Decoding the Language of Lab Data

From diagnosis to treatment, healthcare professionals rely on lab data to make informed decisions. Standardization of lab data ensures consistency in the interpretation of test results across different healthcare settings so providers can easily identify and compare specific laboratory values, regardless of where they were conducted. This consistency not only allows providers to easily compare and contrast specific values but also enables seamless information exchange between different corners of the healthcare world, fostering collaboration and improving patient care.

As part of the Meaningful Use criteria, healthcare providers must incorporate clinical lab-test results into Electronic Health Records (EHRs) as structured data. Physicians must show that at least 50% of the clinical lab tests they order, with numeric or +/- results, are integrated into the EHRs in a structured format. Achieving this goal requires all laboratories to adopt a universal "language" to ensure systematic importation of these results into EHRs – enter LOINC, the coding system designed for this purpose.

LOINC, an acronym for Logical Observation Identifiers Names and Codes, acts as a universal framework for identifying laboratory and clinical observations. Under this system, each distinct lab test type, specimen type, and methodology is allocated a standardized LOINC code. With roughly 40,000 codes presently defined for lab tests, this standardized coding mechanism enables seamless communication across healthcare entities, guaranteeing the precise interpretation and application of lab data.

In principle, standardization sounds fantastic. However, the reality is not everyone embraces LOINC codes. While major commercial labs such as LabCorp and Quest employ LOINC systematically, this covers only a fraction of the labs ordered by physicians. Local hospital labs, on the other hand, exhibit significant variability in their adoption of the LOINC system, often resorting to internally-derived coding for defining lab tests. Consequently, provider organizations face a daunting challenge when attempting to harness lab data at scale. 

How Particle’s platform surfaces condition-specific data elements

Standardized lab data opens the door to a myriad of use cases beyond diagnosis. Particularly for chronically ill patients, this data serves as a vital instrument for tracking disease advancement, managing medications, and evaluating treatment effectiveness. That’s why it's critical that platform solutions providing clinical data are able to standardize and extract specific data elements from patient records at scale.

Particle’s data team assessed a population of 14,958 patients. By parsing through ICD-10 codes, 6,607 chronic kidney disease patients were identified. Of these 6,607 patients, 16.1 million historic lab values associated with a CKD diagnosis were extracted and 74% of those labs were associated with a standardized LOINC code. This comes out to 1,997 CKD-specific labs per patient. Let’s take a look at three labs commonly used to manage patients with chronic kidney disease:

Creatinine

• A waste product of creatine used to assess kidney function
• Successfully identified for 89.22% of CKD patients, averaging ~50 per patient

Estimated glomerular filtration rate (eGFR)

• A metric used to assess disease stage by measuring your kidneys’ ability to filter toxins from your blood
• Successfully identified for 81.53% of CKD patients, averaging ~50 per patient

A1C values

• A test used to measure long-term blood glucose levels
• Successfully identified in 75.1% of CKD patients, averaging ~9 per patient
  

As an organization committed to enabling value-based organizations to achieve better patient outcomes, we are working to move the needle beyond 74% for LOINC associated lab values. How? By improving our data normalization process and understanding how the same lab might be represented in different coding systems.

Access to standardized, historic lab values is important for healthcare providers as it provides valuable context for current test results and enables healthcare providers to make informed clinical decisions. By comparing current results with previous ones, providers can identify key trends, assess disease progression, and determine the effectiveness of treatments. For patients with chronic conditions such as diabetes, hypertension, or kidney disease, regular monitoring of lab values is essential for disease management.

Keeping in mind the expanding application of lab values, evidence shows that standardization of lab data can impact predictive model performance in multi-site datasets. In a study that predicted 30-day hospital readmissions for a set of heart failure-specific visits, models that utilized LOINC performed significantly better than models that utilized local laboratory test codes, regardless of the feature selection technique and classifier approach used. These findings support the need for data standardization in predictive modeling, especially in studies leveraging multi-site datasets extracted from electronic health records.

Particle’s data platform empowers healthcare providers to harness the full potential of data to deliver high-quality, patient-centered care. Standardization not only enhances data accuracy and interoperability but also opens doors to new opportunities in research, population health, and healthcare innovation. Working with a chronically ill patient population and need to boost your insight into key pieces of data? Particle can help. Reach out to our team to learn more.