Dr. Mike Merrill

Screening tests may not make sense from the individual's perspective

There's a big difference between how doctors traditionally think about screening and how patients should think about it. 

Let's say I'm a doctor, and I'm considering all my patients as a group. I have 1,000 men that I'm trying to save from prostate cancer. That's about this many guys: 

Source: NYS Archives

I know that if I do PSA screening on all of them, I will save one from dying of prostate cancer. So it seems like a good idea to me, and I start the screening, and do this for 11 years.  (The following is based on the book Cancer Screening Decisions by R.N. Adler, 2018.)

890 of these men will never have prostate cancer. 780 have negative tests. 100 men have positive PSA testing, and then have biopsies. 32 men have complications of the biopsy. 66 men get prostate cancer, diagnosed either from the screening or symptoms. So OK, I have a bunch of testing in this population, and 32 men have complications from the biopsy, but overall I'm saving one life, so that seems worth it. 

Now let's imagine I'm one of the men in the picture. What are the chances I'm going to be the one person saved by the test? Well, literally it's 1 in 1,000. Let's say I worry about it and I want to get the testing anyway. What is the price of the bet that I'll be that 1 in 1,000? PSA testing for 11 years, and a 1 in 10 chance I'll have a biopsy, and if I have a biopsy, a chance that I'll have a major complication like an infection. 

I can tell you that many men, after this discussion, will decide not to have the test. 

There are decision tools out there now that help providers have conversations with patients about this exact problem: identifying the risks and benefits of screening for individual patients. These tools increase patient knowledge and increase their ability to accurately assess risk. 

 

 

Yeah, I'm diabetic and my sugar gets low sometimes. No big deal, right?

Actually it can be a pretty big deal, especially if you are older and you've had diabetes for a while.

Hypoglycemia causes anxiety, tremors and sweating. In more severe cases, you can get confused, lose consciousness, have a seizure, or even die. Think about driving and having all that sneak up on you. Or if you're asleep and don't wake up. 

It's actually not that uncommon. In medical training, we learn that if a patient is found confused or unresponsive, one of the first things to do is check their blood sugar, or just give them some sugar through an IV. 

Considering the risk, the American Geriatrics Society says that for older adults, the benefit of tight glucose control might not be worth it if you're getting hypoglycemia. They suggest hemoglobin A1C goals of 7.0 to 7.5% if you are healthy and have a long life expectancy, 7.5 to 8.0% if your life expectancy is less than 10 years, and 8.0 to 9.0% if you have multiple medical problems and a shorter life expectancy. 

You should discuss your exact goals with your doctor. But you should also know that tighter control of blood sugar is not necessarily better overall. It's another example of "more is not always better" in medicine.

 

 

 

 

Better medical tests: "normal" becomes "disease"

Human beings are smart, and we keep inventing new things. In medicine, that means more and better tests to analyze us in increasingly sophisticated ways. This trend will continue, so we need to get used to it, and be smart about it. 

Sometimes when tests get better, they are able to identify more subtle things. For example, years ago we would diagnose a pulmonary embolus ("PE," a blood clot in the lungs) mostly by using a ventilation/perfusion scan, which gives a rough idea as to whether a large chunk of lung has a blood clot in it. We know that if there is a large clot, the patient needs to be anticoagulated with warfarin, for at least a few months. 

Then we began diagnosing PE's with CT scans. This was an improvement, because they are easier to do, can be done on less stable patients, and give a clearer picture. On a CT angiogram, you can actually see the clot. The following picture shows a really bad one, called a "saddle PE," that stretches across both pulmonary arteries.

(Source:  https://commons.wikimedia.org/w/index.php?curid=53722435). In a CT scan, dark means less dense and white means dense. A dense "dye" is injected into the blood to make it appear lighter. The dye does not go into the clot, so now you can tell where one ends and the other begins.

These days, CT scans are becoming increasingly excellent, to the point where very small clots can be viewed in the lung. But there is a question whether these small clots are as dangerous as the big ones, and whether the patient actually needs anticoagulation. It's currently controversial. 

There are other potential examples of conditions being overdiagnosed by improved testing: heart attack, breast cancer and lung cancer, for example. 

Does this mean we should stop testing? No! Medical care is quite good and quite helpful, overall. We just need to be careful. 

One answer to the problem of new technology is to do different kinds of research. A new technology that sees subtler things should make us stop and ask ourselves if this is still disease. And in fact, these are the questions that researchers are asking now. 

While we wait for the scientific consensus to evolve, it's important to ask questions and weigh what we know of risks and benefits. 

 

 

 

 

Over time, disease definitions are expanded. Is that good?

A great many things are labelled diseases nowadays. "Caffeine withdrawal" is a formal diagnosis, for example. But there are also examples of existing diagnoses getting expanded, such that more people are classified as unwell. Pre-diabetes and "mild cognitive impairment" (basically pre-dementia) would be examples. 

This all creates higher risk of overdiagnosis and overtreatment. And it's worrisome that many of the doctors adjusting these definitions of disease have ties to drug companies. 

Overdiagnosis is receiving a diagnosis that describes something that never would have hurt you, or caused you any discomfort. Overdiagnosis leads to treatment that, at least in retrospect, you didn't need. It's hard to tell if an individual has been overdiagnosed, but you can see it in statistics. One cause of overdiagnosis is this expanding of the definitions of disease. 

How much is this happening? A study looked at guideline updates created by specialty societies and found that most of them expanded disease definitions. Few of them addressed the risk of overdiagnosis, and most members of the expert panels disclosed financial relationships with drug companies. 

From one perspective, and a rational perspective at that, we have developed a self-perpetuating "medical-industrial complex" that seeks to do more and more good, at greater and greater expense, by identifying more diseases and treating them earlier. The blind spot of this system is not looking at trade-offs. One big trade-off is overdiagnosis. 

Let's look at it from another perspective. During my primary care days, I would routinely see 80-year-olds with 15 chronic conditions. Apart from the fact it's hard to treat 15 diseases in 20 minutes, many of these diseases bordered on insignificant, in the greater picture. These weren't things that were going to kill the patient or cause significant discomfort. They represented risks.

For example, "chronic kidney disease stage 3" is important mostly because sometimes, in a small proportion of people, it progresses to stage 4 and stage 5 disease, and to dialysis. But I had several conversations with patients who, after looking at their records, needed to know what was wrong with their kidneys. Is the anxiety worth tracking that problem and putting a label on it? That's part of the trade-off.

Expanding technology has changed disease definitions. Years ago, you would have a heart attack, and that was that. Then we developed the EKG, and the definition became electrical. Now we have sensitive bio-markers in the blood, and they eclipse the previous definitions. 

But then you use the test on healthier people, finding subtle abnormalities. So when you look at someone with severe pneumonia, those heart biomarkers (e.g. troponin) go up a bit. Has this person had a heart attack now? Some would say, "Yes, they had a non-Q wave MI." 

That's how one branch of overdiagnosis works: using technology to see what it can do, without considering the trade-offs. 

 

 

 

Is it OK to say to your doctor, "I don't know what you're talking about"?

Many patients think they will get better care from doctors if they are nice. Some think it's better to be aggressive. Many just want to get away as soon as possible. 

Of the ones who try to be nice, there's a risk that they will not get what they need. That's partly because their politeness leads them not to ask questions, even when they need to. 

The power dynamic between doctor and patient is lopsided. The doctor almost certainly knows more, and has the authority to do all sorts of things that a patient can't. (I was always horrified that while covering ICU patients, I had the legal authority to literally paralyze people.) What does that do to open communication? 

People respond to power differentials differently. There's a dimension of culture called "power distance" described by organizational behavior expert Geert Hofstede. It can be measured, and the measurement describes the comfort that people have with power being distributed unequally. The U.S. is a low power distance country, but individuals vary quite a bit. And for individuals with this cultural trait, it seems to me that you get a kind of reservedness and formality in addressing people in power. That can lead to poor communication.

Any doctor will tell you that questions are OK. You need to ask questions to build an understanding for yourself. If the doctor winces, or looks annoyed when you ask the questions, you need to ignore that non-verbal behavior, which probably doesn't represent the doctor's final opinion on your questions at all. 

What kind of questions to ask, if you're confused? If you really don't know what's going on, say so. If the language the doctor is using is confusing, say so. The doctor may be surprised, but in the end they will be glad you said this.

These three questions offer a nice framework: 

1. What are my options?
2. What are the possible benefits and harms of those options?
3. How likely are each of those benefits and harms to happen to me? Including ‘What will happen if I do nothing?'

Nobody gets a class or goes through training about how to interact with doctors. Probably we should have that option somewhere. 

 

 

 

Overdiagnosis is an error in prognosis

In overdiagnosis, a clinician identifies an abnormality that would never have hurt the patient, and treats it as an actionable pathology. You can never know who is overdiagnosed in real time; it only becomes apparent after the fact, while examining populations. 

In a way, the term "overdiagnosis" is misleading. It implies an error of judgement in the diagnosis process. But more exactly, it's an error of prognosis. Overdiagnosis is inaccurately predicting the future of an abnormality. It is making the judgement that an abnormality is sufficiently likely to be dangerous that some kind of action is required. 

Overdiagnosis is one of the reasons that a good clinician will be careful about ordering too many tests. If a test is needed to solve a clinical problem, great. If the test is a vetted preventive health measure, great. But if it's not either of those, it's possible that any abnormalities uncovered will not be meaningful, and that chasing them - with further lab tests, imaging and biopsies - might lead to more harm than good. 

Why does overdiagnosis happen? There are many possibilities, and a few of them are emotional, or on an intuitive level. For example, I believe that clinicians and patients with a low tolerance of uncertainty order more testing because the uncertainty makes them uncomfortable. They believe that empirical studies are the only way to achieve peace of mind. One medical commentator says: 

"Yet medical, and especially paediatric, practice takes place in a sea of uncertainty: diffuse or intermittent symptoms, borderline or equivocal tests, unanticipated turns of events and complicated patients and parents whose real issues may not be the ones they first tell you about."

The same author says over-adherence to guidelines might be partly responsible. Rather than observing mild abnormalities like mild neonatal jaundice, providers are encouraged to make them into pathologies and treat them. If they don't, they risk of appearing non-compliant with best practice. 

Guidelines, after all, are often created in the setting of substantial uncertainties. 

Lab tests before surgery: not always useful

Partly out of habit, physicians frequently order lab tests before surgery, to make sure everything is OK. But this is not always useful.

The purpose of pre-operative testing is to figure out what risks there are, and to minimize those risks. If you are going through a low risk surgery, like a cataract, the labs will not provide information that will make you safer, in almost all cases. 

"Preoperative testing for low-risk surgical procedures typically does not reclassify the risk estimate established through the history and physical examination, may result in unnecessary delays, lead to downstream risk from additional testing, and add avoidable costs." That's from the Society of General internal Medicine, on the Choosing Wisely website. 

Chest x-rays before surgery - any surgery - are also unnecessary, if you have no symptoms. They will provide no useful information in almost all cases. That's from the American College of Physicians. 

There's actually a Cochrane review on pre-op testing before cataract surgery. Cochrane reviews are gold-standard evaluations of the evidence respected worldwide. The analysis included three randomized controlled trials, and showed no benefit of the testing. 

And yet this stuff happens. One study from Philadelphia found that 53 percent of low-risk patients had inappropriate testing. Usually this was standard tests like a complete blood count and blood chemistries. 

Why do these tests continue to be common? It might be due to historical habit. After all, it's hard to go against your training. And it may be fear of liability, because much of medicine occurs due to the fear of a malpractice suit. 

 

 

 

How do you explain overdiagnosis?

 

• You can't know if an individual has been overdiagnosed. You only see it at the aggregate level. 
• The decision point is the choice to seek a diagnosis. After that point, overdiagnosis can occur.  
• By definition, overdiagnosis will not improve prognosis - it's not going to make you live longer or feel better. 
• Overdiagnosis will harm individuals, such as through unnecessary biopsies. 
• Overdiagnosis will harm society through costs, including opportunity cost (e.g. spending $1m unnecessarily on a cancer treatment when that money could have gone toward vaccinations).
• The antidote is discussions ("Have a discussion, not a test") that include the possibility of doing nothing. 
• Doing nothing needs more respect. We already use the phrase "watch and wait." 
• To talk about overdiagnosis, you need to talk about uncertainty. Many people have low tolerance of uncertainty. That can be dangerous when interacting with the healthcare system. "I just want to know" can lead to overtesting and overtreatment. 
• Individuals can be presented with information on likely outcomes, based on number needed to treat: "If I had 100 people in your situation, and they all went through the test, this is what would happen to 15 of them," and so forth. 
• Remember that the public, in interacting with the healthcare system, has a bias to action. At least that's what they expect.
• Patient decision aids are useful. Check this out.

Source is here. 

 

 

In overdiagnosis, harm looks like benefit

It's hard to explain the concept of overdiagnosis to the public. Think of overdiagnosis as "pseudo-disease," in which something innocuous is labelled as a disease, and leads the patient down a pathway of unnecessary testing and treatment. An example would be finding and treating a prostate cancer in a 75-year-old man who would never die of the cancer. After treatment, he may end up incontinent, or impotent, or have chronic bloody diarrhea. 

We know this happens in groups of people. But we can't tell which individuals this is happening to in real time. When we find a prostate nodule and it turns out to be cancer and we treat it, it is really a correct diagnosis. But it's also true that only a few of these lesions will progress to dangerous disease, and we can't absolutely tell which ones ahead of time. So if we do this to ten people, and only help one, we have correctly treated the one but harmed the other nine. Meanwhile, all ten will say, "I beat cancer!"

Harm looks like benefit. 

Let's assume that the 9:1 ratio is correct (it's not - I just made that up - I'm totally oversimplifying and I'm not a urologist). Then you're the guy on the exam table deciding if you want to find out if there's a nodule in your prostate. Let's lay it out for you, in this imaginary scenario: 

"OK, if there's a nodule there, and it's cancer, there's a 90% chance that if we treat you, it will have been for nothing. And there's a 10% chance that you will get another five years from the treatment." Would you always choose to go ahead? What about if you were 85 years old and had severe emphysema? That's the kind of decision-making we have to encourage in medicine - because it describes the real situation. And, I think, it's the right way to get informed consent. 

There's a great textbook that helps docs outline these options. 

Why do doctors order unnecessary imaging of the back?

Imaging for low back pain (LBP) is only recommended in certain situations. The Choosing Wisely campaign suggests avoiding imaging during the first six weeks in the absence of red flags, like weight loss, a history of cancer, recent trauma, osteoporosis or age greater than 50.

And imaging for LBP is overused. For example, a meta-analysis in press (https://bit.ly/2P5Ux8X) shows that 32% to 35% of low back pain imaging is inappropriate.

What are the main reasons? It's not that providers don't know what to do. A survey in 2014 (https://bit.ly/2BLCAus) showed the following factors were important:

1. Inability to refer to a specialist without an image
2. Worry that patient would be upset
3. Not enough time to discuss risks/benefits with patient
4. Concern about malpractice risk

Of these factors, the only ones influenceable by the ordering provider involve patient education. This can be problematic in a time-limited office visit. However, the Choosing Wisely campaign suggests these messages:

1. The tests will not help you feel better faster. You're likely to feel better in about a month whether or not you get a test. The test can also lead to surgery and other treatments that you do not need.
2. Imaging tests have risks. Radiation's harmful effects add up.
3. Imaging tests are expensive. The cost is especially burdensome to you if you have a high-deductible plan.

Choosing Wisely has a great handout, also: https://bit.ly/2w3dvWJ

Keep in mind that frequently imaging is necessary, but not ordered. The meta-analysis cited above found that this occurred in 61% to 68% of cases where imaging was indicated, based on the presence of red flags.