Correlations
Also known as Proficiency Testing. One of the accreditation requirements of laboratory and POCT services is to ensure that similar tests give reasonably similar results. It would be of little value to the clinician if two instruments gave two totally different answers for the same test.
POCT operators can minimise errors in a number of ways: an awareness of the pre-analytical phase of testing is most important. Most errors are caused in this phase.
However, while the accuracy and precision of each instrument is sustained by routine QC analysis and maintenance schedules, correlation samples ensure that each instrument is matched against the other and that all give similar results.
The POCT Coordinator analyses correlation samples regularly:
- A sample is selected for suitability. It has already been analysed in the lab for the test(s) in question.
- It is taken to where each POCT instrument capable of measuring the same test, is located.
- The sample is then analysed.
- Results are compared to each other (if more than one instrument), and to the lab.
Most correlations are made on a weekly basis, although some are compared monthly.
Correlations cover the following POCT processes:
- blood gas analysers, including hand held models, eg i-STAT 1
- Clinitek 50 urine reflectance meters
- DCA2000 glycated haemoglobin analysers
- glucose meters
- Haematology analysers, eg, Beckman Coulter AcT-8,
- Hemocue haemoglobin monitors
- Urine pregnancy tests
(Haemostasis analysers, e.g., Hemochrons and INR analysers, are not able to be correlated easily due to the sample requirements: these analysers test the clotting time of fresh whole blood- the sample must be tested straight from the patient's vein or finger. Quality Control samples are analysed monthly on these analysers by the POCT Coordinator).
Laboratory values are considered the true value or benchmark of correlation studies. The results of samples analysed by POCT are compared or correlated to the lab results.
Generally speaking, results should not deviate more than ±5%, i.e., the results should lie within the ranges of 95% to 105% of the lab value. However, for glucose meters, a wider range of 90% to 110%, especially for lower value samples is considered acceptable. (NCCLS recommendations are up to ±15% of the true value, e.g., 85% to 115%).
All results are documented and available for inspection. The POCT Committee regularly reviews QC, external QC and correlation studies. Please contact the POCT Coordinator to view results.
