Screening for Cancer Costs Billions - I Doubt it is Worth it

The study of the week found that Americans pay $43 billion in one year to screen for just 5 cancers

I would like to live a long life. Surely you do as well. You can improve your chances by eating well, exercising, avoiding smoking and too much alcohol. There is also the matter of luck.

The question today is whether modern medical screening can improve on those basic lifestyle choices. People believe it can, obviously. The screening industry is a money-making machine.

Cancer researchers recently quantified the costs of (initial) cancer screening in one year in the US. It’s hard to grasp the idea of $43 billion. And that was the result of screening for just 5 cancers (breast, cervical, colorectal, lung and prostate) in the year 2021.

The largest part of the costs went to colorectal cancer screening at 64% and colonoscopy represented 55% of that fraction. Pause for a second: we spent $23.7 billion alone on screening colonoscopy.

My point in choosing this study of the week is to stop and think what we get for that $43 billion.

But first we should set out, as the researchers did, that this is an underestimate of the costs. For instance, the estimate includes only the initial cost of the mammogram, PSA test, and CT scan. As everyone knows, the downstream costs to evaluate positive tests likely exceed initial costs many times over.

Benefits of Screening

The first point—before considering any data—is that screening does save lives. Randomized controlled trials output average effects. Even if the trial result is non-significant, there will be individuals who have a cancer discovered and treated, leading to a life saved.

What patients, doctors, hospitals and payers want to know is: does that $43 billion spent on screening lead to better outcomes in the average patient.

Another point to pause and stop and think about is the ultimate goal of screening for cancer. We screen for cancer because it is a leading cause of death. Cancer can shorten your life. It’s the same reason we screen for atherosclerotic heart disease—because it can kill you. This leads me to the main question: does screening extend life?

Finally, there is the matter of lethality. If cancer wasn’t deadly, or if we had curative treatments, there would be no reason to screen for it. We don’t screen for gall bladder disease; we just treat it when it occurs.

My Questions About Screening

Screening is clearly popular. Most hospitals have dedicated sections for screening. Breast and colonoscopy centers, for instance. So strong is our belief about screening, that it has been made a quality measure for primary care doctors.

You would think the data would be clear and unassailable. But it is far from that.

I won’t rehash all the work that Vinay Prasad has done on critical appraisal, but let’s just look at one of the biggest trials.

NordICC Trial

In the NordICC trial, Michael Bretthauer and his team studied invitations to colonoscopy in more than 84,000 people in Poland, Norway and Sweden.

In the intention-to-screen group (invited) the risk of colon cancer at 10 years was 0.98% vs 1.2% in the usual care group. Because there were so many patients in the study, that 0.22% difference made statistical significance. It was an 18% lower rate of colon cancer if you were randomized to the invited group.

However, the risk of death from colorectal cancer was not statistically different (0.28% vs 0.31%). Most important was that the risk of dying (from anything) was 11% in both groups.

Nearly a 100,000 people in a trial and at the end of three years, there was no difference in survival. None.

The GI professional associations had the talking points ready when this trial came out. They focused on the fact that this was an invite-to-screen trial.

They wanted everyone to look at the per-protocol analysis. They said you can’t benefit from colonoscopy unless you get it. This sounds correct on principle, but the point of this trial was not to study colonoscopy as an individual procedure, but to study whether invitations to screening colonoscopy improved mortality over basic care in populations.

In the DANCAVAS cardiovascular screening trial, authors have never reported the as-screened population because:

Due to high likelihood of healthy user selection bias, unadjusted comparison of outcomes among those who actually attended the screening and those randomized to usual-care would be highly inappropriate.

NordICC authors did adjustments to account for biases and reported results in those who were actually screened.

The found that the risk of death from colorectal cancer was 0.30% vs 0.15%.

After adjustment this difference had very wide confidence intervals. The procedure reduced the risk of dying from colorectal cancer (in an adjusted analysis) by a mere 0.15%. And the lower bound of that adjusted 95% confidence interval included 0.

Lung Cancer Screening is Even More Humbling

Low-dose CT scans have been recommended for patients with strong smoking history. (At least here there is an attempt to focus screening on the highest risk individuals).

But Vinay Prasad has shown that trials also fail to show a survival benefit.

The Philosophy of Screening

My final thoughts on screening are less empirical.

There are thousands of diseases that can cause illness and death. How in the world has the medical establishment convinced millions of people that avoidance of a handful can change outcomes?

Even if the screening-RCT data were strong for any of these diseases (they are not), that would leave like 9,999 other diseases that could kill you over the next decade.

Medical and surgical therapies of cancer and heart disease have advanced greatly. My take is that you live as healthy as you can, and then pay attention to symptoms. If you have symptoms or signs, you get checked.

This way you avoid the anxiety of screening and the possibility of harm from downstream testing and treatments.

I would wager that if we showed people the actual data behind screening, a lot more would forgo it.

Dr. Gilbert Welch has an editorial on this paper. My thinking aligns well with his. (The essay is open access for another two weeks.)

Medicine excels when we treat people asking for our help. We struggle when we intervene on healthy people. We underplay the role of luck in survival.