Standardizing Clinical Data with Analysis Normal Form
Earlier this month, the Solor team succeeded in balloting the High Level Seven (HL7) Informative Ballot for Analysis Normal Form (ANF), a standardized clinical data framework that allows for useful analysis and reliable decision support. The team, seeking to further develop the knowledge architecture of health care interoperability solutions, received 62 comments from seven international health care IT and informatics organizations and generated interest from multiple stakeholders on the future implementation of ANF for clinical data analysis.
The Current State of Clinical Data
Clinical information systems are continually in need of modernization to eliminate outdated, unreliable, and error-prone processes. Often, this is addressed by advancements in new health care technologies that seek to increase efficiency, lower costs, and provide better patient treatment. One of the prime examples of these advancements is the widespread adoption and implementation of electronic health record (EHR) systems. This modernization effort is a fundamental step on the journey for better care coordination and improvement in health outcomes. However, health care organizations implemented systems expecting interoperability of health data that didn’t exist – This led to data entry issues, unexpected conversions, and repeated errors that reduced the effectiveness of clinical decision support and ultimately caused patient harm.
In a recent study conducted by Porter Research, 52% of health care CEOs, CFOs, and CIOs indicated that data sharing between providers, payers, government, and industry would lead to exponential improvements in patient care. The Solor team knows, however, that achieving this objective requires investments in processes that improve clinical data representation through semantic normalization, the process of reshaping clinical data so that it retains its intended meaning. This concept is largely dependent on health care organizations having a strong foundation of clinical data, and that is where Analysis Normal Form (ANF) comes in.
What is Analysis Normal Form (ANF)?
The Solor team has helped to create Analysis Normal Form (ANF) with the purpose of standardizing clinical data so that it is understandable, reproducible, and useful. ANF was born out of recognition that current clinical data is unpredictable and unreliable: because different EHR systems represent their inputs in different ways, data is often very difficult to analyze and interpret.
For example, in the illustration below, when entering a patient’s blood pressure information into an EHR, Medical Center A may list the “Brachial Artery” under one dropdown box, and “Left side” in another. This means the data was divided into two separate variables and captured as two data points. Medical Center B may group “Left side” and “Brachial Artery” together in one dropdown box separated by a dash. This means Medical Center B’s data is collected as a single data point. Currently, this would mean that informaticists have to rely upon the tedious manual mapping processes to link Medical Center A’s two data points to Medical Center B’s one data point. Now think about the hundreds, if not thousands of different data points across the continuum of health care that must undergo the same process. It is easy to see why an simplified improvement upon clinical data statements is critical to the future of health care interoperability.
Like Solor, ANF seeks to promote a collaborative environment where a patient does not have to worry about their quality of care when they transition from one physician to another. By taking in inconsistently structured data, ANF standardizes inputs that allow for normalization, analysis, and clean queries, ultimately leading to better clinical decision-making. ANF provides data for analysis and insights which creates an environment that can handle healthcare interoperability solutions like Solor.
How Does ANF Work?
ANF was developed with four key design principles in mind: simplicity, consistency, reusability, and reducing variability. But how does it actually work? Let’s continue with the blood pressure example. Currently, clinical data is represented in a complex, tree-like structure, with multiple branches stemming from each segment – resulting in data that is inconsistent, difficult to search, and hard to analyze. Seeking to remedy this situation, ANF breaks up all clinical data into two simple categories: Topic, which defines the action (being performed or requested) or the result of that action, and Circumstance, which defines the how, why, when, and what “result.”
When data is processed using ANF, it can then be analyzed and queried on the back-end by informaticists, data scientists, and health statisticians. ANF creates an opportunity for more accurate decision support for physicians that can lead to more preventative, rather than reactive care. By simplifying the structure and data, ANF opens the door for interoperability efforts across all health care data systems. Adopting ANF’s framework does not replace or compete with any existing input forms or clinical data models, but it does create a new, much-needed standard for developing Health IT systems. It is our hope that by working with health care industry leaders, health record entry will lead to saving lives, not causing undue harm.