These new rankings replace the former “Top 2000+” lists, which had become outdated. Our new data applies rank values for the top 20,000—yes, twenty thousand—LOINC codes. So, while you can still extract a list of the most frequently used 2000 from the LOINC table, we now provide rankings for the next band of usage up to 20,000. As before, the rankings appear as a separate field in the LOINC database (named COMMON_TEST_RANK), and in LOINC searches in a column labeled “Rank”.
What has not changed is that relatively few codes account for much of the typical result volume. In the refreshed listing, the most common 2000 ranked LOINC codes represent about 99% of the test volume from the newly compiled data set.
Put the Pareto Principle to work for you. These LOINC codes represent the vast majority of test volume.
What to know about the LOINC rankings
- The data is derived from multiple sources. Three large research data sets, two from Observational Health Data Sciences and Informatics (OHDSI) and one from the National Patient-Centered Clinical Research Network (PCORnet) are included. Each comprises more than 20 individual site data sources. Additionally, data from three large hospital systems are included in the compilation.
- The time period varies by data source; however, most sources restrict data to a period between 2018 and 2021.
- The data is exclusively from sources within the United States, but we hope to include non-U.S. data in the future.
- The new rankings do not include an “SI Version” specifically. As a result, countries that use SI units may find some of their common tests unranked. In those cases, we encourage you to look for the corresponding mass unit test to derive an approximate ranking. LOINC will no longer provide that abstraction in a separate SI ranking field.
- If you would be willing to donate your institution’s data to Regenstrief Institute so that we can keep this valuable resource up-to-date, please contact us.
Caveats when using the rankings
- LOINC has not verified the completeness nor accuracy of mappings of each site’s local observations to LOINC, so the data is reflective of how individual sites have mapped to LOINC without any Regenstrief input or review.
- Sites tend to focus on mapping their more commonly performed observations first. As a result, it is expected that more rarely reported observations may be even less represented in the data than in reality.
- It is recognized that sites focus on mapping laboratory data to LOINC more than other clinical observations, and some of our sources mapped only laboratory data. As a result, expect that clinical observations may have artificially lower rankings for this reason.
- Above a rank of 10,000, the distinction between ranks becomes less meaningful. For example, the difference between rank 14,000 and rank 16,000 may represent only a small frequency of usage.
- SARS-CoV-2 observations: Keep in mind that the data collection period for most sites included years prior to the COVD-19 pandemic. As a result, SARS-CoV-2 codes may be underrepresented compared to the peak of the pandemic and overrepresented compared to other time periods.
Where to find the rankings
The Top 20,000 LOINC terms have their relative rank stored in the COMMON_TEST_RANK field of the LOINC table, which is handy if you're working directly with that file. You can find this column in search results on SearchLOINC. Also, RELMA has a "common lab tests only" filter that will limit searches to return only codes in this common test list.