By Irene Yeh 

May 15, 2025 | Patient recruitment remains a “never-ending challenge,” as described by Laszlo Vasko, senior director of R&D Therapeutic Enabling Innovation Technologies at Johnson & Johnson. Despite efforts to leverage real-world data to alleviate the difficulties, there has not been a surefire solution. During April’s Bio-IT World Conference & Expo, panelists gathered to discuss the roots of the problem and share ideas that could potentially solve this issue.  

Moderated by Vasko, the panel consisted of Steven Labkoff, vice president of Clinical Operations, Data and Analytics at Bristol Myers Squibb; Yuri Quintana, chief of the Division of Clinical Informatics at Beth Israel Deaconess Medical Center; and Christian Reich, principal investigator of the Observational Health Data Sciences and Informatics Center at Northeastern University. 

The Roadblocks of a Chronic Challenge 

Being able to identify patients within an electronic health record (EHR) system would be a great way to help with patient recruitment, but in practice, this has proven to be a challenge. According to Vasko, this has to do with the different disparate EHRs.  

The challenge around patient identification in EHRs is complicated. “It could be easier, but there is a number of issues that are extrinsic factors to the actual problem,” said Labkoff. Some of these factors include US government policy. Labkoff also cites the lack of a unified unique patient identifier that follows the patient from birth to death. 

Another factor is privacy concerns. HIPAA prevents patient’s medical records and numbers from being shared in order to protect their privacy and personal information. However, Labkoff suggested a different perspective on privacy laws. “It’s hard to say if they’re overblown or not when people are willing to give out their personal information on Facebook,” he said. In other words, are these privacy concerns relevant if patients are openly sharing private information on public platforms already? 

The real question is, according to Labkoff, what are the policy issues that the industry needs to think about that could actually help with patient recruitment? 

A Comprehensive Data Source and Data Harmonization 

Quintana discussed how his organization, Beth Israel Deaconess Medical Center, is a network of 13 hospitals and 17 affiliates, of which most have already migrated to a central Epic medical record, but “not all of the data of the last 50 years is being integrated.” Furthermore, even for any single patient at any one time, their medical record does not contain all of the data on that individual patient. 

To conduct good patient-trial matching, a comprehensive data source is required, and medical records do not meet the requirements. Beth Israel Deaconess Medical Center is looking at how to build a longitudinal record so that patients could have all of their historical data, including current lab values and even social determinants of health. This dataset could then be used to match patients with the best trials. The idea is to also allow both the provider and patient access to the longitudinal record to see what trials are available and give them the opportunity to proactively select them. 

“Now, the challenge is that we need data harmonization as we import all of this data,” said Quintana. Pipelines for integration need to be strengthened, as well as better data models that are specific to the fields that are needed for clinical trial data matching. 

Getting Patients to Trials 

There are usually two ways patients enroll in trials, according to Reich. The first is public ads that have a phone number on them that the patient would call to get recruited. The second way is a provider brings them in. The provider is the investigator of the site and knows the inclusion criteria of the trial. It is up to the provider to identify which patients matches and then the recruitment process ensues. 

Reich, however, equates this with Einstein’s definition of insanity: doing the same thing over and over and expecting a different outcome. 

“Something needs to change,” said Reich. He emphasized the ability to identify the right patients and the need for full visibility of the patient, their clinical history, the makeup of the clinical history, and so on. But there lies a third problem. “There are lots of patients who are not the patients of investigators.” 

As such, these patients do not have access to trials as easily as others. Reich proposed a solution where patients can be found by data, but in order to do so, data must be searchable. In order to be searchable, data must be standardized. Providers can go on platforms, such as Epic, to find patients, but platforms differ across institutions and organizations.    

The Problems of Patient Privacy 

Patient privacy, obviously, must be honored, but HIPAA and other privacy protection laws can limit data access, which may hinder drug development and treatment research progress. Lapkoff proposed a way to make HIPAA “work to your advantage.” 

He recalled an incident during his time at the Multiple Myeloma Research Foundation, where he and his team set up a direct patient clinical registry for myeloma. They managed to work with HIPAA’s policies by having patients self-identify through the inclusion criteria and come into the trial or into the registry. Because they had the patients’ consent, they were able to ask for their medical records. The law does allow this. 

Labkoff elaborated on an idea he had. He proposed having a white hat organization, such as Harvard Medical School, be a central clearinghouse for patients to inform them of their diseases, cancers, and other conditions and give consent for trials to access their medical records wherever it occurs in the healthcare ecosystem. With the patient’s consent, the white hat organization then offers to help the life sciences community and organizations identify patients.  

“If we take the Washington policy and put it on its ear, if the patients say they want to do it…you then have the permission to break through HIPAA,” Labkoff continued. If life science firms subscribe to the service and if this concept is brought to a national level, then there needs to be a way to fund the service. But Labkoff stated that this method could break down the barriers and align incentives. “Patients want to find trials,” he added. 

Bringing This Forward 

Vasko discussed the possibility of using identifiable patient data that anyone could use. At Johnson & Johnson, they are trying to connect healthcare institutions directly and set them up with a recruitment algorithm. While this comes with its own set of challenges, this method at least provides a rudimentary screening algorithm that could identify patients and send an alert. However, “there’s a barrier of entry.”  

While some small nonprofit organizations, usually disease-oriented, have tried to help patients get to certain trials, there are still some missing “building blocks,” according to Quintana. For example, if a universal patient identifier is built and ready to be used, then there is a possible issue of running into multiple profiles of a patient with the same name. Additionally, as previously mentioned, patients often have more than one medical record from different healthcare providers. Gathering the needed amount of data from several different records would be a long and tedious process. There are also different factors to consider, such as patients moving to different places of work and residence. There needs to be a way to directly link the patient with their data. 

Secondly, the data need to be transformed into a common data model, as importing data can result in errors and lead to mismatched trials. “We need to have better data models that are more specific to diseases and specific to clinical trials and transparently talk about those import problems so that we have cleaner data sets,” said Quintana.  

But there are different kinds of laws that make it difficult for life science organizations and healthcare providers to implement this strategy. To navigate these governance issues, a nonprofit entity that is insulated and checks all compliance requirements would probably be the best solution. Instead of having several small initiatives, a handful of larger, national-level initiatives would probably allow higher sustainability and scale, as well as open to more collaboration. 

(Deciding and) Setting the Standards 

“The standards are not compatible between each other,” Reich explained. Different incentives, operations, and purposes can result in “protectionist” attitudes. 

For example, one organization may create their own set of standards for their own people and share it with only one coordinating center to promote these resources. But these standards are incompatible with another company. Reich also cites “a certain amount of jealousy” circulating between companies. What needs to happen is that the healthcare providers and patients must demand the use case. By doing so, the demands will consolidate pressure and force organizations to solve the problem as opposed to arguing over standards and methods. 

Taking Action Now 

“Each of us represent roughly 30 years of experience in this field,” Labkoff commented. “These problems dogged us from the first day we started, and most of us are going to retire soon. And they’re still there. They haven’t been solved. Some of them have gotten more sophisticated.” 

Quintana also commented on AI’s role in data standardization and patient recruitment. Though he acknowledged that AI and generative AI has a lot of potential, it is a “black box” that does not actually explain its matching process. Additionally, uploading medical records and asking to find clinical trials is not a long-term solution.  

To actually solve the issue of getting patients to trials, importing data, using the data to match patients, and standardizing it all, there needs to be clear, transparent matching algorithms that can be validated, be easy to use, and prevent patients from being sent to the wrong places. And, hopefully in time, progress can be made.