age-adjusted cancer incidence rates and death rates have been going down since 1990, primarily due to better social policies like discouraging smoking. Five-year-survival rates have been gradually improving since at least 1970, on average by maybe about 10% though this depends on severity. Although some of this is confounded by improved screening, this is unlikely to explain more than about 20-50% of the effect. The remainder is probably a real improvement in treatment. Whether or not this level of gradual improvement is enough to represent “winning” the War on Cancer, it at least demonstrates a non-zero amount of progress.
Robin Hanson pointed out years ago that various data sources suggested little difference in mortality between populations with extensive medical treatment and populations with less medical treatment. He hypothesized that the cases where medical treatment prolongs lives are offset by cases where treatment hastens death.
How to reconcile Hanson with Alexander? Some possibilities:
1. Perhaps Alexander’s data are more recent and reflect recent progress.
2. Perhaps the difference in cancer survival rate is too small to be significant in overall mortality statistics. You are affecting a small portion of the population, and you are increasing lifespans of the affected individuals by less than 10 years each (they tend to be elderly and prone to dying of other things.)
3. The gains in successful cancer treatment tend to be offset increases in deaths caused by medical interventions that worsen outcomes.
Robin Hanson pointed out years ago that various data sources suggested little difference in mortality between populations with extensive medical treatment and populations with less medical treatment.
This seems like a very exaggerated statement and in my life time the average age of life expectancy moved from 72 in 1970 to 78 today. That is significantly a longer time and the US is on the lower end of spectrum of developed world (Japan is over 80) and say India with less health is closer to 69 and several Africa still below 60. We can question certain aspects of health testing but nations don’t get life expectancy over 70 without good health care to extend life. (I suspect some of this progress on life expectancy slowed down after 1970ish probably because of diminishing returns of new health discoveries. There are no washing hands discoveries anymore.)
1) It takes significant efforts to increase life expectancy by 1 year for developed nations. This is a hard needle to move! Your point on 2 is likely correct and cancer treatments does weaken other health areas. So these cancer discoveries add +.2 years to life expectancy over 10 years.
2) Probably one reason it is so hard to move the health needle is humans like bad behavior. I joke Generation X was great at rejecting smoking but then drank too many Big Gulps (guilty as charged) of soda. Obesity replaced smoking!
3) Is the reverse of US life expectancy the last 5 years been more drug epidemic, obesity and decreased rural health care. (The reason why the latest drug epidemic feels more contained is violent crime is still contained.) The last 15 years maternity deaths are increasing in the US and I do believe some of that is diminished rural healthcare.
I kind of wonder what the data would look like if you excluded North America, as North American metros are uniquely focused on auto-centric land use patterns, and hence more obese and less active than other populations of OECD countries. I suspect, as you suggest, that there is a material difference in life expectancy between developed nations (excluding NA) and developing nations (excluding NA) and that part of that is explained by the greater intensity of medical services.
At the CDC web site one can see cancer death rate trends charted by race from 1975 and projected out to 2020. Cancer is the second leading cause of death in the US. There has been a big overall downward shift in cancer deaths among black males. Males have much higher levels of cancer than females, including black females. The cancer death rate for white women is the lowest of the 4 groups displayed and is flat over the period. https://www.cdc.gov/cancer/dcpc/research/articles/cancer_2020_figures.htm
The greatest drop in cancer incidence rates for black males was between 2003 and 2012 during which time incidence rates dropped 2% per year for all cancers combined.
Hanson looks to be right for white women and can’t be ruled out for black men given the drop could be explained by changes in environmental exposures, smoking and drinking rates, etc. Perhaps Clinton era welfare reform led to increased workforce participation and moved segments of the population out of Medicaid and into employer sponsored care. The Oregon Medicaid study, as we will remember, demonstrated no difference in health outcomes between Medicaid coverage and no health insurance coverage.
Nevertheless the US has had 3 straight years of declines in average life expectancy. A steady rate of decline in heart disease deaths, the leading cause of death, was disrupted by Obamacare penalties for hospital readmissions leading to many preventable deaths. And community rating requirements have penalized diabetics, (diabetes is the 7th leadding cause of death) whose death rates are climbing steeply. All this tends to be swept under the rug by the intellectuals who like to claim it is all the result of working class white drug use. But drug overdoses only killed 63,632 people in the US in 2016 (most recent year available). Nearly two-thirds of these deaths (66%) involved a prescription or illicit opioid. About 2.7 million people died in the US that year so it is difficult to see how one can blame the trend on marginal changes in what amounts to 2% of total deaths.
Well, it would depend on the delta to 2016 back from the point in the past that life expectancy started dropping, right? Also I would guess that overdose deaths from prescription opiates and other opiates skew towards younger people and thus have a disproportionate impact on expectancy.
This emphasises the problems with statistical, one size fits all, medicine rather than exposes fault with particular treatments. Of course the experiment needs to be different for each different patient, which is difficult to achieve at the present time. But if it is possible to determine in advance which patient will be terminated early by particular treatments, this will no longer apply.
Having worked in research in a “world renowned” cancer institute, I would side with Scott Alexander. A few years back there were dramatic outcomes with some of the new targeted therapies. Truly amazing outcomes – cancer disappeared – but in reality, most were really below detection – a few were cured. The treatments did provide the patients with a few more months of life. At first thought the cancer cells were developing immunity to the treatment, and they set about trying to stop the cells from developing immunity. That was not the problem, the main issue was that most tumors are a mix of cells with different mutations (heterogeneous) and the treatments killed off the susceptible cells (usually the most abundant and therefor the cells/mutations that were easily detectable with sequencing technology – leading to the false promise of a cure), but then the other cells took over and killed the patient. There are big issues with figuring this out the first being able to filter out significant low level mutations in a tumor. Sequencers aren’t 100% accurate – is the difference a sequencing error or a mutation? The current method for calling a mutation rely on sequencing the same DNA multiple times (30 times was standard – this is not 1 strand of DNA be sequenced 30x, DNA is extracted from many cells and you want to sequence 30 different DNA sections) and only calling ones that exceeded a threshold (say 5 out of 30), and importantly no 2 genomes are the same, so which differences matter – some are known and I would imagine this is a very small percent), the second is to have drugs that can target the other mutations. Drug discovery is ongoing – never as fast as reported, new sequencing methods are coming online – like single cell sequencing, data analysis is getting better, and finally researchers always add to the known list of significant mutations (COSMIC). I was involved in the data analysis part, so this hardly an expert opinion, but the problem is huge – progress will be slow.
Quality is not the same as quantity. It’s completely possible for the quality of treatment to improve over time, even as at each point in time people are often over-treated. I’d say it’s likely.
Scott Alexander writes: “Also, something has to kill you, so if other issues like violent crime or heart disease get better, it will look like a higher cancer rate.”
The reverse logic of this is that if the US starts a war with Iran and/or North Korea, less Americans will die of cancer.
As an aside, here are the rough figures for cancer deaths each day in the US:
“According to the American Cancer Society, in the United States, about 1,620 people were expected to die of cancer each day in 2015.”
Here’s how many die from opioid overdoses:
“Every day, more than 115 people in the United States die after overdosing on opioids.”
So that’s 115 every day that aren’t dying of cancer. As the opioid epidemic worsens, we should see a lower cancer rate.
Observe that in The Elephant in the Brain, Hanson and his co-author make a good argument that some part of healthcare spending is “conspicuous effort”. If this is combined with the recent paper claiming that healthcare spending in the US is so high simply because of high prices and wages, the combined implication is that % of GDP or $ equivalent per capita, don’t really measure the effects of healthcare effort vs life expectency. They are really measuring rich people going to great lengths to be seen going to great lengths, and perhaps rents captured by the healthcare industry.
There are also accounting problems – on a recent ER visit I watched 2 hospital police staff the ER waiting area – they’re clearly there full time. The behavoir of various drunks and the like suggest they are needed there full time. The cost of staffing that post will show up in the cost of healthcare. But of course that doesn’t inform “healthcare effort vs life results.”
Now also observe that national level life expectency is heavily affected by the genetic and social circumstances and mix of the population – and there are all sorts of well known statistical fallacies that arise from this. So saying that male life expectency in Canada is 80 while in the US it’s only 75, has little meaning given the 10x to 1 difference in population size, and the differences in population mix. (And in my zipcode it’s apparently 81 – can I expect to outlast all of my peers in canada?)
What does this have to do with cancer? The same sorts of confounding effects will apply, but in droves. The disease is often somewhat slow, so people have time to seek out and apply maximal treatment. It’s fatal, so people are motivated to do that. As Arnold notes, even if the cancer is cured the people in question may well die of something else soon anyway.
A short search finds a source reporting that adult survivors of childhood cancer have reduced life expectency compared to those who didn’t have cancer. OK, not happy news. But having a life expectency of 55 is surely better than dying of cancer at 12. Do we call that a “cure”?
Actually, cancer isn’t always fatal because some cancers are so slow growing that something else will get you first. The number of males who have prostate cancer but die of something else is astounding. (If you are diagnosed with prostate cancer, NOT treating it may be a good idea. Since treatments all have risks and side effects, “watchful waiting” is often recommended. )