Making a proper diagnosis is a crucial step for physicians in developing a rational treatment plan to either cure disease or relieve suffering. Sometimes, using modern imaging technology and laboratory testing, diagnosis is a straightforward process. But other times, despite extensive testing, no coherent picture emerges from patients’ complaints. In her book, The Age of Diagnosis, Dr. Suzanne O'Sullivan, a neurologist, explores the complexities and potential pitfalls of the diagnostic process. She tries to make the case that there is much over-diagnosis in contemporary health care. She contends that many widely used designations like long COVID and chronic Lyme disease are labels given to vague and unrelated symptoms that are psychosomatic in origin yet have been used to satisfy patient anxiety and physician need for certainty.
Intuitively, it would appear sensible to secure a specific diagnosis for a clinical problem. We assume that the sooner we make a diagnosis, the sooner treatment can start, and the better the outcome. But very early diagnosis of a clinical condition can also lead to a diagnostic trap. Let's say we have a test that can detect the presence of a few cancer cells in the blood without any overt manifestation of clinical disease (such tests exist at this point). But we also know that many times, a few cancer cells will appear in an organ and then either remain quiescent or disappear with time. The search for those cells can lead to anxiety, potentially invasive tests, and even biopsies when the likelihood is much greater that they may never cause a clinical problem. There are several examples of such conditions. Both thyroid cancer and prostate cancer occur in many more patients than die from either prostate cancer or thyroid cancer. Finding these cancers through screening processes that are ever more sensitive can suggest a growing frequency of such cancers when, in fact, all that is happening is more frequent finding of cancers that previously would have been missed and may never cause a clinical problem. Screening tests can thus lead to over-diagnosis.
Screening tests are best used in patients with a high risk for the disease in question. That is the best way to avoid over-diagnosis. If a person is at low risk for a cancer, say a 1 in 1,000 chance, and if the false positive rate for the diagnosis is 5 in 1,000, then any positive result has an 85 percent chance to be an error. The greater the risk the patient has the disease (for example, lung cancer in a heavy smoker), the greater likelihood that a positive screening test is a true finding. A false positive result on a chest X-ray can lead to a risky and unnecessary invasive procedure like a lung biopsy. This is the danger of over-diagnosis.
Is there a solution to the problem of rational use of screening tests to avoid over-diagnosis? In O’Sullivan’s mind, a close working relationship between a trusted physician and patient will be the key to solving the dilemma. But the dilemma can only be solved on a personal basis. The highly anxious person may never be satisfied without complete knowledge of their clinical condition, and a 10 percent probability that they might die from prostate cancer will not be overcome by the 90 percent probability that they will outlive the cancer. But like most treatments, the therapies themselves have substantial risk and downsides. Prostatectomy is a major surgical procedure that can cure prostate cancer but can lead to chronic problems such as impotence as well as urinary incontinence.
O’Sullivan believes that certain diagnoses—like chronic Lyme disease—that have been promoted as real entities have trapped patients in expensive and potentially dangerous treatment protocols when the diagnosis itself continues to be of dubious value. Long COVID may be another example of such a condition. Most of the patients with long COVID were not those critically ill but rather had mild or even no evidence of acute COVID. Many viral illnesses can produce prolonged symptoms without needing a specific label. Once the label is created, patients may convince themselves of other symptoms of the "disorder" and actually become physically ill. This phenomenon is called a nocebo—the mirror image of placebo. Once labeled, these patients seem to develop a multitude of unrelated symptoms.
Long COVID and chronic Lyme disease each lack a defining diagnostic test or proven therapies. Yet there are clinics that will provide months or even years of intravenous antibiotic therapy for chronic Lyme disease or a variety of unproven therapies for long COVID. O’Sullivan’s lucid discussion of a patient who fell into the trap of self-diagnosis of chronic Lyme disease well illustrates this clinical dilemma. She uses such vignettes to great advantage throughout her book.
I did find it surprising that she did not include the most contentious diagnosis of our time—so-called gender-affirming care in children. This has become the prototypic "new disorder" spawned by social media and transgender activists and suddenly skyrocketing in incidence. And there are other examples as egregious. Several years ago, the kidney disease authorities created a new formula to measure kidney dysfunction, and suddenly 30 million Americans were told that they suffered from an early stage of the five stages of "kidney failure." What they typically identified as kidney failure was basically a "disease" of aging (kidney function simply declines over the years). Imagine creating a new, five-stage muscle disease if a lab test could easily measure muscular decline as we age. Over-diagnosis of "kidney failure" is now a staple of most primary care physician’s practice, yet only a minimal number of patients ever demonstrate progressive kidney failure to the point of needing kidney replacement therapy like dialysis or transplantation.
Genetic testing for hereditary diseases poses a specific problem for both patients and physicians. Some diseases invariably track with a single genetic variant. Huntington's disease is the result of a single genetic abnormality that is passed to 50 percent of the offspring of an individual carrying the gene variant. The tragedy of Huntington's disease is that it does not manifest itself for the most part until later in life. The dilemma is whether to know in advance about whether one carries the gene for this disease even though there is no treatment for the lethal and devastating neurologic disorder. This becomes a real-life version of the curse of Cassandra, who knew when she would die and when Troy would succumb to the Trojan horse, but no one would listen. It's even more complicated: Knowing that one is a carrier and already having offspring means not only that the offspring need to deal with this dilemma, but the parent carries the burden of guilt of passing along this lethal disease. O'Sullivan deals with this tragic quandary through a patient vignette that introduces the dizzying complexity of such decisions. This is not really an example of over-diagnosis, but she writes movingly about the dilemma of making a diagnosis that cannot lead to any therapeutic benefit.
There are other dilemmas of genetic testing where the answer is not black and white. Testing for the gene that conveys an increased risk of breast cancer confronts many Ashkenazi Jewish families since that group is a major carrier of a gene variant that raises the risk of several cancers, including breast, pancreatic, and ovarian. The problem here is that the risk may be as little as 15 percent or as high as 50 to 60 percent depending on the specific gene variant in question. While more research is fine-tuning this information to give greater precision to the genetic risk, again, a difficult and often emotionally devastating preventative measure like bilateral mastectomy is the only way to prevent breast cancer. The patient's tolerance of risk will determine the approach. As more and more genetic risk factors for cancer are defined, this dilemma may involve ever increasing numbers of patients.
O’Sullivan makes a convincing case that over-diagnosis is a major problem. Creating labels for a variety of symptoms that are most likely psychosomatic produces newly identified "diseases" that require expensive treatments and months or even years of disability. Other disorders like ADHD have criteria that are so vague that virtually everyone might suffer from it at some time or another. Her other messages, however, are that screening for genetic diseases for which there is no cure is problematic and there is dubious benefit of early screening for cancers.
The case studies are illustrative and capture the dilemma associated with making a diagnosis—be that of a psychosomatic syndrome labeled as an organic disease or a cancer screening test for a patient with a low risk of a true positive result. Readers who contemplate a yearly physical with multiple screening tests and imaging studies can benefit from her clear explanations of the problem.
The Age of Diagnosis: How Our Obsession with Medical Labels Is Making Us Sicker
by Suzanne O’Sullivan
Thesis, 320 pp., $32
Stanley Goldfarb is an emeritus professor of medicine at the University of Pennsylvania and father of Washington Free Beacon chairman Michael Goldfarb.