, drug product claim, method of use claim), for all drugs reviewed and for drugs with at least one government-interest patent disclosure (excluding drugs with no listed unexpired patents in any Orange Book edition). The number of patents per drug averaged 6.0 for all drugs, and 5.7 for drugs with at least one government-interest patent disclosure (as reflected in the standard deviations reported, there was high variation in the numbers of patents per drug). For drugs with at least one government-interest patent disclosure, the number of patents per drug varied from an average of 2.0 for drug substance patents per drug, 3.2 for drug product patents, and 3.4 for method of use patents. As noted, a given patent may include more than one type of claim (so the sum by type exceeds the total number of patents per drug); calculations exclude drugs without a patent of the specified type.

Table 3. Average number of Orange Book-listed patents per drug, by type of patent.

	Drug substance patents	Drug product patents	Method of use patents	Total Orange Book patents
All drugs reviewed	2.2 (2.1)	3.9 (4.4)	3.8 (4.0)	6.0 (6.2)
Drugs with at least one government-interest patent disclosure	2.0 (3.2)	3.2 (3.6)	3.4 (3.4)	5.7 (5.1)

Notes: Method of use (MOU), drug product (DP), and drug substance (DS) patent claim types are as recorded in FDA Orange Book. Calculations exclude drugs without a patent of the specified type; individual patents may include multiple claim types. Government-interest patent disclosure is either: government-interest statement patent disclosure; or US government patent assignee (or both). Averages exclude drugs with no unexpired patents in any Orange Book edition data. Figures in parentheses are standard deviations.

Source: Author’s analysis of FDA Orange Book, US PTO data (see Notes 11, 15, and 17).

Percentage of drugs having at least one government-interest patent disclosure

Table 4 shows the percentage of drugs reviewed which had at least one Orange Book-listed patent with a government-interest patent disclosure (i.e. either a Bayh-Dole-mandated disclosure in the Government Interest section of the patent, or a US government patent assignee, for any of its Orange Book-listed patents). As described above, calculations were performed at the patent-level, and rolled up to the drug-level, allowing for multiple patents per drug, and the possibility that any one of them (or more than one) may have a government-interest patent disclosure. Of all the drugs in the sample, 10.2% of top-selling 2013–2017 drugs met either criteria for government-interest patent disclosure; 8.6% of drugs had at least one Bayh-Dole disclosure in any of its patents; an additional 1.5% had at least one government patent assignee in any of its patents (no drugs met both criteria).

Table 4. Top-selling 2013–2017 drugs with at least one government-interest patent disclosure.

	Percent with at least one patent with government-interest statement	Percent with at least one patent with government patent assignee	Percent with either government-interest patent disclosure measure
Number of drugs reviewed	8.6% (17 of 197)	1.5% (3 of 197)	10.2% (20 of 197)

Notes: Government-interest patent disclosure is either: government-interest statement patent disclosure; or US government patent assignee (or both).

Source: Author’s analysis of FDA Orange Book, US PTO data (see Notes 11, 15, and 17).

A robustness check was conducted to assess whether patents not reflected in the most recent 2016 edition of the Orange Book provided by the FDA (but which would be reflected in subsequent editions) would have an impact on these findings. As noted, analysis of top-selling drugs launched after January 1, 2016 relied on a current version of the EOB (as of September 25, 2018), thus including data that would be included in future 2017 and 2018 Orange Book editions. Analysis of drugs launched in 2015, however, relied only on data from the 2016 edition. As a check, drugs launched in 2015 that were characterized as not having a government-interest patent disclosure were re-reviewed, using data from the current EOB. This had the effect of providing these drugs the equivalent of two-and-a-half years of additional FDA patent-listing data. All patents listed in the current EOB for these drugs were reviewed using the online USPTO database, and none of these drugs were characterized as having a government-interest patent disclosure as a result.

Discussion

The findings are generally consistent with those of earlier researchers¹¹, who analyzed a sample of 379 new drug applications (NDAs) approved between 1988 and 2005 that were classified as new molecular entities (NMEs) and had at least one patent reflected in the Orange Book. The authors found that 9.0% of these 379 new drug applications had a so-called “public-sector” patent (defined as either having a Government Interest Statement disclosure of government funding, or having the first-named patent assignee a US government agency). The definitions of government-interest patent disclosure were closely aligned between the prior and current analysis. While the current analysis reviewed all listed patent assignees, rather than only the first-named patent assignee, in the current sample, where a US government agency is listed as a patent assignee, it is always listed as the first-named assignee. In assessing whether drugs have a government-interest patent disclosure, the current research retrieves additional patent-year values that may not have been available in the prior research. As a result, it might be expected to yield somewhat higher results.

With regard to the sequencing of public and private sector patent activity, this analysis provides some evidence of interesting variation which may merit further analysis. Of the drugs with at least one patent with a government-interest patent disclosure, the earliest awarded patent was about as likely to have a government-interest patent disclosure as not. In eight instances, a patent with a government-interest patent disclosure had the earliest award date; in ten instances, a patent without a government-interest patent disclosure had the earliest date; in two instances, only patents with government-interest patent disclosures were listed.

Limitations

In addition to the limitations noted earlier (namely, the potential for non-disclosure in the Government Interest Statement), four important types of limitations of this analysis with potential implications are acknowledged. First, the patents characterized as having government-interest patent disclosures could also have relied on sources of funding other than the disclosed federal grants, such as industry investment or other funding sources. The degree to which this is true, and the respective levels of government and private sector funding for patents with government-interest patent disclosures is unknown.

Second, no attempt was made to distinguish among the various patents associated with a given drug with regard to the relative: economic investments associated with them; importance in the discovery and development process; or contributions to eventual clinical value. The difficult task of proposing, much less applying, a systematic method for doing so was not attempted. No attempt was made to distinguish between cases in which the patent with a government-interest patent disclosure was one of numerous patents and/or played a relatively minor role in the drug’s development, approval, and clinical value, vs cases in which it was one of only a few patents and/or played a major role.

Third, the analysis relies on Orange Book-listed patents, and drugs approved through the NDA process. Its findings cannot be extrapolated to biologics, which are likely to exhibit different patent profiles. While the FDA’s new Purple Book (formally, “Lists of Licensed Biological Products with Reference Product Exclusivity and Biosimilarity or Interchangeability Evaluations”) provides information for licensed biological products and whether a biosimilar has been approved referencing that product (and whether it has been licensed as biosimilar to or interchangeable with the reference product), it does not include patent information.^∥∥ Different strategies, therefore, are required to achieve comparable levels of consistency and completeness for studies of biologic patents.

Last, the study is limited to top-selling drugs. The analysis is, therefore, focused on the most economically important drugs, and the results may not be representative of all drugs, which may be less or more likely to have at least one government-interest patent disclosure.

Conclusions

Recent top-selling drugs have maintained rates of government-interest patent disclosure, as measured by having either government-interest statement patent disclosures, or US government patent assignees (or both), broadly comparable to earlier estimates. Of Orange Book-listed patents for the top-selling small-molecule drugs in the sample, 2.6% had a government-interest patent disclosure; 10.2% of the top-selling drugs had at least one patent with a government-interest patent disclosure.

Transparency

Declaration of funding

Analysis Group, Inc. received financial support from the Pharmaceutical Research and Manufacturers of America for this research.

Declaration of financial/other relationships

GL is an employee of Analysis Group, Inc., a consulting company that has provided services to biopharmaceutical manufacturers, both brand-name and generic. The peer reviewers on this manuscript have received an honorarium from JME for their review work. In addition, a reviewer on this manuscript discloses receiving grants and honoraria from companies developing, producing, and marketing NOACs. The reviewers have no other relevant financial relationships or otherwise to disclose.

Acknowledgements

The author gratefully acknowledges the assistance of Michael Kleinrock of IQVIA in providing the data used to identify the top-selling drugs, Heidi Williams of Standford University for making the new Orange Book patent data publicly available and for helpful communications, and Harriet Jeon for valuable data collection and analysis assistance. No other assistance in the preparation of this article is declared.

Notes

* Stevens et al.³; referencing US Government Accountability Office, Federal Research: Information on the Government’s Right to Assert Ownership Control Over Federally Funded Inventions. Washington, DC. 2008 (GAO–09-742).

† 35 USC 202(a)–(f). For a summary of the requirements, see, for example, Loise and Stevens⁵.

‡ See Zycher et al.⁸. More recently, two sets of authors examined the same set of clinically transformative drugs, emphasizing different aspects, but both finding that public–private sector collaboration was key to the clinical innovations studied. See Chakravarthy et al.⁹ and Kesselheim et al.¹⁰.

§ The authors’ original sample totaled 478, or all the new molecular entities approved between 1988 and 2005. Of this total, 99 had no Orange Book-listed patents (one-quarter of which were antibiotics not subject to Orange Book requirements until 2008), leaving 379 for their patent analysis.

∥ 21 CFR 314.53(d)(5).

¶ For example, in some cases the Orange Book reflected differing proprietary names for some drugs by dosage form and strength, whereas the sales data were consolidated under a single brand name. In other instances, there were minor naming differences. One drug was renamed after launch in order to avoid confusion, leading to misalignment in the data sources.

** FDA Electronic Orange Book, as of September 25, 2018.

†† Six of the 197 drugs had no patents listed in the new Orange Book patent database, nor in the current EOB. Unless otherwise noted, they are included in the calculations for drugs, but not for patents.

‡‡ 21 USC 355(c)(2).

§§ For additional detail on “Patents for which information must be submitted and patents for which information must not be submitted”, see 21 CFR 314.53(d)(5). Available from: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=314.53.

∥∥ US FDA, CBER and CDER Lists of Licensed Biological Products. See, https://www.fda.gov/Drugs/DevelopmentApprovalProcess/HowDrugsareDevelopedandApproved/ApprovalApplications/TherapeuticBiologicApplications/Biosimilars/ucm411418.htm.