Now that you are collecting all this data, what do you do with it? Well, one of the first challenges involves normalizing the data collected from disparate sources. A data normalization solution offers healthcare organizations the ability to semantically map between disparate reference terminologies, classification systems, local proprietary coding systems, and unstructured text. A semantic map allows both you and the computer systems to understand what the codes and words from your data actually mean.
The ability to provide that mapping is critical given the expansion of healthcare IT (HIT) applications and the range of terminologies used within those proliferating systems. Data is often confined within standalone systems, each with its own way of representing clinical terms. The lack of a common clinical vocabulary makes the effective sharing of health information difficult, if not impossible. Healthcare organizations, however, increasingly need to share data, especially as new models such as Accountable Care Organizations and health information exchanges emerge. The terminology hurdle needs to be cleared for interoperability and collaboration to become the healthcare norm, rather than the exception.
Medication is one area in which the limitations imposed by disparate terminologies become particularly apparent. Here, the lack of adoption of a single standard makes it challenging to share and aggregate data related to drugs. Providers, pharmacies, payers, and HIT systems vendors may use different terminologies to describe the same medication. Data normalization cuts through this chaos, linking incompatible terminologies to a common standard.
The National Library of Medicine’s RxNorm and the Veterans Health Administration’s National Drug File - Reference Terminology (NDF-RT) are examples of terminology standards that are finding use in HIT applications that incorporate information on medicine. The Meaningful Use requirements, for example, require RxNorm as the drug terminology used to share medication information between systems.
Here are four use cases that call for such terminology standards and data normalization solutions:
1. Clinical Data Repositories
A healthcare organization may seek to pull in medication data from a number of sources to discover patterns such as redundant antibiotic prescriptions -- ordering two antibiotics that address similar bacteria, for example; or determining whether pneumonia patients were receiving appropriate and timely antibiotics. Such data warehouses may be used to determine other types of medication errors as well. But the organizations would need to normalize the incoming drug data for the warehouse to be effective.
If a healthcare provider wanted to study the anti-inflammatory drug Naproxen, it might find that different systems represent the drug under different labels. Naproxen may be described as Naproxen Tab 250 MG, Naproxen 250mg tablet, or NAPROXEN@250 mg@ORAL@TABLET, among other ways, according to NLM. Under RxNorm, however, Naproxen groups the different drug identifiers into one concept and then assigns a normalized name to the concept, NLM noted. So for this drug, you may receive NDCs (National Drug Codes), proprietary codes (from Medispan, Multum, or FDB), plain text strings, and possibly some standard RxNorm codes. Data normalization would allow you to take these disparate codes and strings and normalize them all to RxNorm for example.
2. Evidence-Based Medicine
Evidence-based medicine calls for providers to review clinical evidence -- in the form of research, for example -- before making decisions on treatment programs. Data normalization serves as a key underpinning of evidence-based medicine, providing the means for making a like-for-like comparison of data gleaned from different HIT systems. Standards such as RxNorm help ensure that organizations can aggregate medication data for a more precise picture of patient outcomes. Do you use normalized drug data if all appropriate patients are receiving DVT prophylaxis? Are ventilated patients receiving gastric acid reducing medications such as omeprazole, and do they have increased risk of pneumonia? Did your diabetic inpatients have increased risk of renal failure while on metformin?
3. Drug Safety
Data normalization can boost databases that track drug safety. Researchers found that normalizing data contained in the Food and Drug Administration’s Adverse Event Reporting System (AERS) may improve the data mining capability of the database for drug safety purposes. The researchers normalized AERS data to RxNorm and also used NDF-RT. The project resulted in an open source data resource that “has the potential to assist in the mining of [adverse drug events] from AERS for the data mining research community.” Drug safety and surveillance still utilize chart reviews, self reporting mechanisms, and poor quality claims data to determine potential drug adverse effects.
The federal Meaningful Use initiative has designated the use of RxNorm codes as the way to meet the e-prescribing vocabulary requirement. For physicians and hospitals, normalizing data will help them comply with this regulatory requirement. Compliance can help providers qualify for incentive payments, so this particular use case has a bottom-line benefit. A common challenge faced by hospitals, for instance, is that your emergency room and your inpatient clinical systems may in fact be using different drug terminologies, thus hampering the ability to e-prescribe to clinics or prescribe from the ER to the inpatient setting.
Data normalization can support additional use cases such as improved data exchange, facilitation of emerging healthcare models, improved support for clinical data tasks, and overall ease of use.
Are you considering data normalization as you deal with medication data? Leave your comments below.