Basic Concepts in Laboratory Testing
To understand the appropriate use of any diagnostic test, you need to know some basic terms and concepts. This article reviews that information.
A diagnostic test is any test that indicates whether or not a specific disease is present. Because this is important information, everyone is interested in knowing if a given test is accurate or reliable. Tests that are unreliable or inaccurate cannot be trusted and should not be used to evaluate patients.
Reliability and accuracy have different meanings. Reliability looks at whether results are reproducible, another term for this is precision. Repetitive testing of a specimen should give consistent results. Differences in lab equipment, personnel or procedures may cause minor differences in the results but the variations should be insignificant. If results vary significantly, then the test is imprecise and there will always be uncertainty about its results.
Accuracy is the ability to determine what is "true", to find disease when it is present and not find it when it is absent. The accuracy of a test is based on its sensitivity and specificity. Sensitivity is the ability of a test to identify all patients with the illness in question and specificity is the ability to identify only those who have it. Here's an analogy explaining the difference:
Imagine you're an usher at a popular event, determining who in the lobby will be allowed into the auditorium. Your decisions are based on your boss's instructions. If you're told to let all the blue-eyed people in, leaving none in the lobby, it's likely that those with brown eyes will stay in the lobby and blue-eyed people won't. But, to avoid leaving some blue-eyed people out of the auditorium, you mistakenly let some with green or grey eyes in. In this case you're a sensitive test.
Let's say the boss changes his mind, telling you to only let blue-eyed people in. The brown-eyed will remain in the lobby but so will those with green or grey eyes. And, because your goal is to keep the non-blue-eyed out, when you're uncertain about the color, the person isn't let in. This results in some blue-eyed people remaining in the lobby. In this scenario, you're a specific test.
An ideal test would be both sensitive and specific, producing positive results for every patient who is ill and negative results for the well. There are very few ideal tests.
Sensitivity and Specificity, in detail
This grid shows the relationship between a disease and its test.
|Disease Present||Disease Absent|
|Test (+)||True Positive||False Positive|
|Test (-)||False Negative||True Negative|
Sensitivity is interested in identifying all patients with the disease so it is concerned with column 1. It is usually stated as a percentage and can be expressed as:
% Sensitivity = (True Positives / (True Positives + False Negatives)) x 100
From the perspective of sensitivity, false negatives are undesirable, implying the disease is absent when it really is present. Because highly sensitive tests rarely generate false negatives, a negative result essentially rules-out the disease you're testing for.
Specificity is interested in identifying those without the disease so it is concerned with column 2. It is also stated as a percentage and can be expressed as:
% Specificity = (True Negatives / (True Negatives + False Positives)) x 100
From the perspective of specificity, false positives are undesirable, implying disease is present when it really is absent. Because highly specific tests rarely generate false positives, a positive result essentially rules-in the disease you're testing for.
A test is only as good as its sensitivity and specificity. To calculate specificity, the test is run on those who don't have the disease in question. Samples come from healthy volunteers and people with similar diseases and should give negative results because the disease is absent. The actual results are plugged into the specificity equation, giving an overall value for specificity. Calculating sensitivity requires samples from people known to have the disease; these should be positive. The test is run and the results are plugged into the sensitivity equation, generating a value for sensitivity.
Most lab tests are not simply reported as "positive" or "negative", usually the result is reported as a specific value. This means that a test's sensitivity or specificity can be increased by manipulating the cut-off point for a "positive" test. But increasing one lowers the other. This teeter-totter diagram may help you visualize this concept.
If failure to identify a disease leads to significant consequences and treating it is safe and easy you would be willing to sacrifice specificity to gain sensitivity. But if a disease causes only moderate problems and treatment has the potential for causing significant harm, you would desire a highly specific test so that only those with the illness would take on the risks associated with treatment.
There is one more point to consider. Laboratories want to complete testing and report results back to the ordering physician as quickly as possible. Using tests with faster "turn-around" frees up time, allowing labs to run more tests overall.
A Stepped Diagnostic Approach
There are situations where you want to identify all patients who have an illness while at the same time treating only those who truly have it. Here it makes sense to run a sequence of tests. The test used in Step 1 should be highly sensitive, yielding very few false negatives but potentially producing several false positives. The second step is crucial and only run on the "positives" from Step 1. The test should be highly specific, being positive only for those who truly have the illness; this will eliminate the false positives generated by Test 1. But, if Test 2 sacrifices sensitivity to achieve high specificity, some of the true positives from Step 1 will be falsely eliminated in Step 2 and the approach will not be useful to patients and physicians.
Let's apply this approach to the usher analogy.
Now imagine your boss instructs you to sort the crowd so that when the show starts all the blue-eyed, and only the blue-eyed, are in the auditorium. And, because your boss believes "time = money", you need to do this as quickly as possible. If you sort correctly, there won't be any blue-eyed people in the lobby and no one with green, grey or brown eyes will be in the auditorium. Plus, you might earn a bonus for being efficient.
So, you begin by letting all those who aren't brown-eyed into the auditorium, a quick glance is all you need; then you close the doors. Next, you carefully examine those who made it in, sending the non-blue-eyed out. When you're done, those in the auditorium see the show, those in the lobby are sent elsewhere and you've got some extra cash.
Other test properties will be discussed elsewhere; the material in this article should be enough to help you understand why Lyme disease testing is frequently criticized.