What Do Lab Tests Really Mean?

Lab Test Analysis

All forms of health care practice have made great advances in the last thirty years. Laboratory testing has also improved, both in the range of tests available and the rapid turnaround-time of results. But no lab test is perfect. Tests may miss the presence of the disease or may falsely suggest a patient has a particular disease. Either situation may create serious problems. For example, MRIs are an important tool. They reveal more information than ever before. But this comes with a price. Not every suspicious shadow on an MRI means something. Experience in interpreting MRI images is gained over time. Often technology becomes available well in advance of our ability to accurately interpret the results. Lab tests need to be interpreted with caution. Accurate diagnoses need supporting evidence - in terms of patient history, physical signs, and symptoms - in addition to lab results.

You're at the doctor's office because you think something might be wrong. Rationally, you know tests are probably necessary, but getting the tests done sometimes provokes a lot of anxiety in all of us. We want to know the results, but are very concerned about the outcome."You need blood work" your doctor remarks casually. Your insides do an immediate flip-flop and you feel as if you've just begun hurtling down a very steep roller-coaster.

And when the results come back,  it's tough to understand the medical jargon, particularly if the tests are "positive". We've all had the experience of "going south" and not listening to another word that's said after the initial "your test results are positive". Our minds are racing ahead, imagining all the awful possibilities.

Rarely, a considerate physician will be able to put the lab results in a less-worrisome perspective for the patient. For patients, it's important to know some key facts - facts that will empower you any time you need lab tests done.

First, statistically one out of every 20 tests performed will be reported as "abnormal" - even though the result is "normal" for the person being tested. If you are perfectly healthy and have 20 tests done on your blood sample, one of those 20 test results will be "out of the normal range" based on statistics alone.¹

This is because a normal test range is calculated by eliminating the top 2.5% and the bottom 2.5% of results for a large group of  normal people. This leaves the "normal range" as 95%. So if 20 tests are done, statistically one will be "abnormal", because its results fall in the top or bottom 2.5%. The result is normal for you, but reported as "abnormal".

Other factors to consider are the sensitivity and specificity of the lab test.² Sensitivity relates to the test's precision in detecting the disease when the disease is present. If a test for colon cancer was 90% sensitive, it would miss ten cases of the disease out of every 100 cases. Specificity relates to whether a positive test actually indicates the presence of the disease you're being tested for. If a test is 90% specific, in ten cases out of 100 positive results, the patient does not actually have the disease.

So if a test has low sensitivity, actual cases of the disease may be missed. If a test has low specificity, test results reporting the presence of the disease may be wrong. All these factors need to be considered in accurately interpreting test results. Things are not always as they seem.³

Bottom line - test results always need to be evaluated in the specific context of the patient. An isolated lab result - or set of results - needs to be related to the patient's condition and circumstances. An accurate diagnosis is not based on lab results alone. That could be a big mistake. Similarly, x-ray, ultrasound, or MRI results need to be interpreted in context also.
Trying to interpret an x-ray without any clinical information is likely to lead to a wrong conclusion. Context is everything. If you have questions, we will be able to explain why certain tests are necessary.

¹Thomas SL, et al: How accurate are diagnoses for rheumatoid arthritis and juvenile idiopathic arthritis in the general practice research database? Arthritis Rheum  59(9):1314-1321, 2008
²Friston KJ, et al: Classical and Bayesian inference in neuroimaging: theory. Neuroimage 16(2):465-483, 2002
³Kobayashi M, et al: Intraindividual variation in total and percent free prostate-specific antigen levels in prostate cancer suspects. Urol Int 74(3):198-202, 2005