Trisha Greenhalgh has been one of the drivers in the evidence-based practice movement, and she is author of an excellent book entitled “How to read a paper: the basics of evidence-based medicine.” (1) Her text provides a clinician an overview of the basic concepts of EBP, and it is eminently readable and easy to follow. In one of her chapters she looks at diagnostic tests in specific, and she offers the reader a set of questions which can help you interpret the information in such a paper. These include:
1. Is the test potentially relevant to my practice? This is an examination of the utility of the test. Even if you knew the test was 100% reliable, would it actually help you in any way? Could you use it to identify a treatable disorder? And if it does, would it be better for you to use this test than the tests you are already using? Would it affect your treatment plan? If you can answer no to any of these, it might be best for you to not ready further into the paper.
2. Has the test been compared with a true gold standard? You need to ensure that the test has been compared to some other test which is seen by the profession as being the best possible test to use in that situation; for example, if you are looking at an orthopedic test for diagnosing lumbar disc herniation, your comparison test would be MRI. I should note that in some cases, there may not be a gold standard test, so you need to carefully examine what the new test is being compared against.
3. Did the validation study include an appropriate spectrum of subjects? One challenge here is that often, papers examining new diagnostic tests use patients who are truly ill. These patients represent only one part of the full spectrum of patients in which you can do the test and they are the ones most likely to be found positive. This creates what is called spectrum bias; you need to include a wide range of subjects to best study the sensitivity and specificity of the test. Thus, the paper should include subjects with both mild and severe disease, and those with other commonly confused conditions.
4. Has work-up bias been avoided? This question is actually asking whether or not everyone who got the new diagnostic test also got the gold standard and vice versa. You don’t want to read a study where the gold standard test is performed only on people who have already tested positive for the text being validated.
5. Has expectation bias been avoided? This occurs when the people who interpret a diagnostic test are subconsciously influenced by knowledge of the particular features of the case. An example might be the presence of low back pain and its specific manifestations when interpreting an MRI. Normally, all assessments should be blind and the person interpreting the test should not have any inkling what result is expected.
More to come on this next week…
1. Greenhalgh T. How to read a paper: the basics of evidence-based medicine. London; UK; BMJ Books, 2001:111-113