Background/Aims: Assuring that complete and correct laboratory test results are brought into the VDW and associated with the correct Test_Type is an on going task. Many problems can occur. For example, a particular version of a test can incorrectly be left out when the data are extracted
Results: can be duplicated, or results can be associated incorrectly with a Test_Type. Creating quality assurance checks to locate problems or validate data requires evaluating the same data record from several different angles.
Methods: We conducted quality assurance on numerous laboratory test results that had been incorporated into VDW format across as many as 11 HMORN sites. Quality assurance programs were written to provide counts of test results across sites and over time. These programs also detailed result values, result units, patient location, result location, and other specific metrics. Descriptive statistics and graphical displays were used to facilitate assessment of across-site data variability and identify potential data quality issues. Sites were provided their site-specific results and results for all participating sites.
Results: Masking site information, we will show examples of quality assurance checks and depict what type of problem each is designed to point out. Examples may include incorrectly mapped tests (e.g., “hemoglobin” test results where the result unit was in percent and was determined to actually be glycosylated hemoglobin results) and tests that require investigation by the site because the result unit is possible, but unusual (e.g., total cholesterol units of g/dl require verification of g/dl unit and conversion to mg/dl before use or verification that g/dl was entered or read incorrectly). Other examples can include the volume of test results now contained within the VDW (e.g., total cholesterol total n is over 26.6 million results).
Conclusions: Laboratory result quality assurance requires assessment of problems as well as verifying expectations and standardizing data presentation. Multi-site comparisons, comparisons of multiple associated tests at one site, and in depth cross tabulation of single site data play a part in assuring accurate, complete laboratory data.