I am a former Laboratory Director and Clinical Lab Scientist (a.k.a licensed Medical Lab Technologist). I was in the field for forty years. A little knowledge for everyone about lab testing:
There is inherent "variation" in lab test results no matter the method or instrument (brand and model). It has been generally accepted for decades that any test result that varies from any other test result on the same patient by 10% or less is "clinically insignificant", and can be attributed to that day to day, hour to hour, inherent method and instrument variation. However, test results found to be greater than 10% variation from the previous result on the same patient, are considered "clinically significant," and warrant repeat testing and investigation.
Laboratories have clinical lab information systems (computer systems) with built-in quality control "rules", known as the Westgaard Rules, developed by Dr. Westgaard and now carried on by his son. The rules are built into the computer system, and when a result on a patient varies by greater than 10% or the quality control test results vary by 10% from what is expected, it triggers an alert that results in an investigation as to why the variation occurred. In fact, clinical lab specimen testing is so accurate and reproducible, that it results in the fewest lab errors of all. "Pre-analytic" errors (human handling of the specimen before testing) and/or post-analytic errors (computer systems, or human errors reporting the data) are the cause of the majority of lab errors now and for the passed several decades. The analytic instrument testing phase is nearly perfect.
Two things are continuously monitored in the lab testing environment with quality control materials and methods: Accuracy and Precision. Think of a target with a bullseye. The "bullseye" in the middle of the target is the true result. The test method (chemical reagents and instrument) seek to "hit" that true value "accurately" through calibrating the instrument and method with known chemical "standards" and quality control material that is like patient serum. "Precision" is reproducibility: can the method repeatedly yield consistent results? You can have a test method and instrument with high precision (very reproducible), but if it isn't "accurate" hitting the true value, then it is meaningless. You need a new method or you need to recalibrate the instrument. You could hit the same "outer ring" of that "target" consistently, and say the method and instrument have good "precision" (reproducibility), but it could be very "inaccurate", well away from the "true value" (bullseye). You could also have a method and instrument that produces results all over the target in every ring and all over the target. That method would be neither accurate nor precise. So, you know where this going. You want a method that is "accurate" and hits the true, expected value, AND, you want it do so consistently with reproducibility, which means it is "precise." So, if you had 10 "arrows" (testing events), you would want all ten arrows to hit the middle bullseye to show it is an accurate and precise (reproducible) test method.
Through careful comparison and evaluation, labs select and use a certain instrument and method. They calibrate the instrument at a frequency per the manufacturer's recommendation, as well as data in the lab that evaluates the calibration stability. Then depending on the stability of the method, "high", "normal" and "low" quality control material that mimics human serum or plasma, is run once per day, twice per day, or on each lab shift...sometimes it is run each analytic run of one or more specimens. If the quality control test values are within expected range, then the lab knows that the test run of patient samples will be valid. If the quality control test values are outside expected values, then the test run is rejected based on the Westgaard Rules used by the Lab, and the situation is investigated as to why? It could be an instrument functionality issue (examples): an incorrect specimen sample volume aspirated into the test cup for testing; or...an incorrect amount of chemical reagent was dispensed into the testing cup for the test run; or...the instrument has an electronic malfunction due to heat, humidity, or other issues; or the chemical reagent lot # was bad or the vial exceeded the expiration date. As you can see, there are many potential causes of lab testing errors, and you can see this is a massive issue for every lab. Many labs have quality control techs who analyze data constantly to catch errors or "bias" and recognize that a statistical "trending" or "shift" is occurring that signals a problem. Every instrument has these features built into them as well, along with the Westgaard Rules. Every Lot #, expiration date, date of rehydration or preparation of the quality control material, chemical reagents, or specimens is monitored. Finally...
This gentleman's comparison of the two test values is inaccurate. He had his blood tested by two different labs using TWO DIFFERENT METHODS AND INSTRUMENTS. This is an "apples to oranges" inaccurate comparison even though the test was a "PSA" test...one being the ultra-sensitive method. The only way that comparison would be accurate is if the two different labs - LabCorp and Quest - used the "same exact instrument" and "test method", and ideally were using the same Lot # of chemical reagent, and tested a portion of the "same blood sample" (split the sample and sent half to LabCorp and half to Quest). Only then could as close of a comparison be made. I hope this helps somewhat in your understanding of specimen collection, quality control, instruments, and accuracy vs reproducibility in clinical labs!