Mis-mappings between a producer's quantitative test codes and LOINC codes and an algorithm for correcting them

Area of Science:

• Health Informatics
• Clinical Data Management
• Laboratory Medicine

Background:

• Accurate mapping of laboratory tests to standardized codes like LOINC is essential for data integration in healthcare.
• Existing data integration efforts face challenges due to mapping inaccuracies across different systems and over time.

Purpose of the Study:

• To assess the accuracy of Logical Observation Identifiers Names and Codes (LOINC) mapping to local laboratory test codes.
• To determine the rate of LOINC mapping errors within the PCORnet data.

Main Methods:

• Utilized software tools and manual reviews to evaluate LOINC mapping accuracy.
• Analyzed 179 million mapped test results from two PCORnet DataMarts.
• Reported unweighted and weighted mapping error rates, overall and by LOINC term components.

Main Results:

• A 4.6% LOINC mapping error rate was identified across 179,537,986 mapped results for 3029 quantitative tests.
• Error rates were below 5% for common tests (≥100,000 results).
• Significant variation in error rates observed across LOINC classes, with chemistry at 0.4% and hematology at 7.5%.

Conclusions:

• The overall LOINC mapping error rate in PCORnet data is 4.6%, which is substantial but lower than previously published rates.
• Automatic detection and correction algorithms can significantly reduce mapping errors, decreasing the rate to 0.1% for quantitative tests.