BACKGROUND: Electronic lab reports (ELRs) often contain missing or conflicting data, requiring interpretation and standardization to make them suitable for public health surveillance and action and allow them to be incorporated into surveillance systems, such as the Washington State (WA) implementation of Maven (WDRS). Reporting laws in WA allow laboratories to submit reports that contain very little information. Laboratories vary widely in their ability to interpret and correctly implement both messaging (HL7) and terminology standards (e.g. LOINC, SNOMED). While a human brain can usually correctly interpret a report, automating that process is a resource-intensive task previously solved by writing rules in SAS. New tools and collaboration within the WA Department of Health will allow for standardized interpretation of electronic lab records by disease area.
METHODS: We reviewed 20 months of ELR data to identify as many unique test and result combinations as possible. We were able to create logic rules to be applied by the Disease Reporting Interoperability and Verification Engine (DRIVE) – a project using Orion Rhapsody to transform ELR data. DRIVE will map input data to standard test types, results, units of measure, test interpretations, specimen types, and structured numeric data by reviewing multiple fields and components inside an HL7 message and apply logic provided by the Epi business area owners. These standardized fields are brought into the notifiable conditions database for public health action and data analysis.
RESULTS: During the 20 month period, the WA HIV surveillance program received more than 18,000 electronic laboratory results. A total of 394 unique combinations were used to determine test types and 271 unique combinations to determine results. DRIVE will use 10 components from ELR messages to determine test type and an additional five components in conjunction with the standardized test type to determine the result. Interpretation of both test type and result use requires correct identification of the submitting/performing lab.
CONCLUSIONS: DRIVE will assist in standardizing the hundreds of thousands of electronic lab records we receive each year. HL7, LOINC, and other standards assist in test interpretation but can also be misused, incomplete, or inadequate by themselves. The processes now in use requires many steps, allowing room for error. Standardization of many of these steps will hopefully reduce the potential for error and allow ELRs to be more easily used to improve public health surveillance and produce public health actions.