Suzanne Shelley, Contributing Editor

It’s widely understood that poor adherence to prescribed medication regimens has far-reaching clinical and cost implications for all healthcare stakeholders.

For example, patients who never start or don’t stay on prescribed treatments fail to control their symptoms and often experience disease progression and increased morbidity and mortality. Insurance companies end up paying for expensive medical interventions that could have been avoided by more stringent adherence to therapy. Meanwhile, even pharmaceutical brands fail to recoup the sunk costs associated with drug development and miss the opportunity to achieve competitive success for the brand when their often life-saving medicines—which were proven to be safe and effective in clinical trials—fail to deliver on their promise due to adherence lapses.

“Additional tangible and intangible societal costs arise in terms of job loss and reduced work productivity for both patients and caregivers,” says Pierantonio Russo, MD, chief medical officer for EVERSANA.

Considering how many clinical and financial factors, unmet social needs, mental health challenges, and comorbidities are at work, it’s no surprise that patients have a hard time remaining adherent to one or more complex therapy regimens as they struggle to manage one or more chronic conditions. While the reasons are varied, there is one thing everyone can agree on: No single “one-size-fits-all” intervention will suffice when it comes to helping diverse patients to overcome the diverse adherence challenges they face.

Fortunately, two factors are working in tandem to break the logjam on adherence:

The collection of real-world data (RWD) from both established and novel sources continues to grow throughout the healthcare arena.

The number of artificial intelligence (AI) methodologies—including those based on machine learning (ML), natural language processing (NLP), generative AI, and more—continue to mature, enabling stakeholders to convert RWD into real-world evidence (RWE) that provides actionable insights.

When it comes down to interpreting patient data, specificity is likely to garner better outcomes.

“The goal is to use today’s advanced modeling technology to better understand the complex relationships between patient-level characteristics, adherence-related behaviors, and health outcomes, and then use the data-driven insights to deliver individual level outreach, making them patient-specific (not generalized) to improve outcomes,” says William Grambley, CEO, AllazoHealth.

In fact, being able to connect the source of information to one individual creates a more streamlined process.

“The ability to access many sources and link them to the same patient is really extraordinary in terms of increasing efficiency and helping us to make connections and predictions we could not have made before,” adds Amanda Bruno, PhD, MPH, senior vice president in real-world and late phase at Syneos Health. “It will be exciting to see how this increased sophistication in data will impact the pace and integrity of future evidence generation."

Ensuring quality, accuracy, and reliability of insights and predictions generated by today’s AI models is paramount. Stakeholders must prioritize the development of algorithms using credible real-world healthcare data, while also leveraging expertise from scientific and medical professionals. Validating models of predicted results against real patient outcomes is essential to ensure confidence and trust in AI-driven healthcare advancements, notes Christina Mack, chief scientific officer, real-world solutions, for IQVIA, adding: “Today’s predictive models continue to get better in terms of identifying which patients are at high risk for non-adherence, and that helps stakeholders to create personalized adherence interventions that are informed by the patient’s actual personal choices and barriers.”

Experts believe any opportunity to build an algorithm that accurately reflects patient behaviors in the real world can feed further understanding of the many barriers patients are facing, and from there, "then develop more targeted outreach that helps to personalize each patient journey based on predictability,” says Angela Wlodar, MS, Syneos Health’s senior vice president, medical affairs in business development.

In the clinical trial setting, safety and efficacy is validated in a highly controlled environment where drug regimens are administered and monitored closely by healthcare professionals, so adherence lapses are not a factor that impacts the drug’s performance.

“By contrast, once the drug moves into the real world, healthcare systems, providers, and patients become responsible for many of the factors that impact adherence, so a large amount of control is lost,” says Mack.

She notes that the types of adherence challenges patients face can be organized broadly into the following three buckets:

 Challenges that relate to access to healthcare resources and communication methodologies and format that may not be in the patient’s primary language.

 Obstacles such as cost issues, complexity of the drug-administration requirements, and the impact of adverse effects.

Challenges that connect back to comorbidities, management of side effects, mental health issues, and other considerations related to social determinants of health (SDOH).

“Creating effective interventions requires us to consider all of the factors we can modify to make the patient journey more successful, not just focus on what the patient is or is not doing,” adds Mack. “We see a lot of promise in marrying AI and RWE to meet the challenge, and we are pushing the envelope and investing in targeting these capabilities in a robust and validated way.”

One particularly vexing challenge for brand teams relates to patients who are prescribed complex drug regimens to manage potentially life-threatening conditions that present no real symptoms or discomfort on a day-to-day basis, such as cardiovascular disease, hypertension, high cholesterol, and diabetes.

“Patients are much more likely to remain adherent to their drug regimens if they feel their medications are helping to alleviate symptoms, make them feel better, and improve the quality of life,” says John Doyle, DrPH, president of consulting for Fortrea, and adjunct assistant professor at Columbia University’s Mailman School of Public Health. “Brand teams can be more effective in how they engage with patients by deploying AI to personalize messaging and behavior nudges, helping them connect the dots and really understand how the medications work, how their disease will progress, and what risks arise if they don’t remain adherent to therapy.”

Yet every patient has different needs and preferences related to this knowledge gap.

“If we’re not finding ways to personalize outreach to help patients and their caregivers manage their expectations and clarify how the health journey will continue to evolve at every inflection point, we’re missing a key opportunity to help the patient remain on therapy over the long term,” says Wlodar.

As an industry, “we do a lot of great marketing to physicians to support prescribing practices,” adds Bruno. “But not enough of the outreach is directed at patients to help them truly understand what’s at stake so they can take control and become better healthcare advocates for themselves.”

There’s another often-overlooked aspect when crafting patient-focused outreach.

“The tech industry and other consumer brands do a very good job persuading consumers to believe their products can really improve their lives and help them to be their best selves,” says Doyle. “Drug companies can learn from them and should make such targeted messaging a bigger part of their overall health and adherence intervention campaigns.”

Affordability, or lack there of, is another major contributor to poor adherence. Healthcare programs that implement technology advances on a continuous basis can help. For example, they can aid in streamlining the benefits-verification process or automating the steps required to onboard patients into a health plan. This can "help to instill confidence in access and reduce the burden for patients and healthcare providers by removing one more barrier to therapy,” notes Jordan Armstrong, VP of business development for AssistRx. Meanwhile, complex drug-administration regimens—particularly those associated with self-injected and infused biologic therapies—can be daunting, so quickly identifying which patient needs what type of support can be a benefit.

“Whether the patient needs support to coordinate travel to a doctor’s office or infusion center, or must learn how to self-administer these high-cost, high-stakes therapies at home, both can be difficult—even for the most motivated patients,”adds Armstrong.

Predicting behaviors, tailoring the outreach

All of this becomes even more daunting for patients facing structural health equity issues. Such SDOHs include challenges related to healthcare literacy, education level, socioeconomic status, lack of transportation, childcare challenges, and the inability to take time off work to manage health conditions. “More than ever, addressing these multi-faceted challenges requires tailored interventions, education, and support to enhance adherence and improve outcomes,” says Russo.

One proven strategy uses modeling and predictive capabilities to stratify patients by adherence risk factors as they are being enrolled into patient-support programs, so that tailored outreach can be offered based on individualized risk factors and communication preferences. “Using data-driven insights into the barriers a given patient is facing, pharma companies may choose to provide even more-targeted interventions, such as transportation, childcare, or lodging coordination and services to help them remain on therapy,” says Armstrong. Similarly, he notes that providing remote nursing-support resources is particularly useful to help patients master self-injection routines without the hurdles they may face in traveling to a provider’s office.

Deciding what to send to patients at what time

Pharma brand teams have a lot of options when it comes to how and when to influence and engage the patient, in terms of phone calls, email, and SMS text campaigns that provide reminders, information about adverse effects, access to nurses for clinical information, and benefits counselors. All of these avenues reduce the administrative burden for coverage.

But knowing exactly what to offer at what point in the patient journey so that a given touchpoint will be most effective is still a mystery to most brand teams, contends Grambley, whose company uses AI to make predictions on specific patients in order to tailor their results.

“By optimizing communications for patients at the right time, on the right channels, at scale, we have proven that we can get them on therapy sooner, stay on therapy longer, and get more benefit from the medication,” he says. “For the most part, we do not change the branded program offerings and messaging that the pharma company has already developed. Rather, by studying and modeling more than 500 different variables related to their preferences and behaviors in the past, we can deliver decision intelligence that allows the brand team to proactively target the timing and fine-tune which program elements reach which patients at what time.”

In one example working with a pharma brand that offers a therapy for a major depressive disorder (MDD), AllazoHealth’s AI platform was used to predict which patients were at risk of non-adherence to their MDD medication, determine the most impactful email content to address that risk, and personalize the communications for each patient. According to Grambley, doing so boosted first-fill rates by 8.4%, while average time a patient remained on therapy rose by 13%. Also, emails sent to MDD patients were 2.5 to 4.2 times more effective with AI optimization than those without.

Meanwhile, AllazoHealth is also piloting an advanced approach to personalize “dynamic modular content” with pharma clients. Consider a matrix that involves different versions of the approved subset line, headlines, email content message, and images that may resonate differently with different people, and specific calls to action that relate to different challenges patients may be facing.

“Using our data-driven intelligence, we are able to drive the optimal combination of elements from the matrix that will provide the most personalized and impactful outreach for any given patient," says Grambley. AllazoHealth plans to roll out this feature more broadly later this year.

Mack adds that having patients on the correct treatment pathway to begin with is a key component to adherence, and a place where leveraging AI and RWE can drive improvements. In a recent study in Germany, IQVIA leveraged local RWD to identify and profile unmet needs among asthma patients, based on demographics and other clinical drivers, leading to a 17% reduction in uncontrolled asthma in six months following new treatment initiation.

RWD sources that provide insights into patients’ lives

Traditionally, identifying non-adherent patients has relied on direct inquiries, surveys, and analysis of pharmacy data for missed refills, notes Ramaa Nathan, PhD, VP of data science and RWE insights, EVERSANA. However, both approaches are costly and time-consuming; frequently, they do not allow for the discovery of all non-adherence contributing factors.

By contrast, today’s AI methodologies are able to produce sophisticated pattern recognition and build models based not only on EHR data, lab tests, and phenotypes and genotypes associated with the disease, but can identify relevant patterns and make predictions related to patient demographic data and SDOH.

“We know these factors play a direct impact on drug adherence and clinical outcomes,” says Doyle.

“Today, we are able to build ML predictive models with inputs from extensive datasets, including clinical notes, administrative claims, behavioral health profiles, SDOH, patient-reported data, and social media listening for early identification of at-risk individuals,” adds Nathan. “This provides a superior approach over traditional regression methods, as it recognizes complex, non-linear patterns and its predictive accuracy improves over time with the input of new data.”

According to Nathan, classifying patients into different clusters and recognizing those patient sets that are most likely to encounter adherence issues can help when developing personalized outreach methods.

IQVIA is working to use NLP to simplify medical terminology for patients (removing a common adherence barrier),and has developed an AI-powered patient-relationship management tool, which, Mack says, is being used by several pharma brands.

“The chatbot can digest the patient’s input and route them to the most appropriate person or resource to triage the situation in a personalized and accurate way, allowing both patients and physicians to overcome barriers in access and affordability to promote adherence,” she adds.

Meanwhile, AssistRx’s patient-support programs currently use voice/tonal recognition technology to identify, address, and report on access and affordability challenges patients are facing that can reduce adherence. “As AI becomes more sophisticated, we’re excited for the opportunity to be able to flag patient behaviors and interactions—live during calls with our representatives—that are barriers to access and adherence, then proactively address those challenges,” says Jordan.

Success in this department, however, will ultimately come down to continuity.

“Once we as an industry put these life-saving therapies out into the world, sustained patient adherence is one of the most critical factors to ensure these products can deliver on their promise and potential,” says Mack.

The ability to implement AI, catered to specific patient needs, as detailed, will be a major catalyst in these efforts.

“Any given patient program will always help some people but not all of the people,” notes Grambley. “But when you can apply AI to really tailor the options and help patients to get exactly the type of support they need at the right time, it’s kind of magical.”

Articles

By marrying real-world evidence and AI, pharma stakeholders can identify new opportunities to tailor interventions in a way that is most impactful to patients—helping them overcome roadblocks in initiating and maintaining therapy. 

Combining RWE and AI to Bend the Adherence Curve 

Real-world evidence (RWE) plays an essential—and increasingly viable and visible—role in enabling evidence-based medicine. Such data-driven insights, developed after a therapy has already entered the market, help to address critical clinical and safety questions that arise between the end of the randomized clinical trial (RCT) and longer-term use of the drug in day-to-day clinical practice. Evolving regulatory precedents and frameworks that signal increased receptivity among regulators are creating momentum for pharmaceutical companies to use RWE-based initiatives to drive both clinical and financial objectives and, where possible, demonstrate competitive advantage for their products.

Analyzing real-world data (RWD) from a variety of sources—including electronic health records (EHR), claims and pharmacy data, wearable monitoring devices, patient and product registries, patient-reported outcomes (PROs), and more—helps drug manufacturers to validate how a treatment (prescribed on- and off-label) performs in patients who are decidedly more heterogeneous than those who participated in the RCT.

Scientifically validated RWE findings are able to demonstrate the safety and efficacy of the drug in patient subgroups that did not participate in the underlying trial, especially those with comorbidities; provide evidence that supports different dosing strategies or methods of drug administration; validate that the drug provides superior comparative effectiveness against a competitor product or the standard of care; verify that the therapy improves quality-of-life for patients; and articulate the overall value or competitive advantages.

Outcomes such as drug adherence, hospitalizations, physician visits, and total costs of care are much more easily collected through administrative claims data sources rather than through randomized trials, yet they still represent important information for payer decision-making, notes Nicole Ferko, SVP and general manager, value and evidence for Eversana.

To date, the most commonly explored use cases for RWE-based efforts include oncology, precision medicine and niche indications, and orphan drugs for rare diseases. "Typically, the approval of these therapies within these indications is based on small clinical trial sample sizes (sometimes single arm and small patient numbers) and so it is difficult to make an accurate judgment of how representative the data is in relation to how the product will perform over time in a real-world patient population,” says Mike D’Ambrosio, VP, head of RWE solutions for Syneos Health, adding: “It is not uncommon for pricing/reimbursement discussions and future agreements in these situations to come with a commitment to demonstrate comparable effectiveness/value in a real-world patient population.” Today, health technology assessment (HTA) bodies are increasingly asking for RWE to help address uncertainty

 and develop more longitudinal data that verifies the durability of the clinical response over time. Today, health technology assessment (HTA) bodies are increasingly asking for RWE to help address uncertainty

 and develop more longitudinal data that verifies the durability of the clinical response over time.

This makes sense because it is not uncommon for drugs approved in these 

narrow indications to be used broadly off-label, where there may be no other therapeutic alternative. “Such high-cost specialty therapies have the most to gain when stakeholders can study RWD related to how the therapy is faring among diverse patients who were explicitly excluded from the underlying trial,” notes John Doyle, DrPH, formerly group vice president at Exponent, now president, consulting at Fortrea, and adjunct assistant professor at Columbia University’s Mailman School of Public Health.

Staggered approvals also provide opportunities for RWE insights to be developed. For instance, “in oncology in particular, on multiple occasions, drugs that have been approved initially in the US provide opportunities for other countries to learn from ongoing RWD collected in the US and generate evidence that can help to answer the health organization’s questions in those countries,” notes Javier Jimenez, chief medical officer, Flatiron Health.

In the US, the use of RWE in regulatory submissions has grown significantly in recent years, climbing by more than 120% since FDA published its Evidence Program framework, notes Kim Riggs, VP of value & access strategy at Cencora (formerly AmerisourceBergen), according to a paper she recently co-authored.

“In any crowded therapeutic space, the safety and effectiveness of the approved therapies have generally been established, so how do physicians choose which drug is the most appropriate for any given patient?” adds Christina Mack, chief scientific officer, real-world solutions, for IQVIA. RWD enables greater understanding of how factors such as demographics, comorbidities, geography, and lifestyle can help healthcare providers (HCPs) determine the right product to prescribe for specific subpopulations.

Meanwhile, data-driven findings can demonstrate incremental benefit, overall effectiveness, and comparative effectiveness—helping stakeholders to articulate why it is so important for the health plan to provide coverage for a particular treatment. “Through retrospective database studies, drug developers can uncover data that helps to articulate the overall disease burden, and demonstrate product value via increased adherence, fewer hospitalizations, reduced physician visits, and superior safety profiles,” says Riggs.

By way of example, Trinity Life Sciences recently conducted a RWE study on behalf of a client that showed that, after switching to the client’s therapy, patients no longer needed other supporting drugs, so their overall use of concomitant medications came down.

 “Reducing overall medication use and pill burden is a big factor in adherence and a huge win for patients, so positive news for our client,” says Nandini Hadker, partner and head of evidence strategy at Trinity Life Sciences.

“Analyses of trial data, ongoing PROs, differences in adherence and adverse event data in different patient subpopulations, more robust payer claims data over time—all of these can build support for reimbursement and in negotiations,” says Alan Crowther, general manager, products, Eversana. He notes that even if these factors don’t result in a therapy switch or a change in overall formulary placement, the effort can impact the nature of step edits and prior authorization criteria. In turn, that could translate to a more favorable drug formulary placement, thereby boosting patient access.

“Outcomes such as adherence, long-term safety, durability, and changes in healthcare resource use and costs can help to inform decision-making by payers and HTA organizations, which is especially important in markets involving constrained budgets and limited healthcare resources,” adds Ferko. One recently published, national US payer survey

 demonstrated that RWE was widely used in oncology decision-making for formulary placement of products in the absence of clinical trials.

However, the receptivity of RWE does vary by HTA body and by case scenario; there is no one-size fits all approach in this evolving landscape. Ferko also notes that the list of examples of successful reimbursement that is based primarily on RWE continues to grow.

“The use of RWE in initial negotiations for drug pricing and reimbursement has much more potential across the US compared to other parts of the world,” adds Crowther. “This is due primarily to a number of challenges that exist globally and the variable acceptance of RWE and RWD in these regions.”

Experts are quick to acknowledge the value of RWE across the product-development lifecycle—from early opportunity assessments to launch planning, evidence generation, brand management, commercial optimization, and post-marketing assessments. “Pre-launch, [clients] are also using RWE to assess unmet needs in the 

current care paradigm and identify gaps,” says Hadker. “Post-launch, they are using RWE to showcase how their product is providing value—not just clinically and economically, but humanistically as well.”

Meanwhile, RWE is also being used to inform field sales force, detailing efforts that are used to increase brand awareness among HCPs. RWE strategies and insights can help to not only inform the content of the targeted messaging, but identify and prioritize call targets, as well. “Patient-level and/or prescription data can be used to identify those HCPs or practices that represent the highest opportunity, so that pharma field representative territories and strategic call plans can be designed accordingly,” notes Adrienne Lovink, partner and head of RWE at Trinity Life Sciences, adding that this is an application that is shifting quickly to take advantage of artificial intelligence (AI) and machine learning (ML) tools to unlock more nuanced insights.

RWE studies also provide an important tool for supporting long-term safety monitoring and informing on the occurrence of rare events. “Clinical trials typically focus on clinical effectiveness endpoints, but due to their small size and limited duration, they may not be large enough to detect small differences in adverse events that are uncommon or rare,” says Ferko. “By comparison, sources of RWD can involve thousands to millions of patients, which provides a broader foundation for exploring safety outcomes.”

Biosimilars are another class of products that often typically enter crowded therapeutic spaces. “RWE is not yet being used as widely used to support the first-wave biosimilars but will likely play a bigger role to help the more sophisticated immunotherapy biosimilars—such as the PD1 and PDL1 therapies that are approaching the patent cliff—gain traction as they enter the market, adds D’Ambrosio.

Many providers and patients tend to prefer the use of the more costly originator biologics, due largely to good clinical awareness and brand loyalty, “so the ability to develop RWE that verifies the long-term safety and clinical effectiveness of lower-cost biosimilars may be what is required in some instances to support greater uptake,” explains Ferko.

As all stakeholders continue to grow more comfortable with the use of RWE, it has become clear that most of the ongoing challenges are not necessarily philosophical or ideological in nature, but are science-based challenges, notes Doyle, adding: “It is still of paramount importance that high-quality RWD be gathered and analyzed using proven, state-of-the-art methodologies and the highest scientific standards.”

“When you don’t have randomization, it increases the risk of confounding and other forms of bias, as there may be missing data, there may be lag time for data availability, and so on,” continues Doyle. He adds that RWD collected under pragmatic trials that randomize 

patients at baseline and then convert to an observational design offers a uniquely effective approach to producing regulatory-grade RWE. These trials, however, are often expensive and time-consuming.

As a result, Ferko suggests that RWE be complementary to RCTs rather than a replacement for them—due to the hype surrounding growing data sources and AI technologies that can create blind spots to the rigor of strong trial designs.

According to Doyle, one pressing challenge with using EHR data as a primary source of RWD is that roughly 80% of the data is unstructured, via narrative text, clinical notes, medical imaging data, and lab reports. However, he adds that progress in the use of natural language processing (NLP), machine learning (ML), and other advanced data-analytics methodologies allow teams to extract and standardize important information from such unstructured data sources.

Insights gleaned from complex unstructured patient data can not only provide unique perceptions into patient care and disease states but can help clinicians to identify at-risk patients based on social determinants of health. These can encompass socioeconomic, behavioral, and environmental factors that contribute significantly to most health outcomes. These insights can then be used to develop targeted interventions, notes Mack of IQVIA.

“The ongoing technological advances that are being used to analyze RWD are incredibly exciting, and we are watching this unfold every day,” adds Ferko. “These techniques have the power, speed, and accuracy that allow stakeholders to move from raw to actionable data more quickly, across diverse scenarios, to effectively harness and integrate such data at much larger scales.”

Today, the use of advanced modeling and data-analytics techniques continue to demonstrate many interesting breakthroughs, says Nabha Subramanya, VP of data science at Trinity Life Sciences. Examples include using:

While machine learning can help to generate structured data from the unstructured fields, the details regarding clinical context, intentions, and the process of data collection and data empathy can be lost,” cautions Jimenez. “Applying the right framework to maximize the integration of human abstraction and machine learning is critical to enabling effective and accurate use of RWD for research purposes.”

More recently, large language models (LLMs), especially medical LLMs, are helping to address this challenge. “Qualitative studies of clinical notes that would have been previously too large to accomplish in a reasonable amount of time can be done with relative ease, quickly, and at scale, and LLMs can be used to pull specific information directly out of clinical notes in a structured format or can summarize large notes into a single paragraph,” explains Ramaa Nathan, VP, data science and RWE insights, Eversana. “The realm of data from wearable monitoring technologies, and the use of advanced technology to transform digital images into structured data, has opened vistas for disease screening, particularly in scenarios where specialist accessibility is limited.”

“Substantial innovation in LLMs such as OpenAI’s GPT-4, ChatGPT, and Google’s LaMDA and Bard stand to have numerous benefits when potential risks are properly mitigated,” adds Mack. “As experts in healthcare and research methods, it is crucial for us to understand how these techniques perform when applied in specific healthcare settings.” IQVIA recently published a white paper4 that shares benchmark results of the accuracy of these techniques when applied in specific healthcare settings, to help users understand the benefits, risks, and tradeoffs in accuracy—and tie those considerations into value where these techniques are used. 

1. Dabbous, M., et al. A Review of Real-World Evidence in HTA Decision-Making of Gene Therapies in the US, Canada, UK, EU4 and Japan, Poster presented at ISPOR US. May 2023. 

https://www.ispor.org/docs/default-source/intl2023/ispor-us-evidera-2023rwe-in-htas-for-gtsposterhta34final127079-pdf.pdf?sfvrsn=ad2f91fc_0

2. Tormey, K.; Gaiser, A.; Danavar, A.; Campbell, D.; Riggs, K. RW1: A Review of Real-World Evidence Use in United States Regulatory Submissions. ViHRI. 2020. (23) 1, 383-384. 

https://doi.org/10.1016/j.jval.2020.04.1491

https://www.valueinhealthjournal.com/action/showPdf?pii=S1098-3015%2820%2931679-X

3. Faraone S. et. al., A Retrospective Real-World Analysis Demonstrating the Value of Azstarys for Patients with Attention-Deficit Hyperactivity Disorder, Poster 112, Psych Congress 2022. 

https://corium.com/pdf/adhd-treatment-patterns-using-azstarys.pdf

4. Navigating New Large Language Models in Healthcare NLP. IQVIA White Paper. July 2023. 

https://www.iqvia.com/-/media/iqvia/pdfs/library/white-papers/navigating-new-large-language-models-in-healthcare-nlp.pdf

How drugmakers can better harness real-world evidence to drive their clinical and commercial objectives, while also providing greater clarity for healthcare providers and improved access to therapy for the patients they serve.

The Business Case for RWE

Inconsistent adherence to prescribed medications results in clinical and financial implications for all stakeholders, including patients, prescribers, payers, and the overall healthcare system. For patients managing certain chronic conditions, such as diabetes, rheumatoid arthritis, and many types of cancer, failure to stay on therapy makes it harder for these individuals to manage their symptoms and slow their disease progression. For patients managing other conditions, such as hypertension, high cholesterol, and cardiovascular disease, which often present no symptoms on a day-to-day basis, failure to consistently take their treatments as prescribed increases the patient’s risk of experiencing acute medical emergencies, hospitalizations, or even death that may have been prevented.

Payers end up shouldering the cost of hospitalizations, emergency room visits, and other medical and therapeutic interventions that could have been prevented by consistent adherence to the prescribed medication regimen. Meanwhile, poor adherence among large patient groups over long time periods also dilutes and confuses the real-world evidence (RWE) that life sciences companies develop to help articulate the therapy’s overall safety and efficacy profile. Mixed clinical signals coming from real-world patient experiences can damage not only the brand reputation among prescribers and patients but, importantly, can undermine the pharma company’s bargaining power when it comes to the data-driven negotiations they engage in with drug-pricing and formulary-placement decision-makers.

Meanwhile, as the Inflation Reduction Act (IRA) is likely to create pricing pressure for many branded therapeutics, “it’s more important than ever for drug companies to double down on adherence-related initiatives,” says Kim Plesnarski, VP, market access and patient services solutions for Syneos Health. “Successful interventions that are able to keep patients on therapy help to close the valve on related revenue losses.”

Potholes and speed bumps impede the patient’s journey

A.J. Triano, EVP, customer experience practice, for Syneos Health Communications, notes that it's a "different mountain to climb" when patients don't necessarily feel the impact of their disease on a day-to-day basis. In such cases, it's still vital for these individuals to remain adherent to potentially costly and complex drug regimens in efforts to prevent the occurrence of some abstract problem in the future.

“Unlike cancer patients who typically acknowledge the crucial role of medication in their survival, patients managing long-term chronic conditions with fewer symptoms may feel healthy and not perceive the imminent threat from their disease,” adds Andrew Hourani, founder and CEO of EveryDose. “As a result, they may underestimate the importance of their medications, creating an environment that is prone to non-adherence.”

In the US, roughly half of medications are not taken as prescribed by physicians.

 This non-adherence leads to preventable healthcare costs, hospitalizations and deaths that cost more than $500 billion (roughly 16% of the total) in U.S. healthcare costs each year.

The tendency for any given patient to fail when it comes to taking their drug as prescribed cannot be blamed on forgetfulness alone. Rather, a number of factors are at play. These factors include:

Fear of discomfort (such as from self-injection) or potential side effects

Misunderstanding related to the importance of taking medications as prescribed

Complexity of the drug regimen or self-administration routines

Mistrust of healthcare providers (HCPs) or healthcare systems

Developing a holistic view of the patient experience

According to John Doyle, DrPh, group VP at Exponent and adjunct assistant professor at Columbia University’s Mailman School of Public Health, the adherence concept has been around for quite some time—along with the resources that are used to help address it—but these tools could use improvement.

“The entire adherence challenge has been ripe for re-examination, because many of the early tools have been around for a while, achieving only moderate benefits,” he says. “Version 1.0 of adherence support really focused on tactical tools to provide one-way push reminders. There’s still room for that strategy, but we’ve all learned that that’s not enough.”

He notes that Version 2.0 of adherence support is pursuing a much broader set of goals. “Today, the focus is on driving improved adherence through better upstream engagement by educating patients about the disease and about managing the medication," explains Doyle, noting especially the implications of the disease's progression, the role the therapy plays, and helping patients understand the complex drug-administration requirements and potential side effects. “The next generation of adherence support also catalyzes downstream engagement to track and analyze clinical, humanistic, economic, and even social outcomes.”

Importantly, today’s adherence-related tools are increasingly designed to provide two-way communication, rather than just the static “push” reminders of yesteryear. When adherence-support tools are designed to be more interactive, drug companies can collect insights actively, for instance, through regular questionnaires for users, or passively, such as through data collected by wearable health devices or sensors that monitor blood glucose levels or blood pressure on the go, for example. Companies can then use the data-driven insights to tailor the support and outreach and inform ongoing improvement.

“Some patients may respond better to digital or self-service interactions versus more traditional human coaching, and these same patients may have different attitudes toward rewards or peer interventions,” says Scott Snyder, EVERSANA’s chief digital officer. “Understanding the behavioral profile of each patient is just as critical as their clinical profile to combat non-adherence, so using predictive analytics and behavioral segmentation programs can be critical to determine the best intervention strategy for each individual.”

In one recent example, EVERSANA used its proprietary ACTICS artificial intelligence (AI)-driven platform to tailor adherence outreach for an oral anticoagulant from the patient services hub to serve up customized messages to specific patient segments based on real-world data (RWD). The model leveraged RWD related to claims, demographic information, prescription drug costs, and total cost of care.

“Using predictive analytics, we were able to help drive a 50% increase in adherence while reducing the cost of a hub program by nearly 40% over a three-month period for a cohort of nearly 1,000 patients,” says Snyder. “This shows the significant impact of personalizing adherence programs by focusing on the whole patient experience [including their behavioral profile].”

However, too often, efforts to assess the patient experience are overlooked. “Pharma companies should be reaching out directly to patients and physicians, asking, ‘Tell me about how and why the experience of using my product and dealing with this disease are difficult,’” says Triano. “These two questions need to get asked more often when developing adherence-support interventions."

He adds that while it's not always possible to measure the impact, this type of qualitative input can help to establish a growth roadmap for a brand and inform direct, just-in-time messaging to improve patient engagement and adherence to therapy.

“Mining your own call center data will provide a really good jump on this process of gathering up such qualitative insights from users of your product,’” adds Plesnarski.

Meanwhile, to drive patient engagement with support tools or a program, patients often resonate with anecdotes and motivational insights from those in the same position. “After the initial trial, patients want to confirm that their own experience matches the promise of the promotional claims and the experience of other patients," says Triano. "A well-timed testimonial or case study can provide a make-or-break moment to cement adoption and motivate long-term adherence.”

Understanding patient behavior—and then changing it

The hub concept has grown in popularity in recent years, giving patients and providers a centralized, one-stop access to a range of high-touch and high-tech support programs to help patients get on and stay on therapy. “For instance, a patient who is struggling with cost or affordability concerns may require patient services support to help navigate to the best reimbursement options available, while at the same time, a patient that stops experiencing symptoms needs more educational or awareness-focused experiences on the risks of stopping their therapy regimen,” explains Snyder.

However, from a brand perspective, involving multiple vendors or partners to provide different parts of the offering can create tremendous operational burden to ensure they are seamlessly bridging all of the connections, according to Syneos Health's Plesnarski. “Brands should strive toward simplification and streamlining, reducing the number of separate partners and vendors involved," he says. "The more partnerships and bolt-on program elements you have, the harder it can be to pivot in the face of new information or changing market dynamics.”

Experts believe today's digital adherence-support tools have markedly improved, with more customization and better integration with electronic health record (EHR) systems and care plans. In turn, as Doyle points out, this enables stronger networking of the various components in the overall tool. For example, integration of wearable devices with a patient’s EHR can result in a more comprehensive conversation with the individual's doctors and earlier interventions where needed.

“Finding out why a particular patient is non-adherent and then working with the patient to overcome that specific barrier is critical,” adds Hourani. However, he notes that historically, such an approach was a costly and labor-intensive undertaking, as it required live communication between a care provider and the patient.

EveryDose, according to Hourani, uses a conversational AI via a feature its calls Maxwell to help uncover patients’ unique barriers to adherence and deliver a more personalized intervention. The AI engages with patients to reveal potential issues that can then be relayed to care teams through a clinical portal.

Consider this example: If Maxwell notices a discrepancy between a patient’s adherence to a statin and its adherence to other prescribed medications, it will automatically initiate a chat with the patient via conversational AI. “In just a few back-and-forth exchanges, we are able to determine that the patient is having severe leg cramps when taking their statin and has responded by skipping doses,” says Hourani. “We notify the prescribing physician, prompting them to work with the patient to find alternative therapies.”

Another important finding, according to Hourani, is that engagement with the EveryDose app has been the highest with older patients, thanks to what he says are simpler features and an easier-to-understand layout. “Specifically, Medicare patients have been engaging at a rate three times higher than our median user,” notes Hourani. “This disproves the common misconception that Medicare patients don’t engage with technology.”

Traditionally, over-the-counter (OTC) medications have not received as much attention when it comes to organized efforts to keep patients on therapy. However, the OTC smoking-cessation space is unique in that users of nicotine-replacement products are typically not patients under the care of a physician. Rather, they are consumers who have made a conscientious decision to reduce or quit smoking to improve their health outcomes and seek out an OTC drug that can help. Nonetheless, even when a consumer has decided to quit smoking, a mosaic of factors still undermines that individual’s adherence to therapy—yet few programs exist to help consumers using nicotine-replacement products to stay on therapy, and there has not been much innovation in new product development over the past two decades, notes says Ashish Girotra, head of marketing for the consumer care business of Dr. Reddy’s Laboratories.

Dr. Reddy’s, which acquired the Habitrol nicotine-replacement product line from Novartis in 2014, recently conducted a longitudinal study of patients to gain a deeper understanding of when, where, and why people reach for their tobacco, in an effort to more effectively support their objectives when it comes to reducing or eliminating their tobacco use. “The findings were very telling,” says Girotra. “Our study found that cigarettes solve at least 11 different needs that spread over physical, social, emotional, and psychological realms. For some people, it satisfies hand-to-mouth habits or an oral fixation; for others, it’s deeply connected to their feelings of social or psychological well-being.”

According to Girotra, the Habitrol timed-release transdermal patch requires a 12-week commitment, as the nicotine-replacement patches work by successively stepping down to a new set of patches with a lower dosage every four weeks, and despite the simplicity of the design, he notes that adherence challenges are “shockingly pervasive.”

In fact, the company recently partnered with CVS to study how many people moved successfully through all three stages of the Habitrol patch, and the study found that only 7% of users made it all the way through—in part because reliance on the nicotine-replacement patch alone does not address the other 11 needs that smokers battle with when they are attempting to quit.

To close that gap, the Habitrol brand has introduced an eight-week version of the product and it has been well received, says Girotra. The company has also piloted an app called AIMI by Habitrol. Launched in 2022, the tool provides gamification to engage users, educational and motivational content, reminders, and easy product re-ordering (to receive the next set of patches days before the current set runs out). To date, the app has only had a “soft launch,” and Dr. Reddy's is currently working on version three to incorporate some of the findings learned from early user experience; it is also partnering with other digital therapeutic companies to help users overcome the range of social, emotional, and psychological hurdles that make it difficult to quit smoking.

Girotra adds that Dr. Reddy’s will likely become involved in the diabetes and metabolic market in the next half of the decade as generic products start entering the space. Thus, he says: “There’s a big opportunity for us to leverage the experience we’ve gained from our Habitrol adherence efforts in terms of digital tools, patient education, behavioral issues, and more to address the multifaceted social, emotional, and behavioral factors that are at work in that sector, too.”

1. Phillion, M. The Impact of Cost on Medication Adherence. Patient Safety & Quality Healthcare. May 9, 2022. 

https://www.psqh.com/analysis/the-impact-of-cost-on-medication-adherence/#:~:text=This%20non%2Dadherence%20leads%20to,U.S.%20healthcare%20spend%20every%20year

2. Watanabe, J.H.; McInnis, T.; Hirsch, J.D. Cost of Prescription Drug-Related Morbidity and Mortality. 

https://pubmed.ncbi.nlm.nih.gov/29577766/

Creating a user-centric approach that harnesses both digital tools and behavioral science is key to getting—and maintaining—patients on therapy, and improving clinical, financial, and, ultimately, brand-success outcomes for pharma manufacturers.

Doubling Down  on Drug Adherence

The cardiovascular disease (CVD) pharmaceutical treatment market has traditionally involved a mosaic of modalities and medications, as, after all, heart disease is not just a single affliction but rather a group of disorders.

 This segment includes conditions such as:

Atrial fibrillation or arrhythmia (abnormal heart rhythm)

Chronic ischemic heart disease (reduced blood supply to the heart)

Stroke (interruption of blood flow to the brain)

Heart failure (inability to pump blood adequately)

Angina (caused by insufficient oxygen flow to the heart)

Myocardial infarction (interrupted blood flow damaging heart muscle tissue)

As such, no single issue defines the complex landscape in CVD. Despite the pace of generic entrants and the impact of poor adherence to therapy on clinical outcomes, other forms of innovation are continuing to generate enthusiasm in the space. For instance, drug developers continue to bring breakthrough therapies to market, while the use of artificial intelligence (AI) and machine learning (ML) techniques are helping physicians to make more accurate and timely diagnoses and are helping to inform efforts to repurpose existing CVD drugs for new indications or pair them in opportunistic drug-drug combinations.

According to IQVIA Institute’s recent publication “The Global Use of Medicines 2023: Outlook to 2027,” CVD medications are projected to rank No. 4 in the list of top 20 therapy categories in 2027 (in terms of global spending), behind oncology, immunology, and diabetes. CVD therapies generally fall into several buckets, including:

Statins to lower low-density lipoprotein (LDL) cholesterol

Beta blockers to lower blood pressure, heart attack, and heart failure

Angiotensin-converting enzyme (ACE) inhibitors to reduce heart failure and blood pressure

Calcium channel blockers to reduce the workload of the heart

Angiotensin II receptor blockers to reduce blood pressure

It could be argued that the stakes are high in CVD. Heart disease remains the leading cause of death in the US, according to January data published by the American Heart Association (AHA).

 Meanwhile, according to the report, between 2018 and 2019, direct and indirect costs of total CVD were $407.3 billion.

While the CVD space was once the place where blockbuster drugs routinely planted a flag—with heart medications such as Crestor, Lipitor, Vytorin, and others practically being household names—the proportion of generic medications in CVD is expected to reach 44% of the market by 2028 (up from 32% in 2018), according to IQVIA. And the number of generic launches continues to outpace new product launches in CVD. IQVIA projects 137 loss of protection and generic entry events in CVD between 2019 and 2028—compared to 25 new CVD therapies approved in the same period.

Despite asymmetry in the proportion of generics to branded therapies entering the arena, “the past decade has also been associated with a boom in new agents and refined drug capabilities in cardiovascular and cardiometabolic care,” notes Alla Shargorodskaya, MD, senior director, therapeutic strategy and innovation, at Syneos Health.

Recent groundbreaking therapies that target the molecular basis of the disease are helping to transform the CVD arena, in terms of their ability to reduce morbidity and mortality, improve quality of life, and help some patients to avoid costly, invasive surgical interventions, trips to the emergency room, and hospital re-admissions. One recent example is the 2022 approval of Camzyos (mavacamten) from Bristol Myers Squibb (BMS). It is the first-in-class—and to date the only—therapy approved in the new cardiac myosin inhibitor class for the treatment of obstructive hypertrophic cardiomyopathy (oHCM), the most common hereditary CVD.

Prior to the approval of Camzyos, patients with severe oHCM typically required an open-heart surgery or ablation procedures, notes BMS. The company acquired Camzyos as part of its $13.1-billion buyout of MyoKardia in 2020. A second investigational therapy in the cardiac myosin inhibitor class—aficamten from Cytokinetics—is currently undergoing clinical trials.

“Sodium-glucose co-transporter inhibitor therapies (SLGT2) for the treatment of heart failure have also given many a cause for celebration,” says Pierantonio Russo, MD, chief medical officer, EVERSANA. They are designed primarily for patients with heart failure (with and without diabetes) who have a history of frequent hospitalization. “This class of drugs are proving to reduce hospitalization in these patients, and they’re offering a lot of hope for these patients,” he adds.

Meanwhile, renewed attention to resistant systemic hypertension is also encouraging. “Results from the recent BrighHTN trial (involving investigational drug baxdrostat) provide evidence that selective inhibition of aldosterone synthase leads to a substantial dose-dependent in blood pressure among patients with resistant hypertension,” explains Russo. In January, AstraZeneca entered into a definitive agreement to acquire CinCor Pharma, which produces baxdrostat.

In a parallel lane, it has been more than 30 years since a new antihypertensive agent with a new mechanism of action has been approved. In December 2022, Swiss pharma company Idorsia submitted a new drug application to FDA for its experimental therapy aprocitentan, a dual entothelin receptor antagonist (ERA), for patients with difficult-to-control hypertension.

“Aprocitentan is a potential novel and effective agent targeting a currently unopposed pathophysiologic pathway, and market authorization is expected as uncontrolled blood pressure is one of the most common risk factors for major cardiovascular events,” notes Shargorodskaya.

She adds that the "highly efficient" PSCK9 inhibitors are also making a significant impact in treating hyperlipidemia, without the feared muscular side effects or potential concerns of diabetes and dementia. There are currently three drugs in this class: Praluent (alirocumab) from Regeneron, Repatha (evolocumab) from Amgen, and Leqvio (inclisiran) from Novartis.

Physicians often have trouble identifying and diagnosing patients with a particular form of heart disease called wild-type transthyretin amyloid cardiomyopathy (wtATTR-CM), an under-recognized, life-threatening condition that results in progressive heart failure. Pfizer has developed a novel prediction algorithm

 (working in conjunction with researchers at the Northwestern University Feinberg School of Medicine and the Brigham and Women’s Hospital) that can help to stratify patients and identify those who are at risk of wtATTR-CM.

Pfizer’s novel model leverages AI and ML—and has been validated using medical claims and electronic health records (EHRs) data—to identify symptoms of wtATTR-CM earlier. Unlike conventional statistical approaches, ML can more efficiently evaluate complex interactions across multiple 

input predictors, and in this particular model, was used to automate the suspicion of wtATTR-CM based on associated clinical conditions, according to the company.

 reports that the model performed with 87% accuracy in predicting patients with heart failure associated with wtATTR-CM, enabling healthcare providers to identify and appropriately diagnose patients with the condition wtATTR-CM earlier and with greater accuracy. This is particularly important to allow for initiation of therapy earlier in the trajectory of this life-threatening condition.

To make the novel prediction model more easily accessible to physicians, Pfizer also developed and launched EstimATTR, a web-based educational tool that US healthcare providers can use to help estimate the probability that a hypothetical patient with heart failure may have wtATTR-CM.

AI and ML are also being explored to support faster, more accurate diagnosis in CVD by identifying clinically relevant patterns from large, complex datasets (not just EHR and claims data, but data from wearable sensors and monitors, CT scans, and more) and creating actionable outcomes.

“In the near future, it will not just be some companies—it will be most companies­—that develop analytic algorithms to support their drugs,” says John Doyle, DrPH, group vice president at Exponent and adjunct assistant professor at Columbia University’s Mailman School of Public Health.“I expect that we’re going to see a lot more of that—developing predictive models based on AI and ML that are approved like a companion diagnostic—to support CVD drug development and market access.”

Putting real-world data and evidence to work

Separately, a recent EVERSANA study used real-world data (RWD) and advanced analytical techniques helped to identify patients with heart failure and fluid overload suitable for subcutaneous furosemide infusion in the outpatient settings rather than inpatient settings.

“This data allows us to more precisely identify the right patient population, something that was challenging in the past,” says Russo. “What makes this so meaningful is that subcutaneous furosemide infusion may substantially improve the care of patients with heart failure by keeping them at home rather than treating them with frequent visits in an emergency room or hospital.”

Using advanced techniques and RWD sources, “we can now prove in real-time just how effective the results are for patients,” continues Russo. “In this new climate of value-based care and outcome-based reimbursement, pharmaceutical companies must continuously demonstrate the impact of their CVD assets on reducing the total cost of care per patient, and reduce the need for more costly interventions.”

“Real-world studies help maintain access and demonstrate continued value of the product by providing evidence on comparative clinical effectiveness and costs, long-term outcomes data, patient safety data, and patient adherence and utilization data,” adds Shargorodskaya. “Continual observation of disease epidemiology, treatment patterns, and outcomes in the real world can also help to accelerate evidence generation that is needed to add indications and support label expansion for specific cardiovascular products.”

In CVD and other disease settings, real-world evidence (RWE) insights are increasingly being used to effectively articulate disease burden and demonstrate the long-term value of a therapeutic intervention and help justify premium pricing and coverage decisions.

“To do this, drug developers must strive to connect medication adherence with tangible outcomes, such as reduced hospitalizations or cardiac events, and develop comparative clinical effectiveness and cost data,” says Doyle.

The BMS/Pfizer alliance that brought the direct-acting anti-coagulant Eliquis (apixaban)—a therapy to reduce blood clots and the risk of stroke—to market was considered a pioneer in the use of RWE to not only expand the scientific understanding of a drug after market launch, but to use the clinical findings to demonstrate the full value of the product in every day clinical practice. By linking long-term adherence to the ability to offset other, more costly interventions, the companies were able to provide an economic case for switching from warfarin (a leading alternative therapeutic option) to Eliquis, and then use the findings to inform pricing, reimbursement, and market access negotiations.

The work of the BMS/Pfizer alliance “remains the gold standard for how a CVD therapy can differentiate itself among regulators and payers based on RWE generated post-launch, and still serves as an exemplar for other drugmakers in other therapeutic classes,” says Doyle. “It was novel at the time, and many companies are still trying to emulate what this partnership has done: they proved with long-term RWE that Eliquis could lower the risk of stroke, with fewer incidents of bleeding compared to warfarin. That’s been a massive differentiator that helped to extend the lifecycle of the product.”

As part of the BMS/Pfizer alliance, the companies also partnered with FitBit to develop an algorithm to be used with their sensor that could identify patients who are at risk of atrial fibrillation (Afib). “In this context, sensor-derived insights can help to stratify patients based on risk and identify potential candidates who have elevated risk of Afib earlier (particularly those in disadvantaged communities),” notes Doyle. “While such findings are still subject to medical follow-up to verify and validate the finding, the opportunity to accelerate diagnosis is beneficial to patients, providers, payers, and manufacturers.”

Efforts to pair various CVD medications with mobile sensors and wearable devices provide novel opportunities to generate actionable data that can help to improve diagnosis, support adherence objectives, and generate RWE to strengthen the brand’s value proposition and likelihood to grow. “Progressive pharma companies are leaning into developing and deploying these digital health technology tools to differentiate their core portfolio, as well as add new products to the CVD ecosystem,” says Doyle.

Drug repurposing—whereby approved therapies are able to secure label expansions into different therapeutic indications—offers another route for makers of CVD medications to prolong the lifecycle of their branded therapies. Historically, such opportunities have been driven by knowledge of underlying disease mechanisms or by serendipitous observation of clinical benefits in other patient populations—rather than by systematic efforts to prospectively match an existing drug to a new indication.

However, there is ongoing innovation in computational analysis and predictive modeling that helps stakeholders to systematically screen drugs for disease-modifying activities, ones that are in search of repurposing opportunities that could address unmet medical need and create new opportunities for aging drug franchises. The newer approach relies on extensive phenotyping of a drug’s effect at the level of patient care—gleaned from EHR data—and using modeling systems to identify compounds that ameliorate disease phenotypes.

A recent example of drug repurposing is helping patients with another form of CVD, heart failure with preserved ejection fraction (HFpEF). “HFpEF has recently been recognized as the single greatest unmet need in cardiovascular medicine,” says Shargorodskaya of Syneos. “People with HFpEF also have a higher prevalence of hypertension, diabetes mellitus, chronic kidney disease, atrial fibrillation, and non-cardiovascular comorbidities (such as obesity and chronic obstructive pulmonary disease), compared with those with heart failure with reduced ejection fraction (HFrEF).”

The SGLT-2 inhibitor Jardiance (empagliflozin) from Boehringer Ingelheim—originally approved to lower blood sugar levels in patients managing type 2 diabetes—earned an expanded FDA approval in 2022, for the treatment of patients with heart failure independent of left ventricular ejection fraction (LVEF), and for the treatment of patients with HFpEF.

Shargorodskaya points out that empagliflozin was initially a drug for the treatment of patients with diabetes, but has demonstrated notable and nearly unprecedented benefit for cardiovascular risk reduction in HFpEF patients.

“In cardiology, one of the most recent celebrated examples of drug repurposing is the use of new antidiabetic drugs, specifically SLGT2 inhibitors, as they have been shown to reduce hospitalization in patients with heart failure even if they don’t have diabetes,” adds Russo of EVERSANA. The company is collaborating with Every Cure to carry out drug repurposing studies.

While generics and adherence challenges are keeping drugmakers up at night, breakthroughs in the pipeline coupled with innovative uses of artificial intelligence and machine learning techniques are bringing optimism to the cardiovascular disease space.