By now you’ve no doubt read that cancer death rates (age-adjusted) are down 27% since 1991. The better sort of news articles point out that this is a bit of cherry-picking, as cancer death rates peaked in 1991, and current progress has brought us back to the rates that prevailed in the 1940s. This graph in the source article, Cancer Statistics, 2019, makes the point clearly:
A few sources reported out that cancer incidence also is down (more so for men than women) and thus reduced incidence accounts for some fraction of the drop in deaths:
And some publications even took note of this figure, which highlights disparities due to race and income:
So I’m not going to write about any of those issues; they’ve been covered and you probably know about them already anyway. But it’s my duty as a blogger to say something interesting that hasn’t been said, so I’ll give it a try.
The whole point of collecting statistics is evaluation–to see what’s working and what’s not, what needs are being met and which are being neglected. And “evaluation” means just that, assigning appropriate values to different approaches.
The key divide in our approach to reducing suffering and death from cancer is between prevention and treatment. There’s no reasonable debate that both are important, but there certainly is room for disagreement about how to allocate resources between them.
It won’t surprise you that we pay way more for cancer treatment than for cancer prevention. US spending on cancer treatment is about $150B per year. That number seems low to me; the spend on cancer drugs alone was about $40B in 2014, over a quarter of the total, whereas drugs account for about 10% of healthcare spending overall. But let’s go with the official numbers. At $150B per year and an incidence of 1.7M per year, that works out to a spend of $88K per year per case.
Mortality is 600K per year, equal to 36% per new case, a number that has not changed much in the last decade, despite all the ballyhoo about personalized medicine and other purported breakthroughs.
$88K of treatment thus prevents 0.64 deaths, for a cost of $138K per death prevented. Actually, this is too generous, as it assumes every diagnosis eventually results in a cancer death even without treatment. So $138K is the minimum cost of preventing a cancer death via treatment.
How much does it cost to prevent a new case of cancer? No one knows, which is itself a tell. I spent several hours trying to find a decent answer to what in fairness is a very complicated question, and struck out completely.
So let’s ask a simpler question. How much does it cost to get a smoker to quit? According to the CDC, it’s about $480. Smoking raises the lifetime risk of cancer death from 14% to 26%, so that $480 buys about a 12% risk reduction (although it takes several years until former smokers risk matches that of non-smokers). So let’s call it a 10% risk reduction, meaning that the CDC’s ad campaign prevented cancer deaths at a cost of $480/0.10 = $4800. That compares very favorably to the $138K required to prevent a death through treatment. Quitting smoking prevents all kinds of other diseases and deaths as well, whereas cancer treatment, even when successful, causes significant morbidity.
Although smoking accounts for about 30% of the US cancer burden, there are plenty of other preventable causes of cancer. Estimates of cancers attributable to diet/obesity, infectious diseases and alcohol range from 20% to 60%. The potential cost savings here–not to mention the reduction in human suffering–are enormous. Reducing preventable cancers by 20% would save anywhere from 60,000 to 100,000 US cancer deaths per year. It would be the equivalent of finding a cure for both breast and prostate cancer.
So you’d expect a pretty substantial fraction of cancer R&D to be directed at prevention. Prevention is effective, it is cheap, and it has all sorts of ancillary health benefits.
Ha. Just 5.7% of the National Cancer Institute’s 2018 budget of $5.6B is directed toward cancer prevention. $320M is not nothing, but it’s less than the Cancer Moonshot initiative alone ($400M), and a trifle compared to the tens of billions of dollars spent yearly by pharmas to develop oncology drugs.
So it’s great that cancer deaths are down by 27%, but the real story is that most of that drop is due to a reduction in cancer incidence, mostly due to smoking reduction, while our cure rate (deaths/diagnosis) has stalled out per the graph above. All those new and pricey cancer medicines are doing just about nothing to reduce cancer deaths.
You won’t find that story coming out of our Cancer Industrial Complex, nor from the sycophantic journalists eager to report on the latest gee-whiz breakthrough in therapeutics. Preventing cancer is boring; curing it is thrilling.
And lucrative.
A $100B market with nearly 10% annual growth is a trough that everyone wants to dive into. Oncology drugs are the modern gold rush: as of 2015, there were 586 late phase therapies being developed by 511 companies.
With that kind of cash sloshing around, it’s not hard to suborn “key opinion leaders” whose ego-driven needs for validation and recognition make it easy to rationalize grabbing a golden ticket. After all, even if you are caught, the consequence appears to be a prestigious and lucrative job in pharma.
It’s time to call bullshit on this charade. It’s time to recognize that cancer prevention is far more effective than cancer therapy in saving lives, and its time to put our money into prevention efforts rather than high-priced therapies that are doing nothing to reduce deaths from cancer.
First rate!
How much of that drop in deaths is due to early detection finding things that would never have killed the patient? I understand that most supposed prostate cancers are false alarms.Are those numbers corrected for that bias?
That’s a tough question but a good one.I don’t think a definitive answer is possible. There is too much disagreement as to how to define and detect those false alarms.
The most I think we can hope for from these numbers is that they are consistent, even if biased, over the time period reported. That at least gives us some info with regard to trends. I worry that my graph which shows death per diagnosis is biased in a way that makes the apparent flattening out of progress just an artifact. But that would mean a drop in diagnostic yield in the last 7-8 years, while deaths from older cases continued apace, which hardly seems possible.