Dr. Harriet Hall at Science-Based Medicine recently reviewed the book Over-diagnosed: Making People Sick in the Pursuit of Health by Dr. Gilbert Welch. I was immediately interested enough to buy the book and read it, largely because it promised to address a subject I’ve discussed here before, the dangers of unnecessary or inappropriate screening tests and the irrational and counterproductive reaction that often occurs when scientists try to dissuade people from employing or taking such tests.
The book was well worth reading. Dr. Hall’s review covers the book’s contents in detail, so I won’t repeat all of that. In brief, Dr. Welch makes a strong, evidence-based case for several fundamental concepts:
1. The more sensitive our criteria for using a lab test or imaging method to diagnose a disease in someone without actual symptoms, the more likely we are to identify abnormalities that will never cause actual clinical problems.
2. This then leads to people being identified as ill who aren’t, and these people are then subjected to testing and treatment which will not help them, and may actively harm them.
3. There are a number of factors which drive and perpetuate this over-diagnosis:
b. Poor understanding of the meaning, use, and misuse of statistics
c. Uncritical, superficial, and excessively dramatic media coverage
d. Money (industry profits, political contributions, research grant allocation practices, fundraising of non-profit lobbying organizations etc)
e. Lawyers and litigation, which punishes missing a diagnosis but not over-diagnosis and unnecessary or harmful testing and treatment
f. Uncertainty about the risks and benefits of specific tests and interventions
e. All the usual cognitive biases we have, especially confirmation bias and the post hoc ergo propter hoc fallacy
4. Only objective, systematic data from randomized clinical trials and good quality epidemiological studies can truly tell us whether screening tests for a specific condition help or hurt more people. As usual, anecdotes about individual patients are emotionally compelling and totally unreliable in making individual or public health decisions about disease screening.
The examples Dr. Welch provides are based on enviably comprehensive data and analysis. A surprising number of people without disease will have detectable abnormalities on blood tests, x-rays, and other screening diagnostics. And for almost every condition studied in detail, including many cancers, the majority of these people will never become ill from these abnormalities. The more sensitive our screening tests become for cancer, for example, the clearer it becomes that we all get cancer repeatedly and never know it. Just like we all take in potentially disease-causing organisms all the time. Most of these cancers never develop just like most of the bacteria and viruses we are exposed to never make us sick. So finding these cancers early and treating them does us no good at all. And apart from the anxiety of knowing we have cancer, procedures such as biopsy and surgery and cancer treatment can do very real harm.
Of course, the difficulty is that we do not, as yet, have a trustworthy way to know when we find a small cancer whether it will ever develop into a real problem if we would benefit from treating it. This makes it difficult to resist looking and treating such abnormalities, even when the evidence is overwhelming that most people don’t benefit and that more are harmed than helped. Psychologically, we are far more afraid of the error of failing to respond to a true threat than we are of over-reacting to a harmless abnormality. Yet if we look at the issue rationally, we are far better off not employing screening tests that are clearly shown by objective data to lead to more over-diagnosis and harm than benefit.
The issue arises in veterinary medicine just as it does in human healthcare, though as always the quantity and quality of the evidence is far less and the degree of unavoidable uncertainty far greater than in human medicine. And, the limitations on resources and lesser significance of litigation alter the equation somewhat. But the general principles are still broadly applicable to the veterinary setting.
Just the other day I had a client with an older cat in for an annual exam and a followup on a known heart condition. The client asked me to perform whole-body x-rays of the cat as part of my workup. When I asked why, he showed me a book on “natural health” for dogs and cats by a prominent “holistic” veterinarian that recommends annual screening x-rays of older pets.
Apart from the obvious question what is “natural” about x-raying your cat every year, the recommendation is irrational and likely to lead to more harm than good. According to Dr. Welch, 86% of people with no signs of illness will have a lesion on a whole body CT scan. Chest x-rays will often find lung masses, and CT and MRI will find even more. And the vast majority of people without symptoms who have such lesions on random imaging screening will never experience any illness from them. But a clearly larger number of people will be harmed by additional testing and treatment, experience painful and unnecessary anxiety, and use up limited healthcare resources pursuing such findings. And though the radiation risk from a single x-ray is minimal, annual radiographs can expose the patient to small but unnecessary risk from radiation.
Similar data is not available in companion animal species, but there is no reason to believe the fundamental rules of biology are different for our pets than for us. The little research that has been done, such as studies of pre-anesthetic screening blood tests in dogs, does not generally support the value of such tests, but the data are still terribly limited. Nevertheless, the general principles and issues Dr. Welch illustrates so well certainly need to be considered in making decisions about screening tests in veterinary species.
I explained all of this to my client, who opted not to have the x-rays done. A surprising amount of my communication with clients involves convincing them not to do things, in contrast to the image CAM proponents often paint of conventional doctors aggressively selling unneeded drugs and treatments. And it is surprisingly difficult to reassure people with facts and statistics and convince them not to pursue unneeded diagnostics and treatment. And, unfortunately, reassuring them by performing meaningless rituals or providing placebo therapies works much better, which simply perpetuates ineffective therapies.
I highly recommend this book, both to potential patients (which is all of us) and to doctors. The conventional wisdom, that it is always better to look for trouble “just in case” is being shown more and more clearly to be wrong. Diagnostic testing, when there are symptoms or a sound rational reason to suspect disease, are absolutely vital. But screening tests for the apparently well are often far less likely to benefit us than is generally supposed, and they do real harm. The burden of proving such tests are a good thing has clearly shifted to those advocating them.
It seems unlikely that there will be a significant shift in the attitude of the public or doctors towards screening tests in the near future. The U.S. Preventative Services Task Force (USPSTF) is a government agency devoted to examining the scientific evidence concerning preventative healthcare testing and intervention, and they often provide sound, rational guidelines. These guidelines, are also often ignored by politicians, the public, and sometimes even healthcare providers. Our fears are still generally more persuasive than our reason. Still, reason and facts can glacially wear away at our fears and misconceptions over time, which is why I still write this blog and why I am happy to find and recommend excellent books such as Dr. Welch’s.