Questions Answered: Clinical EHRs

In this blog, Dr. Shannon Silkensen answers questions about real-world data and electronic health records (EHRs) in cancer research. Dr. Silkensen is a health science administrator in NCI CBIIT’s Clinical & Translational Research Informatics Branch. She also serves as NCI’s Real-World Data Program Leader on the USCDI+ Cancer Team, which creates guides for academic institutions and industry partners to implement and exchange data concepts as defined by the USCDI+ Cancer Program.

Question: What is real-world data?

Answer:

As defined by the U.S. Congress in 2016, real-world data can be either data relating to a patient’s health status or the delivery of healthcare routinely collected from a variety of sources. Simply put, real-world data can be anything from data in clinical notes to data in wearable devices. Congress also defined real-world evidence as clinical evidence regarding the usage or potential benefits or risks of medical products.

Question: How is real-world data applied in cancer research?

Answer:

It can be applied in a variety of ways. Probably the most well-known example concerns the breast cancer medication Ibrance® (palbociclib). Researchers leveraged real-world data from EHRs and post-marketing reports to demonstrate that the drug was both safe and effective in the fight against male breast cancer. This was different from the traditional method for calculating product safety, and it became a testimony to the usefulness of real-world data in clinical research settings.

Another way we can use real-world data in cancer research is as an active comparator. In this case, investigators can compare investigational treatments to the standard of care, found in the EHR, and assess the impact of the investigational treatment. One of the best examples of this is from Franklin et al. in 2021. In brief, the investigators asked if rigorously designed real-world evidence studies using insurance claims data could yield results that match those of completed RCTs (randomized controlled trials) well enough to support the same regulatory decisions. The team’s paper definitively showed that active comparator studies work well with real-world claims data because both treatment and comparator data are observable in routine care.

Question: Why is there a focus on real-world data for biomedical cancer research?

Answer:

I think there’s a focus on real-world data because we’re human and not perfect. There can be a gap between the “perfect patient” data in clinical trials and patients in complex, real-world settings. Clinicians, patients, and researchers all want to know how medications or devices function in the real world, and real-world data can help them find the answers.

Question: What is the role of real-world data and real-world evidence in AI-driven cancer research?

Answer:

We can use real-world data and evidence to model treatment responses, especially for rare diseases when a cohort of patients is hard to find.

Keep in mind too that good AI models require training on large volumes of data. In oncology, the best sources for this training are rich, real-world data sets built on EHRs.

In 2023, the NCI Real-World Data Program offered administrative supplements to establish a privacy-preserving federated learning network for researchers to run AI models on large volumes of multi-omics data. My colleague Dr. Umit Topaloglu can share more about this NCI-funded project as well as the importance of federated learning networks in modern research ecosystems.

Question: What are the key challenges of using real-world data and EHRs in cancer research?

Answer:

Data standardization and interoperability are two major challenges with utilizing real-world data and EHRs in cancer research. Institutions may use units of measurement (for instance, pounds vs. kilograms) that equate to one another but are different expressions, and such differences can make EHRs less coherent. Data quality levels can also vary from EHR to EHR; fields may be incomplete or missing, and this brings into question the accuracy of the record. It makes data exchange challenging. My colleagues and I wrote an article recently about the challenges of data quality across a sample of real-world data sources.

There are also challenges from a policy perspective. In the U.S., states regulate consent (that is, my given permission for my personal data to be shareable). Then there are also federal privacy laws. So, managing data across state lines can be difficult because we want to be respectful of both state and federal privacy regulations. This can be a big administrative hurdle, especially for researchers studying rare diseases and who need to look across the country (or the world) for clinical trial participants.

Question: How is NCI helping to overcome the challenges in real-world data for cancer research?

Answer:

The biggest way NCI CBIIT is overcoming some of these challenges is by partnering with other federal agencies to set up what’s called the “USCDI+ Cancer” data elements list. By defining a minimum set of cancer research-specific data concepts, institutions can exchange information and more easily identify which patients have cancer, whether they may be eligible for a clinical trial, and whether they have experienced an adverse event. Standards can mean better and more efficient communication, which means less time piecing together a patient’s information.

This is important in oncology because the average cancer patient goes to six different clinics, which could be across multiple states and multiple health networks. So, by having the standard set of data that you could exchange between health systems (assuming you can get the state and federal privacy laws to align), the uniformity would allow a patient to walk into a clinic and have their entire medical portfolio outlined and ready for a clinician or care team to assess.

Question: Where can I find NCI-funded, real-world data?

Answer:

I think the best place to find that kind of data would be the Cancer Research Data Commons. It’s such a big source of data that is shareable with appropriate permissions.

Question: How are wearable devices and remote monitoring technologies changing clinical trials?

Answer:

There are a couple of different areas where this has been very useful. One is in what are called “PROs” (patient reported outcomes). Think of a watch or a smartphone app that you’re inputting data into (for example, you’re reporting that you feel nauseous after taking your medicine). Your input is monitored by your care team, and they address your concern by talking to you over the phone or having you come into an office. This constant communication between patients and their care team builds trust and improves patients’ quality of life. The second area is wearables, which are good for tracking quality of life and basic information like step count and average heart rate. Generally, this can help a clinician judge the health of a patient.

Have another question?

If we missed your question about clinical EHRs, real-world cancer data, or USCDI+ Cancer, please email NCI CBIIT. We’ll connect you with a contact. If it is a commonly asked question, we’ll update the blog with an answer.

Author

Shannon Silkensen, Ph.D.
Dr. Silkensen is a health science administrator within NCI CBIIT’s Clinical & Translational Research Informatics Branch.