Understanding Electronic Medical Records and Why Analytics Are Needed


 Understanding Electronic Medical Records and Why Analytics Are Needed

Joseph Dudas, Division Chair, Enterprise
Analytics, Mayo Clinic

Background

When I entered the healthcare industry I did so knowing there were a lot of issues which also means a lot of opportunities. After spending the last decade observing this industry, it is clear to me that unnecessary waste and far less than optimal results are mainly due to excessive complexity. Sure, the human body is truly complex (and an engineering marvel unto itself). Physicians that have understanding of just part of this system truly deserve due recognition. However, the administration of healthcare should not be nearly as complex and opaque as it is. Mature businesses have mastered core systems that optimize vast enterprises for years and there is no reason that healthcare cannot do the same.
“Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius—and a lot of courage to move in the opposite direction.” - Ernst F. Shumacher, well known economist. (Source: El-Erian)
I would not define our industry as violent but it has and continues to upset many of our consumers and the main reason is complexity. Electronic medical records (EMR) were intended to put a core system in place that would unwind some of this complexity (including basic communication and coordination) however many would say that it has fallen well short of this objective. If fact, for years doubts have been raised about cost savings associated with EMRs by researchers at Harvard University, the Wharton School of the University of Pennsylvania, Stanford University, and others. This same doubt continues today. (Source: Health Affairs)However, it does not have to be this way and it is not the fault of the systems. It is the intent of this article to lay out the basics of how electronic medical records work (particularly for those outside of healthcare), point out where the primary opportunities remain and suggest a basic framework for EMR Analytics.

What a medical record is and how it is structured.

It is often reported that 1991 was the beginning of EMR. It was at this time that the Institute of Medicine (IOM) sponsored several studies that led the way toward the modern concepts that we have in place today for EMRs. The IOM called them computer-based patient records. At Mayo Clinic electronic medical record development dates back to 1960s with what was described as problem-oriented medical records. (Source: Health Technology Review) However, it is in the early 1900's where the true roots of the EMR lie. It was at that time that the concept of a patient record was developed by Mayo's own Dr. Henry Plummer.
To understand what an EMR is, you first really need to understand the contents and the basic workflow of a patient record within a practice office (which is not fundamentally all that different than what Dr. Plummer experienced). Most of us are very familiar with this as we have all been patients at one time of another. The first step is an appointment request, where insurance eligibility is confirmed and the appointment is scheduled. When arriving at the office (or prior) demographic and medical history are confirmed and updated. Then our symptoms are collected as well as vitals in preparation for a medical examination (this is referred to as a "subjective" note). The physician then performs an exam and lists all findings (referred to as an "objective" note). An assessment is then done where the physician arrives at one or more diagnosis (referred to as the "assessment" note) and finally, a plan of care is developed in which further tests, medications and treatment is prescribed (referred to as the "plan" of treatment note). In its most basic form an EMR simply represents this workflow and data capture in an electronic form.
As described, EMR is not just what is stored but also what can be done with the information and for this reason, IOM put forth a set of eight core functions of an EMR that include the following: health information and data, result management, order management, decision support, communication, patient support, administration and reporting.
However, from an analytics perspective it is the data portion that we are mostly concerned about and the bulk of the novel information (even in today's modern EMR) is within the four previously mentioned notes which is often referred to as SOAP (subjective, objective, assessment and plan). Now that does not appear as complex as you would think, right? Wrong. This is actually where today's complexities lie.

Putting into context the issues surrounding EMR data quality.

Until the recent government sponsored "Meaningful Use" program it was estimated that only 20%-30% of healthcare offices used EMRs. "Meaningful Use" is a CMS (Centers for Medicare and Medicaid Services) program that is administered through the ONC (Office of the National Coordinator for Health IT). It calls for EMR data capture (phase I), data exchange (phase II) and data outcomes (phase III) for all institutions caring for Medicare and Medicaid patients. (Source: Health IT)
While there have been delays, "Meaningful Use" has been a very successful program, with an estimate of 80%-90% of offices now using EMRs. However, few health care organizations have invested in practice standardization and data quality programs. So while able to meet minimum requirements for "Meaningful Use", the best structured data that is widely available is that of claims. Given that claims processing is the last activity in the healthcare value chain, it is not timely nor of sufficient detail to be truly transformative (although many are trying).
The problem is not a standards nor tooling issue. While standards can always be improved healthcare has plenty to work with such as HL7, LOINC, SNOMED, ICD and RXNORM just to name a few. As for data quality tools, there are also plenty that are now used in other industries. However, what other industries (such as retail and manufacturing) learned when they bought data quality tools is that they had to back track and learn about data quality as a discipline. Healthcare will need to learn from other industries and pursue data quality as a discipline in order to achieve similar results. (Source: IT Business Edge).
So, for these reasons, as of today, much of the point of care data, described previously that could be used to significantly reduce complexity in the system, is not standardized and in many cases not structured. Said differently, SOAP data is mostly as it was prior to EMR (now in an electronic form).

Knowledge and how can be applied in healthcare.

A step in the right direction and gaining momentum is clinical pathways (also known as care paths). These are standard operating procedures (SOPs) for healthcare and guide physicians as well as other care providers through proven best practices and processes. Used effectively they can significantly reduce variation and complexity as well as ensure organized and efficient care based on evidence and data. When deployed through an EMR they can systemically be used to manage the processes described earlier as well as capture standardized data with high degrees of accuracy. (Source: Health Knowledge)
So what is the downside of care pathways? Care paths have disadvantages with the first and most frequently fear being the dehumanization of care where the relationship between the health professional and the patient results in less personal care. Another concern often voiced is one of mistrust. Codification of protocol as well as checking for errors and defects associated with SOPs brings the profession in line with the position of being controlled. While potentially beneficial in creating repeatable and predictable care, it could easily effect trust among stakeholders (depending on the implementation strategy). Furthermore, rigid process control could lead to limitations for patients with poor physical or unique conditions. These patients may also have a greater risk, for example, of complication and death. (Source: National Center for Biotechnology Information)
Many experts would propose that while concerns are valid, the potential advantages of care pathways far outweigh the disadvantages. What they suggest is an implementation approach that is flexible and improvement focused, as opposed to being rigid and punitive. This is supported by a recent publication which suggests that today full-service hospitals operate as "solution shops". They are structured, managed and incentivized to diagnose and recommend unique solutions to unstructured problems. However, in this same study the authors conclude that 67% of the patients (within their limited sample) would actually be served well by a "focused factory" model which is characterized by a uniform and standardized approach. (Source: Health Affairs).
Other experts suggest that what is needed is "smart" care pathways. A well-defined SOP that is supplemented with analytics which allows for appropriate deviation, evidence based professional judgments and simplicity in administration. And, that is where EMR Analytics come in.

How analytics can be used to improve knowledge and healthcare efficiency.

Analytics is one of the most rapidly growing markets within Healthcare, growing at 25% or more per year with no slow down anticipated in the near future. However, while this market is rapidly growing, it is highly immature and solutions are often designed based on overly complex processes and poor quality data (both with a high degree of variability). Stepping back from this complexity may warrant a different perspective where a relatively simple analytics model could be coupled with a balance of data quality discipline (through care path deployment) and state of the art data mining tools such as those used to optimize the Internet (mining huge text files to identify unique patterns of data). Below is a depiction of what the analytics model might look like.

More Stories