WS MoreOrLess: Bad Luck and Cancer

Summary

More or Less (BBC Radio 4): "Bad Luck and Cancer"

Date: January 12, 2015
Host: Tim Harford

---

Episode Overview

This episode investigates the widely reported claim that "most cases of cancer are the result of sheer bad luck," which originated from a new scientific study. Tim Harford and the More or Less team dissect the core findings of the research, assess its media coverage, and clarify what the numbers actually mean, with help from expert interviews and listener queries.

---

Key Discussion Points & Insights

1. The Origin and Content of the Research

• Background:
  Headlines around the world claimed that most cancers are caused by bad luck, based on a study from the Johns Hopkins School of Medicine (Bert Vogelstein et al., published in Science).
  [00:35] A: “Most cases of cancer are the result of sheer bad luck.”

• What is Cancer?
  [01:04] C (PZ Myers): "Cancer is actually a pretty complicated disease… what you’ve got is a set of cells that are undergoing uncontrolled replication and are often doing things like being invasive and penetrating into other tissues they shouldn’t.”

• Main Finding Explanation:
  The research focused on the correlation between the number of cell divisions in various tissues and the likelihood of cancer in those tissues.

  • Tissues with more cell division (e.g., intestinal lining) are more prone to cancer than stable tissues (e.g., muscle, brain).
    [01:48] C: “What it showed is a correlation between the number of cell divisions to produce a tissue and ... the likelihood that it would become cancerous.”

• Core Question of the Study:
  Instead of asking why some individuals get cancer, the research investigates why certain cancers are common and others rare, based on tissue type.
  [02:41] B: "They were instead asking why some types of cancer are common and some are rare. ... the randomness of cell renewal, or cell division gone wrong."

2. Interpreting the 2/3 Statistic

• Common but Incorrect Interpretation in Media:
  Many outlets said “2/3 of cancer cases are the result of bad luck,” suggesting they can’t be prevented. [03:25] A: “The researchers say they've calculated that 2/3 of cancer variation is down to bad luck... many media reports have simply concluded that this means that 2/3 of cancer cases are the result of bad luck.”

• Problem with Media Reporting:
  This simplification is misleading.
  [03:47] B: “And that's just wrong... We've been scratching our heads trying to figure out what this 2/3 number really means.”

• Clarification from Johns Hopkins:
  The Johns Hopkins website offers an analogy: cancer risk is like the risk of a car accident, with "length of the trip" as a proxy for "number of cell divisions."
  [04:18] B: “We would estimate that two thirds of the risk of getting into an accident is attributable to the length of the trip... two thirds of the variation is attributable to the random mutations that occur in stem cell divisions...”

• Interpreting the Statistical Relationship:
  The 2/3 refers to the correlation across tissue types, not proportion of preventable/inevitable cases. [05:36] A/B: “The rate of cell division is about two thirds correlated with the rate of cancer. So we think maybe that's what the researchers mean.”

3. Expert Critiques and Statistical Nuance

• George Davy Smith's Critique:
  Professor of clinical epidemiology at Bristol University, he finds the comparison unhelpful and the conclusion simplistic.

  • [06:04] D: “Explaining the difference in risk between the leg and the lung is of no interest to anyone... It’s like getting two statistics... and applying that number to some other domain.”

• Who is to Blame for the Confusion?
  Smith does not blame journalists but the original journal and press release. [06:40] D: “That subheading is just not supported by the data... it’s both in the journal and in the press release.”

• Risk Partitioning is Misleading:
  Risk cannot always be cleanly divided into "bad luck" and "preventable" causes.

  • Example: Russian roulette is 100% preventable, yet involves luck if the person survives. [07:45] D: “There's just a fundamental misunderstanding about whether one can partition risk in the way that's done.”

4. What’s Useful About the Study

• Takeaway for Cancer Patients:
  The important message: cancer is not the patient’s fault. [08:15] C (PZ Myers): “If you have cancer... that's not something that you should blame yourself for.”

---

Notable Quotes & Memorable Moments

• On the confusion in reporting:
  [03:47] B (Tim Harford): “And that's just wrong. But, you know, this is hard. We've been scratching our heads trying to figure out what this 2/3 number really means.”

• Academic scepticism:
  [06:04] D (George Davy Smith): “Explaining the difference in risk between the leg and the lung is of no interest to anyone...”

• On the need for more explanation:
  [08:43] A: "He says he's writing a technical paper to explain his previous paper."

---

Important Timestamps

• [01:04] – Biologist PZ Myers: What is cancer?
• [01:48] – Myers explains the study’s findings
• [03:25] – Media misinterpretation of the '2/3' statistic
• [04:18] – Johns Hopkins car accident analogy explained
• [05:36] – Data correlation between cell division and cancer
• [06:04] – George Davy Smith critiques the study’s usefulness
• [06:40] – Smith on media reporting vs. original sources
• [07:45] – On the flawed concept of risk partitioning
• [08:15] – PZ Myers: Cancer patients shouldn’t blame themselves

---

Summary

The More or Less team critically examines the viral claim that "most cancer is bad luck." The original research illuminates a correlation between tissue cell division rates and cancer incidence, but it’s mischaracterized as proof that cancer is mostly random and unpreventable in most cases. Experts point out that this statistic is about variation in incidence between tissue types—not about preventability or individual risk.

The discussion highlights the difficulties in translating nuanced research findings into media headlines, the dangers of oversimplification, and the value (and limits) of such studies for patients. The episode makes clear: cancer’s causes are complex, and while chance plays a role, most cases cannot be neatly partitioned into “bad luck” or “preventable” boxes.

---

Transcript

[A]

This is the short edition of More or Less first broadcast on the BBC World Service.

[B]

Thank you for downloading from the BBC. For details of our complete range of podcasts and our terms of use, go to bbcworldservice.com podcasts hello and welcome to More or Less on the BBC World Service. I'm Tim Harford. Headlines and news bulletins around the world just couldn't get enough of a new study of cancer published at the start of the year year. Here's the independent newspaper from the uk.

[A]

Most cases of cancer are the result of sheer bad luck.

[B]

But more or less listeners. Philip, Rhiann, Simon, Alex and David each raised a quizzical eyebrow, spotting that there was some confusion in the reporting. They asked us to investigate and we discovered that to understand this research, it helps to go back to basics. What is cancer? We asked PZ Myers, a biologist and associate professor at the University of Minnesota, Morris, in the United States.

[C]

Cancer is actually a pretty complicated disease. I mean, there's a lot of different kinds of cancer, but basically what you've got is a set of cells that are undergoing uncontrolled replication and are often doing things like being invasive and penetrating into other tissues they shouldn't. So it's basically a small population of cells that have gone nuts and have decided to take over.

[B]

Right. So what about this new research that says cancer is largely a matter of bad luck? It was published in the journal Science, and the researchers, one of whom is a very eminent cancer expert, Bert Vogelstein, are from the Johns Hopkins School of medicine in the U.S. sadly, they were too busy to hop on the phone to us. But PZ Myers explained what he thought they'd found.

[C]

What it showed is a correlation between the number of cell divisions to produce a tissue and. And the likelihood that it would become cancerous. So what we're seeing in this paper is a quantitative description of what proportion of the errors that lead to cancer are caused by cell division rather than other factors like environmental causes or genetic causes. Some tissues are fairly stable. So for instance, muscle and brain tissue does not divide once it's done developing. So those tissues have a very, very low likelihood of coming down with cancer, whereas things like the lining of your intestine is constantly being regenerated and sloughed off. And so those cells have a high proliferative output and they're much more likely to become cancerous.

[B]

So what the researchers were trying to do here wasn't to say what determines why some people get cancer and others don't. They were instead asking why Some types of cancer are common and some are rare. Why do lots of people get lung cancer, but few people get cancer of the leg? And the answer to that question is this issue of cell renewal, or more precisely, the randomness of cell renewal or cell division gone wrong. Got it, Ruth, Got it. This is Ruth Alexander, our producer.

[A]

So there are genetic factors behind cancer, there are environmental factors, smoking, diet, pollution, and so on. And then there's just the luck of the draw. Because one of these zillions of cell divisions in your body goes wrong.

[B]

Yes, I think that's it.

[A]

And the researchers say They've calculated that 2/3 of cancer variation is down to bad luck. Now, many media reports have simply concluded that this means that 2/3 of cancer cases are the result of bad luck, a random cell division gone wrong, and that these cases have no genetic component and can't be prevented by stopping smoking or improving your diet.

[B]

And that's just wrong. But, you know, this is hard. We've been scratching our heads trying to figure out what this 2/3 number really means.

[A]

We've read countless articles and blog posts by statistically savvy people and experts in this field. And the problem is they don't all agree.

[B]

But then midweek, there was a glimmer of hope in the form of a clarification on the Johns Hopkins University website.

[A]

So they're using the analogy of getting cancer is like getting in a car and going on a trip.

[B]

They're saying it's like getting into a car accident. And so they're saying, using this analogy, we would estimate that two thirds of the risk of getting into an accident is attributable to the length of the trip. The rest of the risk comes from bad roads, bad cars, etc. In terms of cancer, we calculate that 2/3 of the variation is attributable to the random mutations that occur in stem cell divisions throughout a person's lifetime, and the rest is environmental.

[A]

So, but, yeah, but look, this paragraph, the next paragraph, they say we can't say that 2/3 of accidents are caused solely by the length of the trip.

[B]

That's just what they did. Say that's just what they did. Isn't that what they said? Our best guess is that it's about correlations. The correlation between the parts of the body that have a lot of cell going on and the parts of the body that are prone to cancer. If you imagine plotting a graph listing all these different types of cancer, and you have the frequency of cell divisions on one axis and the frequency of cancer on the Other axis. Now, imagine the dots on the graph were scattered all over the place. There was no correlation. You'd say that there was no particular relationship between cell division, which is what these researchers call bad luck, and cancer.

[A]

And if all the dots lined up perfectly, there'd be a 100% relationship between cell division and cancer. And the answer is somewhere in between. The dots on the graph line up pretty well. So the rate of cell division is about two thirds correlated with the rate of cancer. So we think maybe that's what the researchers mean.

[B]

Maybe. Although if that's what they do mean. George Davy Smith, who's a professor of clinical epidemiology at Bristol University, is not impressed.

[D]

Explaining the difference in risk between the leg and the lung is of no interest to anyone. Amenden says nothing about the contribution to the cancers in the population. It's like getting two statistics, two estimates which bear no relationship to each other. And because you've got a number applying that number to some other domain.

[B]

Now, there were lots of media reports with extremely simplistic views about the relationship between cancer and. And bad luck. So should we blame the media for getting this story wrong? George Davy Smith doesn't think so.

[D]

The subheading in Science's editorial, which was headed the bad luck of cancer, and the subheading was analysis, suggests most cases can't be prevented. That subheading is just not supported by the data, by that analysis. So that is what's misleading. And in the press release, the authors say that they've come up with a method which allows them to quantify the. The contribution of these stochastic or chance factors, which their method doesn't. So it's both in the journal and in the press release. And it's just not fair to attribute the misreporting of this to journalists. They've just copied what's in the Journal and in the press release.

[B]

And there's another problem here, and that's the idea that you can add up various contributions to risk, the part that's preventable and the part that's pure luck, and get a total of 100%. I mean, if I play Russian roulette and I shoot myself in the head, that was 100% preventable. But surely it was partly bad luck too, because after all, I might easily have survived. So here's George Davy Smith on the subject of luck and preventability.

[D]

I think the issue is that there's just a fundamental misunderstanding about whether one can partition risk in the way that's done. There's a great difference between the risk of an individual getting cancer, you know, you getting cancer and your brother not getting cancer, for example, as against the actual preventability of cancer. And one doesn't say anything about the other.

[B]

The headlines may have been misleading and the study itself may have some serious critics, but PZ Myers thinks it's useful.

[C]

What's important about the study is that it does say that if you have cancer, and I think this is something that people who have cancer would like to hear, that's not something that you should blame yourself for.

[B]

PZ Myers of the University of Minnesota. Morris.

[A]

Ah, Tim, I've got news.

[B]

What?

[A]

I've just received a reply to my emails from one of the authors, Christian Tomasetti.

[B]

Fantastic clarity at last. What does he say?

[A]

He says he's writing a technical paper to explain his previous paper.

[D]

Brilliant.

[B]

Maybe then we can do another more or less item to explain this one. In fact, I reckon we could spin this out into a multi part series. What do you reckon?

[A]

I really don't think so. It'll never catch on.

[B]

Well, that's all we have time for this week, but please keep your questions and your comments coming in to more or lessbc.co.uk. our website is bbcworldservice.com more or less. There are dozens of different podcasts now available from the BBC, including news, documentaries, science, business, arts and sports. The details of them all go to bbcworldservice.com podcasts.