Benefits of delegating the patient screening process for oncology trials

Emily Akin

Lindsay Bramwell

Barry Leybovich

In recent years, breakthroughs in our understanding of cancer biology have led to many important innovations, and research centers and biopharma companies around the world are now hard at work developing promising new drugs for a growing number of indications.

But the clinical trials necessary to bring those innovative treatments to market are getting increasingly complex. They collect more data, require a greater number of sites, and experience more protocol deviations and amendments than trials in other therapeutic areas.¹ In modern cancer studies, recruitment is often based on narrow biomarker and genomic inclusion and exclusion (I/E) criteria, and it can be a big challenge for busy sites to identify (and enroll) enough eligible patients to produce statistically significant insights within the time constraints of the study. Unsurprisingly, oncology trials take 30% to 40% more time to complete enrollment than other clinical trials.²

The growing cost and burden associated with oncology trials run the risk of derailing promising new developments. Thankfully, recent advances can help alleviate that burden and keep scientific progress on track.

The EHR paves the way

When the HITECH Act passed in the U.S. in 2009, it incentivized physicians and hospitals to adopt EHR systems and procedures, and for the past ten years, the vast majority of them have indeed embraced a certified EHR in their daily practice.³

The impact of EHR adoption on patient safety (e.g., flagging drug-drug interactions or prescription errors), workflow efficiency, medical coding and billing accuracy has been well documented.⁴ But outside of the physician’s office, one of the most important promises of EHRs was that they could be used as a comprehensive data source for researchers to analyze health data across large patient populations, and as a result produce crucial insights relevant to public health, biosurveillance and clinical research.⁵ We saw how useful EHRs were during the COVID-19 pandemic to understand the disease dynamics and the efficacy of new interventions. However, when it comes to the conduct of clinical trials, much of that potential has yet to be fully realized.

Vexing challenges at the point of care

While the EHR is an invaluable tool for clinicians to keep track of a patient’s history and streamline their interactions with medical partners, it’s not an analytical platform. At busy community sites, oncologists could see over 25 patients a day, monitoring progression, interpreting results, ordering new tests and developing new treatment plans. They don’t have much time to keep up with all available trials.

Site-based clinical research teams, for their part, are facing historic turnover rates, especially among staff with the most tenure.⁶ With limited resources on hand, it can be a major challenge to screen patients for new potential trials on top of managing regulatory obligations for existing trials. The average oncology trial today has 38 I/E criteria⁷ and it’s not unusual for sites to turn down trials because they don’t have the bandwidth to screen patients. The problem is particularly acute for rare diseases where they might do significant work to open and maintain the trial and only find a very small number of eligible patients.⁸

Technology can help the screening process by making it easier for the research staff at the site to analyze structured and unstructured EHR data, but it can be enormously taxing if it’s not well integrated with the site’s existing workflow or interfaces poorly with its external partners. In many cases today, the only way to evaluate a patient for a targeted therapy trial is to match I/E criteria by manually digging through PDF files like pathology reports, internal genomic reports, external vendor reports, or worse, by scanning those reports into the EMR. Technology for technology sake isn’t the answer.

Why delegating is the answer

We think that the solution is a hybrid model where sites entrust the bulk of the screening process to specialized health data companies like Flatiron Health. Those companies can (I) work with sponsors and CROs to help them define and optimize their study eligibility criteria; (II) establish data pipelines with sites to analyze their patient records in near real-time; and (III) run the matching process between patients and active clinical trials.

At Flatiron, our database currently covers 3.5 million patients and 22 tumor types across 80+ experienced research oncology sites — 80% of them community practices. And our tools are integrated into every site’s existing workflow, so we can make sure that eligible patients get identified at relevant treatment decision points, and that their treating physician receives a timely notification that they’re a match within the EMR platform they’re already familiar with.

Delegating large parts of the screening process comes with substantial benefits for participating sites:

Having access to a massive EHR-based dataset is also a big opportunity for sponsors to test ahead of time whether their I/E criteria are adequate or prohibitively restrictive,¹² and optimize their protocols accordingly. They can also conduct upfront analysis to select the right sites for their studies, set patient enrollment expectations and make sure they don’t introduce bias based on race, ethnicity, socio-economic status, insurance status or any number of other key determinants.

Data, technology and human expertise

We’ve found that the best screening solutions combine data, technology and human expertise.

Despite dramatic developments in natural language processing and machine learning over the past few years,¹³ there are still a lot of nuances in the patient recruitment process. For instance, a trial protocol might call for a lab value within a specific number of days after the patient’s last visit, or based on the last time they received a particular line of therapy. Those data points may not be readily available or easy to extract from the EHR. A human abstractor provides an extra lens to boost the matching algorithm’s accuracy — and give sites and sponsors the confidence they need to entrust the screening process to an outside party.

Where does the clinical research community currently stand on the use of RWD to improve patient enrollment?

We still have a long way to go. We commissioned a survey in the summer of 2023 to understand what processes biopharma companies and CROs tend to use to select suitable sites, identify eligible patients and design their study protocols, and we found that many still rely primarily on internal company reporting, past studies and the reputation of their principal investigators to achieve their objectives. If it worked in the past, it should work again, or so the thinking goes. Only 16% use RWD consistently to identify sites with relevant patient populations, and 22% use it systematically to optimize their protocols. And many still invest substantial funds to recruit patients on social media or in print magazines, which actually adds to the site workload, perpetuates the lack of diversity in clinical trials, and often leads to more post-marketing work for all involved.

But with the right outside partner and an enrollment strategy that includes robust EHR-based RWD, seamless point-of-care technology and expert human supervision, sites and sponsors can alleviate much of the complexity associated with patient recruitment and keep operational issues from getting in the way of scientific progress.

¹ Tufts Center for the Study of Drug Development, Impact Report Vol 23, Number 3 (May/June 2021)

² Ibid.

³ National Trends in Hospital and Physician Adoption of Electronic Health Records, Office of the National Coordinator for Health Information Technology (2021)

⁴ The digitization of patient care: a review of the effects of electronic health records on health care quality and utilization, Annual Review of Public Health (2019)

⁵ Title XIII of the American Recovery and Reinvestment Act of 2009 (Pub Law 111-5)

⁶ Now is the time to fix the clinical research workforce crisis, Society for Clinical Trials (June 2023)

⁷ Source: internal Flatiron Health analysis

⁸ The great resignation: its impact on clinical research and where we go from here, WCG (June 2022)

⁹ The great resignation: its impact on clinical research and where we go from here, WCG (June 2022)

¹⁰ Scaling real-world data curation through machine learning and large language models, Flatiron Health (Dec 2023)

¹¹ Guardant Health, Flatiron Health announce integration of Guardant genomic profiling tests onto OncoEMR platform (Oct 2023)

¹² Evaluating eligibility criteria of oncology trials using real-world data and AI, Nature (Apr 2021)

¹³ A natural language processing algorithm to improve completeness of ECOG performance status in real-world data, Applied Sciences (May 2023)