When Testing Becomes Overtesting

May 14, 2015

Billionaire Mark Cuban spurs a debate about overtesting.

On April 1, Mark Cuban, billionaire entrepreneur, NBA team owner, and actor, took to Twitter, sharing his philosophy on how patients can best be engaged with their health care. In a series of tweets, he advocated – among those who can afford it – for having a quarterly blood test in an effort to establish an individualized health baseline.

Not only would this strategy give patients and their children more knowledge about their health care over time, he said, but it would also allow patients to be more proactive about their medical treatment. No more waiting to get sick – patients could, potentially, identify problems early and ask for intervention.

The wake of those tweets has been filled with mixed responses from the medical community. Within radiology, a specialty constantly concerned with the specter of overtesting, providers are in heated disagreement over whether Cuban’s advice is reckless or revolutionary for health care.[[{"type":"media","view_mode":"media_crop","fid":"37738","attributes":{"alt":"","class":"media-image media-image-right","id":"media_crop_5631356982381","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"3733","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"height: 240px; width: 160px; border-width: 0px; border-style: solid; margin: 1px; float: right;","title":"Mark Cuban","typeof":"foaf:Image"}}]]

To determine the pros and cons of a quarterly blood-test strategy, Diagnostic Imaging interviewed two industry experts. Saurabh Jha, MD, MS, assistant professor of radiology at the Hospital of the University of Pennsylvania, and Stephen Hunt, MD, PhD, an interventional radiologist and co-director of the Penn Image-Guided Interventions Laboratory shared their thoughts.

Diagnostic Imaging: When it comes to the concept of frequent testing, what makes it appealing? What makes it something we should shy away from in health care?

Jha: It’s appealing because it gives you a sense of determinism. You’ve got these numbers, and you think you can do something about them. You can mold your health as a result of them. So, it’s very seductive but gives you a false sense of control over what is, really, very much a process that we have very little understanding about. Why this is dangerous is any testing yields a number which can be blatantly normal when one is well, or blatantly abnormal when one is ill. But, it’s when you’re not ill and you’re given a number that falls into that grey zone of normal/ abnormal that you could end up pursuing it when there was really no need for that pursuit to happen in the first place.[[{"type":"media","view_mode":"media_crop","fid":"37739","attributes":{"alt":"","class":"media-image media-image-right","id":"media_crop_6577266399019","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"3734","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"float: right;","title":"Saurabh Jha, MD, MS","typeof":"foaf:Image"}}]]

Hunt: Rather than speaking from generalizations, let’s look at an example from the news, such as overdiagnosis during screening for breast cancer in younger women. You see articles in the New York Times or the Wall Street Journal on whether we should screen for breast cancer in women aged 40-49. There’s been a whole series of academic research in this area. However, even though there’s a whole lot of people knocking on breast cancer screening, we know that it’s saved a lot of lives. The number we save has been incontrovertibly shown. We’ve reduced breast cancer deaths by 30% since 1990 – that’s about 20,000 lives a year. You can do the math on how many lives that have been saved since we started mammography in the 1970s. It’s been literally hundreds of thousands of women.

A lot of women get told they have breast cancer, and the breast cancer may not have affected their life span. The estimates are that out of every 200 women that get told they have breast cancer, only one dies from it, because many people diagnosed are a lot older, and they die from other causes. But, regardless of how you look at the equation, those lives getting saved come out to 20,000 a year because of breast screening. These are women who have come in asymptomatic for screening. They haven’t felt anything in their breasts. It’s different from a diagnostic work-up of the breast because of a lump you feel. They literally don’t think anything is wrong, and they get an X-ray. It shows they have cancer, so they move on it and get it treated. So, we have a 30% reduction in death since 1990. That’s the kind of thing [Cuban] was talking about. It’s about getting a baseline. If you don’t have a baseline, if you come in and get your breasts screened, we might see something suspicious in there. But, we don’t know what it is, so we end up bringing you back and doing a biopsy. If you come back every year, over time, we’ll have a baseline. We’ll have multiple data points. We can track those data points over time and act on that information.[[{"type":"media","view_mode":"media_crop","fid":"37740","attributes":{"alt":"","class":"media-image media-image-right","id":"media_crop_328965030057","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"3735","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"float: right;","title":"Stephen Hunt, MD, PhD","typeof":"foaf:Image"}}]]

Diagnostic Imaging: Clearly, this is something that works for breast screening, but is it something that’s applicable to all situations?

Jha: Many things are being conflated here. Dr. Hunt talks about a baseline, but what Cuban is suggesting is that you test yourself for everything every three months. Does that mean blood selenium levels? Selenium toxicity gives you garlic breath, but so does Chicken Dopiaza. I know I’m being ridiculous, but the idea is so ridiculous that it invites ridicule. What about your baseline potassium? How do you determine it? How do you know the first test is normal? You have to look at laboratory results for comparable demographics. That’s the entire point, it’s a contradiction in itself at the outset – you need the population results to reassure the individual.

Hunt: But, you’re talking about a single data point. And, this is where [Cuban] clarified his idea with many following comments. He wasn’t talking about acting on a single number that you get. A baseline can’t be a baseline in terms of a single point. That’s why it’s called a base line. You look at multiple data points – measure your potassium once, twice, three times, while you’re asymptomatic. Maybe you fall outside the normal population statistics. But as you trend it over time, if you have sudden changes, you can see how they change from the baseline – not a base point. Dr. Jha is talking about a single point. I’m talking about a baseline – about collecting a series of data points early when you’re asymptomatic.

Cuban didn’t say the government should pay for this. If you have the money and you can collect the data points, it still may not be something you act on. And, that’s the main point of the conversation. You have to collect the data points and have the fortitude to not act on a meaningless one.

Jha: So, you’re nervous enough to collect this data because, “Oh, my God, what’s going to happen to me,” but you must have the fortitude to not act on it. It’s a contradiction in terms. If I may, when you’re talking about collecting for baselines, how do you know the first test is abnormal or normal? You look at the population demographics and reassure yourself.

Hunt: But you don’t care about the first data point, you’re looking for the change in it. It’s not a contradiction. It’s gathering pieces of data that aren’t actionable when you’re gathering them. You’re just pulling everything into an Excel sheet so you can track it over time. So, if you suddenly have big bumps, you can go to the doctor and have him or her investigate. If you’ve suddenly deviated from whatever your personal baseline is, you’ll know it’s a deviation from a line – not from a point.

Jha: But how do you define deviation? Is it 10 standard deviations? Fifteen or 20?

Hunt: In the example of TSH, the American Association of Clinical Endocrinologists provides those guidelines.

Diagnostic Imaging: To look at this from a radiologist’s perspective, what’s the good, bad, and the ugly? Radiologists are always thinking about testing and overtesting, so where does that point-of-view come in here?

Jha: The main thing is that whenever you measure, the measurements are not gospel truths. There’s going to be a true positive and a false positive. And, a false positive is when you call something a disease when it really isn’t a disease. Whenever you measure, and you know in radiology we do this all the time, if you measure the aorta and it’s beyond a certain size, you can call it an aneurysm and beyond another size say, “Let’s go in and repair it.” A threshold will create false positives. It could, of course, be a true positive as well, but that’s where the trade-off lies. The point with us understanding overtesting is understanding where the risk of the false positive lies the most. The false positive occurs when we fish for problems in a pool in which the likelihood of that particular problem is very, very low. If we fish for that problem in a pool where the likelihood is very high, then the number of false positives really dramatically comes down. This is where we really make that call as imagers with regards to testing.

Hunt: To provide an example, let’s look at lung cancer screening. It’s now approved by CMS as of February of this year. You have to meet specific criteria– you have to be between ages 55 and 77 and have a 30-year smoking history. But, if you look in that group, you get a 20% risk reduction in cancer deaths by screening with a CT of the chest versus just an X-ray. In fact, doing this saved so many lives, they stopped the $260-million clinical trial halfway through because it became unethical to randomize people to the chest X-ray because too many people were being saved by the chest CTs. Now, if you started looking in 20-year-olds for lung cancer and took CT scans, it’s extremely unlikely that you’ll find lung cancer. You’re going to find things in there that you aren’t sure if it’s cancer or not, but it’s very unlikely that it is. You’re going to find something in there that if you act on it, it would probably hurt the person more than it would help them. That’s where it becomes this trade-off. This is Dr. Jha’s point about how you cannot take a normal person who’s asymptomatic and start doing tests on them. The point here is should we be doing it to collect data points because we’re trying to build a big database of what is the normal variations on these tests. What we’re talking about here and what he’s talking about is literally a lab test – getting blood drawn. The risk of a blood draw is so low that it’s considered not even above minimal risk. I sit on an IRB committee and you stamp that through if someone says they want to just do a blood draw. The point is that what Cuban is recommending is getting yourself tested for various things so that you have some information going in. He didn’t say act on that information- because you don’t want to act on it when you’re in an asymptomatic state. But, if you suddenly develop a symptom of your leg hurting (say from a blood clot) and now you’re looking at a lab that shows there’s inflammation, you have pieces of data that can provide insight into why your leg is hurting. You have these positive lab results when they’ve always been negative. The doctor can arrive at a conclusion of what’s wrong much quicker because you have a bunch of baseline data versus shooting around in the dark.

There are two sides to the equation. I’m an interventional radiologist which is a bit different from a diagnostic radiologist. I use imaging to treat disease. So, when Dr. Jha sees all the screening chest CTs, he’s looking at this big haystack, seeing normal chest CT after normal chest CT. He rarely sees a lung cancer. When patients walk into my office, they’re coughing up blood. They already have lung cancer. I see the needles, and he sees the haystack. That’s the real difference here in our approach. I see the people coming in with end-stage cancers – what happens when we don’t go looking for these things soon enough. He just sees a whole lot of normal chest CTs. So, he’s looking at it from the perspective of this huge haystack.[[{"type":"media","view_mode":"media_crop","fid":"37751","attributes":{"alt":"","class":"media-image media-image-right","id":"media_crop_8538079588206","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"3738","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"height: 500px; width: 250px; border-width: 0px; border-style: solid; margin: 1px; float: right;","title":"©VectorShots/Shutterstock.com","typeof":"foaf:Image"}}]]

Jha: I do want to make some counterpoints to that. The first thing to appreciate over here is that when you’re testing and when you’re testing in a serial manner, there has to be some point of doing that. I don’t think Dr. Hunt has entirely read the, now deleted, tweets from Cuban. What he said is that we shouldn’t wait to get sick before we see the doctor. We should anticipate it and preempt it. Preempt is a very dangerous word. We all recall the preemptive strikes of the Iraq war – remember collateral damages? What is the collateral damage of doing this? The collateral damage is that person who looks at the baseline and says, “Wow, I see this 10% change. I better see a physician.” Say it’s a 10% change in the thyroid stimulating hormone (TSH) level, and they see a physician. The physician could say it’s probably all right, but that the patient is worried– and can pay. So, the doctor orders a thyroid ultrasound that shows a 1-cm cyst that is biopsied. It reveals papillary carcinoma. It’s been shown over and over again by researchers at Dartmouth that these make no difference at all, statistically speaking, in survival. But, part of the thyroid comes out. That’s a false positive. What Dr. Hunt does not look at is the false positive. The needle that’s not really a needle – it’s part of the haystack. These false positives – you can’t think of them as foreigners like North Koreans - they’re not. They’re Americans. They’re like you and me. They’re individuals, they’re harmed. They’ve lost part of their thyroid or kidney for no benefit whatsoever. For the medical profession not to object to this, in my opinion, is unethical.

Hunt: I think Dr. Jha makes very good points about the risk and costs of screening. I think he’s, unfortunately, being paternalistic in the way he views healthy patients keeping track of their own health. I wish my patients would follow their health so closely and follow their labs to this extent. I wish my patients were that involved with their health. They’re not. They walk in with multiple health risk behaviors – obesity, smoking – all of the things that we, as Americans, unfortunately, tend to do. And, what Dr. Jha’s dissuading them from doing is caring for their own health by saying, “Don’t do that. Don’t be engaged in your health care. Just trust me. I’m the doctor, and I’ll tell you when you need a test.” Patients need to be very engaged. And, if they want to get a test, as long as their doctor has the presence of mind to say, “Don’t get this thing biopsied. It’s a cyst,” or if you’ve got a scared patient, counsel them. Say this is the risk. You’re going to get this biopsied, even if it comes back as cancer, give them the speech Dr. Jha just gave. You think people can’t make informed decisions? That’s paternalistic – it’s saying, “Trust me. I’m the doctor. Take off your clothes.” That’s not the way medicine should be practiced. We now have an engaged public who has access to the Internet and can look up national guidelines. They can go to the endocrinology website to see what the guidelines are and when to do something. I try to make my patients more engaged, not less engaged.

Jha: I don’t think telling someone who’s jumping out of the plane that they need to use a parachute is paternalistic. The idea that being engaged in health care requires a blood draw every three months makes one ask why not every week, every hour? I don’t think being engaged in health needs bloodletting. I think selling that as engagement, endorsing that as engagement, is disingenuous. Being engaged in your health means eating salad. It means running and looking after your weight. It means moderation. Not nervousness and frequent testing so that you have a baseline and end up using population-based results, the reason why you wanted the baseline in the first place, to reassure. It is rather sneaky of Dr. Hunt, on the one hand, to exculpate himself from paternalism. Yet when it comes to results, he holds the doctor responsible for the downstream testing. Does that responsibility not imply paternalism?

In Dr. Hunt’s universe, the idea of perfect information so that people can make trade-offs between let’s go get tested and get the thyroid removed to avoid that very low risk of consequential thyroid cancer and going on a holiday, is utopian and delusionary. It does not exist. You can call it paternalism, but when the plane doesn’t take off because the pilot doesn’t think it’s a good idea because of a storm, you don’t go to the passengers and ask for their opinions. You don’t say “let’s make take-off a customer-centered experience because otherwise it would be paternalism.” That’s utter absurdity. This word is so overused, so vilified that one doesn’t realize that what paternalism really is, is for me to tell you what you’re doing is your right, for sure, but it’s rather unwise. Call it paternalism. Call it whatever you want. Facts are facts.

Hunt: I think the absurdity of equating a TSH test to jumping out of a plane without a parachute is self-evident.

Jha: I’m not equating, I’m just giving an example. You have to define paternalism. Where do you draw the line? Would you call it paternalism to tell the patient with a ruptured spleen that they need to have surgery? Are you twisting their arm? What you’re calling paternalism is actually asymmetric information. It’s based on our greater training and experience with false positives. To pretend asymmetric information doesn’t exist is disingenuous.

Hunt: You have the opportunity to address the question with the patient at the beginning of the whole process. You can say if they’re testing their TSH not because they’re going to act on any individual number, but because you want a baseline and have that information about your health, I understand that. Of course, your insurance won’t pay for it. But if you want to collect the data, and you feel you’re an informed consumer, you can do that. It’s called informed decision-making. You can’t suddenly say informed decision-making is paternalistic. Now you’re at a stage where you can engage with your patient and tell them if they act on it, they could end up with a bad outcome based on the information you have on how many of these things are meaningful in terms of cancers. There are opportunities for engagement all along the way here, and what I want to do is keep the patient engaged in their own health.

Diagnostic Imaging: Switching from patient engagement, how will this change how providers – radiologists, general practitioners, whoever – practice medicine?

Hunt: For example, I’m here in Washington, D.C. at the moment for a Translational Science meeting with representatives from the National Institutes of Health and the Food & Drug Administration. Francis Collins, the NIH director, is walking around wearing a Fitbit, and discussing its implications for health and health care delivery. He’s collecting information about his physiology as he walks around. There’s curiosity about how all this collecting data and information will engage people in their health. He’s already seen it. It tells him how active he is for the day, what his heart rate is, and whether it goes up when he’s upset or in a meeting. He said people are wearing these devices, there’s all this opportunity to collect this information. Can we use it to a useful purpose? We’re collecting all this information on patients. Can we find evidence that it impacts their health? If the onus is on us to go and collect that evidence, in the meantime, do we tell our patients not to wear a Fitbit? Not to collect data on their heart rate because who knows what they might find? They might find their heart rate is too fast for the norm, and now you’re going to act on it. No, I don’t stop them from engaging in their heath. As far as we know, there are no side effects from wearing these things and collecting this data.

Patients and consumers are moving more and more toward engagement. One of the things the FDA did is a huge survey about informed consent and research. Overwhelmingly, patients’ drive is for us to get them answers to their health problems, their diseases, their cancers. Overwhelmingly, patients report we’re moving too slowly. Our future is bogged down by this paternalism that is essentially telling patients, “Let us decide for you what we’ll do.” What other industries act that way? If the lines are too long at check-in, the TSA hires more people or loosens the laws. That’s the reality of other industries. That’s what we need to do. We need to allow patients to be engaged. It will change the way we practice medicine.

We can’t act on every single data point that comes in because a lot of these data points are going to be meaningless. But it does allow for a huge amount of big data screening to look for patterns. What is the pattern of your TSH when looking at 200 million people? There’s actually a program of 178 million people called the Mini-Sentinel study funded through the FDA. It’s looking at data across half of Americans. This is the kind of huge program where you can mine the database of what’s going on with patients. It’s possible to see the low-hanging fruit. Everyone likes to go through these databases and say, “Here’s a signal,” but in a collection of a lot of data, you see false signals. You see outliers. We have to be smarter than that. We have to have hardcore biostatisticians come in and determine what’s a normal variation. We have to be smart in how we parse the data, but this will fundamentally change the way we practice medicine.

Jha: So, you must’ve noticed the tense in Dr. Hunt’s speech. It is future tense. What’s going to happen 150 years from now? I don’t know. Maybe we might be living in a place where the moment one cancer cell starts mutating, robots will zap it out. That’s not where we are presently, and to think we’re not there because of paternalism or a guild really is wishful thinking. We’re not there because we don’t have the information or the science.

The original question was how will this impact imaging? I try not to make predictions beyond two to three years. I think if an engaged public must test themselves frequently, then it will lead to a parallel market. To his credit, Cuban said this testing would be done on a person’s own dime and own time. Any further tests induced by this test should also be done on an individual’s own dime and time. So, potentially, this could lead to a parallel market of imaging of the worried well or the worried sick. It could lead to discovery of new pathological substrates that aren’t here yet, and that could be one silver lining.

The key is how will we deal with small deviations in numbers. For example, with genetic testing, you could be told you have a greater than 6.25% chance beyond the population mean of ovarian cancer. You might ask how the test came up with that percentage – why not 8%? The pseudo-precision of these numbers is no different than snake oil. A patient could say with a 6.25% added chance that it’s time for a pelvic US or abdominal CT. An engaged patient will increase utilization of imaging. I can make that prediction with a certain amount of certainty. It will create a parallel market.

It will be interesting to see what happens there. Tests show possible disease. Imaging shows possible disease, and biopsies show possible disease. Even pathologists disagree with each other – not even biopsies are the gospel truth. We’ll have this merry-go-round. People will not be living any longer. People will feel empowered, and they will feel they have more control. And, they’ll still die at the time they’re supposed to die – that is to say, all-cause mortality will not change. That’s fine. Honestly, our economy needs stimulation.

Hunt: My short-term predictions are this. Look at lung cancer from which 160,000 people die yearly. Finally, we have screening for these folks. We see the death rate of lung cancer will drop dramatically, as happened when we started mammography screening. We’re saving tens of thousands of lives by doing this. This is what I want to see. I want to see people having more birthdays. That’s where my whole focus is. It’s got to be how are we getting to these patients? This is an opportunity for engagement. They come into the doctor’s office, as CMS has forced them to do. In order to get the screening, it has to be part of a counseling session about smoking. The counseling session is free. The CT is $400. There will definitely be greater imaging utilization. Whether you trust Dr. Jha’s estimate of $9 billion or mine of $4 to $5 billion, there’s definitely going to be some significant cost every year. But when you look at what lung cancer treatment is right now – it’s $12.6 billion in direct costs and $38 billion in indirect costs. We can move that $50 billion that’s being spent on allowing people to die and shift it to allowing people to live.

Jha: Right, and we’ll all be Lazarus and live forever. I think Dr. Hunt can be seen as implying that if we can save people from dying from lung cancer, we’ll save them from dying. That will not be the case – we extend life, not prevent death. In this case, we’re talking about people who are well, who don’t smoke, and in whom we aren’t looking for a specific disease.

Why quarterly blood tests? I ask why not daily? Why every three months? It seems arbitrary.

This will lead to an epidemic of anxiety. An epidemic of overtesting. Organs will be taken out in healthy people and will be shoved into the surgical pathology bin. If this is what society calls progress, quite honestly, send me back to the 19th century.

Diagnostic Imaging: Cuban is a data guy. His approach is to accumulate data to look for patterns of change, then predict an illness or ailment before it happens. He might have been extreme in his suggestion of a blood test every three months, but the interest in data, in general, continues to grow. People continue to understand that they can access the data that’s available. What’s the in-between? Is there a compromise?

Jha: Dr. Hunt and I discussed this. Any information you collect, you have to really ask yourself how actionable that data is going to be. I wear a Fitbit. My wife bought it for my birthday to get me off my butt. It tells me when I’ve done 10,000 steps. That’s my goal. I have to admit that it has worked in that regard. What I would say is any time you collect data, you have to ask what you’re collecting and why and what you intend to do with the data.  The best data I used to collect, I stopped collecting, was abdominal circumference. It upset me when I collected it. It was actionable. It was a mirror to my lifestyle. Anything measurable that is external to the patient that has the least chance of causing a false alarm and downstream testing is the way to do it. I don’t think blood tests fall into that category. I think there’s an extreme possibility of real harm, particularly when you collect tumor-specific antigens and inflammatory markers for no better reason than “baseline” or “anticipation of illness.”

I’m all for empowerment. It is possible to be empowered without having your blood drawn every three months.

Hunt: With all the medical advances, Dr. Jha wouldn’t really want to go back to the 19th century. Life is getting longer and we’re getting healthier into our old age. Some of that is due to screening in asymptomatic folks to establish baselines and look for early stage disease. I predict patient engagement and screening will continue to have positive effects on human health.