Built-in proof: How affected person information derived from digital healthcare supply can drive innovation


For a long time, doctors’ offices, large or small, operated under a seeming contradiction; even as modern medicine pushed the frontiers of innovation, providing new treatments and sometimes cures for devastating diseases, so much of our healthcare system remained stuck in the pre-digital past. The most obvious example is one that many of us have experienced – the floor to ceiling “walls” of file cabinets full of paper patient records. Fortunately, this innovation gap is closing (and these file cabinets are disappearing!) faster than ever.

In fact, today, some of the most exciting innovations in healthcare extend beyond new therapeutic approaches and breakthrough medical devices. The digitization of all aspects of healthcare delivery is dramatically changing how healthcare is provided across the U.S. and around the world. From electronic medical records (EMRs) that document a patient’s healthcare status during a visit to their doctor, to virtual/telehealth visits and consumer healthcare apps that monitor all aspects of an individual’s health status – digital healthcare delivery is transforming healthcare.

A consequential output of healthcare digitization is the generation of electronic patient data – and enormous amounts of it — of which clinical researchers have taken notice. For decades, clinical research relied almost exclusively on capturing patient data de-novo, either on paper or electronic forms in clinical trials for often narrowly selected patient populations and highly prescriptive treatment protocols. These trials have undeniably produced high-quality evidence related to the safety and efficacy of the investigational drugs studied; however, the generalizability of clinical trial results to “real world” care settings are often not well understood. Additionally, the lack of patient diversity in trials remains a significant problem — primarily because clinical trials are typically executed within large academic medical centers that are inaccessible to uninsured, or rural populations.

The patient data generated by digital healthcare delivery as part of routine care has begun to bridge these gaps between trials and everyday care. As digital healthcare data has accumulated, researchers have been using this rich source of real-world patient data, in addition to clinical trial data, to better understand disease and treatment outcomes. In fact, as Covid-19 spread across the globe, analyses of patient data from electronic medical records were used to substantially increase our understanding of Covid in near-real time.

While analysis of electronic medical records data is not new, the continued digitization of health information — and thus the availability of complementary data from other aspects of healthcare delivery — is heralding a new era. Referred to as “integrated evidence,” this concept or discipline provides a significant opportunity to accelerate our ability to study treatment and associated patient outcomes across diseases. Integrated evidence is the result of carefully generating, combining, and analyzing healthcare data from multiple data sources, to produce evidence not otherwise possible from any of the component data sources alone.

In recent years, data-linking technologies have enabled researchers to match, and subsequently enrich, patient records across diverse data sources.  While linking is a basic operation or “tool” by which data is physically integrated, it is just  the first step in the discipline of integrated evidence.

An example of integrated evidence in practice involves a composite mortality variable used in several recent cancer studies.  For context — data from electronic medical records is often insufficient for cancer studies because survival is an important outcome in cancer, and roughly 35% of actual deaths (represented as death date) are not captured in EMR data.  To address this deficiency, integrated evidence — in the form of a brand new mortality variable — was developed.

To do this, researchers first linked the EMR data to data from funeral homes and obituary websites. This provided a dataset that included the patient mortality information to meaningfully assess survival. And then the “heavy lifting” so to speak began – before this data could be used reliably for research, disagreements between the different data sources needed to be resolved. To address these disagreements, a new method had to be developed to sequentially incorporate each potential source, and then a number of different sources and sequences had to be tested until we hit a performance target relative to a trusted benchmark.

The results have been astounding – we have seen this composite mortality variable perform nearly as well as the National Death Index, which is the gold standard. And we can use this composite variable across many studies where overall survival is a critical endpoint. By characterizing the idiosyncrasies of each datasource and then developing a method to carefully combine them, we are left with a brand new thing – integrated evidence in the form of a composite mortality variable, that wasn’t possible using any of the individual data sources alone.

The composite mortality variable is a clean, illustrative example of integrated evidence at work.  As healthcare becomes more digitized, and more patient data becomes available, we are further expanding the playing field of integrated evidence to include more kinds of information, from genomic datasets, to radiographic images, to data from continuous patient monitoring devices (e.g. HbA1C, heart rate, sleep disturbances, etc.), to patient socio-economic status, and – going back to our original example – clinical trial data. Integrating clinical trial data with data captured as part of everyday patient care may in part help to address the clinical trial generalizability gap previously mentioned, and also provide rich evidence that’s not possible with either approach alone.

There is a growing volume and variety of data being generated as an artifact of this era of digital healthcare delivery, which has the potential to transform diagnosis and treatment of disease. And the more inclusive real-world populations where these data originate, means the evidence has the potential to provide more diverse and generalizable science.

With that said, this is not just a technology challenge. Successfully generating and harnessing the emerging discipline of integrated evidence will require uniting the best practices and advanced methods of various fields and leaders including: clinical science and operations, biostatistics, data science, epidemiology, technology, regulatory sciences and patient privacy experts, to name a few. If we are able to foster this cross-disciplinary collaboration and fully realize the potential of integrated evidence, we will be able to truly learn from the healthcare experience of every patient.

Photo: tonefotografia, Getty Images



Source link

Leave a Reply

Your email address will not be published.