As a backdrop, most people do not understand the concept of "tests" in medicine until they have a problem and realize that the problem is not reflected in the tests ordered by their doctor. That is a very common experience. Some studies show that up to 30% of patients presenting to a clinic for investigation of a symptom never find out what the cause of that symptom was. That is true even after they were given the usual panel of blood tests, imaging studies, and electrophysiological studies. The assumption that symptoms and disorders in medicine are all diagnosable by a "test" is incorrect.
The second problem occurs at the level of test interpretation. When a doctor orders a test they have to interpret it correctly and in many cases the idea of an "abnormal" test is blurred by biological variation. The evaluation of back pain using imaging studies like MRI scans is a good case in point. As people age there is a greater likelihood that an abnormal MRI scan of the spine is not necessarily the cause of their back pain. That has very important implications for treatment and the physician interpreting the test may will definitely be influenced by their specialty training, their own personal experience, their knowledge and examination of the patient, and possibly treatment guidelines that they may be mandated to follow. There is also the question of false positive and false negative testing. The recent controversy about the utility of prostate specific antigen (PSA) testing for prostate cancer is another good example. The current guideline says that this test has too many false positive results to use for treatment planning and further invasive procedures. Even in the case where the diagnosis is made by a specific number there is always the question of whether the test number is accurate or not. I have frequently repeated thyroid function tests that seemed to show hyperthyroidism only to see them in the normal range on repeat testing. It is obvious to physicians that the so-called biological tests in medicine have their limitations and always need to be interpreted in the context of a comprehensive evaluation of the patient.
How are biological tests currently used in psychiatry? It turns out that there are a lot of applications similar to the rest of medicine.
1. To detect a medical cause of a psychiatric disorder. The DSM classification has an entire set of disorders that are caused by underlying neurological illnesses, endocrine disorders, and infectious diseases that need to be recognized and treated. They often present as psychiatric disorders. In my experience of treating people with severe problems, up to 15% of the psychiatric presentations had an underlying medical illness that either was a direct cause of the "psychiatric" symptoms or it made a psychiatric disorder worse. In that case the psychiatrist has to be trained to order the appropriate tests, make the diagnosis and refer the patient for treatment of the underlying disorder.
2. To screen for medical illnesses that complicate the psychiatric disorder or its treatment. A good example here is screening blood tests and electrocardiograms based on the clinical assessment of the patient and the likelihood that a disease is present. At times patients present with significant problems that require urgent treatment that they are unaware of. A good example would be detecting complete heart block on an ECG because of a patient's responses to the cardiovascular review of systems and the fact that an antidepressant was going to be prescribed.
3. To monitor the safety of biological treatments. There is probably no better example than the FDA focus on cardiac conduction and how that can be affected by medications. The most recent warning occurred with citalopram. This antidepressant has been used for over a decade by psychiatrists and was widely considered to be a very safe medication. Both the FDA and the Mayo Clinic have guidelines about how this problem needs to be assessed and that is a combination of clinical assessment and electrocardiograms. In some cases electrocardiograms and referrals to electrophysiologists are required. In light of this information psychiatrists need to have access to these ECGs and a plan to address any abnormalities. As specialists, it is common to see patients who are referred taking doses of citalopram that exceed current FDA guidelines and that may involve testing and a plan to modify the dose of antidepressant.
4. To identify medical emergencies in patients who are being followed for a psychiatric disorder. Many patients who see psychiatrists either do not have primary care physicians or are very reluctant to see them. A psychiatrist in this position needs to make every effort possible to encourage the patient to establish primary care, but even then medical emergencies need to be recognized and appropriately triaged. That can happen more quickly if testing is available to facilitate the referral. If a patient presents with jaundice, medical consultants are more likely to see him quickly if some basic testing is done that can be discussed with the consultant.
5. To identify neurobiological correlates of psychiatric disorders. A common example is an abnormal brain imaging or electrophysiological study that was ordered because of an acute or progressive behavioral change.
6. For heuristic purposes. The classic example of a test done for heuristic purposes was the dexamethasone suppression test. At various times it was suggested as a test for various forms of severe depression and suicide risk. The test is rarely done today because of the false positive errors but it helped generate a couple of decades of research on the neuroendocrinology associated with psychiatric disorders.
Some of the articles currently out there on the internet deny the existence of psychiatric disorders because there is no biological test for these disorders like depression or schizophrenia. That really does not differentiate psychiatric disorders from neurological or rheumatic disorders that may have some supporting markers but that in general depend more on pattern recognition and less on a gold standard or pathognomonic test. From a paper that addresses that subject (1):
"Although the results are often useful, they can be misleading. Few tests yield results that are pathognomonic for particular diseases. For these reasons, test results for autoantibodies alone are insufficient to establish the diagnosis of a systemic rheumatic disease; they must always be interpreted in the clinical context. Positive results for tests such as the ANA test are seen quite commonly in patients with nonrheumatic diseases and even among normal, healthy persons..."
The key sentence here is: "Few tests yield results that are pathognomonic for particular diseases." That would mean that in fact there are few so-called gold standard medical tests that confirm or deny the existence of a diagnosis. Up to 30% of people presenting to a medical clinic for an evaluation of symptoms will never have a diagnosis to explain their symptoms no matter how many tests they have. People with real illnesses often are frustrated by the fact that the test results they get are often equivocal. Obvious conditions that have no biomarkers range from migraine headaches to Alzheimer's Disease. I don't think that any rational person would suggest that neither of these conditions exist. People who have first hand experience with severe mental disorders know that the profound emotional and cognitive changes that they see in their friend of family member is real - whether there is a biomarker or not.
George Dawson, MD, DFAPA
Reference:
1: Arthur Kavanaugh, Russell Tomar, John Reveille, Daniel H. Solomon, Henry A. Homburger; Guidelines for Clinical Use of the Antinuclear Antibody Test and Tests for Specific Autoantibodies to Nuclear Antigens. Arch Pathol Lab Med 1 January 2000; 124 (1): 71–81. doi: https://doi.org/10.5858/2000-124-0071-GFCUOT
Amen to what you say here, Dr. Dawson. It is a good corrective to the foot-in-mouth statements of Thomas Insel and David Kupfer about psychiatric diagnoses lacking because of the lack of laboratory tests.
ReplyDeleteThe existence of disease is not predicated on having a biological test. There are many areas in medicine where there is no conclusive diagnostic test. Think migraine. Think multiple sclerosis. We identified many diseases before lab tests came along.
Laboratory tests are not the automatic gold standard of evidence for validity that Insel and Kupfer imply. Indeed, in many areas diagnostic tests cause clinical mischief. That’s what the whole debate around indiscriminate diagnostic screening is about. Laboratory measures are the servants of clinical science, not the other way around. The reason for this emphasis is that the great majority of laboratory tests are probabilistic rather than pathognomonic – which means that clinical judgment enters into their interpretation.
Clinical science operates through a process of convergent validation. That is how disease constructs take form, through iterative attention to signs, symptoms, course of illness, response to treatments, family history, and laboratory data. This is an area where DSM-III and DSM-IV failed us by limiting their focus to signs and symptoms. Had the framers of DSM-III understood the original Washington University position, they would have incorporated course of illness, response to treatments, family history, and laboratory data into the manual. Instead, Spitzer rejected these potential validators, saying they might introduce circular logic, and that misstep has been carried forward. However, a positive family carries great diagnostic weight in Huntington’s disease; likewise neurologists routinely use response to l-DOPA as a diagnostic test of Parkinson’s disease (to distinguish it from look-alike disorders).
This process of convergent validation has given us an A-list of psychiatric diagnoses that are candidate brain diseases. Here is the list: psychosis, mania, melancholia, vascular depression, crippling anxiety, panic disorder, dementia, autism, obsessive-compulsive disorder, delirium, catatonia, and more.
The fact that we have not nailed the pathophysiology of these conditions does not invalidate the diagnoses, however. We knew about Huntington’s disease and correctly diagnosed it for 110 years before its genetic basis was discovered. If you want the objective validity that Insel and Kupfer naively aspire to, then by all means tell the magistrate at your next commitment hearing that you were serious when you commandeered an airliner, prevented the scheduled passengers from boarding it, declared yourself the owner of the airline, announced that you were going to fly your entire extended family to London to meet with Margaret Thatcher, and that the psychiatrist who said you suffer from mania must be wrong because he hasn’t shown the court a laboratory diagnostic test for mania.
There is no problem with the clinical validity of most psychiatric diagnoses.
Many thanks for your post Dr. Carroll. The misstatements about the reliability and validity of psychiatric diagnosis associated with the release of DSM5 have been at such a high frequency that for a moment it seemed like I had slipped into an alternate universe. The A-list is certainly valid and interestingly is probably what Dr. Insel's scientists should investigate. It a curious state of affairs to consider that psychiatrists successfully treat thousands of patients with severe problems over the course of their careers. The media and some of our experts seem to be suggesting that is an entirely random process. If psychiatry really was the application of ineffective treatments to persons with diagnoses that lack validity - I am sure I would have become an Internist or a Neurosurgeon about 25 years ago.
Delete