Originally published in August 2019, updated in May 2022
This is the third installment of a four-part series dedicated to the importance of data mapping. Be sure to check out the previous two blogs where we discuss the importance of lab and medication mapping.
The problem: Messy and disparate allergy data
Allergies are linked to numerous chronic diseases and serious illnesses. Obtaining allergy information is a critical step toward safe prescribing, prevention of adverse drug events (ADEs), and reducing the cost of care. To share this vital information across the continuum and ensure the highest level of patient care, providers must fully document allergy information within electronic health records (EHRs) at the point of care.
Unfortunately, stakeholders face challenges when attempting to share allergy information due to incomplete and inconsistent healthcare data. In fact, many clinicians choose to enter allergies as un-coded free text because they are unable to reconcile language used by an EHR with familiar allergy terminologies.
In addition to the challenges of extracting allergy information from free text, some inhalant allergies (such as asthma, hay fever, and allergies to dust and mold) can be difficult to locate because they are typically found in the Problem List rather than the Allergy List of an EHR.
Documenting allergies will be increasingly important as food and environmental allergies become more common. If this information is not shared across institutions and providers, the potential for ADEs becomes much greater.
For example, peanut oil is found in a few injectable medications, and glucosamine tablets are sometimes derived from shellfish—two areas where an uptick in allergies has occurred in recent years. If providers unknowingly prescribe these medications to patients with allergies, there will be negative downstream consequences. Propofol is another example of a medication that could result in an ADE for patients with egg or soy allergies.
Incomplete and inconsistent allergy data is a significant industry problem. Prescribers override more than 90% of allergy alerts intended to protect patients from ADEs due to the belief that many of the documented EHR drug allergies are incorrect, and therefore inconsequential to patient care.
Mapping opportunities and challenges for data
Healthcare organizations can achieve a framework of data normalization by implementing systems that support complete extraction of all disparate allergy information and then mapping the data formats to the appropriate industry standard. Complete aggregation of allergy data requires clinical natural language processing (cNLP) techniques, a powerful branch of artificial intelligence that can efficiently locate allergy information in free-text fields and turn unstructured documentation into shareable data that can be analyzed and acted upon. The efficacy of cNLP was proven for a large, well-known health system: after conducting their own analysis, they found that the cNLP technology successfully detected 96.7% of allergens captured within their organization’s EHR.
Once healthcare organizations have systems in place to support complete aggregation of allergy data, the next step is determining the correct standard to use for mapping—an ongoing industry challenge due to inconsistent and conflicting industry protocols. The Health Language solutions use the Office of the National Coordinator (ONC) Health IT Interoperability Standards Advisory manual as our guide. In the 2022 release of the manual and the new requirements in the USCDI, environmental allergies and food allergies require codification to SNOMED CT®. In the case of medication allergies, the standard of choice is RxNorm with medication class allergies mapping to either SNOMED or MED-RT.
It’s a complex undertaking—one that many healthcare organizations do not have the resources or expertise to address.
The solution: Data normalization
Much like lab and medication data, the business case for leveraging an infrastructure that automates the mapping process for allergies is an easy one to make due to the volume of information that exists in disparate medical data systems. The Health Language Data Normalization Solution combines the efficiency of machine learning and cNLP with the clinical expertise of our team of informaticists to help organizations address the burdensome, error-prone processes often managed across numerous spreadsheets and departments.
Specifically, our web-based application allows healthcare organizations to collaboratively map, search current standards, and distribute healthcare data throughout the enterprise. Clinical auto-mapping powered by domain-specific algorithms ensures the highest map rates and accuracy, and dashboards alert teams to maps that require manual review.
The right allergy mapping strategy powered by an advanced infrastructure of automation and machine learning has potential to deliver significant value across a wide variety of use cases.
To learn more about how the Health Language Data Normalization Solution can help your organization, speak to an expert today.
In our next and final installment of this Data Mapping Blog Series, we will explore the opportunities inherent in mapping problems and diagnoses.
SNOMED CT® is a registered trademark of the International Health Terminology Standards Development Organisation (IHTSDO).
LOINC® is a registered trademark of Regenstrief Institute, Inc.
