Cancer screening keeps spreading to more groups of people, pushed by a widely shared conviction that more and earlier must always be better. As the shadow of cancer widens to cover ever more people, and lengthens to cover longer stretches of their lifespans, cancer angst spreads far and wide, too.
Barry Kramer wants to counteract irrational fear and actions by helping us get more rational about cancer screening. He’s currently Director of the Division of Cancer Prevention at the National Cancer Institute (NCI). A decade ago, Kramer started an annual evidence boot camp for journalists called “Medicine in the Media”.
This year’s course with 48 journalists from broadcast, digital and print media started last Sunday with the topic of over-diagnosis of mental disorders. On Wednesday Kramer took on cancer over-diagnosis, and the triple whammy distorting understanding of cancer screening: selection bias, length bias and lead-time bias. Here’s how he explained these critical 3 concepts.
1. Selection bias. Studies of cancer screening are often skewed by a 'healthy volunteer’ effect. Many people who follow through with regular screening also do other things that make them less likely to get, or die, of cancer. You need to stick to rigorously randomized trials to find out the effects of screening alone, said Kramer, “because then you have just as many health-conscious people being screened or not screened.”
2. Length bias. For common cancers, no one generally knows how to tell at an early stage if it will become life-threatening. Even if you lived for 120 years, many cancers would be growing so slowly, they’d never make you sick.
Screening, unfortunately, is better at finding those slow, unthreatening cancers than at finding aggressive more lethal cancers that appear suddenly in the time between screening. As Kramer put it, “We're ‘curing’ people who didn't need curing in the first place.”
3. Lead-time bias. Considering survival rates rather than mortality data leads people astray. In screening, that’s not quite the same thing. Even most doctors fall into this trap. Kramer led us through this thought experiment to explain the concept. Imagine a hypothetical cancer that will kill absolutely every person within 4 years from the day they have symptoms. That means their 5-year survival rate is 0%.
If you develop a screening test that detects everyone with this cancer a few years before symptoms start, you can still improve their survival rate without adding so much as a day to anyone’s life. Why? Because if the cancer clock starts ticking earlier, everyone will live longer than 5 years with the diagnosis: a 5-year survival rate of 100%. You increase your time with cancer, while decreasing the amount of your life you don’t have cancer: that’s not the same as living longer.
Screening only works when there’s a way to help more people than are already helped when they come to the doctor with symptoms. Kramer pointed out that ineffective screening is like being tied to a train track with a set of binoculars: you can detect the train that’s coming to hit you earlier, but it won’t change the moment of impact.
Meanwhile, cancer treatments can do major damage on a large scale to people who can’t benefit from them. Kramer’s view: "Japan's national screening tragedy was neuroblastoma. Ours is prostate cancer screening." The discoverer of the test used for prostate cancer screening, Richard Ablin, expressed a similar view in an op-ed this week: “I never dreamed that my discovery four decades ago would lead to such a profit-driven public health disaster.”
Of course, screening particular groups of people for some types of cancer does save lives. You can check official US recommendations here and look for good evidence and info here. Can’t get to a course, but want to learn more about understanding health research results? Kramer’s fellow musketeers over the years at “Medicine in the Media” are Lisa Schwartz and Steven Woloshin, and reading their work is a great start.
Woloshin and Schwartz are general internists and part-time academics from Dartmouth who wrote a book intended to help people better understand risk statistics – and then ran a randomized trial to see if people who read the book really benefited. They did. So “Know Your Chances” is a book that’s actually proven to be clinically effective!
Still want more? There are some books and articles here. And the US Cochrane Center at Johns Hopkins University provides an online course. All of these can help you learn how to know if research constitutes strong evidence. And as Kramer said, that’s the key: “Strong evidence of benefit is important when putting large numbers of people in harm’s way.”
Interests: I presented at the NIH Medicine in the Media course and while I work at the NIH’s National Center for Biotechnology Information, I am not part of the organization of this event. The views expressed here are my own. Medicine in the Media is an annual event organized by the NIH’s Office of Disease Prevention.
Image: By the author at Statistically Funny (CC-BY SA license)