In dealing with many cases, doctors lack comparative real-time evidence and are forced to make decisions in spite of unknown variables that can dramatically alter outcomes. Such evidence gaps happen every day, particularly for patients with multiple conditions, complex medical histories, and diverse ethnic backgrounds. Breakthroughs in academic research, including a major project at Stanford University School of Medicine, have led to technological innovations that allow clinicians to generate on-demand evidence drawn from research data and anonymized medical records, bridging the evidence gap so doctors can make informed decisions that improve outcomes.
When clinicians are asked if a patient’s case they’re managing has a corresponding care guideline, the answer is usually “no.” As those in medicine can attest, only about 20% of patients are linked to a standard care guideline. Pair that with the fact that just shy of 20% of existing care guidelines are backed by high-quality evidence, and we arrive at a shocking conclusion: Roughly only 4% of the patient care situations a doctor must manage have guidance derived from randomized, controlled clinical trials. In other words, research is almost always lacking.
Such evidence gaps force doctors to make treatment decisions without data, based solely on intuition and experience. This situation is exacerbated when physicians are caring for patients with multiple conditions, complex medical histories, or diverse ethnic backgrounds.
At a time when every click can be tracked and medical records are fully electronic, physicians should be able to digitally reference the decisions made by other clinicians to find out: What happened to other patients like mine? While the technology to perform such a digital consultation of data exists, incorporating the results into routine care requires new solutions. One model is emerging: services staffed with physicians and data scientists, all of whom have access to longitudinal patient records and can return the required evidence in under 24 hours.
At Stanford University School of Medicine, the academic research project the Green Button Informatics Consult Service, was developed to deliver clinical consults backed by technology. Informatics departments at some of the country’s best-known institutions have since followed suit and have experimented with similar services. Examples include City of Hope, Columbia University, and the Mayo Clinic, which developed the Mayo Clinic Platform to securely organize data with the goal of being able to answer even the most complicated medical queries as quickly as possible and immediately impact patient care.
The purpose of this article is to explain how these services work and how they can both improve patient care and reduce costs.
The Evidence Gap
Consider this real-life and personal example (coauthor John Halamka’s mother, to be exact): An elderly female who presents with an impaired mental state, fever, and a low level of serum sodium. She is hospitalized and seen by a primary care physician, who recognizes the patient likely has a urinary tract infection (UTI) and begins treatment with antibiotics and fever reducers. However, a UTI does not entirely explain the patient’s low sodium levels. While low sodium could be a result of the renal clearance of sodium, patients with UTIs rarely present with low sodium, leaving the physician seeking answers.
Unfortunately, there aren’t many answers available because a clinical trial involving 80-year-old females with impaired mental status and abnormally low sodium hasn’t been conducted. But, with millions of electronic patient records available, a database consult could allow the physician to accurately diagnose and treat the patient rather than just best-guess it.
Even highly trained and skilled physicians experience an evidence gap that impairs their ability to accurately diagnose and treat certain patients, which is one reason this analysis needs to be done routinely. Regular database consults can answer critical clinical questions like:
- What is the right diagnosis?
- What diagnostic tests should be ordered?
- What is the implication of this abnormal lab result or genomic marker?
- What is the typical prognosis for patients like this?
- What medications or other treatment modalities should be pursued, in what order, to optimize outcomes?
- Will this procedure be worth the risk and/or cost for this patient?
- Can the patient’s life be extended or improved with alternative treatments?
A Four-Decade Effort
The idea of consulting the medical record to learn what happened to similar patients is not new. Arguably, it began in 1972 when famous physician researcher Alvan Feinstein published “Estimating Prognosis with the Aid of a Conversational-Mode Computer Program,” an article in which he described a computer system where clinicians, medical data, and technology combine to create real-world evidence to support clinical decisions for treating patients with lung cancer. Since that landmark idea, several efforts have been made to realize this vision.
One example is the Duke Databank for Cardiovascular Disease, which was launched in 1975 and produced reports called prognostigrams that summarized what happened to similar patients when different treatment choices were made. The hope was for the Duke Databank to become part of clinical practice. However, the cost of obtaining the data in electronic form, the focus on just one medical area (in this case cardiology), and limitations around payments constrained the effort.
Duke’s prognostigram reports ultimately ended, but the concept wasn’t abandoned. Other universities and institutions have worked to overcome the logistical hurdles and foster the technology and supportive policy required.
In 2011, pediatricians at Stanford University School of Medicine faced a critical decision about treating a patient with systemic lupus erythematosus (SLE) with an anticoagulant. It was not a standard practice but was one the physicians felt was the best course of action given the complications. However, there were no related studies to confirm this treatment option was best given the risk. To guide them, the doctors used the school’s clinical data warehouse to estimate the risk of blood clots. In less than four hours, they were able to review data on an SLE cohort that included pediatric patients with SLE cared for by clinicians between October 2004 and July 2009 and make the data-informed decision to administer an anticoagulant. Their success was later published in the New England Journal of Medicine article “Evidence-Based Medicine in the EMR Era.”
A Feasible Solution
In 2018, a group of physicians and data scientists from Stanford University School of Medicine piloted the Green Button Informatics Consult Service, which used routinely collected, de-identified data from millions of individuals to provide on-demand evidence in situations where adequate evidence was lacking. The results could then be analyzed immediately by the attending physicians and data scientists to inform patient care right away. During the pilot, the service responded to 100 consultation requests by 53 users from multiple specialties. The consultations informed individual patient care, resulted in changes to institutional practices, and motivated further clinical research — establishing the feasibility of on-demand evidence generation to close evidence gaps.
This new capability was the catalyst for Atropos Health, which has successfully provided over 1,800 consultation requests to date. A number of health tech startups are committed to harnessing the power of electronic health record data to improve patient care. Lucem Health, nference, OMNY Health, and Atropos Health are just a few examples. Health care giants such as Mayo Clinic and Cleveland Clinic are taking notice too. Seeing the potential of database consults at the point of care, Mayo Clinic partnered with Atropos, leveraging the company’s technology to enhance the Mayo Clinic Platform.
All of this is made possible in part thanks to federal rules requiring patient data to be migrated from paper charts and stored in electronic form — meaning 10 to 15 years of past medical histories on hundreds of millions of patients are now routinely accessible. The Mayo Clinic Platform, for example, improves care delivery through insights and knowledge derived from the de-identified data from 10 million patient records that include laboratory values, diagnosis codes, vital signs, medications, and clinical notes. This data is necessary to enhance the effectiveness of database consults; it gives data analysts and machine-learning platforms the basic building blocks needed to turn data into insights.
Expert-in-the-Loop Services
Accessing and storing data to analyze makes the lives of data scientists and physicians looking to improve care delivery easier, but developing new best practices for care delivery requires building an evidentiary basis. Since data is developed through patient encounters, the generation of such evidence takes a long time and costs a lot of money. Further, finding patterns in data is a manual process for many institutions, adding an additional time and labor cost to the equation. Even at institutions with well-staffed analytics teams, such evidence generation often requires over 300 hours and $300,000 just to answer clinical questions and develop a new care guideline. This artisanal approach cannot scale to serve the needs of every patient.
The solution is employing novel search technology like the one used by the Green Button Informatics Consult Service and a staff of physicians and data scientists who have access to on-demand datasets and the expertise to interpret the search results. This approach can achieve the accuracy and speed required for the under-24-hours turnaround time needed for closing evidence gaps at the bedside. In addition to dramatically reducing the time and labor required to produce this tailored clinical information, this approach will almost certainly generate significant cost savings; ongoing research aims to quantify exactly how much.
In the last 40 years, despite the tremendous investment in technology and policy to drive digital innovation in care, physicians are still making a plethora of clinical decisions without data. Expert-in-the-loop services that use now-available technology and data can help change that, delivering on the initial promises of electronic health records to turn real-world clinical insights into improved patient care. They can bring us closer to realizing Feinstein’s vision and close the evidence gap at the point of care.
