Higher US expenditures on cancer patients do not result in improved mortality.

But you'd never know that reading AEI's highly dubious contribution to the literature in this week's Health Affairs (lay Reuters article here). Consistent with their free-market solves everything and can do no wrong (cover ears and yell "nananananananana") attitude towards the broken US healthcare system, they have managed to contaminate the literature with a paper that suggests our higher expenditures on cancer are generating significant returns in patient survival. Except that it doesn't show this, and to her great credit, Reuter's Sharon Begley nails it:

Cancer patients in the United States who were diagnosed from 1995 to 1999 lived an average 11.1 years after that, compared with 9.3 years for those in 10 countries in Europe, researchers led by health economist Tomas Philipson of the University of Chicago reported in an analysis published Monday in the journal Health Affairs.

Those extra years came at a price. By 1999 (the last year the researchers analyzed), the United States was spending an average of $70,000 per cancer case (up 49 percent since 1983), compared with $44,000 in Europe (up 16 percent). Using standard figures for an extra year of life, the researchers concluded that the value of the U.S. survival gains outweighed the cost by an average $61,000 per case. The greater spending on cancer care in the United States, they conclude, is therefore "worth it."

Experts shown an advance copy of the paper by Reuters argued that the tricky statistics of cancer outcomes tripped up the authors.

"This study is pure folly," said biostatistician Dr. Don Berry of MD Anderson Cancer Center in Houston. "It's completely misguided and it's dangerous. Not only are the authors' analyses flawed but their conclusions are also wrong."

I am amazed, this is real science reporting here, because rather than just doing a press-release rehashing of the AEI-authors' dubious assertions, she found some real cancer experts and shredded it.

The mistake they made, or possibly ignored, is lead-time bias. By using survival from diagnosis, rather than the patient's actual mortality, they create the appearance of improved outcomes. But the reality is, earlier diagnosis is creating a false survival signal. The US isn't actually extending lives in comparison to treatment in other countries, the overall mortality rates are the same.

Further the paper only showed the benefit in prostate cancer and breast cancer, and, if anything, worse survival for colon and uterine cancer given the amount of expenditure. The reason is pretty straightforward, and consistent with the lead-time bias issue. In the US, we probably over-screen for breast and prostate cancer, which means more people live with these diagnoses than do in other countries. It's been a topic of debate among medical professions and discussed extensively by other medical bloggers like Orac because it's quite possible, especially for breast and prostate cancers, that screening protocols are too inclusive. The result is there are more patients in the US that are given a cancer diagnosis, but they have disease that may never progress to being a life-threatening illness. Excessive screening may even result in unnecessary procedures and treatments when it comes to these two diseases, and we are still trying to work out what protocols will include the most patients with serious disease, while excluding as many false-positive patients as possible. This is acknowledged by Begley's expert reviewers:

Even more problematic, said Berry, is a problem cancer experts have only recently recognized: overdiagnosis. Because cancer screening is much more widespread in the United States than in Europe, especially for breast and prostate cancer, "we find many more cancers than are found in Europe," he said. "These are cancers that tend to be slowly growing and many would never kill anyone."

Screening therefore turns thousands of healthy people into cancer patients, even though their tumor would never threaten their health or life. Counting these cases, of which there are more in the United States than Europe, artificially inflates survival time, experts said.

"As long as your calculation is based on survival gains, it is fundamentally misleading," said Dr. H. Gilbert Welch, a healthcare expert at the Dartmouth Institute for Health Policy & Clinical Practice.

In the new analysis, the survival gains in the United States compared with Europe were greatest for prostate cancer, at more than triple the gains for breast cancer, the cancer with the second-greatest U.S. survival edge. "These are the two cancers where screening has raised the most serious issues about lead-time bias and overdiagnosis," said Welch.

For melanoma and colorectal and uterine cancer, survival gains over the period analyzed were greater in Europe than the United States.

If anything the opposite is true based on the correct analysis which is based on mortality statistics, not survival.

Other calculations cast doubt on the superiority of U.S. cancer care. For instance, breast cancer mortality fell 36 percent in the United Kingdom from 1990 to 2006, calculates MD Anderson's Berry, and fell 30 percent among whites in the United States. (The U.S. figure would be even lower, he said, if it included African-Americans, who generally have less access to health care.)

Cancer mortality in the United States is higher than in 11 countries reporting to the Organization for Economic Co-operation and Development, and lower than the rate in 14. Mortality is lower in Switzerland, Sweden, Japan and Finland, among others, but higher in Hungary, Slovenia, France and Britain, in the latest years for which OECD has data.

The reduction in cancer mortality in the United States since 2000 also puts it toward the middle of OECD countries. It is less than in Israel, Japan, Switzerland and some others, for instance, but better than in Britain, Estonia and Poland.

To sum up, the authors successfully identified lead-time bias in two cancers which the US is known to screen more for than other nations with universal healthcare systems that have health expenditures of roughly 50% less than ours, per capita. They then falsely attributed this lead time bias to our excessive health expenditures, then generalized this biased finding to suggest our excessive expenditures overall provide benefit to our population. This is despite a large body of literature, using appropriate measures of health outcomes, that suggest many other countries do better than us in cancer and medically-preventible disease mortality in general.

It is amazing that this paper passed peer-review, and that such a substandard analysis by authors with clear ideological biases was not detected and rejected. And one could have predicted this considering the author's affiliation with AEI and Manhattan Institutes, think tanks which routinely engage in denial of science like global warming. This is a comparable mistake to accepting a paper from the Discovery Institute, or Peter Duesberg without approaching it with a hefty dose of skepticism. Anti-science ideologues tend to write shoddy, unscientific papers, and some basic peer review should have prevented this analysis from contaminating the literature.

Shame on Health Affairs for publishing this garbage, but let's a salute Reuter's for doing an excellent job in appropriate post hoc review of this flawed paper.


More like this

Last week, the Washington Post took Rudy Giuliani to task for an ad where he claims that his chances of surviving prostate cancer -- which he had about 6 years ago -- were much higher in the US than in the UK. The ad is meant to indict those who wish to modify the health care system. He says: "I…
by Kim Krisberg Another study, another support beam in the argument that access to insurance coverage matters — a lot. In a study published this month in the journal Health Affairs, researchers took a look at rates of amenable mortality deaths — in other words, deaths that shouldn't happen in the…
by Kim Krisberg When it comes to good health, America is far from top dog. Yes, we may spend the most, we may have some of the most advanced medical technologies and we may produce some of the best doctors. But when it comes to the ultimate measure of a health care system's success — the health of…
Friday, the NY Times described the relatively paltry efforts in cancer prevention, compared to those for heart disease. Not that researchers haven't been busy figuring out how to prevent various cancers: Then, in 1999, he had a chance to do another breast cancer prevention trial, this time of an…

This is an interesting post, but I'm not sure I agree with your solution. Cause-specific mortality rates reflect not only that cause, but all other causes, yes? In particular, mortality rates due to one cause will increase if the hazard due to other causes is reduced. So when you look at cancer mortality, what you see reflects changes in the hazard of mortality due to cancer, and also changes in the hazards of mortality due to everything else.

In contrast, changes in post-diagnosis survival are specific to the cause in question and to a population that has that cause. I'm not sure how to deal with the differences in detection that you mention, but switching to cancer mortality does not seem like a good solution.


By ecologist (not verified) on 13 Apr 2012 #permalink

Shouldn't earlier detection for those cancers lead to better mortality rates? I thought finding cancer earlier gave you a better chance of surviving. Or are we simply finding it so early that it doesn't actually make a difference? Sorry if this is obvious to you, but I don't know that much about medicine.

By CaptainBooshi (not verified) on 13 Apr 2012 #permalink


It's heavy reading if you follow the link to Respectful Insolence that MarkH provides discussing the subject in more depth, but definitely worth it.

One of the reasons early detection is not as helpful as one might think is that some portion of the tumours detected are not growing at all or are growing very slowly, such that the people in whom they are detected are much more likely to die of some other cause long before those tumours could become a threat.

The problem is that given the current state of medical science we can't yet tell which tumours we find when screening are redolent (I think that's the right term) and which are going to cause problems - so we check and remove them all, putting patients at risk of the side effects of biopsies, surgery, chemo, and radiotherapy without them receiving the benefit that they would get if we were treating genuinely dangerous cancers.

Other reasons get discussed by Orac, of course. That's the main one that springs to mind for me.

If anything the opposite is true based on the correct analysis which is based on mortality statistics, not survival.

That's a bit overstated. Improvements in palliative care for incurable cancer is still a valuable accomplishment in medicine, after all.

As Orac has pointed out, it's hard to come up with a good metric for this and mortality is certainly a better statistic than survival -- but it's a long way from "correct." More like, "less misleading."

By D. C. Sessions (not verified) on 13 Apr 2012 #permalink

composer99, the word you want is "indolent."

However, a the first-order problem with survival rates and earlier detection is that even if treatment does absolutely nothing to alter the course of the disease, detecting it a year earlier increases survival by a year. The standard remedy for this problem is "stage-adjusted survival:" grouping cancer diagnoses by the stage of disease progress at the time of diagnosis.

Any discussion of survival (as distinct from mortality) that doesn't at least use stage-adjusted rates is not even wrong.

By D. C. Sessions (not verified) on 13 Apr 2012 #permalink

Looking for samples? I received the sample quickly. Thanks to "Official Samples" for the samples. I can't wait to get another in the mail!

By robertzbaker (not verified) on 14 Apr 2012 #permalink

I'm not going to argue the statistics here, but point out one additional fact. If anything, these results are a gross underestimation. In 1999, molecular therapies (TKIs, mTOR inhibitors, etc), were still at least two years away. Biologics (with the exception of rituximab and maybe one or two others), were not available.

Why does this matter? Because these are REALLY expensive therapies. A year's worth of Soliris (used to treat a hematologic disorder, but which may get lumped into onc therapy) costs upwards of $400,000 dollars. Brentuximab (an antibody conjugated to a microtubule inhibitor, used for refractory Hodgkin lymphoma and a few other related diseases) costs upwards of $15,000 per dose for up to 16 doses. Provenge (a T-cell antibody for prostate cancer) costs $97,000 for a median survival benefit of 4 months. Avastin, Erbitux, Gleevec, Tarceva - all effective, but at a substantial cost that would have been unthought-of a decade and a half ago when those numbers were generated. And for a substantial subset of patients - think Her2/neu overexpressing breast cancer or virtually any metastatic colon or lung cancer - these therapies have become standard of care in the US. In many other places, the subtle benefits that they provide (i.e. increased survival by weeks or months, not years) have been determined to not be worth the cost in dollars (or euros, pounds, whatever).

My point? Yes, these drugs work. That's why they have been built into the regimens that we use. But they come at a substantial financial cost. And these costs have exploded in the last five years in a way that is not reflected in the data here.

By Brandon K (not verified) on 14 Apr 2012 #permalink

Just wanted to point out that 'improved mortality' is not really what we're looking for in cancer treatment. Reduced mortality is more the point. Improved mortality rates would work, but as it is this headline sounds like it's referring to more comfortable or more efficient death, and not to better rates of survival.

Thanks for responding to my question, composer99 and D.C. Sessions. I think I understand a little bit more what the article is referring to now. Mainly, the study was done so poorly it was "not even wrong," and better-done studies have shown that there is no real improvement based on our higher spending.

By CaptainBooshi (not verified) on 15 Apr 2012 #permalink


>>I agree that BC and P cancer are overdx in the U.S. More than you say. However I add the issue of how cancer survival is measured.


David Emerson

But you don't - you quote the study approvingly - stating that Philipson accounts for the bias when measuring diagnosis to survival, and thyen don't even discuss the fact that lots of people are unnecessarily given treatment.

However, a the first-order problem with survival rates and earlier detection is that even if treatment does absolutely nothing to alter the course of the disease, detecting it a year earlier increases survival by a year. The standard remedy for this problem is "stage-adjusted survival:" grouping cancer diagnoses by the stage of disease progress at the time of diagnosis.

Any discussion of survival (as distinct from mortality) that doesn't at least use stage-adjusted rates is not even wrong.