- Avik Roy, whose columns I recommend if you’re interested in health care policy, debunks the myth that Americans have a poor life expectancy. As he notes, this claim is often used to argue in favor of socialized medicine, but it’s false.
- The New York Times published a piece according to which support for gender equality in the household among millennials was significantly lower than among people the same age 20 years ago. Interestingly, it has increased in Europe during the same period, which the author of that piece attributes to the absence of family-friendly work policies in the US. This may or may not be the case, I haven’t checked the studies she cites, but in any case this is interesting.
- For April 1, the Russian Foreign Ministry posted a fake automated message on its Facebook page, supposed to be for Russian embassies:
“To arrange a call from a Russian diplomat to your political opponent, press 1,” the recording begins, in Russian and English. Press 2 “to use the services of Russian hackers,” and 3 “to request election interference.”
I personally think it’s pretty funny, but journalists are so stupid that I wouldn’t be entirely surprised if some of them actually took that seriously. I have been thinking to comment more on the Russia/Trump nonsense, which has gotten even more nonsensical, but 1) I really don’t have time to write another 16,000 words post to debunk the media’s narrative on this and 2) at this point it’s becoming increasingly clear that people who take this seriously are so lacking in critical thinking that nothing I could say would change their mind.
- On the same topic, Michael Tracey from The Young Turks (a well-known fascist news organization), explained the recipe for every shocking Trump/Russia “revelation” after the non-story about Sessions’s meeting with Kislyak. I wish my friends on Facebook would read this, when they’re not busy being manipulated.
- In case you missed it, Joseph Heath published a very good post on what he call’s “me” studies a while ago, in which he gives what I take to be a very good explanation in terms of ideological uniformity of the poor quality of so much of the scholarship that focuses on gender/race and various forms of “oppression”. He also wrote a follow-up which I also recommend.
5 thoughts
Comments are closed.
The Roy article is pretty bad. Here are the three main clams from the article:
1. “If you really want to measure health outcomes, the best way to do it is at the point of medical intervention… In 2008, a group of investigators conducted a worldwide study of cancer survival rates, called CONCORD. They looked at 5-year survival rates for breast cancer, colon and rectal cancer, and prostate cancer. I compiled their data for the U.S., Canada, Australia, Japan, and western Europe. [The US] came out number one[.]”
5-year-survival-rates for cancer are actually a terrible measure of health care outcomes, due to lead-time bias. All the 5-year-survival-rate tells you is what proportion of patients diagnosed with the disease are still alive five years after the time of detection. This means that you can inflate a 5-year-survival-rate just by performing lots of unnecessary tests to find cancers when they’re still in their preliminary stages, even if detecting the cancer early doesn’t help to prolong the patient’s life at all. For instance, if both of us have an inoperable tumor which will kill us at 70 no matter what, but yours is diagnosed when you’re 60 and mine when I’m 66, you’ll count as a having “survived” 5 years while I will not. Clearly, though, this isn’t because you’re receiving better treatment.
Last time I checked, lead-time bias turned out to be a large part of why the US does well in international comparisons of 5-year-survival rates. The fee-for-service model combined with patient litigiousness leads doctors in the US to perform lots of needless tests, which means they tend to catch cancers earlier. But, as the recent controversies over colonoscopies, mammograms, and PSA tests suggest, detecting cancers earlier is not, on balance, a good thing for patients, because the number of false positives from universal screening is so high.
More on lead-time bias:
https://theconversation.com/when-talking-about-cancer-screening-survival-rates-mislead-30395
2. “[W]hat happens if you remove deaths from fatal injuries from the life expectancy tables? Among the 29 members of the OECD, the U.S. vaults from 19th place to…you guessed it…first. Japan, on the same adjustment, drops from first to ninth.”
On its face, this claim is fairly implausible. Homicide is a blip in overall mortality, so its impact on life expectancy should be negligible. Driving fatalities are a little more common, but I would still be very surprised if they took an average of 2 years off of people’s lives, as the table Roy presents suggests. Here’s an article which discusses these points in more detail:
http://theincidentaleconomist.com/wordpress/how-flawed-is-life-expectancy/
The author notes that life-expectancy in the US at 65 is still lower than in many other OECD countries– but auto fatalities should be rare after 65, and homicides effectively zero. So something extremely fishy is going on here.
But it doesn’t really matter, because life expectancy is not a good measure of the quality of a health care system in any case. Rates of obesity and tobacco use, unlike car accidents and homicides, are going to account for a lot of the difference between countries in life expectancy, and medical providers have only very limited control over their patients’ smoking and dietary habits.
3. “As I have noted in the past, health outcomes for those on government-sponsored insurance are worse than for those on private insurance.”
There must be something I’m missing here, because I can’t believe Forbes would publish something this stupid. In the US “those on government-sponsored insurance” are the poor, old, and permanently disabled, while “those on private insurance” are the young, healthy, and affluent. Medicare and Medicaid also control costs by paying doctors less than most private insurers, which significantly restricts patient access to health care. Obviously you’re better off with private insurance under the current regime, but this tells us absolutely nothing about what health outcomes would look like were we to switch to single-payer.
1. This is a good point, which I hadn’t thought about, but frankly I would be surprised if the bias were large enough to fundamentally alter the ranking between countries. If you know of something which shows that, when you take this into account, the US falls behind other developed countries, I’m interested.
2. I think you’re right that something weird is going on and I think I know part of the explanation. Based on what Roy says in his note at the end, and what the blog post you link to says in passing, I think what the study he uses did was run a regression about life expectancy with the rate of fatal injuries as one of the independent variables, then use the effect-size for the rate of fatal injuries to estimate what life expectancy for various countries would be if they had a rate of fatal injuries equal to zero. I can’t be sure because I didn’t read the study and I don’t have time to do so, but I’m pretty sure that’s what they did. It’s a very indirect way of making that comparison, but it would probably be very complicated to do it directly, which I’m guessing is why they used that method. I’m really not sure what to make of this. But in any case, I think the important point is the one you make at the end, namely that the health care system probably explains very little of the difference in life expectancy between developed countries.
3. You misunderstood his argument. If you read the post he linked to, you’ll see that he discusses a study which shows that, for a number of outcomes, people on Medicaid and Medicare do significantly worse than people on private insurance, even when you control for income, background illnesses, etc. (This may not take into account cost-effectiveness, but I don’t have time to read the UVA study he is using, so I’m not sure.) His argument is that, even if life expectancy in the US was worse than in other countries because of its health care system, the evidence suggests that something like Medicare for all would probably make things even worse. (But as you pointed put yourself, the health care system has probably little to do with life expectancy, at least past a certain level of economic development.)
1. Here’s a NYT piece about how lead-time bias distorts survival rates (coincidentally, by the same guy who runs the incidentaleconomist blog):
https://www.nytimes.com/2015/04/14/upshot/why-survival-rate-is-not-the-best-way-to-judge-cancer-spending.html
“The differences in mortality rates between the United States and Western Europe are nowhere near as large as the differences in survival rates. Even so, the United States often outperforms Europe. From 1982 to 2010, it’s estimated that we averted almost 67,000 deaths from breast cancer compared with Western Europe. We averted almost 60,000 deaths from prostate cancer and almost 265,000 deaths from colorectal cancer.
But at what cost? The researchers found that the incremental cost of each year of quality adjusted life, or QALY, gained for colorectal cancer was $110,000. For breast cancer, we spent more than $400,000 per QALY gained. For prostate cancer, we spent almost $2 million per QALY gained.
We often focus on breast, colorectal and prostate cancer because we do better with those diseases. But we don’t with all cancers. Over the same period, the United States had more than 1.1 million more deaths from lung cancer than Western Europe. Because we still spent more on care for this disease, we had a negative cost of about $19,000 per QALY gained. We also had negative costs per QALY gained for other cancers, including melanoma (about $137,000) and cervical cancer (about $855,000).”
I also want to emphasize a point I mentioned earlier: even if the US does manage to prevent a few more deaths from prostate and breast cancer than other countries, there’s good reason to think that this is not, in fact, in patients’ best interests. The US Preventive Services Task Force and other groups have been urging for years that we roll back universal screening for various types of cancer, because, for each handful of lives prolonged by early detection, there are thousands of false positives, pointless biopsies, and unnecessary procedures. And these recommendations are made on purely medical grounds, not even taking into account the added financial burden for patients and the government. So, even if five-year-survival-rates were a good way of measuring quality of cancer treatment, which they’re not, the US outperforming other countries would not actually show that patients here are, on balance, better off than patients in other countries.
2. “I think you’re right that something weird is going on and I think I know part of the explanation.”
It’s not just that something weird is going on, it’s that excess mortality due to the higher rate of traffic fatalities in the US couldn’t possibly come close to accounting for the data Roy presents. Here’s an actuarial table showing the lifetime odds of dying in various types of accidents:
http://www.iii.org/fact-statistic/mortality-risk
The lifetime odds for an American of dying in a car crash are 1 in 113. If we assume that a traffic fatality takes off, on average, 50 years of the victim’s life, this means that all car crashes in this country together reduce our life expectancy by only around 6 months. Yet the table in Roy’s article has life expectancy in the US rising from 75.3 to 76.9 once “fatal injuries” are factored out of the data, while life expectancy in Japan plummets from 78.7 to 76, for a total of a 5 year swing. But Japanese people die in car accidents, too (at about 1/3 the rate of Americans), which means that car accidents can account for, at maximum, about 4 months of those 5 years. So the “explanation” Roy offers falls short of explaining the data, by more than an order of magnitude.
I suspect what’s happened here is that the largest category of “fatal injury”, by an enormous margin, turns out to be (roughly speaking) old people falling down, and what Roy’s data actually shows is that US life expectancy is higher than other countries when you factor out old-people-falling-down. But there’s no principled reason to think that mortality from old-people-falling-down doesn’t tell us something about the quality of a country’s health care system, so if life expectancy in the US shorter than life expectancy in other countries because of this sort of “fatal injury,” that may mean that health care in the US is inferior after all. If my hunch is correct, this also raises serious concerns about how causes of death are being classified in different countries. We can expect some degree of consistency across countries in attributing deaths to car accidents or homicide, but I’m not at all confident that Japan, the US, and Germany will have comparable classification practices when it comes to death-by-fatal-fall.
Let me reiterate, though, that I agree that crude international comparisons of life-expectancy are so thoroughly confounded that they essentially tell us nothing about the quality of a country’s health care system.
The link you provided neglects deaths by suicide. Might that be a reason for part of the discrepancies? (There the question arises if the healthcare system is or is not responsible for mental health)
What you didnt mention in your critique and seems important to me is about the graph first shown in the article. First I do not really understand why it is in %. % of what?
Then more importantly, I do not see a strong correlation, but a rather weak one.
For poor countries its vertical, nearly no correlation at all.
For richer countries there is a correlation (obviously), but it seems rather weak with a lot of noise.
1. What we are wondering is how the US could gain so many years of life expectancy and Japan lose so many when “fatal injuries” (including auto accidents, homicide, and suicide) are factored out of the life expectancy data. To explain this, we need to find classes of fatal injuries which occur at a much higher rate among Americans than among the Japanese. Suicide isn’t a candidate, because Japan has one of the highest suicide rates in the world, while Americans commit suicide at a rate right around the average for OECD countries.* Hence, excluding deaths by suicide from our calculations will increase Japanese life expectancy more than it increases American life expectancy, making the shift we see when the remaining types of fatal injury are also factored out even larger and more inexplicable.
*See https://data.oecd.org/healthstat/suicide-rates.htm
2. As far as I can tell, the vertical axis in Roy’s graph of life expectancy versus GDP per capita is just mislabeled; it should be in years, not percentages.
I don’t think it makes much sense to disaggregate the poor countries from the rich countries and look for a correlation between wealth and life expectancy separately in each. What would motivate this? The whole point is that inhabitants of poor countries don’t live as long as citizens of wealthier nations, and you obviously won’t be able to see this if you don’t compare the two groups.