- Pushing beyond the tipping point for real-world evidence (RWE)
- Proliferating sources of real-world data (RWD)
- Anxiety over data privacy and security
- Harnessing the power of AI
- Expanding platforms for HCPs’ consumption and circulation of evidence
May 10, 2022
Now that real-world evidence has garnered industry-wide acceptance and the FDA has issued draft guidelines on RWE use in regulatory decisions, life sciences leaders cannot afford to wait passively for further direction or support. Progressive organizations will invest more aggressively in using sources of real-world data to support clinical development, payer/provider engagement strategies, and a range of market access programs (including outcomes-based contracts.)
The industry is at a tipping point in its comfort with and use of real-world data across the product lifecycle. Diagnosing, treating, and vaccinating against COVID-19 has required health systems and governments to make public health decisions via the near-real-time collection and analysis of real-world data. In December 2021, the FDA issued draft guidance on use of RWD and RWE in regulatory decisions and has publicly signaled that it will continue to release additional guidance on RWE sources and study design. Additionally, FDA appears to be signaling a greater openness to conversation and collaboration with industry stakeholders. In fact, in the draft guidance, FDA suggests that “Sponsors should engage with FDA in the early stages of designing a non-interventional 135 study intended to support a marketing application.”
Beyond recent regulatory changes, investors continue to pour billions of dollars into technologies and consortiums that can aggregate, integrate, and safely de-identify disparate real-world data sets for more sophisticated analyses. And although buzz around real-world evidence has existed for years, it’s now fully in the public spotlight—and top-of-mind for all health care leaders.
As a result, real-world evidence is past its tipping point on acceptance. Hesitation to invest beyond claims and registries is getting harder to justify, as regulators are beginning to offer more clarity, impactful use cases are proliferating, technology platforms are enabling greater ways to link disparate data sets, and stakeholders are raising the bar for demonstrating value.
Life sciences leaders can no longer take a passive or reactive approach to investment in and use of real-world evidence.
However, as investment and interest in RWE intensify, the bar for real-world evidence is evolving. Increased scrutiny on data quality, scale, and relevance—coupled with increased pressure from stakeholders to see ROI on RWD investments—means that life sciences leaders need to proactively (and aggressively) evolve their RWE strategy. Life sciences leaders must identify opportunities to invest in tools and platforms that support secure and lawful data linkages for longitudinal analysis, trusted AI applications, and near-real-time analytics. They must work with payers and providers to make better sense of the abundant cost, utilization, and clinical data available. And they must put that real-world data to use in ways that better align stakeholders on improving outcomes, lowering total cost of care, and delivering value. Medical and HEOR leaders must continue to work with their R&D colleagues to identify appropriate use cases for utilizing RWD earlier in the product lifecycle and for leveraging the wealth of data they already have. As a case in point, several progressive life sciences organizations have started to use RWE to inform clinical pipeline prioritization, protocol design, market access strategies, and business development decisions.
Yet a comprehensive RWE strategy isn’t just about data collection and evidence generation. Life sciences leaders must change how and when they communicate real-world evidence with key stakeholders. Clinicians and other medical product gatekeepers will demand answers to questions about the validity of data and analytic algorithms, assumptions underlying data analyses, and any possible holes in data due to disrupted care during COVID-19. Stakeholders may also need help making sense of the sheer volume of real-world data that continues to emerge so that they can curate data based on quality, business need, and context. Some customers may even ask life sciences leaders for help analyzing their own data, which could create new opportunities for personalized, data-driven engagements between sales or medical liaisons and their key customers.
To avoid falling behind competitors, life sciences leaders should be actively engaging regulators and key customers to discuss real-world evidence needs, adapting and preparing for upcoming regulatory guidance, and identifying opportunities to expand use cases for existing data sets.
The RWE landscape will need to account for delivery and clinical innovations such as whole genome testing, telehealth, and home-based care. This will raise the bar for integrating data from increasingly fragmented care sites and highly varied data sources into reliable, trusted, quality data sets.
The broad acceptance of RWE and the growing interest in additional applications of data are happening concurrent with an unprecedented shift in care delivery from inpatient to outpatient settings, and from outpatient to home/virtual settings. 2021 saw unprecedented investments in digital health, home-based care, and innovations that support care delivery in non-acute settings.
While many experts have touted the benefits of these innovations, especially from the patient perspective, few people are talking about the ripple effects of this care fragmentation. Notably, this transition to “everywhere care” makes care coordination and data integration much, much harder. Manufacturers will face new challenges in ensuring that the longitudinal real-world data sets they use are sufficiently robust across multiple sites of care. And they’ll also face new customer demands to measure and demonstrate value of their products when used in a wider range of clinical (or non-clinical) settings.
Life sciences leaders need to understand how new care models impact clinician decision-making, especially about diagnosis, prescribing, and product use. This information is particularly important given the efforts of payers and primary care innovators to influence clinicians’ decisions as a way of reducing unnecessary downstream utilization. But it’s not just about the provider. Life sciences leaders should also strive to understand their top integrated delivery network (IDN) customers’ long-term site-of-care strategies. Which IDNs are investing in ambulatory surgery centers? Which are building out infusion centers or partnering with home care agencies to enable more acute care at home? These plans will have significant implications on product purchasing, distribution, use, and real-world data collection (e.g., for symptom/side effect or adherence tracking).
These site-of-care shifts also raise questions around the safety, quality, and cost of providing care in atypical settings. Life sciences leaders are asking questions like: “What are the right metrics to track?” Or, “What are the appropriate benchmarks?” Such data does not readily exist for many treatments and interventions, thus requiring manufacturers to gather additional RWD they can share with providers, payers, and IDN leaders. Doing this important work requires access to data sources that these stakeholders trust as accurately reflect real world practice and outcomes.
There is also an emerging need for organizations to gather new and different kinds of data, either because the source is novel (like wearables or connected devices) or because the data itself is relatively novel and untested (like social determinants of health or internet search histories). Life sciences leaders must gather input cross-functionally, and across key customer groups, to ensure that they are investing in data sources and evidence-gathering initiatives that meaningfully contribute to conversations about value in a world of fragmented “everywhere care.”
Life sciences leaders must balance their thirst for connecting disparate real-world data sets with very real institutional and individual responsibilities for ensuring the privacy and security of the underlying patient information.
In the search for a more holistic understanding of patient journeys and diseases themselves, the life sciences research community has sought to leverage a diverse array of deidentified information sources by linking together traditional medical data (like EHR and claims data) with emerging resources like social determinants of health (SDOH), patient-generated wearables data, genomics, and consumer data. New data interoperability mandates coming into effect in the next few years will further reduce the barriers that hinder the creation and maintenance of longitudinal patient histories.
When tasked with solving a business problem, well-intentioned researchers design analyses and identify their data needs. In many cases, they may desire to connect different forms of data together—for example, by combining deidentified information from a clinical trial with a publicly-available data source. Researchers must accompany this type of data linking with careful due diligence to assess the resultant dataset and ensure it does not inadvertently increase the risk of reidentification, because as more attributes are known about a deidentified person, the risks of re-identification increase.
When individuals’ health data is exposed, they may confront reimbursement fraud, personal financial risks or unwanted stigma. For life sciences manufacturers, data reidentification could violate their own protocols for IRB-approved studies or their contractual obligations with third parties. They may also face scrutiny and penalties from a variety of state and Federal regulators for any resulting breach from an exposure as well as financial exposure from the individuals impacted by a breach. Business leaders can safeguard the outputs and protect against misuse or privacy breaches by making sure the right compliance and governance processes are in place.
Chief Information Security Officers and Privacy Officers cannot be the only ones who are concerned with privacy and data security. The risk of patient re-identification, coupled with ever-more-sophisticated cyberattacks, means that life sciences organizations must take steps to protect their own reputations along with the security and privacy of the patients, payers, and provider organizations generating and sharing real-world data for research purposes. Leaders must cultivate a culture of preserving confidentiality: everyone touching data must understand his or her obligation to ensure the information remains deidentified and secure.
Recent miscues by social media platforms and Big Tech cast a spotlight on the misuse of personal information, and that spotlight is unlikely to fade away anytime soon. Several times each year, the newswires decry the latest data breach or ransomware attack on a health care organization. Cultivating a cross-functional data governance team that includes data, analytics, legal, security and privacy experts can help your organization acquire, curate, and deploy data securely to keep your organization out of the headlines.
Leaders must be aware of the regulatory obligations and public perceptions around data access, because cross-sector partners will be protective of their own institutional obligations and reputational risk. This risk aversion may make it more difficult to pursue innovative projects that depend on multiple forms of real-world data.
These challenges are particularly acute outside of the U.S., where European privacy laws and regulations make decentralized trials or remote data collection for research purposes particularly difficult.
Massive investments in data science partnerships suggest that pharma leaders have bought into AI’s promise to make drug discovery more efficient—but life sciences companies won’t realize the full potential of this technology unless they take deliberate steps to embed AI applications into day-to-day workflows in ways that empower employees and exercise caution to minimize unintended bias.
Some of the biggest players in the life sciences space have placed nine- or ten-digit bets on companies that promise to use artificial intelligence (AI) to discover new medicines more efficiently, leading to better returns on R&D dollars and less time spent on potential treatments that are ultimately unsuccessful. Should these endeavors prove successful, they have the potential to mark a turning point in the history of drug development, as the promise of these capabilities to reduce waste and increase speed to market finally comes to fruition.
Applying AI to identify druggable targets or model molecular structures is an activity that occurs well before clinical trials in actual humans begin. The hypotheses formed from AI-driven insights are tested in a lab, where researchers can readily observe the outcomes of controlled experiments. If and when these initial tests are successful, the rigors of the regulatory approval process require that any treatment coming out of an AI-informed process meets safety and efficacy thresholds.
The use cases for AI expand well beyond drug discovery, though. Life sciences manufacturers can adopt well-established forms of AI to personalize consumer experiences, as the retail and banking industries have done. They can select appropriate administrative processes to automate which, when paired with appropriate human oversight, can streamline operations and help employees be more productive. And they can analyze real-world data in new ways: for example, a machine learning algorithm could analyze medical claims data and potentially identify label expansion opportunities. Researchers can also use natural language processing (NLP) to transform unstructured data from clinical notes into research-ready discrete data that other forms of AI can ingest and examine.
While all these applications have the potential to streamline operations and increase both consumer and employee satisfaction, they nonetheless come with a cost—and in many ways, it’s a harder one to swallow than the high-dollar investments in pre-lab discovery. Laws and regulations are emerging in this space, and organizations must be mindful to have proper controls and governance models in place to reduce risk of unintended consequences, such as the introduction, persistence or exacerbation of bias. The recommendations or predictions made by AI-driven models must be presented to the human end users in ways that build trust, fit seamlessly into workflows, enable responsible use and arrive at moments when they can take action to influence outcomes. It takes sustained effort, a commitment to building a culture that embraces technological change and the awareness that human oversight is needed to minimize any unintentional negative impacts to the people affected by the model’s predictions or recommendations.
Many of these data science partnerships are in their infancies, and we are only months removed from DeepMind’s decision to make their AlphaFold protein-folding prediction technology publicly available. The industry seems to be on the cusp of significant scientific breakthroughs that could create meaningful changes for patients, providers and payers—but those advances won’t be realized for many years to come.
More immediately, two macro trends are converging that force life sciences leaders to examine their AI strategies: first, the emphasis on equality and equity, and second, the societal distrust of AI. Much of the media focus on AI in health care has scrutinized the ethical or practical constraints of AI in clinical practice. As such, the burden of proof is understandably high when it comes to illustrating the benefits of any program that uses AI-driven recommendations. When collaborating with providers or other patient-facing entities on clinical programs that incorporate these insights, decisionmakers must take steps to limit unintended consequences.
Other parts of health care have already discovered that it takes an incredible effort to close the last mile between the potential and the practical when it comes to embedding AI into operations, even when there’s broad agreement that AI can offer observable value (either monetarily or by improving patient outcomes). That means it’s likely easier to judge the ROI on efforts to increase efficiency in early discovery—despite the high price tag—because the downstream gains of AI applications show up in ways that are harder to measure.
As leaders assess the ROI on downstream AI applications, they should evaluate a mix of hard costs (i.e., dollars and time) and soft costs (i.e., sustained effort and persuasion). That appraisal is further complicated by the time lag between when the technology investment is made and when the benefits accrue to patients and staff. But in an increasingly competitive landscape, leaders must examine every lever they can pull to offer exceptional consumer experiences, operate as efficiently possible and deploy human talent where it can make the most impact. Organizations must consider how different forms of AI can help achieve those goals.
The growing online presence of clinicians, coupled with their heightened demands for real-time consultation and evidence, are changing the evidence dissemination paradigm from the traditional one-way push of information to a real-time circulation of knowledge.
Online clinician communities for medical information sharing – both open social media channels (e.g., Twitter, LinkedIn) and physician-only digital platforms (e.g., Doximity, Sermo, epocrates, Figure 1) – have evolved to become top destinations for clinicians to discuss clinical evidence, network with their peers, and extend their own reach and “influence” within the health care community.
Clinicians are becoming more comfortable seeking and reviewing clinical information (including, but not limited to, peer-reviewed journal articles) and anecdotes online – especially as their traditional information access from pharma representatives, conferences, and traditional channels have been restricted due to Covid-19. Further, the pandemic spurred questions about drugs, vaccines, and conditions faster than researchers could generate evidence, so clinicians relied on crowdsourced answers from experts all over the globe. While the digitization of medical information has existed for years, Covid-19 and the acceleration of online information exchange have made medical consensus-building more transparent and accessible than ever.
As HCPs are increasingly debating evidence studies and engaging in rich discussions with their peers online, those conversations are directly informing treatment selection and care decisions -- becoming part of a dynamic body of evidence in the process. This is creating fundamental shifts the evidence communication paradigm, moving it from one traditionally focused on evidence dissemination to one focused on the circulation of evidence at unprecedented scale. As a result, evidence dissemination is no longer a static, one-way street from life sciences organizations to HCPs.
Changes to the evidence generation and dissemination paradigm will require many life sciences leaders to re-think their traditional medical information dissemination strategies – including publications, conference presentations, use of key opinion leaders (KOLs), and use of MSLs. No longer is evidence dissemination a static, one-way activity – and life sciences leaders need to understand and capitalize on the discussions surrounding their studies, and the subsequent consequences these discussions have on how clinicians practice medicine.
As interest in online clinician communities grows, life science leaders must recognize that such discussions can create new opportunities for real-world evidence generation and insight about unmet medical needs, physicians’ clinical decision-making processes, and gaps in research/clinical evidence. For example, online debates can provide insight into physicians’ perceptions of standards of care and treatment options, how clinicians decide what medical products to use, and how the current evidence base (or lack thereof) informs actual treatment decisions.
However, life science leaders must also prepare for the unintended ripple effects these platforms create. With Covid-19 accelerating the pace of evidence generation, researchers and HCPs are now demanding new data and evidence at an unprecedented pace. Whether life science leaders can keep up with these heightened demands, or will need to temper their customers’ expectations, remains to be determined.
Further, the rapid pace and proliferation evidence discussion means that conversations are happening outside of life science leaders’ control – making medical misinformation or disinformation more likely to arise. As clinicians typically congregate online by specialty or background, some clinicians may start to resist changing their perspective or opinions as online “echo chambers” of discussion can amplify preexisting biases or opinions. To adapt, life sciences leaders need to not only actively monitor these communities and the influential voices on them, but also understand how discussions impact product use, perceptions, and decisions.