Data normalization seeks to harmonize data from different sources so that data can be made available to, and consumable by, a provider’s IT system. Alerts are based on aggregating data from across disparate systems and applying rules to that data - if the interrogated data meets certain conditions, the alert is fired.
As data is aggregated from disparate systems across the continuum, one of the first steps in using this data rests with the ability to normalize clinical and claims data into standardized terminologies to ensure an accurate picture of health across identified patient populations. Simply put, if your data is not normalized, the alert may miss potentially dangerous situations.
For example, let’s say your system has a rule that fires if a patient who is allergic to Codeine is prescribed Codeine anywhere within an interconnected health system. Now, let’s assume you have a rules engine that interrogates your data for occurrences of Codeine prescriptions based on the RxNorm code for Codeine (248550). However, all your IT systems don’t ‘speak’ RxNorm. You have a claims system that uses NDC codes and you have a prescription ordering system that uses proprietary codes and free text. If you don’t normalize this data to RxNorm, you may miss a potentially dangerous situation.
The field of real-time care alerts represents an emerging use case for data normalization. Care alerting is a key function of clinical decision support systems, which warn healthcare providers about harmful interactions between drugs or other conflicts such as drug-food and drug-allergy interactions. The use of clinical decision support is one of the government’s criteria for electronic health record (EHR) system deployment under Stage 2 Meaningful Use.
According to HealthIT.gov, clinical decision support encompasses tools such as “computerized alerts and reminders to care providers and patients; clinical guidelines; condition-specific order sets; focused patient data reports and summaries; documentation templates; diagnostic support, and contextually relevant reference information.”
EHRs may include clinical decision support functionality or a provider may deploy a standalone clinical decision support system that integrates with an EHR product. Clinical decision support systems may also be used in conjunction with computerized physician order entry (CPOE) systems.
The alerting capabilities of a clinical decision support system can benefit from data normalization. Here are three normalization uses for care alerts:
1. Translating Evidence-Based Guidelines
A clinical decision support system typically consists of a knowledge base of rules that embody best practices and help trigger alerts when conflicts arise. Evidence-based clinical guidelines, which may come from government or academic sources, can help inform a clinical decision support system’s rule base. But those guidelines are generally text-based and must be translated into a machine-readable format so clinical decision support systems and EHRs can use them. This is a role for data normalization. Work on this challenge is ongoing. The CDS Consortium, based at the Vanderbilt University Medical Center, has identified “knowledge translation and specification” as one of its research objectives. In addition, the Consortium for Healthcare Informatics Research, a research program backed by the Department of Veterans Affairs and located at the University of Utah’s School of Medicine, is investigating “tagging codes in unstructured text and mapping to standardized terminology.”
2. Advancing Clinical Decision Support Among HIEs, ACOS
Health information exchanges (HIEs) and accountable care organizations (ACOs) are in a position to gather patient data from participating healthcare providers. That data can help support a clinical support system and its alerting function. Patient data on prescription drug use and allergies, for instance, is a critical component of clinical decision support. But that data must first be normalized if it comes from disparate IT systems using different coding methods. Normalization could help promote evidence-based medicine among HIE members and ACOs.
3. Improving CPOE Deployments
CPOE relies on clinical decision support to flag drug-drug and other harmful interactions. Furthermore, a health system must translate its order sets into standard terminologies so that they can be properly loaded into CPOE systems. Data normalization can help ensure the clinical decision support aspect of a CPOE system is working with the fullest possible set of patient data.
Time Is Critical
Real-time alerts can help providers avoid mistakes in patient care. Those alerts are part and parcel of the clinical decision support systems that will probably become more widely adopted in light of Meaningful Use. Data normalization can play a part in making sure those alerts are based on a solid foundation of machine-readable data.
Do you plan to introduce clinical decision support and alerting in your healthcare facility? Have you considered a data normalization strategy to make that happen? Leave your comments below.