Making the Learning Health System a Sustainable Reality

When clinicians have no definitive guidelines or empirical evidence on which to base their decisions, they still must come up with a care plan that meets their patients’ needs. Combining data from the available published studies with intelligence gleaned from electronic patient records can often address the dilemma.

By Paul Cerrato, MA, senior research analyst and communications specialist, Mayo Clinic Platform and John Halamka, M.D., president, Mayo Clinic Platform

In our soon-to-be-released new book, Redefining the Boundaries of Medicine, we describe the three “ages” of medicine. During Medicine Past, physicians had limited instrumentation and technology to assess cardiac function, metabolism, and a host of other physiological metrics.  The physician/patient relationship was also less than optimal, with most physicians maintaining a paternalistic “Doctor knows best” mindset.  Similarly, diagnostic and treatment protocols were largely based on expert opinion and clinical experience. This combined expertise and experience were often collected by consensus panels, summarized in official sounding statements, and canonized in medical textbooks.

Medicine Present, evidence-based medicine 1.0, relies heavily on randomized clinical trials (RCTs), probability, and statistical analysis. It has become the primary standard currently used to judge the efficacy of diagnostic and therapeutic regimens. The rationale for placing the most trust in RCTs, systematic reviews, and meta-analyses stems from two critical issues that reduce the reliability of observational studies: unrealistic expectations and confounding.

Medicine Future, evidence-based medicine 2.0, recognizes the weaknesses of RCTs, including the tendency to fall victim to Type II statistical error or beta error, which occurs when a study concludes that there is no real difference between treatment and control groups when, in fact, a genuine treatment effect exists. A Type II error can occur when too few subjects are included in the study. With this and similar limitations in mind, clinicians would be wise to view RCTs as one component of a set of tools needed to guide diagnosis and treatment.

Observational studies, including both retrospective analyses and prospective cohort studies, can play an important role in the decision-making process. Thomas Frieden, MD, MPH, a former director of the Centers for Disease Control and Prevention (CDC), has pointed out the real-world advantages of cohort studies, which have been used to assess the prognosis and treatment of various types of cancer. That, in turn, has led to better treatment protocols.

Similarly, such cohort studies have successfully been used to evaluate survival among pediatric cancer patients and made clinicians aware of the “increased risk of post-treatment cardiac complications, enabling better clinical care.” Frieden summed up the controversy this way, “Elevating RCTs at the expense of other potentially highly valuable sources of data is counterproductive. A better approach is to clarify the health outcome being sought and determine whether existing data are available that can be rigorously and objectively evaluated, independently of or in comparison with data from RCTs, or whether new studies (RCT or otherwise) are needed.”

While the combined benefits of observational studies and RCTs enable clinicians to gain access to many valuable insights, they leave one important resource untapped: the data of past patients that can inform the care of future patients. A growing number of progressive-minded provider organizations are taking advantage of this resource, folding it into an even larger toolbox: the learning health system. 

The U.S. Agency for Healthcare Research and Quality defines a learning health system “as a health system in which internal data and experience are systematically integrated with external evidence, and that knowledge is put into practice. As a result, patients get higher quality, safer, more efficient care, and health care delivery organizations become better places to work.”  

Edward Shortliffe and James Cimino shed more light on this paradigm in their biomedical informatics “bible”:  “The ultimate goal is to create a cycle of information flow, whereby data from local distributed electronic health records and their associated clinical data sets are routinely and effortlessly submitted to registries and research databases. The resulting knowledge can then feed back to practitioners at the point of care, using a variety of computer-supported decision support delivery mechanisms.”  

At Mayo Clinic Platform, we have created Mayo Clinic Platform_Discover, an online system that allows users to query a massive data set of de-identified clinical data from our EHR system. It can be used to supplement the intelligence gained from published studies to help clinicians make more informed decisions. That’s especially important when there are no clear guidelines available from professional organizations or RCTs.

Similarly, there are third-party services to help providers extract intelligence from patient encounters, which in turn can guide patient care. Atropos Health,* for example, employs a three-step process to generate what it calls a prognostogram. First, a physician creates an inquiry within their existing workflow, through an EHR system or an online portal. Atropos’ informatics team then cleans up the data, which typically includes clinical notes, procedures, lab results, demographics, and vital signs. Using its proprietary search engine and statistical analyses, it generates an automated report, which is often based on aggregated data from millions of patient charts. Finally, the prognostogram is delivered to the clinician to help guide the decision-making process.

Stretching the boundaries of medicine continues to challenge thought leaders and clinicians “in the trenches.” Providing the best sources of data, whether from published guidelines or aggregated EHR systems, is in our patients’ best interests.

*Footnote:  Mention of commercial services does not indicate endorsement by Mayo Clinic.