Comment by tokenadult | Hacker News Reader

Comment by tokenadult | Hacker News Reader

The numbers mentioned in this article are numbers of cancer CASES, for the most part, not numbers of cancer deaths. The chart lower down in the article, captioned World Cancer Cases 2012, at least shows countries in different shadings based on cancer RATES (standardized to cases per 100,000 population, as is typical for international health statistics), but that mostly shows which countries have higher rates of diagnosis of cancer (or much lower rates of diseases that kill people when they are young) rather than countries where cancer is the main cause of death. Rates matter more than the raw number of cases in a world that still has a growing population.

If we look carefully at charts of death rates by all causes for either the United States or the developed world, we see that death rates from cancer and from most causes of death are still steadily falling at all ages. Life expectancy at age 40, at age 60, and at even higher ages is still rising throughout the developed countries of the world.

http://www.pinterest.com/pin/195977021257786739/

http://www.scientificamerican.com/article.cfm?id=longevity-w...

Yes, we still need to do more research on cancer prevention and cancer treatment. Yes, the article is correct that smoking, infectious diseases (mostly viral, some of which are already preventable by vaccines), overuse of alcohol, and obesity all increase cancer risk, but you and I can do something about our use of tobacco, alcohol, and fattening food-and-activity lifestyle combinations today--we don't have to wait for more medical research to reduce our risk of cancer. The article kindly submitted here makes clear that making people more aware of what increases their cancer risk is one of the best interventions for reducing cancer rates and cancer mortality. As other comments have already correctly pointed out, a rise in the number of cancer cases worldwide first of all indicates that there are more people in the world than ever, and secondly indicates that fewer people are dying young of diseases linked to stark poverty. That's good news, on the whole.

I find it particularly frustrating that doctors of all people are so bad with statistics. I know people in general are bad with statistics, and when journalists misunderstand them I have some sympathy, because hey they probably got through college without seeing a number, but I expect more from doctors.

...but I expect more from doctors.

Is that really reasonable, though? Doctors don't have a use for statistics in their daily professional lives, and they almost certainly have less free time in which to immerse themselves in a field of math that isn't going to help them.

Journalists, on the other hand, often should or need to rely on statistics to convey useful and pertinent information in a meaningful way. That they often obviously aren't taught and otherwise usually don't learn statistics, or at least not effectively, strikes me as a much bigger problem than doctors who don't have that knowledge.

Now, the people who write reports like these, they should have and use a thorough working understanding of statistics. I'd say what's really missing is effective collaboration between the people who do the work and the people who should be doing and writing the analysis.

1. Some large percentage, possibly more than half, of what a doctor does is diagnosis and prognosis. This is all pure probability. Failure to understand the difference between Prob(A|B) and Prob(B|A) has nontrivial consequences.

2. Frontier areas of medicine -- oncology, psychiatry -- deal increasingly with uncommon situations that not necessarily happen with enough frequency for a doctor to develop a doctorly intuition. There are some great resources -- meta-analyses, for example -- out there, but if we keep cutting slack to practitioners and let them follow the word of "thought leaders", medicine becomes the prey of ideological waves and conceptual fashions.

> Doctors don't have a use for statistics in their daily professional lives...

You can't give patients good advice that allows them to make the trade-offs inherent in life without a solid grasp of statistics. Saying an IUD has 98% effectiveness is useless without citing the time span over which that probability is measured (2% will get pregnant per year, per 5 years, ever?) Saying that doing X increases the risk of outcome Y by Z% is useless without citing the base rate. Would you stop eating meat if it increased your risk of a rare cancer by 300% if the base rate was 0.01% (over a lifetime) to begin with? I don't know how you make diagnoses without knowing the relative statistical probability of conditions that may present with the same symptoms.

> Doctors don't have a use for statistics in their daily professional lives

I really think that depends on what they are doing. My girlfriend is a neurologist investigating new treatments from MS. She uses statistics packages daily. Her colleague has gone of to London for a year to study a statistics masters. Sure your average GP probably doesn't have much need for stats, but there are a lot of fields within medicine.

Unless they are practicing cargo cult medicine, doctors actually do need statistics regularly in their professional lives -- medicine isn't a static field and you can't keep up with the new results in the field without being able to understand and apply statistics.

Historically doctors have been the most educated people. Not many people went to Uni and very few had a Masters. Doctors had 5-7 years of rigorous education and did quite a bit of lifelong learning, including following scientific research. To most people, they were by far the most educated person they interacted with.

Today, a lot of people have degrees and even the ones who don't have the opportunity to engage with a lot of self education through the internet, a lot of it informal (like HN). Doctors are also more knowledgable than they were but the gap has narrowed. The difference between doctors and the median (or 75 percentile) person has narrowed. Obviously, they still know a helva lot more about medicine, especially in their specialization but when it comes to Science as a discipline or statistics/math/etc., they aren't that far ahead of the pack.

*BTW, oncologists are probably the most versed in science, research methodology & statistics because they participate in a lot of research and choosing a course of treatment may be a choice between treatments based of studies with samples of insufficient size or duration to make absolute statements about. One study follows remission rates of patients segmented by the presence of specific genes in the cancer. Another remission rates for a slightly modified treatment and slightly different segmentation. etc.

Anyway, I don't think cases is irrelevant or incorrect thing to look at here. The number of cases is the number of people that need to be treated. If you are talking about prevention, the number of cases is a relevant indicator of success.

Reply on news.ycombinator.com