...
A third year house staff in emergency medicine is preparing a quality of care project to meet certification requirements of the ACEM. In collaboration with a dozen of their colleagues at sister institutions they will compare opioid testing results on patients presenting to the ED with recurring abdominal pain. They wish to be inclusive of any urinary or blood results for any opioid-receptor agonist agent. They plan to identify whether any testing was done within 24 hours of the ED visit and to classify the results as COMPLETE, NOT DONE, OPIOD PRESENT or OPIOD NOT DETECTED. The project will evaluate the effects of the screening data on the expense and outcomes of the ED evaluation.
Clinical Scenario 3 Use Case
A patient with severe COVID-19 is transferred from an outside hospital. Laboratory results from the outside hospital, available through the EHR vendor’s HL7 integration engine, included a troponin I of 4.0. The troponin result was trended with previously reported troponin results from the in-house laboratory. After 24 hours, cardiac decompensation is observed including venous distension and increased peripheral edema. A follow-up specimen is collected and (high-sensitivity) troponin I is measured and reported as 7,000 ng/L.
Numerous clinical assays lack analytical harmonization. Consequentially, test results, reference intervals (normal range), and measuring units are often not comparable between various assays and subsequently between laboratories. In the scenario described above, the troponin of 4.0 was using a 4th generation assay that reports troponin in ng/mL. In contrast, the high sensitivity, 5th generation troponin assays typically report in ng/L. By harmonizing the units, the patients initial troponin would be reported as 4,000 ng/L. Using high-sensitivity assays, 4 ng/L are inconsistent with myocardial injury, and results >1000 ng/L are consistent with a considerable myocardial infarction. Because the external and internal test results were co-mapped in the EHR, the units could have been absent and assumed, displayed deceptively as that of the other test, or displayed appropriately but missed by the busy clinician. Interpretation of results in this case are further complicated by lack of analytical harmonization between troponin assays. Even if the same units were used, the physician did not know the method for the previous result; in many clinical scenarios this would prevent a clinician from discerning if an acute myocardial injury had occurred (rapid change in troponin), or if these differences were simply assay dependent. Overall, in this case it was easy for the treating physician to miss the apparent cardiac injury because the test method and units were not clearly defined.
In the clinical scenario described above, the treating clinican would have clearly benefitted from knowing: 1) the method used for assessing troponin concentration (including the generation of assay used) 2) the units that the results were reported in 3) the reference interval from the reporting laboratory.
Because of the risk of this specific error for this specific test, much has been done to decrease the change of this outcome, including the creation of distinct LOINC codes for 4th generation and 5th generation assays and the generation of specific guidelines for troponin assay implementation. However, there are many other laboratory tests that are poorly analytically harmonized that do not have such safeguards in place.
Research Scenario Use Case
A research faculty at U of Anywhere is designing a study to compare the sensitivity, specificity and positive predictive value of two novel screening tests for COVID-19 respiratory infection in patients with diabetes mellitus. The test strategies are nasal swab for SARS-CoV-2 N gene versus salivary SARS-CoV-2 N gene. Diabetes mellitus is defined by a history of random non-stimulated blood glucose >= 200mg/dL or Hemoglobin A1C/Hemoglobin ratio >6.5%. The research team plans to conduct this as an observational trial across the N3C network and they are struggling to define the value sets of coded test results which they will incorporate into their SAS query for cohort identification and outcomes assessment.
Quality 1 Use Case
Goal: Real-time surveillance for test performance deficiencies (proficiency testing)
Current state
Methotrexate is a chemotherapy commonly used to treat cancer and autoimmune disease. It has a narrow therapeutic range and is frequently monitored to adjust dosing and avoid toxic concentrations.
To help ensure accurate test results, clinical laboratories participate a few times a year in external proficiency testing (PT) whereby they compare their test results to those of peers. On one laboratory’s most recent PT survey, three samples of five showed a clinically significant negative bias relative to their peers. The root cause of the negative bias was ultimately determined to be incorrect automated dilution. The identification and correction of this deficiency occurred more than 30 days after its onset. In the interim, many patients results were falsely increased, leading to potentially toxic methotrexate dosing.
Situations like the one described above are not uncommon in laboratory testing and PT represents an important way for laboratories to detect systematic errors. Unfortunately, results from most PT surveys take weeks to months to process.
Potential state
Improved interoperability of laboratory data could enable much more rapid detection of such testing deficiencies, preventing considerable downstream patient harm. Some such errors could be detected by systematically surveying multi-site patient test results and changes in these results, comparing across laboratories performing the same exact test. Some such errors could be detected by comparing results for QC samples, which are tested as part of standard of care. Some such errors could be detected more rapidly by modernizing the PT process by leveraging the aggregation of clinical information streams. Such approaches would reduce the burden of paperwork, potential human error in recording results, improve efficiency and ultimately improve the standardization of laboratory results.
These surveillance approaches would require collation of laboratory test results inclusive of UDIs for instruments and assay. Using patient test results would benefit from clinical and ordering context, such as provided by standardized order and result codes. Using QC samples would require transmission of QC lot information. All of these approaches would benefit from test component (e.g. reagent) lot information.
Quality 2 Use Case
Goal: Earlier and more sensitive detection of laboratory test component (reagent, calibrator, container) deficiencies. Some testing device defects are not readily detectable by existing, standard manufacturing and process controls. Surveillance of aggregated laboratory testing data could speed the detection of such defects, thus decreasing the potential harm. In addition, this higher resolution labeling of test results would facilitate better cleaning of data for additional secondary uses.
...
Class 2 Device Recall of specific collection transport media for SARS-CoV-2 Antigen testing (Recall Number: Z-1266-2021). Specific viral transport media led to false-positive results.
Class 2 Device Recall of specific Unconjugated estriol reagent lots (Recall Number: Z-3006-2020; 08/26/2020). Specific formulation of reagent lot led to falsely elevated results in a subset of patients, potentially leading to false-positive prenatal screening results for Down Syndrome.
Class 2 Device Recall of specific Parathyroid hormone reagent lots (Recall Number: Z-1342-2021; 11/23/2020). These specific reagent lots had potential to ‘produce falsely elevated…results’, potentially leading to false-positive diagnoses for hyperparathyroidism.
Class 2 Device Recall of specific Rheumatoid Factor calibrator lots (Recall Number: Z-0283-2020; 08/09/2019). These specific calibrator lots caused falsely low results, potentially leading to missed clinical diagnoses.
Recall of specific Creatinine reagent lots (11/2020). These specific lots led to false-increase in results for serum (but not plasma) specimens, which directly led to inappropriate referrals and biopsies.
Explanation:
Laboratory tests are performed using several components, including specimen collection devices and reagents, test reagents, and test calibrators. Components are labeled with device identifiers and lot numbers, which indicate the batch in which they were prepared. Together these identifiers comprise the Universal Device Identifier (UDI). Not all potential combinations of devices can be evaluate pre-market. Performing laboratories should locally verify specific collection devices and reagents, but these practices are variable and limited. Specific component lots are not evaluated pre-market. For some assays, batch-to-batch variability can be substantial and clinically impactful. Manufacturers verify each batch, but typically using few or no patient samples. Clinical laboratories verify batches, but practices are variable and use few to no patient samples. These processes are not designed to detect errors that affect some but not all patients.
...