Real-world evidence

The Food and Drug Administration has worked to build up its use of real-world data for evaluating COVID-19 diagnostics, participating in a partnership exploring how to use such data to improve testing and help understand the virus' spread.

Using real-world data and evidence in regulatory decision-making has been on the FDA's agenda for years. But the pandemic sped up the adoption of newly generated data so that guidelines could be updated for coronavirus tests that were cleared for emergency use based on limited information, Commissioner Stephen Hahn said at an event hosted by the Medical Device Innovation Consortium in June.

That month, the FDA joined the COVID-19 Diagnostics Evidence Accelerator, an initiative organized by the Reagan-Udall Foundation, a nonprofit that supports FDA, and Friends of Cancer Research. Among other efforts, the diagnostics accelerator will tackle a key question in understanding of the virus: whether the presence of antibodies signals protection against future infection, and if so, which ones help.

Real-world evidence provides insights on the benefits or risks of a medical product taken from real-world data, which is information derived from registries, electronic health records, claims and billing data, as well as mobile devices and other sources that reflect patients' actual use of products.

For drugs, FDA leaders have previously signaled that, while companies shouldn't rely too heavily on real world evidence, the agency is open to evaluating this information and allowing it to be used for expanding a drug's label. The agency has also said it is advancing real-world data tools to help speed up development of COVID-19 treatments. The FDA said these tools will help analyze outcomes and illness patterns, and could help to supplement placebo-controlled trials.

COVID-19 has lent urgency to the FDA's efforts on RWE, both for therapeutics and for diagnostics.

In mid-May, the agency engaged Aetion, a New York-based healthcare technology company that specializes in analyzing real-world evidence for payers, providers and regulators, to address questions about tests and medications — as well as risk factors for COVID-19 complications in various patient populations. The agency also said it's leveraging Sentinel, its electronic system used to surveil medical product safety.

But the FDA's focus on real-world evidence will continue beyond the current COVID-19 pandemic, Hahn said.

"Now as we look forward, we need to think about how real-world evidence fits into the initial authorization or approval of a product," he said. "But on the [post-approval end], what data can we collect to look at really how these devices, how the tests operate, how they perform, and then incorporate that back in?"

With the agency leaning heavily on the emergency use pathway during the COVID-19 public health crisis, the need to revisit authorizations once new evidence emerges is arguably more apparent now than ever before.

The agency has shown that it's willing to reverse course on the products it allows during the public health emergency, as additional insights on efficacy come to light. In June, for example, the FDA revoked the emergency use authorization for a COVID-19 antibody test from Chembio, one of the earliest of such tests to have received an EUA, or emergency use authorization.

"You take the data you have, you take the information you have, you make decisions, and then you update it as you go based on new information," Hahn said during the June event. "And I think that's what we've seen here during this pandemic. It's what the agency has always strived to do, we've just done it in an accelerated timetable."

Hahn said the agency has tried to balance the need for speedy access to products that can help treat the new coronavirus while continuing to ensure safety and effectiveness, as well transparency about its processes.

Still, "in a fast-moving pandemic, that's been a challenge," he said.

The FDA has been criticized for getting that balance wrong in some cases. For example, it allowed scores of antibody tests onto the market without agency review under an umbrella-type policy aimed at increasing the volume of tests available. Many of those tests were found to have high rates of false results, prompting the agency to later order some off the market and require emergency use authorizations for all test makers wishing to continue selling a product.

The agency was also scrutinized for its handling of hydroxychloroquine, which it cleared for emergency use early on only to later revoke the authorization once more data showed the drug unlikely to benefit COVID-19 patients.

Article top image credit: Dufour, Tia. (2020). "White House Press Briefing" [Photograph]. Retrieved from Flickr.

The Institute for Clinical and Economic Review, a drug price watchdog, in February partnered with health data company Aetion to add real-world evidence to its reviews of some new drugs as well as periodic re-analyses of marketed medicines. 

The announcement comes weeks after ICER said it would use such evidence, which pharma companies argue can show the value of their products beyond data from the randomized, placebo-controlled studies that are the gold standard for regulatory approval. 

Use of real-world evidence has stirred controversy in health policy circles, however, as many argue such data can exaggerate the benefit of a new treatment due to uncontrolled variables and lack of placebo comparisons. 

ICER's value assessments of new drugs are frequently criticized by biopharma companies, which in many cases argue the group's reviews don't capture enough data about how medicines benefit patients. The non-profit organization has been making adjustments to how it prepares its assessments, and using real-world evidence represents a big step forward.

New York-based Aetion has both high-powered backers as well as big customers. Sanofi has also announced a partnership with Aetion, and New Jersey insurer Horizon is relying on Aetion's data platform to analyze healthcare utilization by patients with diabetes.

Meanwhile, former Food and Drug Administration Commissioner Scott Gottlieb has joined the Aetion board, and the company has received financial backing from the venture capital firm founded by Andy Slavitt, the former acting administrator of the Centers for Medicare and Medicaid Services.

Aetion's analyses of real-world data would be used to review new drugs when data on comparator treatments or the natural course of a disease are needed to calculate the value of a new medicine, ICER spokesman David Whitrap said.

For drugs receiving FDA accelerated approval based on single-arm trials and biological activity, re-analyzing an initial ICER assessment using real-world evidence can determine whether the benefit has been as good as expected at approval, he said.

"Randomized clinical trials remain the gold standard for developing unbiased evidence, but as the FDA is increasingly approving treatments based on thinner [trial] data, the entire health industry will need to fill larger knowledge gaps about how well a new drug actually works," he said.

"Real-world data reflects settings and populations not represented in clinical trials," Aetion CEO Carolyn Magill said in an interview with BioPharma Dive. By analyzing that data, regulators and insurers can make decisions about dosing or use in additional indications and patients.

ICER will continue to seek data from other partners, Whitrap said, although he commended Aetion for both the breadth of its data as well as a "mindset" that takes into account the strengths and weaknesses of real-world evidence.

Real-world evidence usually refers to information derived from analyses of data gathered from "real-world" sources like electronic health records or drug registries, rather than controlled clinical studies. 

"We want to be as transparent as we can be as to when it is a no-brainer to use real-world data," Magill said, "as we are when the data is not there yet, or the endpoint isn't correct, or we can't adjust for confounding variables."

Real-world evidence's weaknesses were outlined in a New England Journal of Medicine article written by University of Oxford researchers this year.

"Because of the potential biases inherent in observational studies, such studies cannot generally be trusted when — as is often the case — the effects of the treatment of interest are actually null or only moderate," the researchers wrote.

"In those circumstances, large observational studies may yield misleading associations of a treatment with health outcomes that are statistically significant but noncausal, or that are mistakenly null when the treatment really does have clinically important effects," they added. 

Article top image credit: Getty Images

IDegLira may be used as an intensification strategy for patients with type 2 diabetes across a range of different baseline therapies, according to a real-world evidence study conducted by researchers at the Medical College of Wisconsin, Novo Nordisk, and Veradigm.¹ Initiation with IDegLira was shown to lower glycated hemoglobin (HbA1c) with minimal impact on weight. Among patients simplifying therapy, reductions in HbA1c were maintained despite reductions in the daily number of injectable antidiabetic medications.

Published in "Diabetes Therapy", this retrospective observational study used de-identified ambulatory patient data sourced from a large US electronic health record platform in use at over 25,000 clinical sites (mostly single provider or small group practices) across the United States.

IDegLira is a fixed-ratio combination of insulin degludec, a long-acting human insulin analog, and liraglutide, a glucagon-like peptide-1 (GLP-1) receptor agonist. Clinical trials have demonstrated the efficacy and safety of IDegLira in patients with type 2 diabetes; however, studies demonstrating its effectiveness and safety in real-world populations are needed. This study is the first to report reduction in HbA1c among patients in US clinical practice who initiated treatment with IDegLira.

Over 34 million Americans have diabetes, with an estimated annual economic burden of $327 billion.² ³ Clinical practice guidelines recommend use of a combination of basal insulin and GLP-1 receptor agonists for glycemic control in the management of type 2 diabetes.⁴ ⁵

The authors conclude IDegLira offers a simplified approach to therapy for patients who previously were receiving more than one injection daily of antidiabetic medication. Complete the form to the right to download the complete article now.

---

References:

1. Egede LE, Bogdanov A, Fischer L, Da Rocha Fernandes JD, Kallenbach L. Glycemic control among patients newly prescribed IDegLira across prior therapy group in US real-world practice. Diabetes Ther 2020. https://doi.org/10.1007/s13300-020-00850-w.
2. Centers for Disease Control and Prevention. National diabetes statistics report, 2020. https://www.cec.gov/diabetes/data/statistics/statistics-report.html.
3. American Diabetes Association. Economic costs of diabetes in the US in 2017. Diabetes Care 2018;41:917-928.
4. American Diabetes Association. 9. Pharmacologic approaches to glycemic treatment: standards of medical care in diabetes – 2019. Diabetes Care 2019;42:90-102.
5. Garber AJ, Handelsman Y, Grunberger G, et al. Consensus statement by the American Association of Clinical Endocrinologists and American College of Endocrinology on the comprehensive type 2 diabetes management algorithm – 2020 executive summary. Endocr Pract 2020:26:107-139.

Article top image credit: iStock.com/simpson33

Drugmakers want to prove their products' worth more quickly and at less cost. The demand is so great that contract research providers, often called CROs, are building out their business to meet it.

In 2018 and 2019, Thermo Fisher, Cambrex and Charles River Laboratories each spent hundreds of millions of dollars on deals meant to grow their offerings.

Data services, in particular, have become newly attractive. Pharmaceutical and biotech companies want to know what CROs can do with it, from collection to analysis to validation.

"The CRO industry started out as people, bodies for hire. That's what we were good at," said Dawn Anderson, managing director of R&D life sciences consulting at Deloitte and former employee of inVentiv, Cognizant and Quintiles.

Now, the business model is "to not rely so heavily on people, but rely on their ability to aggregate that data."

Real-world data a catalyst

The drug industry sits on mountains of patient data, and the peaks are only getting higher. Clinical trials have grown exponentially in number over the past two decades, as has the amount of information those studies generate.

Where a typical Phase 3 trial conducted 20 years ago might yield about 3 million data points, the adoption of wearables and rise of real-world evidence, according to Anderson, is allowing researchers to collect more than a million data points per patient each day.

Pharmas are looking to put this extensive data to better use, and to do so earlier. They appear particularly interested in the ways CROs can help out with real-world data, which encompasses the various information available outside the confines of a clinical trial.

That data is starting to hold greater influence over drug approvals, payer decisionmaking and even label expansions. Electronic health records, for example, helped Pfizer secure another approval for its marketed breast cancer therapy Ibrance.

Further fueling the interest is signaling from regulators that they're receptive.

In 2019, Ned Sharpless, formerly the acting head of the Food and Drug Administration, said the agency is "open for business" when it comes to considering real-world evidence. The FDA has also been working to streamline and standardize how clinical trial investigators gather and analyze real-world data.

As a result, CROs are now spending more time and money on data services. Jonathan Shough, chief information officer at Covance, noted in an interview with BioPharma Dive how his company's venture fund is investing in others that can leverage mobile health technology — a common tool in real-world data collection — in clinical studies.

"It improves the interaction with the patient, it improves the value and enrichment of the data that we get," Shough said. "I think that's going to continue to grow, our involvement in that is going to be significant."

Data drawbacks

Data services aren't a guaranteed success story, however. PRA Health Sciences, one of the largest CROs in the U.S., recorded a small dip in its data solutions business unit early in 2019, for instance. 

Collecting and analyzing data requires considerable upfront investment in technology, which can be difficult to integrate, as well as employees who can operate it. Covance, for instance, is searching for data scientists who can complement the company's existing bioinformatics team, according to Shough.

On the people front, healthcare companies have had a tough time wrestling away talent from big tech.

"CROs are going to have to evolve to survive," Deloitte's Anderson said. "The big players, the Apples, the Verilys, the Amazons, all have incredible technology capabilities around data. Those perhaps could be the new players in the future."

Operational challenges also come from drugmakers.

Michelle Hoiseth, chief data officer at Parexel, said it can be tricky figuring out what pharma clients mean when they ask about diving deeper into real-world data, because the answer largely depends on to whom they plan to submit that data. From her experience, Hoiseth contends that progression free survival data for a cancer drug means much more to regulators than payers.

"Our clients are like, 'Come talk to us about real-world data.' And we're frustrated," she told BioPharma Dive. "It's not about the data, it's what you do with it."

Regulators can be a source of frustration as well. Though the FDA seems more agreeable toward non-clinical data, there's still ambiguity about what tools and validation methods it will accept.

A hungry market

Despite the risks, hunger for both the data and broader clinical services offered by CROs should remain high. Grandview Research expects drugmakers will continue to outsource R&D operations, spurring the global healthcare CRO market to reach $55 billion by 2025.

The demand already facilitated some of the sector's largest transactions, including tie ups between INC Research and InVentiv, Labcorp and Covance, and Quintiles and IMS.

Young biotechs can be especially reliant on CROs to provide the technological and clinical platforms necessarily for drug development, much of which now targets smaller patient populations.

Big pharma, meanwhile, is slowly but surely bringing data and tech know-how in-house. That, in turn, could put additional pressure on CROs to demonstrate their value.

"Their own clinical trial data, plus whatever real-world data they buy and put into the platform, plus whatever other investments they make in tech — our job is to try to complement that and increase the R.O.I.," Hoiseth said. 

"It's not to compete with it or think that we're going to stand up something adjacent that's going to have more value than what they've already put in."

Article top image credit: Getty Images

The Food and Drug Administration is "open for business" in considering real-world evidence to clear expanded uses of new cancer drugs, the agency's former acting head Ned Sharpless said last summer, but warned industry against relying too heavily on alternative sources of data.

Speaking at last year's BIO conference, the former National Cancer Institute director acknowledged the rising cost and complexity of clinical trials, and indicated the FDA is aware of how difficult enrolling studies of rare disease subtypes can be.

"The FDA is always going to want to see really well designed beautiful trials with convincing endpoints," he told conference attendees. "If you can do a large [randomized controlled trial] with overall survival as your outcome, there is nothing better."

Still, "we realize there are circumstances where there is a need for flexibility," he added.

For example, the FDA approved Pfizer's Ibrance for male breast cancer, granting the pharma an expanded indication in last April. In making its decision, the regulator used data from electronic health records and from real-world use of Ibrance, information gathered from Iqvia, Flatiron and Pfizer.

"Male breast cancer is a rare cancer and it's not easy to do a clinical trial," Sharpless noted. "Real-world evidence was supporting material — it wasn't the sole factor."

"The FDA is open for business to consider those kinds of data," he said, while noting industry's case can't be made just by arguing it's cheaper and faster than doing a clinical study.

The increasing segmentation of cancer into genetically defined subtypes gives drugmakers a better chance of proving efficacy with a targeted drug. But recruiting for such trials requires further steps, such as genetic testing, that can add time and expense.

"When I started my career as a cancer researcher we had one kind of breast cancer. Now we have 10 kinds of breast cancer," Sharpless said. "That makes clinical trials much harder."

The FDA's willingness to consider alternate sources of data isn't new. Drugs chief Janet Woodcock told a separate BIO session the agency has used real-world evidence for years in reviewing new treatments for rare diseases.

"We have vast experience with real-world evidence on the safety side," she said.

Real-world evidence comes with concerns, both in quality and in interpretation. Still, for follow-on indications of already approved drugs, companies can make a plausible case.

"I think we have seen some of those particularly in oncology, where [drugmakers] are pulling some of that [data] from treatment use in rare cancers," Woodcock said. "I think that is a very reasonable thing to do."

Peter Marks, head of FDA's biologics unit, echoed the theme, while sounding a note of caution.

"It is a really powerful technology to use these large databases" Marks said. "Unfortunately, the electronic medical record as it has grown up in the U.S. is highly fragmented and not correct a lot of the time."

Marks encouraged industry to help ensure more accurate data in the records. "We have to get doctors to do a better job of what they put into the EMRs," he said.

Article top image credit: Food and Drug Administration

Last April, Pfizer secured an expanded U.S. approval for its breast cancer drug Ibrance in men. 

Rather than relying on clinical study, however, the Food and Drug Administration OK'd Ibrance for the new use based on data from electronic health records and postmarketing reports detailing patients' real-world experience with the therapy. 

Breast cancer in men is rare, accounting for less than 1% of all cases of the tumor type. As a result, clinical studies testing breast cancer drugs often exclude men, as was the case with PALOMA-2 and PALOMA-3, the two Phase 3 trials supporting Ibrance's approval. 

Under former chief Scott Gottlieb, the FDA sought to accelerate use of real-world data in its regulatory review of medicines. In December of the year previous, the agency had released fresh plans which, among other priorities, highlighted the potential of real-world data in supporting label expansions. 

The new approval for Ibrance didn't move the needle much for Pfizer commercially, but the OK added to a string of regulatory successes the pharma has had in front of the FDA.

In the fall of 2018, the FDA cleared four Pfizer oncology drugs, including two for certain types of lung cancer and one for breast cancer. 

That spate of approvals capped something of a turnaround for Pfizer in a therapeutic field that has fast become one of the most active in all of pharma. Last year, Pfizer's cancer drugs — led by Ibrance — nearly outsold its heart and primary care medicines, which would have been a first. 

Ibrance's commercial success has been rooted in its approvals for hormone-positive, HER2 negative metastatic breast cancer. Until the April 2019 approval, first-line use of the drug was limited only to postmenopausal women.

Now, Ibrance is approved for the estimated 2,670 men who will be diagnosed with breast cancer each year.

To secure the FDA's nod, Pfizer submitted an analysis of real-world data from EHRs as well as reports from male patients treated with Ibrance sourced from databases including Flatiron Health's, Iqvia's and Pfizer. 

The FDA has used real-world data in reviewing drugs before, but frequently such evidence has been used to construct a "historical control" to compare results from single-arm studies of a drug. That was the case with Amgen's Blincyto, first approved in 2014 for a type of acute lymphoblastic leukemia. 

The promise of using real-world data to create virtual control arms was part of Roche's rationale for buying Flatiron in February 2018 for $1.9 billion.

Despite that acquisition, Flatiron has continued to operate as a seaprate legal entity and to work with external partners like Pfizer. 

Article top image credit: Norbert Nagel, Mörfelden-Walldorf, Germany

Drug regulators are beginning to look at more than primary endpoints when evaluating new products.

Particularly in the last few years, the Food and Drug Administration has shown an interest in patient-reported outcomes, meaning the information patients can provide about their disease burden, the drugs they take, and associated feelings and experiences.

The information is seen as a valuable tool for improving clinical trial designs, guiding drug approvals and supporting label decisions. A six-minute walk test, after all, doesn't necessarily indicate whether a patient could make it down the hall to the bathroom in his own home.

Yet the reach of patient-reported outcomes, or PROs for short, has been limited, in good part because of the varied ways clinical trial investigators collect and assess them.

Two studies of chronic migraine medications, for example, may ask participants to answer differently-worded questionnaires about their pain level. The studies might also use different tools — a wearable versus neural imaging — to measure and validate the responses.

The disparities have troubled regulators, who are tasked with evaluating the merit of clinical data. And they've presented obstacles for trial sponsors and operators. Michelle Hoiseth, chief data officer at Parexel, noted researchers aren't generally going to gamble on a multi-million-dollar clinical trial with data collection methods they don't think will be accepted.

As a result, PROs rarely get top-billing in the hierarchy of study endpoints.

"That's often a missed opportunity: when a well-developed endpoint, a patient-centric endpoint isn't given the proper attention that it deserves," said Elektra Papadopoulos, associate director of clinical outcome assessments staff in the FDA's Center for Drug Evaluation and Research.

Papadopoulos, along with other FDA leaders, went over the agency's most recent efforts to guide and understand PRO use during a panel discussion at the Drug Information Association's annual meeting, held last June in San Diego. 

From 2013 to 2017, CDER staff met two dozen times with patients who had certain chronic illnesses to discuss treatment burden and unmet needs. Input from those meetings helped streamline disease-specific guidances, according to Papadopoulos.

CDER and its equivalent biologics division, CBER, rolled out in 2018 the first in a four-part series of guidances about collecting and submitting patient experience data. The first in the series, while still a draft guidance, focused on from whom and why researchers or sponsors should gather data.

It recommends tapping statisticians and other subject matter experts when deciding what sampling method to use for PROs. The FDA also nudged trial organizers to meet early and often with the agency to iron out whether the intended data collection methods and strategies align with their studies' objectives. 

Broadly, regulators are trying to make the PRO process consistent. The FDA recently launched a pilot program aimed at creating publicly available, standard sets of Clinical Outcome Assessments (COAs) measures for distinct indications. The program already has buy-in from the National Institutes of Health, which is offering grants for qualified participants.

The agency's argument was that there were so many independently developed PRO protocols and tools, it created a "continued learning curve" as well as higher costs and less long-term sustainability. The variability also threatened the quality of COAs, which "both limits their value to decision makers and can potentially undercut confidence in these data," the FDA said.

Not everyone is sold on this push.

One audience member noted the problem of buy-in during a Q&A with the panel. He said a potential corporate partner of his doesn't feel the need to submit PROs for a drug it's working on because the drug has breakthrough status and only needs to show biomarker data to gain approval.

During a separate DIA panel, an audience member who identified herself as a licensing specialist from Merck explained how, at times, the level of competition in the pharma industry can be counterproductive. She voiced concerns about whether standardization would exacerbate the issue, especially if the FDA chooses a go-to COA that uses expensive data collection and measurement tools.

"I might pay $400,000 for a measure that's the gold standard. And that's where companies start using those measures less and less," she said.

Panelists in both sessions acknowledged there will be problems left to solve even with the implementations of PRO programs. Still, they maintained that doing something is better than leaving the system as is.

"If every company develops their own outcome set, it will just not fly," said Anton Hoos of Patient Focused Medicine Development.

"As an industry, we have to start agreeing on what could be standards that we measure our drugs against, because this is ultimately what the patients will ask from us and regulators will ask from us," he added. "It will make it easier for all of us."

Article top image credit: Getty Images

Pharmaceutical companies amass troves of data. But for most large drugmakers, the executives in charge of those data stores don't make the C-suite.

Only seven of the nearly 30 pharma and biotech companies valued at more than $10 billion as of last summer ranked a chief digital or information officer alongside their heads of science and research on their executive committees.

Six of those individuals, however, were appointed to top management within the previous two years — a shift that came as tech companies like Amazon and Apple have inched further into medicine.

Such placements afford greater influence for organization-wide changes, industry watchers say, and signal drugmakers are taking seriously the potential of digital technologies to change how they do business. For some companies, the role previously existed at lower levels of management, or responsibilities were spread throughout different business units.

"Whether it's warranted or not, every company is becoming a data and technology company," said Aarti Shah, chief information officer at Eli Lilly since 2016 and a member of the pharma’s executive committee since 2017, in an interview.

Yet for all of pharma's biomedical innovation, drugmakers don't enjoy a similar reputation on the digital front. Healthcare has been slow to adopt even basic tools like electronic record-keeping and, outside of areas like imaging, use of more advanced technologies like artificial intelligence remains more flash than substance.

The newly appointed chief digital officers will be in charge of changing that state of affairs. As direct reports to the CEO, they are in a better position to push forward projects and secure resources to do so, said Shah.

"The pharma industry is behind compared to other industries in terms of digital disruption," said Bertrand Bodson, chief digital officer at Swiss pharma Novartis since mid 2017, when he joined from U.K. retailer Sainsbury's Argos.

"But, gosh, the impact could be bigger than in most industries."

CDOs moving up

It may be the digital age, but that hasn't meant a CDO in every boardroom.

About a fifth of the top 2,500 public companies globally have an executive in the role, according to results from a 2019 survey by PwC.

CDOs are most common in industries like insurance, banking, communications and retail. As a broad group, pharma and healthcare ranked roughly middle of the pack by CDO representation, PwC found.

Until recently, though, CDOs weren't on many drugmaker executive committees. That's now changing, as the recent spate of executive committee appointments in pharma have put CDOs or their equivalent in top positions at half a dozen of the industry's biggest names.

Since summer 2017, GlaxoSmithKline, Novartis, Pfizer, and Merck & Co. each tapped executives formerly at retailers like Walmart and Nike to run their digital operations. Eli Lilly and Sanofi, meanwhile, promoted from within, naming company veterans to the higher-profile posts. (Merck's top digital officer has since left the post since this article was published, as has Sanofi's.)

For all six pharmas, the appointments were the first time chief digital or information officers have sat on the companies' executive committees. Bristol-Myers Squibb has had a chief information officer on its senior leadership team since 2016.

"It's not surprising that they're allocating specific resources to people in these seats," said Vamil Divan, an analyst at Credit Suisse who covers large pharmas, in an interview.

"They're all talking about the importance of big data, and they're all to some extent behind in using data in their drug development and commercial efforts."

Gold mines of data

Over the past two decades, Novartis estimates it has collected 2 million patient-years worth of clinical trial data. Its labs, meanwhile, house 1.5 million chemical compounds.

It’s a similar story elsewhere in pharma, where researchers toil in laboratories to solve the chemical structures of molecules; clinical studies track results from tens of thousands of patients each year; sales teams meet and log interactions with physicians across the globe.

"We sit on gold mines of data," said Shah, who rose through Lilly's ranks since joining as a statistician in 1994.

Putting that data to better use has become a top priority for pharma, one made more pressing by faltering R&D productivity and rising clinical trial costs.

"Over time trials get more complex, the requirements from regulators get more complex and, because the science gets more complex, we actually measure more things," explained Novartis CEO Vas Narasimhan earlier this year on a podcast recorded by Andreessen Horowitz, a venture capital firm.

"What our industry has really not been great at is deploying technology to make this more efficient," he said, noting estimates that suggest better use of data could cut as much as 20% of a study's cost.

A good deal of that expense is rooted in operational issues: enrolling patients into studies, ensuring protocols are followed and data is properly collected.

Under Narasimhan, the pharma has launched 12 digital projects across its business, including one designed to use machine learning to predict the performance of the roughly 500 trials the company has underway. (Machine learning, broadly speaking, is an AI technique in which computer models progressively learn from experience.)

For inspiration, Novartis sent staff to observe the electricity supplier Swissgrid as well as aircraft control towers, incorporating lessons on how to better manage clinical trial operations in real time.

Behind the scenes at our new ‘control tower’ to track, analyse and predict the status of all our clinical studies. 500+ active trials, 70+ countries, 80 000+ patients - transformative for how we develop medicines. pic.twitter.com/8lP0zZpQAl

— Vas Narasimhan (@VasNarasimhan) December 11, 2018

Lilly's Shah, meanwhile, points to the two years the Indianapolis company claims it has recently cut from the average time one of its drug candidates spends between first human dosing and commercial launch.

Culture shock

As is the case elsewhere in healthcare, though, drugmakers face an enormous challenge in translating all information they generate into something usable.

"For each disease area, units run their own trials in their own ways," said Novartis' Bodson, in an interview. "We spent quite a lot of energy harmonizing all those data sets."

And empowered leadership doesn’t always yield improvement, as the track records of many corporate restructurings can testify.

"If you're a big company, change is hard because it’s cultural," said Alexis Borisy, a partner at Third Rock Ventures who has helped launch biotechs like Insitro, Celsius Therapeutics and Relay Therapeutics.

Notably, several of pharma's newest CDOs don’t come from a drug discovery or clinical science background, joining instead from roles at consumer companies.

"If you're thinking about a pharma company as an organization that discovers drugs, the phenotype that you need is very different than the chief digital officers that companies have been appointing," said Daphne Koller, CEO of the drug discovery startup Insitro.

R&D leaders who have deep backgrounds in artificial intelligence aren't easy to find, of course — a challenge that extends beyond the boardroom, too.

"We really need a whole new generation of people that straddle the world of machine learning as well as life sciences," said Borisy.

Article top image credit: Novartis