Abstract
How corporations surveil and influence consumers using big data tools is a major area of research and public debate. However, few studies explore it in relation to physicians in the USA, even though they have been surveilled and targeted by the pharmaceutical industry since at least the 1950s. Indeed, in 2010, concerns about the pharmaceutical industry's undue influence led to the passing of the Physician Sunshine Act, a unique piece of transparency legislation that requires companies to report their financial ties to physicians and teaching hospitals in a public database. This article argues that while the Sunshine Act has clearly helped expose important commercial influences on both prescribing and the scale of industry involvement with physicians, it has also, paradoxically, fuelled further commercial surveillance and marketing. The article casts new light on innovative pharmaceutical marketing approaches and the key role of data brokers and analytics companies in the identification, targeting, managing, and surveillance of physicians. We place this analysis within the political economies of the pharmaceutical industry, surveillance-based marketing, and transparency, and argue that policies to promote increased transparency must be tightly tied to policies that impede the commodification and use of transparency data for surveillance and marketing purposes.
Introduction
Critical social science scholarship has explored the pharmaceutical industry's power across a range of health, social, political, economic, and scientific spheres (e.g. Abraham, 2008; Busfield, 2006; Light et al., 2013). In this scholarship, a particular concern has been the industry's strategic management of financial relations with other health actors, especially health professionals (Sismondo, 2013), patient organizations (Rose, 2013), and regulators and policymakers (Lexchin and O’Donovan, 2010), for the purpose of influencing them and their environment, often out of public sight. This concern resonates with influential policy critiques of corporate marketing (Avorn, 2017) and of financial conflicts of interest in medicine and their limited transparency (Field and Lo, 2009).
However, about a decade ago, drug companies started routinely disclosing vast amounts of data on their financial relations, giving unprecedented insights into corporate payments to other health actors and counteracting complaints about limited transparency (Grundy et al., 2018). By far, the most recognized transparency initiative internationally is the US Physician Payment Sunshine Act. Since 2013, this legislation has required companies to report, in a public database, payments to as well as ownership and investment interests of physicians and teaching hospitals. Yet research and commentary have often concluded that the Sunshine Act falls short in delivering enough transparency to allow an understanding of the full scope of the industry's influence, and have therefore called for even more extensive disclosures (e.g. Kang et al., 2019; Kanter et al., 2019; Lexchin and Fugh-Berman, 2021; Ornstein, 2017).
The main argument of this article is that this line of critique overlooks one major problem with the routine disclosure of volumes of granular payment data: this data deluge can also be used to inform and further intensify the commercial surveillance and manipulation of physicians, especially when combined with existing streams of commodified physician-level data such as prescribing data. Put bluntly, the Sunshine Act is predicated on a critique of corporate influence, but the public release of payment data can have the paradoxical effect of sharpening companies’ marketing tools by helping them improve the identification, targeting, managing, and surveillance of physicians. Furthermore, corporate uses of disclosure data are only likely to increase as pharmaceutical marketing becomes ever more dominated by big data practices that allow the greater ‘personalization’ of sales tactics.
Below, we analyse commercial surveillance of physicians before and after the implementation of the Sunshine Act to illustrate how companies can capitalize on the public release of industry-wide data on physicians’ financial relations. Combining insights from critical scholarship into, on the one hand, the political economy of the pharmaceutical industry and, on the other hand, surveillance-based marketing and transparency, we argue that advocates of industry transparency – among whom we count ourselves – should broaden their field of vision to encompass the problem of commercial uses of ‘transparency data’. In the final section of the article, we argue that policies to promote increased transparency must be tightly coupled to policies that impede the commodification and use of transparency data for surveillance and marketing purposes.
The Sunshine Act: Background and policy debates
The last decade has seen a global trend towards increased transparency in the pharmaceutical industry (Fierlbeck et al., 2021). Among the most far-reaching innovations are frameworks to monitor health actors’ financial relations with drug companies through the public release of data on industry payments (Grundy et al., 2018). In the USA, this enhanced transparency has emerged against a background of high-profile court cases in which current or former employees of many large pharmaceutical companies acted as ‘whistle-blowers’ to help uncover a range of marketing tactics (Ornstein, 2017) – some unmistakably fraudulent and medically risky (Kesselheim et al., 2011). Whistle-blower complaints have informed prolonged investigations by the Department of Justice as well as other federal and state bodies that involve the interrogation of complainants, subpoenas for documents or electronic records, witness interviews, expert consultations, and sometimes search warrants to obtain further evidence (Vilhelmsson et al., 2016). An extensive examination of company documents and witnesses in the course of these investigations has provided extraordinary insight into the complex range of marketing strategies that companies engage in, and into the coordinated and planned nature of their campaigns (Kesselheim et al., 2011; Steinman et al., 2006).
It was in 2010, as evidence of widespread maleficence was rapidly accumulating, that Congress passed its Sunshine Act, formally section 6002 of the Affordable Care Act, requiring pharmaceutical and medical device companies to, from the year 2013, report on payments over USD 10, or annual cumulative payments over USD 100, to named physicians and teaching hospitals in the Open Payments Database. Subsequently, a few European countries, including France and Portugal, have enacted Sunshine Act-type legislation (Grundy et al., 2018), but most rely, in one way or another, on industry self-regulation – meaning disclosure rules implemented by industry trade groups rather than governments (Ozieranski et al., 2021). Suffice it to say here that no country, regardless of the regulatory solution, compares with the USA in terms of the quantity, granularity, and user-friendliness of the disclosed data (Ozieranski et al., 2021). For 2019, for example, the Open Payments Database included a staggering 10.96 million total records attributable to 615,000 physicians, 1194 teaching hospitals, and 1601 companies, amounting to USD 10.03 billion in payments, ownership, and investment interests, including for hospitality, travel, lodging, and entertainment reimbursements as well as consulting fees, royalties, and research grants. Each database record contains information that contextualizes the payment, such as donor and recipient identification; the value, nature, and date of the payment; and the drug or device product or products pertinent to the payment (CMS, 2021a).
There has been considerable research and debate on the benefits of the Sunshine Act for various stakeholders (e.g. Agrawal and Brown, 2016) and on possible unintended negative effects (Loewenstein et al., 2012). According to the official policy narrative, the Open Payments Data primarily seeks to benefit patients as health care consumers by allowing them to become better informed about the potential influence of industry ties on their physicians; moreover, payment transparency is also hoped to deter physicians from accepting payments that patients might view as suspect (Grundy et al., 2018). However, Kanter et al. (2019) used experimental evidence to show that Open Payments has failed to deliver increased knowledge to patients of their doctors’ ties to companies. Instead, its main benefit seems to lie in exposing the scale of industry involvement with and influence on US physicians. Open Payments has triggered extensive research on the distribution of industry payments among different medical specialities (e.g. Tierney et al., 2016) and on the links between these payments and drug prescriptions (e.g. Hadland et al., 2019) and costs (e.g. Mejia et al., 2019), showing consistent dose–effect relationships with prescriptions that benefit industry (Mitchell et al., 2021) but that can be detrimental to the patient and public health (Hadland et al., 2019). Furthermore, Open Payments has helped prosecutors identify unusual payment patterns, sometimes indicating undue influence or corruption (Department of Justice, 2016).
At the same time, several researchers have maintained that payment disclosure alone cannot solve larger underlying problems of the ‘institutional corruption’ of medicine (Light et al., 2013), namely, ‘that the pharmaceutical industry has a disproportionate influence on medical opinion, which weakens medicine's ability to promote individual and public health in ways that are independent of the industry’ (Sismondo, 2013: 636). Furthermore, it has been noted that many of the larger payments that companies make to physicians are primarily intended not to affect their prescriptions, but rather to purchase their influence on other physicians (Sismondo, 2013), meaning that studies tend to underestimate the effects of these payments (Winn et al., 2021).
Still, even among sceptics, payment disclosure is touted as ‘ethically desirable’ (Sismondo, 2013: 640), and as providing useful data that can support arguments for broader policy reform (Fierlbeck et al., 2021). Indeed, a recent critical appraisal of the ‘false solution of transparency’ concluded that existing disclosures ‘should be expanded to include a wider range of healthcare professionals, institutions, and organizations’ (Lexchin and Fugh-Berman, 2021: 3). In fact, in response to critiques of incomplete transparency, disclosures will be expanded in 2022 to also include nurse practitioners, physician assistants, clinical nurse specialists, registered nurse anaesthetists, and certified nurse-midwives (CMS, 2021b). Furthermore, some scholars have called for an extension of the Sunshine Act to also encompass patient organization recipients of payments (Kang et al., 2019), which, if accepted, would produce additional streams of payment data (Ozieranski et al., 2019).
However – and as the remainder of this article seeks to show – if researchers can use this data to expose both important payments that influence prescriptions and the scale of industry involvement with US physicians, so can the pharmaceutical industry with its unrivalled financial, informational, and analytic resources. This points to a separate and arguably bigger problem with transparency-based strategies in the pharmaceutical industry: it is not only that transparency alone cannot address power asymmetries, but transparency may actually exacerbate them by handing over information to companies that can be turned into surveillance and marketing assets.
Political economy of pharmaceutical industry surveillance and marketing
In developing our argument about this problem of transparency-based strategies, we will draw on, and connect and contribute to, three related strands of critical political economy literature. The first strand is situated within broader scholarship on the pharmaceutical industry that is concerned with industry power over research, innovation, regulation, and medical practice (e.g. Busfield, 2006; Davis and Abraham, 2013; Dumit, 2012; Fierlbeck et al., 2021; Fisher, 2008; Light et al., 2013; Mirowski and Van Horn, 2005). A key insight from this work is that power may not be visible to most, both because power is actively concealed – as in the case of the ‘ghost management’ of science and prescribers (Sismondo, 2007) – and because it is often normalized or hegemonic – as in the case of the industry's financial relations with and influence over many health system actors (Sismondo, 2017). Another key insight from this work is the need to consider not only pharmaceutical companies but also many auxiliary industries with which drug companies are tightly networked, and that together exert pervasive commercial influence (Busfield, 2020; Fisher, 2008; Mirowski and Van Horn, 2005). These include data broker, business intelligence, and marketing companies that collect vast amounts of data on US physicians and monetize them by selling marketing-related services and products to pharmaceutical companies that seek to shape medical opinion and practice, often without physician knowledge (Applbaum, 2011; Greene, 2007; Sismondo, 2017).
Concerns about the pharmaceutical industry's capacity to track physicians ‘behind the scenes’, and use the data gained to surveil and influence medicine, resonate with critical marketing scholars’ research into the opaque dynamics of the commodification of consumer data within pervasive commercial surveillance (Arvidsson, 2003; Crain, 2018; Darmody and Zwick, 2020; Pridmore and Lyon, 2011; Pridmore and Zwick, 2011). This second strand of literature argues that the commercial surveillance of consumers increased significantly starting in the 1980s, as a result of the over-production of products and services, hyper-competition, and rapid improvements in big data practices. As can be exemplified by pharmaceutical marketing (Applequist, 2017), key players are the data broker companies that collect and monetize a broad range of consumer data on a massive scale (Crain, 2018), and whose products and services have a great influence on businesses and governments and, ultimately, individuals’ choices and behaviour (Fourcade and Healy, 2017; Sadowski, 2019). Modern marketing theory and technical discourse have sought to justify this intense commercial surveillance by arguing that it is needed to align company production with consumer preferences. Thus, critical marketing scholars Pridmore and Zwick (2011: 270) have noted how, according to this theory, surveillance ‘is akin to providing an important public service because making consumers happy with products they desire … depends on the best possible intelligence about those same consumers’. Perhaps most radically, Zuboff (2015) has rejected this ‘consumer empowerment’ frame (Darmody and Zwick, 2020), and instead sought to diagnose an entirely new species of capitalism, ‘surveillance capitalism’, which is characterized by marketers’, companies’, and governments’ use of big data practices to ‘predict and modify human behavior to produce revenue and market control’ (Zuboff, 2015: 75).
The third strand of literature is concerned with the political economy of transparency and, as Crain (2018: 88) has put it, seeks to ‘diagnose the limits of transparency and interject a theory of commodification into policy debates’. One of the oft-cited limits of transparency is that it creates ‘the illusion of reform while leaving basic power imbalances intact’, as shown, for example, in Crain’s (2018) case study of the US data broker industry, but this argument is also applicable to the pharmaceutical industry (Lexchin and Fugh-Berman, 2021). Relatedly, transparency has been criticized for often being ‘conceived as a means to minimize government interference with the market’ consistent with dominant pro-business notions of the regulation (Pozen, 2018: 136). However, a more radical critique of transparency was offered by Mirowski (2018), who suggested that data transparency in the current context of ideological and economic corporate dominance in the biosciences not only fails to address power imbalances but actually, and very intentionally, risks enhancing them. Specifically, using ‘open science’ as his case, Mirowski (2018: 178) suggested that ‘what is often misrepresented as “openness” and “transparency” is in fact a posteriori private expropriation’. This critique resonates with complaints about ‘open government data’ agendas being used to advance the privatization of major public assets (Bates, 2014) and how they help ‘fuel new data-driven businesses’ (Pozen, 2018: 142). Indeed, although transparency is often justified with reference to consumer empowerment and the need to enhance public trust and government accountability, an analysis of requests under the Freedom of Information Act (FOIA) to six US federal agencies revealed that most requests were made by ‘a cottage industry of companies whose entire business model is to request federal records under FOIA and resell them at a profit’ (Kwoka, 2015: 1361).
Taken together, there is now a large literature on the political economy of commercial surveillance that describes how personal data are extracted, combined, mined, and, in general, commodified within largely hidden circuits of ‘surveillance capitalism’ in which data brokers and related firms play a pivotal role, using increasingly sophisticated big data practices. There is also a growing literature suggesting that ‘transparency data’ can and will be easily extracted and commodified by such companies. At the same time, research on the pharmaceutical industry's political economy predicts that this industry will invest in data-saturated surveillance products and services because it is a marketing-driven industry that seeks to bolster profit and shareholder value.
These insights are evaluated empirically below, where we indicate areas of continuity and change in physician surveillance before and after the Sunshine Act based on the synthesis and analysis of peer-reviewed (i.e. scientific) and ‘grey’ literature (e.g. company white papers and reports, and drug industry news). An initial body of material was identified through PubMed and Google searches using combinations of the following search terms: ‘Sunshine Act’, ‘Open Payments’, ‘Transparency’, ‘Pharmaceutical’ or ‘Pharma’, ‘Marketing’, ‘Surveillance’, ‘key opinion leaders’ or ‘KOL’, ‘Artificial Intelligence and ‘Machine Learning’. We then used snowballing techniques to identify additional material, specifically reference checking and citation search in Google Scholar. We also conducted searches of websites of major marketing and health data broker and analytics companies, which we identified through our initial searches and literature review. In reviewing the material, we focused on understanding the evolution of tools and data streams used by companies to surveil physicians in the USA, as well as how companies explained and justified this surveillance. This was an iterative process that involved initial review followed by re-reviews of material seeking to refine accounts and interpretation of surveillance and marketing techniques, supported by illustrative examples (Steinman et al., 2006). One important problem in constructing accounts of surveillance and marketing was that descriptions in the ‘grey’ literature are likely to be selective and non-exhaustive, and sometimes inaccurate. Indeed, a key lesson learned from the whistle-blower-initiated court cases described above is that many deceptive marketing schemes only became known to outsiders after the release of internal company documents (Mulinari, 2016). However, despite this important caveat, the reviewed material is likely to provide informative ‘glimpses’ into physician surveillance.
Commercial surveillance of physicians before the Sunshine Act
In a series of articles, Sismondo (2011, 2013, 2015, 2017) characterized how pharmaceutical companies in the USA, or their agents in the form of consultancy and marketing companies, nurture and manage high-profile key opinion leaders (KOLs) and physician members of ‘speaker bureaus’ to promote particular drugs, for the purpose of shaping medical and public opinion. Significant for the purpose of this article, Sismondo (2015: 759) drew attention to how ‘companies engage in detailed data analytics to establish the effectiveness of their speaker bureaus and sometimes measure the number of prescriptions written for a drug before and after a talk’. Similarly, he cites a KOL management conference speaker in 2012 describing a new approach to network analysis, saying: ‘So it's really very, very interesting and starts to give us the tool and the power to be able to actually look at these network maps and start to think about the implication in terms of the things that we are doing commercially’ (Sismondo, 2017: 126).
Although important new techniques have been added to companies’ commercial surveillance toolboxes, the pharmaceutical industry in the USA has been collecting and using physician prescribing and practice profiles for marketing for several decades. Greene (2007) tracked the historical origins of the commercial surveillance of physicians to the early 1940s and 1950s. During this time, pharmaceutical companies and the American Medical Association (AMA) together powered the rise of the health data broker industry, as this industry's products and services served the marketing interests of drug companies while providing substantial revenue to the AMA, which sold to data brokers unique data on physicians, for example, via their computerized registry using IBM punch cards of all physicians living in the USA. By the late 1950s, Greene (2007: 747) explained, this data broker industry had developed a ‘robust network of surveillance technologies’ that allowed pharmaceutical marketers and sales representatives to begin tracking prescribing practices and use these data to influence physicians.
However, some decades later, from the late 1980s and early 1990s, a major innovation occurred when data brokers began purchasing de-identified patient prescription records with limited physician identifiers directly from pharmacies to obtain real-time, individual-level prescription datasets (Steinbrook, 2006; Fugh-Berman, 2008). While pharmacies cannot release patient names, they can provide data brokers with anonymized records that allow them to follow individual patients’ drug trajectories. This information could then be strategically linked to specific physicians by purchasing a comprehensive physician database (the ‘Physician Masterfile’) from the AMA. Linked datasets enabled the data broker industry to build prescriber profiles and sell them to pharmaceutical companies, which use them to identify sales targets and plan sales visits. Such databases became available for license to drug companies in 1993 (Hunkler and Musacchio, 2006).
Access to detailed prescriber profiles in the early 1990s coincided with what Applbaum (2011: 273) has described as a ‘break’ in the US drug industry's relationship with marketing, ‘insofar as at that time marketing became a total institutional fact in the industry’. This novel marketing-driven organizational culture was focused on fostering ‘blockbusters’, such as psychotropic, anti-diabetic, and anti-inflammatory drugs, by deploying aggressive marketing techniques throughout the drugs’ trajectories from bench to bedside (Applbaum, 2011). Direct evidence of companies incorporating physician-specific data to surveil and manipulate prescribing practices emerged from high-profile whistle-blower-initiated court cases, many pertaining specifically to blockbuster drugs. A case in point involved the drug company GlaxoSmithKline (GSK), which in 2012 agreed to plead guilty and pay USD 3 billion to resolve its criminal and civil liability arising from unlawful promotion and failure to report safety data, as well as its civil liability for alleged false price reporting practices (Department of Justice, 2012). Among other things, between 1998 and 2003, GSK unlawfully promoted its antidepressant Paxil to treat depression in patients under age 18, despite emerging evidence of increased suicide risk among adolescents. Of particular relevance to commercial surveillance and marketing, the legal investigation revealed that, in 2000, GSK sent child psychiatrists to ‘lavish resorts’ for three-day Paxil Forum events and then measured their Paxil prescriptions to monitor the efficacy of marketing. An internal GSK memo from the Paxil marketing Director confirms that attendance ‘had a significant impact on Paxil's market share in the months after attendance’ (District Court of Massachusetts, 2012: 18–19). Specifically, GSK found that the percentage of Paxil prescriptions relative to other selective serotonin reuptake inhibitors prescribed by psychiatrists who attended the Paxil Forum events in 2000 (‘test physicians’) increased significantly when compared with the percentage prescribed by psychiatrists who had not attended the Paxil Forum events (‘control physicians’). The memo concluded that increased Paxil prescriptions due to the Forum 2000 meetings resulted in at least USD 900,000 in additional revenues in 2000 alone.
In the early 2000s, drug companies would usually buy their prescription data and analyses from IMS Health. Founded in 1954, IMS Health had become the leading health data broker, reporting annual revenue of more than USD 1 billion by 2000 (Greene, 2007). A key data asset for IMS Health was their Xponent™ proprietary database, touted as ‘the first true physician level database for the health-care industry’, which, company representatives revealed in 1999, ‘in the last couple of years, has grown extremely large [i.e., terabyte scale], and currently maintains prescription information by prescriber, product and payment type [i.e., cash, Medicaid, and Health Maintenance Organization]’ (Kallukaran and Kagan, 1999: 3). By the late 1990s, IMS Health was already using artificial intelligence to mine Xponent™ ‘to detect various marketing related phenomena like brand switching, brand loyalty, and brand performance’ (Kallukaran and Kagan, 1999: 4). Specifically, deploying neural-network models, IMS Health was able to offer clients a list of prescribers who had switched from the client's product to competitors’ products, information that could be used when targeting the ‘right’ prescribers to counter switching behaviour, promising large returns on investments: As this database grows, it becomes extremely difficult to identify physicians changing their prescribing behaviors. The neural-network model provides a method of analyzing times-series data and identifying physicians that have changed their prescribing behavior over time. The method provides a tool for the sales force to use in identifying physicians to target when making sales calls. Research has shown that winning just one more prescription per week from each prescriber, yields an annual gain of $52 million in sales [based on the anti-ulcer market]. So, if you’re not targeting with the utmost precision, you could be throwing away a fortune (Kallukaran and Kagan, 1999: 3).
By 2006, IMS Health revenues had grown to almost USD 2 billion, with 13% compound annual growth over the previous five years. According to the 2006 annual report, the company was now able to mine Xponent™ using ‘patented statistical methodology to project the prescription activity of nearly 1.4 million individual prescribers on a weekly and monthly basis’ (IMS Health, 2007: 22). Similar databases were now also available for many European countries, the company wrote. Furthermore, IMS Health offered ‘a sales optimization solution’, Early View™, which, among other things, could highlight ‘competitive prescribing trends for clients’ key prescribers directly to clients’ sales representatives electronically’ (IMS Health, 2007: 23).
Pharmaceutical companies embraced the health data broker's increased capacity to amass, analyse, and commercialize physician-level data because it helped them increase the cost-effectiveness of marketing (Heesters, 2008). Furthermore, according to defenders of this practice in the drug industry and the medical profession, the fact that promotion became more ‘relevant and specific’ as a result of better physician profiling was also in the interest of physicians, since it would spare them from ‘being bombarded with extraneous promotional materials and sales calls’ (Hunkler and Musacchio, 2006). Nonetheless, growing concerns about physician privacy and the negative effects of marketing on health care costs provoked a backlash from some physicians and lawmakers. In 2004, several national and state medical societies, led by the American College of Physicians, formally requested that the AMA prohibit the release or sale of physician prescribing information. And, in 2006, in response to this call, but also in an attempt to deter legislative interventions underway in several states, the AMA launched the Physician Data Restriction Program (PDRP) (Hunkler and Musacchio, 2006). This program allowed (and still allows) physicians to opt-out of sharing data that may reveal prescribing habits for any particular product with pharmaceutical sales representatives and their direct supervisors, but leaves these data available to the company for marketing, compensation, and research purposes. However, only a small percentage of physicians (only 4% by 2011) enrolled in the PDRP (Mello and Messing, 2011).
Furthermore, although the PDRP may have had a cooling effect on some state legislators, it did not deter three states, New Hampshire, Maine, and Vermont, from passing statutes that either prohibited (New Hampshire in 2006) or significantly reduced (Maine and Vermont in 2007) the sale, exchange, or use of physician prescribing information for marketing purposes (Mello and Messing, 2011). However, all three states were swiftly challenged in court by IMS Health, two other data broker firms, and by PhRMA, the trade association of pharmaceutical companies. IMS Health took the complaint all the way to the Supreme Court, where it won a landmark victory in 2011 against the state of Vermont on corporate ‘free speech’ grounds by a vote of six to three, effectively blocking state legislators’ attempts to curb the commercial use of prescribing data (Mello and Messing, 2011).
In 2014, a few years after their Supreme Court victory, IMS Health published a company prospectus that gave new insight into the growth of its data and analytics resources (IMS Health, 2014). The company now described itself as having ‘one of the largest and most comprehensive collections of healthcare information in the world, spanning sales, prescription and promotional data, medical claims, electronic medical records and social media’ (IMS Health, 2014: 1). Altogether, this was said to amount to ‘over 10 petabytes of proprietary data sourced from over 100,000 data suppliers covering over 780,000 data feeds globally … which includes 85% of the world's prescriptions by sales revenue and approximately 400 million comprehensive, longitudinal anonymous patient records’ (IMS Health, 2014: 6).
All of the top 100 global pharmaceutical and biotechnology companies were said to be clients in 2012, and in the USA alone, data and analyses worth over USD 2 billion were sold to pharmaceutical companies in the first nine months of 2012 (Ornstein, 2014). This included ‘commercial applications supporting sales operations, sales management, multichannel marketing and performance management’ (IMS Health, 2014: 3). Specifically, the prospectus offered examples of questions clients were able to answer with its data and analytics portfolio, including: ‘Which providers generate highest return on rep visit? Does my sales rep drive appropriate prescribing?’; ‘How much should I pay my sales rep next month?’; and ‘Is my brand gaining market share quickly enough to hit revenue forecasts?’ (IMS Health, 2014: 3).
Commercial surveillance of physicians after the Sunshine Act
Arguably, these ‘glimpses’ into physician surveillance from before the launch of the Open Payments Database in 2013 should worry critics of corporate marketing, as they reveal the existence of markets in commodified physician data and behavioural control that pharmaceutical companies were harnessing, markets with clear potentials to be further powered by information on physicians’ receipt of payments.
It comes as little surprise, then, that sales and marketing teams quickly seized the opportunity to use Open Payments data in the identification, segmenting, and optimized targeting of physicians. Direct evidence comes from a 2017 article in the Journal of the Pharmaceutical Management Science Association, co-authored by several marketing executives from the large drug company Bayer, entitled ‘Leveraging CMS [i.e., Centers for Medicare & Medicaid Services] Open Payments Data to Identify Channel Preferences and Gather Competitive Intelligence, Thereby Improving HCP Targeting’ (Anand et al., 2017).
The article begins by explaining how companies in the past had been using a range of data sources and services ‘to better understand market potential, improve HCP targeting, and potentially determine competitive loyalists’ (Anand et al., 2017). This includes: (1) ‘Leveraging traditional third-party data’, such as IMS Health's Xponent™; (2) ‘Measuring the share of voice’ of different products through market research; and (3) ‘Identifying office accessibility or channel preferences of HCPs’ using various licensed data sources ‘such as AccessMonitor™ and AffinityMonitor™’, offered by the marketing consultancy firm ZS. AccessMonitor™ looks at whether individual physicians meet with sales reps. In 2017, it captured data about the activities of more than 40,000 pharmaceutical reps and more than 400,000 prescribers. AffinityMonitor™ looks at how health care providers engage with specific marketing ‘channels’, for example, email, mobile apps, websites, telesales, and peer interactions such as speaker bureaus, and captures data from more than 250 companies, 878,000 health care providers, and 254 million individual interactions. For example, AffinityMonitor™ data analyses can show how likely certain physicians are to open an email containing one of 18 predefined ‘content types’, for instance, ‘clinical trials’, ‘KOL/peer opinions and insight’, or ‘invitations to speaker programs, conferences, etc.’ (Sturgis, 2017: 11).
However, the Bayer marketing executives pointed out that all of these sources and services have weaknesses, especially related to coverage and costs. To improve its analytics and cost-effectiveness, the company first created a database that links Open Payments with internal sales/call activity data. This allowed the company to identify the overlap between prescribers receiving payments and the ones targeted by the company as well as to understand the relationship between different payment categories used in Open Payments and their values (e.g. food and beverage >USD 25) and internal marketing categories, or ‘contact channels’ (e.g. speaker program – attendee). Using this new database, the company was able to gather ‘competitive intelligence’, that is, see which competitors were paying which physicians for what. ‘For sales and marketing teams, having a deep understanding of competitors’ level of interactions with HCPs across different channels of influence gives a huge competitive advantage’, they wrote (Anand et al., 2017).
The database was also used to identify new ‘targets for call plan’, that is, new physicians who could be targeted by reps. First, the ‘HCP universe for the market’ was established using the created database. ‘This was overlaid on the existing call plan to exclude HCPs who were already present in the plan’. In addition, health care professionals (HCPs) who ‘participated in competitor clinical trials, or were called historically’ were also excluded from the analysis (Anand et al., 2017). Finally, data on the number of patients and prescriptions per HCP were used to narrow down the number of ‘new targets’ to those HCPs that were commercially most relevant to the company. The analysis concluded: Over time, the richness and accuracy of data will further improve, providing companies the opportunity to monitor existing and newly acquired targets over time. Given that there is no cost associated with the data, if analytical rigor is appropriately applied, pharmaceutical companies can only derive more consumable insights, or at the least, directional insights. It's still a win–win situation. (Anand et al., 2017: 64, emphasis added)
Parallel to such in-house uses of Open Payments to identify and monitor physician ‘targets’, several commercial actors similarly use Open Payments as ‘raw material’ in commodity offerings or services to pharmaceutical companies. For example, there is an app, PowerCMS™, that can be used by sales reps to better track KOLs: The goal of PowerCMS™ is to provide all companies with access to a powerful analytical tool that companies can use to review the complete Open Payments database to answer specific questions, such as what other companies is Dr. Smith working with, how much is my company spending on meals versus my competitors, or should we consider Dr. Jones a regional or national Key Opinion Leader (‘KOL’)? As Jeff Cohen, Chief Compliance Officer and Regulatory Counsel at Globus Medical, Inc. commented, ‘These insights are the kinds of insights we have been trying to accomplish internally but it wasn't easy and very labor intensive. Now we have a tool to easily analyze not only our data, but data from other companies across the industry’ (Sullivan, 2019).
The extent to which companies monitor cross-industry payments for competitive commercial insights, and how they view payments as a proxy for commercial influence and for physician commercial value, may not be something that companies wish to reveal to audiences of researchers, physicians, or policymakers. However, this discourse sits in plain sight in the drug marketing ‘grey’ literature. For example, a 2018 article in the Journal of the Pharmaceutical Management Science Association entitled ‘Deploying Machine Learning for Commercial Analytics’ (Tsang, 2018) explains that one lesson from using machine learning to predict physician prescribing is that payments can be used to build a company-specific ‘allegiance index’ for physicians and to measure physicians’ ‘star power’: The reluctance of a physician to prescribe a drug may have to do with the physician's financial involvement with other pharma companies, which is described in the Open Payments database (Sunshine Act). By looking at payments a physician receives from pharma companies, we can develop an allegiance index that indicates if the physician is strongly tied to one company or is open to developing new relationships with other companies.
It's always helpful to know who are the sought-after physicians. One way to do so is to look at the number of trips a physician takes on pharma's dime, and even at a breakdown of these trips by in-town, domestic, and international. Looking at year-on-year changes, we can also define features that describe how the star power of the physician is trending: rising, falling, steady, or wobbly. (Tsang, 2018)
More industry-wide evidence of how companies leverage Open Payments for commercial surveillance and marketing can be found in a 2019 report from the data analytics company Axtria, ‘Perspectives on Pharma Company Use of Open Payments Data’ (Chressanthis and Padbidri, 2019). It reports on ‘an informal internal analysis within Axtria of people knowledgeable of client activities’ (Chressanthis and Padbidri, 2019: 5). The report emphasizes two main areas of current use – ‘commercial operations’ and ‘commercial analytics’ – and it notes quite critically that, overall, companies currently use Open Payments primarily to enhance prescription volumes rather than ‘any applications that are patient-centered and tied to better clinical practice and ultimately outcomes’. The commercial operations applications include improving the selection, evaluation, targeting of, and access to HCPs in marketing, specifically: (1) HCP target refinement. (2) HCP target identification for pre/new launch of a drug. (3) Understand competitive share of voice. (4) Analogue for physician potential importance and/or measure of value. (5) Influence of competitive voice on brand TRxs [i.e., total prescriptions]. (6) Accessibility of physicians/accounts (i.e. willingness to engage pharma companies) (Chressanthis and Padbidri, 2019: 5).
In contrast, the commercial analytics applications entail collecting, assessing, and using the information to devise commercial strategies, specifically: (1) Pre-launch analytics (e.g., go-to-market (GTM) strategy). (2) Commercial model design (e.g., segmentation, customer valuation, and targeting). (3) Call planning (e.g., target refinement). (4) Marketing analytics (e.g., spend benchmarks, guidance on promotion-mix, competitive promotion assessment, part of the marketing-mix resource allocation decision). (5) Prediction modeling (e.g., predict future adopters, predictor of uptake) (Chressanthis and Padbidri, 2019: 5).
Unsurprisingly, the successor of IMS Health – the global health data analytics company IQVIA, formed after the USD 17.6 billion merger between IMS Health and the clinical research organization Quintiles in 2016 – offers solutions to companies in both these commercial areas using Open Payments in combination with their proprietary databases and analytics. In 2019, the company published a client-facing ‘white paper’ entitled ‘IQVIA insights leverage CMS data for strategy development’ (Jones et al., 2019). The report aligns itself with the current ‘hegemonic technical marketing discourses’ (Ball and Webster, 2020: 3) by promising clients ‘insights’ and value creation using artificial intelligence and other big data tools. Specifically, it highlights the following types of analyses for future clients (Box 1), summarized as giving ‘real-time, granular visibility’ into:
HCP engagement patterns over time and across field territories. Where and when your competitors have concentrated their resources. How to capitalize on HCP interest in a particular product. Which specific KOLs to target for a particular program. What attributes will make a KOL a valued brand ambassador, and how their compensation should reflect that (Jones et al., 2019: 7). Examples of IQVIA analyses using Open Payments (1) (2) (3) (4) (5)
Discussion
Our study adds fresh perspectives to debates on the benefits and drawbacks of the Sunshine Act (Lexchin and Fugh-Berman, 2021) and of transparency-based strategies in the pharmaceutical industry more broadly (Fierlbeck et al., 2021). While previous critiques of transparency have mainly focused on its limitations in addressing structural power imbalances (Lexchin and Fugh-Berman, 2021) and in rooting out corporate and professional maleficence (Loewenstein et al., 2012), our study shows how drug companies and associated industries are capitalizing on transparency by incorporating Open Payments data into their pre-existing ‘network of surveillance technologies’ (Greene, 2007: 747). Specifically, with the Sunshine Act, a unique, rich stream of data on physicians’ industry-wide financial relations has become available that companies can use to further monitor and seek to influence physicians for commercial gain. Ironically, while pharmaceutical companies have generally opposed transparency demands (Fierlbeck et al., 2021), these companies can also benefit greatly from those demands, consistent with some political economy critiques of transparency and ‘openness’ in the biosciences (Mirowski, 2018).
Our study also adds fresh perspectives to scholarship into the political economy of the pharmaceutical industry. This literature has emphasized the ‘financialization’ of the pharmaceutical industry associated with drug companies’ increased reliance on financial actors, instruments, investments and incentives (Busfield, 2020). As part of its financialization, the industry has undergone significant restructuring since the 1980s, including more outsourcing of research and development (R&D) to reduce financial risk (Mirowski and Van Horn, 2005), alongside heavy spending on marketing to maximize revenues (Busfield, 2020), with marketing almost doubling R&D costs (Gagnon and Lexchin, 2008). Significantly, companies’ explanations of how they use Open Payments data cast new light on industry marketing, and especially the key role of data brokers and analytics companies in improving the identification, targeting, managing, and surveillance of physicians. Where recent scholarship into the political economy of the pharmaceutical industry have tended to focus on the intermingling of ‘big pharma’ and ‘big finance’, our study therefore emphasizes the risk, for health and health systems, when ‘big tech’ is brought into this equation (Ornstein, 2014).
Importantly, our analysis of commercial uses of Open Payments is not exhaustive. By both nature and design, pharmaceutical sales and marketing is an opaque practice (Mulinari, 2016), meaning that there are certainly other, and possibly even more questionable, practices that we simply do not know about. Pharmaceutical sales and marketing are also rapidly evolving amid novel data streams, for example, physician activity on social network sites (e.g. IQVIA's ‘DocNet’), and technical and commercial innovations, for example, artificial intelligence tools for improved prediction and physician targeting (Yuan and Zhao, 2018), which may open new possibilities for commercial uses of payment data. This is all the more reason to be concerned about the unearthed uses of Open Payments for commercial surveillance and marketing.
However – and arguably only to a small degree counterbalancing these concerns – pharmaceutical sales and marketing are also inherently hype-based, meaning that some claims about what marketers can do with the data may also be hyperbole. This may be particularly true of claims about new technologies such as artificial intelligence that are often presented as ushering in paradigmatic changes (Ball and Webster, 2020). It therefore remains an empirical question to what degree the descriptions in this article accurately represent ‘marketing in action’. However, we anticipate that hyperbolic claims will be much more frequent among new marketing and consultancy companies seeking to attract investment capital than among pharmaceutical companies with no apparent reason to exaggerate their analytics and marketing capacity or among large, established data broker, analytics, and marketing firms that need to be accountable to their industry clients. This is important because our study draws mainly on descriptions from the two latter classes of companies. Furthermore, sociologists have shown how expectations, even those viewed as hyperbolic in retrospect, play important roles in shaping drug research and markets by directing actors’ actions towards certain goals (Williams et al., 2008) – in our case, for example, they would justify further data collection, commercial surveillance, and attempts at physician manipulation (Darmody and Zwick, 2020).
We find it quite remarkable to observe the incongruities between discourses on physician payments in the ‘grey’ drug marketing literature analysed here and the ‘grey’ drug industry ethics and compliance literature, for example, in industry codes of ethics (Francer et al., 2014). As we have shown, in the former, payments are viewed rather bluntly as proxies for industry commercial influence and for physicians’ commercial value. Hence, payments are said to be useful for measuring phenomena such as physicians’ ‘allegiances’ and ‘star power’ (Tsang, 2018), for identifying ‘competitive loyalists’ (Anand et al., 2017), ‘brand ambassadors’ (Jones et al., 2019), and the most effective ‘channels of influence’ (Anand et al., 2017), and, in general, for ‘monitor[ing] existing and newly acquired targets over time’ (Anand et al., 2017). Furthermore, payment patterns are represented as, in the best case, resulting from a top–down, conscious marketing strategy that is part of a comprehensive drug commercialization strategy. Clearly, this instrumentalist discourse that sees physicians primarily as means to enhance sales resonates with findings from the extensive academic literature showing consistent effects of payments on prescription volumes (e.g. Mitchell et al., 2021).
In stark contrast, in industry codes of ethics and much official drug industry commentary on transparency, payments are represented as proxies for ‘professional exchanges’ or ‘collaboration’, said to be ‘designed to benefit patients and to enhance the practice of medicine’ (PhRMA, 2020). However, if this were true, corporate analyses of Open Payments would not primarily serve commercial ‘operations’ and ‘analytics’ functions within companies, but rather, as the Axtria report acknowledged, companies would use Open Payments to show positive value for research and for patient and public health (Chressanthis and Padbidri, 2019). In light of the evidence presented, it is easy to dismiss the apparent incongruities between the two industry discourses (i.e. marketing vs. ethics) as reflecting an attempt at the ideological concealment of the true marketing purpose of payments, or at least as reflecting some sort of ‘strategic ignorance’ (McGoey, 2012) among the industry's ethics and compliance personnel. Although ideological explanations are surely relevant, an alternative or at least complementary, explanation of these incongruities would emphasize that the pharmaceutical industry, as Quirke (2014: 656) has pointed out, ‘is neither monolithic nor static, but—just as modern medicine—flexible and diverse’. Still, even after acknowledging that the pharmaceutical industry is complex, contradictory, and harbours many professions, values, and priorities (Mulinari, 2015), it also remains true that pharmaceutical companies are, as Abraham (2008) argues, hierarchical organizations that have an objective, though not always over-riding, commercial interest in profit maximization. Arguably, it is this commercial interest that sits in plain sight in the ‘grey’ marketing literature.
Furthermore, where a clear match of discourses does seem to exist is between the ‘grey’ marketing literature and the discourse described in other commercial marketing contexts by scholars studying the surveillance of consumers (Crain, 2018; Darmody and Zwick, 2020; Pridmore and Zwick, 2011). Still, curiously enough, physician surveillance seems largely absent from debates about ‘surveillance capitalism’ and related concepts (Fourcade and Healy, 2017; Sadowski, 2019; Srnicek, 2017; Van Dijck, 2014; West, 2019; Zuboff, 2015), which focus primarily on big tech-companies’ ubiquitous exploitation of individual users’ everyday activities and digital traces, and these companies’ modification and commoditization of behaviour for profit. Arguably, this omission may be problematic for both historical and contemporary analyses of commercial surveillance, because physicians, as Greene (2007: 747) noted, have been ‘one of the most easily defined and efficiently studied sectors of consumers in the U.S. economy’. From this perspective, one key – and perhaps the most surprising – finding of our analysis is the absence of evidence for the integration of data on physicians’ professional roles and behaviour, such as receipt of payments, with their non-professional, personal data, for example, family situation, personal finances, or online behaviour. It is possible that this absence simply reflects efficient concealment – indeed, there is at least one company in programmatic marketing that boasts of using an artificial intelligence tool to build physician profiles that combine professional (e.g. brand loyalty and value index) and personal (e.g. credit score, mortgage, and social platform activity) measures (Doceree, 2021). However, an alternative interpretation is that pharmaceutical companies and associated industries, at least so far, have consciously avoided the full-blown surveillance of physicians because of the large reputational risks involved given physicians’ social standing and professional power. Taken together, this underscores the need for more research on the political economy of physicians’ professional and personal data, including on what types of physician data are being extracted, and how the data are combined, commodified and used.
On balance, we believe that the evidence presented here shows the need to tightly couple policies that increase transparency in the pharmaceutical industry with policies that impede the commodification and use of transparency data for surveillance and marketing purposes. Too often, these two issues have been kept separate in the academic literature and policy debates. Thus, 10 years ago, there was much debate anticipating the implementation of the Sunshine Act and much debate on the commercial surveillance of physicians after the US Supreme Court blocked state legislators’ attempts to curb the commercialization of prescribing data, but few appeared to have made the link between the effects of the two concomitant policy developments. For example, at first sight, one obvious way to curb commercial uses of disclosure data, while still allowing for important non-commercial uses, would be to ban commercial uses, but this would surely violate the Supreme Court's affirmation that pharmaceutical companies’ ‘commercial speech’ in marketing is protected under the constitution (Liu et al., 2021; Mello and Messing, 2011). Still, it is precisely laws restricting pharmaceutical sales and marketing practices, as well as data protection, privacy, and similar laws that restrict the accessing, use, and disclosure of information, that are most likely needed to limit the commercial surveillance of physicians and to ensure that their professional and personal data are kept separate in the future.
The USA may want to look to Europe for inspiration on this. For example, France's version of Open Payments prohibits the use of the data ‘for strictly commercial purposes’ (Transparence-Santé, 2017). Depending on how ‘strictly commercial’ is interpreted it may restrict the commodification and use of transparency data for surveillance and marketing purposes. More critically perhaps, the European Union's General Data Protection Regulation prevents health care companies from linking prescriptions back to the prescriber (Yuan et al., 2019) which in theory should block much commercial surveillance of physicians – although, illustratively, IQVIA promises clients in the pharmaceutical industry that ‘with machine learning algorithms that can analyse accessible data without compromising privacy laws, companies can confidently estimate physician potential’ also in Europe (Yuan et al., 2019: 8).
Footnotes
Declaration of conflicting interests
SM's partner is employed by ICON, a global Contract Research Organization whose customers include many pharmaceutical companies. PO's PhD student was supported by a grant from Sigma Pharmaceuticals, a UK pharmacy wholesaler and distributor (not a pharmaceutical company). The PhD work funded by Sigma Pharmaceuticals is unrelated to the subject of this paper.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Vetenskapsrådet (grant number 2020-01822).
