There’s been a new talking point in the health care debate in the US; the cancer survival rate. Many bloggers (site, site, site, and site) and research institutes have discussed the results of a 2007 article in the medical journal Lancet Oncology that compared cancer survival rates in several different countries. The blogs and news sites claim the US is the best in the world for cancer survival and the medical journal article seems to back that up. But the United States has the lowest longevity of the industrialized nations, so there’s some cause for skepticism. Does the US have the best cancer survival rate in the world?
Death rates and survival rates are tricky things to consider because, as they say, life is 100% fatal, no one gets out of here alive, etc. Since we’re all going to die we probably don’t care so much about how we die, but rather when. Cancer researchers use two different statistics to track this: mortality rate and survival rate, each has its own strengths and weaknesses. Before we examine the study, lets take a moment to understand the terms used.
Five Year Survival rate.
You get cancer and then within five years after you’re diagnosed you die. This includes deaths for any reason. The benefit of this is that if one gets lung cancer (for example) and dies from pneumonia because ones lungs were weakened by the cancer then it seems correct to attribute this death as being caused by the cancer. But this rate has 2 main problems. One is the “lead time bias” which is illustrated in the graph to the right.
We don’t really care how long we live after the cancer has been diagnosed, but rather how long we live since we actually got the cancer. Consider a cancer that kills everyone within 8 years of onset. Normal screening may be such that it is only caught 5 years after onset. Three years later (8 years after onset) everyone with this cancer would be dead. A 0% 5 year survival rate. If improved screening allowed doctors to detect it four years earlier then suddenly there would be a much higher five year survival rate as everyone now survives 7 years after detection. But everyone still dies 8 years after onset. The earlier detection didn’t actually change the course of the disease, the doctors just detected the cancer earlier. There is no easy way to adjust for this effect in the five year survival rate statistic.
The other problem occurs when one wants to compare survival rates between 2 different countries. If these countries are the same in all other ways except their cancer survival rates then the “observed” survival rate could be used. But this is hardly ever the case. People in different countries die at different rates due to heart attacks, automobile accidents, pneumonia and other causes. Medical researchers adjust for this by computing the “relative” survival rate, which “subtracts out” the “background mortality” as illustrated in the graph below.
To illustrate, if 3 out of 1000 people aged 65 die from pneumonia in a given year (the “background” rate), but 7 out of 1000 cancer patients die from pneumonia in that same year (the “observed” rate) then it would be safe to assume that 3 of those 7 cancer-pneumonia deaths were due to natural pneumonia and 4 were due to cancer-induced pneumonia (the “relative” rate).
You die, the doctors determine the cause of death was cancer. You’re now a statistic in the mortality rate for that cancer. One problem with this statistic is it doesn’t account for cancers that weaken but do not kill you. The lung cancer-pneumonia example above would be counted as a death due to pneumonia by this measure. However, the mortality rate is able to avoid the “lead time bias” problem of the survival rate statistic.
The reason is that the longer you live the higher your chances that something else will kill you. So if more people are dying from other causes and fewer people are dying from this cancer it’s probable that the updated detection and treatment are indeed leading toward longer life.
The Lancet study looked at 3 (or 4) different cancers, breast cancer in women, prostate cancer and colorectal cancer (sometimes separately colon cancer and rectal cancer). Below are their results.
- The difference in survival between most of the top countries is small, but despite that the US is clearly in the top ranks.
- The US is also significantly ahead of the European average.
While it’s hard to say that the US is #1 or #2 in this measure since, statistically many countries are tied for the #1 and #2 positions, it is the case that of the 31 countries studied the US has better cancer survival rates than at least 25 of them. So the articles that cite the US’s superior cancer survival rates are accurate, at least for these cancers.
That’s right, many comments in the news merely state that the US has the best cancer survival rate in the world. Not strictly true for at least 2 reasons. One is we’re actually tied for first or second place, not uniformly nor uniquely #1. Also we’re #1 or #2 only for the cancers studied above. We may have inferior survival rates for other cancers and furthermore the cancers studied above may not be the most significant cancers in the US or world. Let’s analyze these concerns.
Are these the cancers that matter?
Who cares if we have a high survival rate for cancers that don’t tend to kill people or that not many people contract in the first place? Is this the case with the 3-4 cancers studied above? The American Cancer society has a report that lists the top causes of death due to cancer around the world. For developed nations their chart is shown below.
|2007 Estimated Deaths in Developed Countries|
|Total Rank||Cancer||Male Deaths||Female Deaths||Total Deaths|
|1||Lung & Bronchus||465,540||173,842||639,382|
|2||Colon & Rectum||175,774||165,480||341,254|
Combining the genders where appropriate to get total mortality, lung cancer is the leading cause of death by cancer in developed countries (all developed countries combined). The cancers studied in the Lancet article are ranked #2, #4 and #6 in terms of total mortality in developed nations. Additionally, breast cancer is #2 for women (#4 over all), and prostate cancer (#6 over all) is #3 for men. So the cancers studied are a reasonable cross section of the most significant cancers that affect the developed world.
The cited study in Lancet Oncology is the first to compare cancer survival statistics across many different countries. The cancers they chose were a reasonable cross section of the most significant cancers among developed nations. Strictly speaking the blogs claiming the US is #1 in cancer survival are correct. But stated in plain English, without the benefit of charts or graphs, it can be misleading. The graphs above, largely the same graphs as published in Lancet Oncology, clearly show many countries are also very close to the US in terms of cancer survival.
The implied question here is, would our #1 ranking in cancer survival be significantly harmed as a side-effect of health care reform? This is for you to decide from what’s been presented. For my part the answer appears to be “no”. The reason is that there are many countries that are very close to the US in cancer survival rates and these countries have some form of government involvement in health care. While it’s plausible that our #1 ranking in spending is related to our #1 ranking on survival, it’s important to note that we could cut our spending on health care by 40% and still be #1 in the world on spending. It’s hard to believe that such a change would significantly affect the survival ranking. Indeed, there is some recent data that supports that, regarding prostate cancer at least, the US may be over-treating (and hence over-spending on) cancer.
Cause and Effect are tricky things to argue. When you have a big cause and a big effect most folks take the conservative route and assume cause & effect until proven otherwise. But when it’s big cause and little effect, such as the case with spending and cancer survival, extreme conservatism doesn’t seem to be warranted.