Steven Brill's opus on the exorbitant costs of U.S. health care is quite informative. His central point-that medical prices are much higher in the U.S. than elsewhere-has been known to health policy scholars for a long time. Nevertheless, Mr. Brill has done a great public service in bringing this message to the attention of the general public with eye-opening concrete illustrations that left many readers shaking their heads in disbelief. Regrettably, his piece also amplified some important myths about the American health care system. The confusion he sowed was abundant enough to leave some arguing that whether intended or not, his piece actually makes a case for a single-payer health system.
So let's examine the most egregious myths peddled by Mr. Brill. In today's piece I will focus on 3 important myths related to international comparisons. Later this week I'll address two other myths related to the functioning of the American health economy in general.
"U.S. Outcomes Are No Better and Often Worse Than in Other Countries"
Mr. Brill surely would have been safe in saying that some outcomes are worse in the U.S. than elsewhere, but he instead makes a much stronger, flagrantly inaccurate claim:
Yet in every measurable way, the results our health care system produces are no better and often worse than the outcomes in those countries." (emphasis added)
Because his lengthy tome is focused on the high cost of U.S. health care, he does not elaborate in detail. But the graphics accompanying his piece illustrate that the U.S. ranks lower than many other major countries in the metrics of life expectancy at birth and ranks 50th worldwide in infant mortality; the latter graphic takes further pains to point out that this ranking put us "nine spots below Cuba, 2012."
There's three major problems with this "analysis" of U.S. performance.
- First, the metrics cited have serious flaws in terms of how they are measured. When we do our best to correct for these flaws, the U.S. ranking is far higher than advertised.
- Second, the metrics in many ways are inappropriate for measuring health system performance; they do not account for social and cultural factors for which no reasonable person would hold doctors and hospitals responsible.
- Third, Mr. Brill completely ignores some far more appropriate metrics on which U.S. performance is superior.
Let's start with life expectancy. The U.S. has many more violent deaths-from homicides, suicides, automobile accidents-than the other major industrialized nations to which it is compared. Such deaths arise from social causes, lifestyle choices or imperfections in public efforts to reduce such deaths, such as highway safety. These fatality rates nothing about the quality of U.S. medical care. When life expectancy figures are appropriately adjusted to account for violence-related deaths, the U.S. ranks number one among OECD nations in life expectancy at birth. How can we draw reasonable inferences about U.S. health system performance without knowing that basic fact?
Infant mortality likewise has serious measurement flaws. Preterm birth (that is, births at less than 37 completed weeks of gestation) is a key risk factor for infant death, yet the United States is one of only eight countries that categorize extremely premature infant births as "live births," despite these babies' very low odds of survival. Specifically, "many nations do not report any live births at less than 23 weeks' gestation, or less than 500 g, despite the presence of vital signs." This may sound like a minor reporting difference, but a Philadelphia study found that when all deaths of infants delivered at 22 weeks' gestation were excluded from its birth statistics, that city's measured infant mortality rate declined by 40 percent. Imagine how the U.S. ranking were to improve were everyone to measure things the same way.
In reality, infant mortality has serious limitations as a measure of health system performance, insofar as it is driven in large part by prematurity/low-birth-weight rates. These rates in turn, are heavily influenced by poverty, maternal lifestyle (e.g., smoking, drinking and drug use) and teen pregnancy. Are these factors for which we should hold doctors and hospitals principally responsible?
Of equal importance, the aggregate statistics also mask this important reality: if we categorize births by length of gestation, the U.S. ranks second, third or fourth as compared to major European countries, in that it achieves the lowest infant mortality rates for every birth category examined prior to full-term (22-23 weeks, 24-27 weeks, 28-31 weeks and 32-36 weeks). Only Norway and Sweden (whose populations are much more homogenous and physically fit than America's) achieve consistently better results. In short, our medical system actually does a superior job at handling the poor birth distribution hand that it is dealt, but you would never know that from reading Mr. Brill's shorthand summary of health system performance.
Which gets to the final point. What about all the other outcome measures that are relevant to evaluating the performance of a medical system and on which the U.S. ranks very high?
- Cancer is the second leading cause of death (so I am not cherry-picking some obscure, irrelevant metric), yet cancer patients live longer in the U.S. than in any other country in the world.
- Despite our having many more uninsured, cancer screening rates for adults 50 and older are much higher in the U.S. than in Europe.
- And if we think that doctors are responsible for lifestyle choices such as smoking, then shouldn't they get credit for the fact that adult smoking rates are much lower here than in any of the other G7 countries?
- The U.S. population is one third smaller than that of the European Union and Switzerland combined. Yet we have produced approximately 50 percent of what experts have designated as the top medical innovations. Likewise we are responsible for more of the top pharmaceutical innovations.
As a general proposition, the U.S. has superior medical outcomes for conditions in which medicine makes a difference, but Mr. Brill mysteriously ignores this reality entirely.
The U.S. Spends 27% More per Capita Than Other Countries
Mr. Brill is not a health services researcher. So he can be forgiven for relying on a flawed study by McKinsey and Co. which purports to show that even after accounting our relatively high per capita income, the U.S. spends 27 percent more than other developed countries.
It makes a terrific story-which is why this statistic is reliably trotted out by those wishing to bash the U.S. health care system. (Even the Congressional Research Service got into the act with a 2007 report claiming that the U.S. spent 60 percent more than it "should" given its level of GDP per capita.) Unfortunately, it's not true. The "standard analysis" simply runs a linear prediction line through a "cloud" of data that arrays countries on a graph with GDP per capita on the X axis and health spending per capita on the Y axis. But as Ohsfeldt and Schneider point out in their book (published in 2006!), if the identical data are analyzed using a statistical model that actually better fits the data, U.S. spending is exactly where it should be given its higher GDP per capita.
But the U.S. is a huge country that dwarfs many of the industrialized countries used in this comparison. Indeed, if U.S. states were countries, six would rank among the top 20 countries in terms of GDP and an even larger number would be included using GDP per capita. In my book, I show that when sub-national areas are taken into account (e.g., states in the U.S. and Australia, Canadian provinces etc.), U.S. health spending is almost exactly where it is expected to be, given U.S. GDP per capita. Again, this revised model better fits the data than the standard model.
In short, using two completely different statistical methods, we arrive at a conclusion that is the polar opposite of McKinsey's: the U.S. does not necessarily spend more than it should compared to other countries. Shouldn't a fair and balanced presentation of the facts have pointed this out? I hasten to add that this conclusion does not imply that there is no waste and excess in the U.S. health system: I am merely objecting to the claim that virtually all other countries do a better job in controlling costs simply because the U.S. is an outlier in the standard model. The standard model is wrong.
"Drug Prices in the U.S. are, On Average, 50% Higher Than in Other Developed Nations
Like so many other American health care myths, this has a germ of truth. Americans generally do pay more for drugs on patent than other nations in the G7 do. But what Mr. Brill forgot to tell us is that Americans routinely pay less-often much less-than our counterparts do for generic and over-the-counter products. For example, the Japanese (whose health spending as a percent of GDP is the lowest of all the G7 nations) face prices for generic drugs that are twice those paid by Americans, while for over-the-counter medications, the prices are more than three times as high.
Fierce price competition in the U.S. and greater regulation elsewhere (among other factors) help drive the U.S. advantage in generic medication prices. Generics account for 70 percent of pharmaceuticals by volume (though only 20 percent by sales). When we take a weighted average of prices paid for prescription drugs-that is a "market basket" of medications that reflects what share is for brand name and generic pharmaceuticals-Americans pay about the same prices as the Japanese. U.S. prices relative to Germany, UK and Canada are about one-third higher, not 60 percent. More importantly, once we taken into account the higher income of Americans, U.S. drugs are more affordable than in Japan, Germany and Canada. Among the G7, only Italy and the UK have more affordable drugs than in the U.S. (by about 6-7 percent).
So why don't we let fierce competition (or government monopsony, which is what Mr. Brill recommends) drive down the prices of patented pharmaceuticals? Because it's a really bad idea! Remember, we have designed our patent system to stimulate discovery and innovation. We give innovators a time-limited monopoly on pricing to encourage, in this example, pharmaceutical companies to invest the $1 billion it takes to bring a single drug to market. To all appearances, this has worked remarkably well. As noted earlier, the U.S. accounts for more of the top pharmaceutical innovations than the EU and Switzerland combined even though the latter countries have a population that is two-thirds bigger.
The adverse effects of pharmaceutical price regulations have been studied extensively be economists. For example, John Vernon has calculated that if pharmaceutical prices in the U.S. were regulated as they are currently regulated, on average, outside the U.S., the result would be a decline in industry R&D of between 23.4 and 32.7 percent. This would be tragic for the average American. If we measure health gains and mortality reductions in monetary terms, the estimated return-on-investment in pharmaceutical research is 18 percent!
Who would want to kill or cripple this golden goose? To put this in terms of human lives, economist Benjamin Zycher has calculated that permitting Medicare to negotiate pharmaceutical prices would result in the loss of 5 million life-years annually! The well-respected consulting firm Lewin Group likewise has the drug price controls/regulatory regime in Europe produced delays in introducing statins that have led to the loss of literally tens of thousands of lives over a five year period.
And for what it's worth, the Congressional Budget Office long ago examined the impact of giving the Secretary of HHS authority to negotiate Medicare drug prices. CBO concluded that any federal budget savings would be negligible as "the Secretary would be unable to negotiate prices across the broad range of covered Part D drugs that are more favorable than those obtained by PDPs under current law." Thus, Mr. Brill's claim that Americans could save $94 billion through government-controlled drug prices appears seriously off the mark. In reality, Americans will never pay the same drug prices as other countries for the same reason that first class passengers will never pay the same prices as those flying standby. Airlines can offer deep discounts to get the last few seats filled only because others have paid enough the fixed costs to get the plane off the ground. This basic understanding of the difference between marginal and average costs seems to have eluded Mr. Brill's grasp.
As I concluded in my first post, Mr. Brill has nicely codified much of what is wrong with American health care. But he also has contributed to some of the very same misconceptions that resulted in Obamacare, a very misguided prescription for what really ails the American health care system. Until we get the diagnosis right, we have no reasonable prospect of curing the disease.
Here's a nice chart showing U.S. ranks #1 in the developed world in its rate of deaths-due-to-violence.
David Hogberg at the National Center for Public Policy Research e-mailed: "You attributed differences in infant mortality to differences in the fact that the U.S. Is "one of only eight countries that categorize extremely premature infant births as ‘live births.'" Are you aware that Dr. Aaron Carroll has tried to control for these differences and still finds that our infant mortality numbers aren't that good? I'm not saying he's necessarily right, but I would like to hear your perspective on it. (See here: http://theincidentaleconomist.com/wordpress/the-zombie-infant-mortality-explanation/)"
The CDC study alluded to by Dr. Carroll is here. The 4th slide shows the U.S. ranks 30th in overall infant mortality, while the 7th shows the U.S. ranks 18th among a smaller list of European countries when births under 22 weeks gestation are excluded. Note that when such births are excluded, it reduces the U.S. infant mortality rate by 16%, whereas for a country like Norway, this adjustment reduces the rate by only 3%. Moreover, this adjustment does jump the U.S. ahead of Hungary and Poland in the rankings. The assertion in Aaron Carroll's piece that "country rankings remained identical" after excluding preterm births is simply wrong. (Not to quibble, but the same is true of Dr. Carroll's assertion at CNN.com that "when this report recalculated the rates to exclude such births equally in all countries, we still ranked last." [emphasis added])
But the far more important point is this. The Palloni and Yonker study cited by Dr. Carroll observes: "The most important determinants of infant mortality are birth weight, prematurity, and maternal background and conditions-including maternal health status, maternal age, and behaviors." Notice that list does not include access to medical care. Admittedly, the about-to-be-released IOM study includes access to health care among a long laundry list of factors that contribute to infant mortality and low-birthweights. What one should take away from this is that access to care matters relatively little compared to all those other factors.
In contrast, medical care "matters" much more in the survival odds of an infant that is born with a low birthweight. And readers can go to slide 6 of the CDC study to see the superior outcomes achieved by American doctors. We rank #2 in survival rates for babies born at 22-23 weeks, #3 for those born at 24-27 weeks, #4 for those born at 28-31 weeks and #2 for babies born at 32-36. How did such a lousy medical care system produce such remarkable results? My only point is that critics have greatly overstated the poor performance of our medical care system by cherry-picking the data and using measures inappropriate for the task of fairly evaluating that performance.
Which gets to David's second point "You also state that cancer patients live longer in the U.S. than elsewhere. But isn't that in some ways like life expectancy in that it is impacted by many factors-social and cultural-beyond the control of the health care system? Obviously rates of tobacco use and probably factors like diet impact cancer survival. Thus, how good is cancer survival as a measure of the efficacy of a health care system?" I concur that cancer survival rates likewise are not a perfect measure of health system performance. I simply think they are a better metric than the ones too often seized by critics of American health care. To put it more succinctly, lifestyle and social factors are much more important determinants than medical care in terms of whether one gets cancer (or is a low birthweight infant). But medical care is a much more important factor in determining your odds of survival for those in the unfortunate position of getting cancer or being born prematurely.
 The figure showsn is from my book on The American Health Economy Illustrated. As you can see, without this adjustment, the U.S. ranks 15th in life expectancy among the 16 OECD countries compared. Details of this calculation are in an excellent book by Robert Ohsfeldt and John Schneider on The Business of Health.
 This is shown in Fig. 19.10 in my book, which is based on a Centers for Disease Control study.
 See Fig. 19.11a and 19.11b in my book, which elaborates on the factors such as more pervasive screening and access to the newest oncology drugs that drive the U.S. advantage.
 This is shown in Fig. 19.14a in my book, which is based on official OECD statistics. The latest data from 2010 show that Germany, UK, France and Italy all have adult smoking rates in excess of 20 percent (adults defined as age 15 and older). Japan and Canada's rate is below 20 percent but still higher than in the U.S.
 This is shown in Fig. 18.1a and 18.1b in my book, based on an excellent analysis by researchers at the Cato Institute.
 Curious readers can see the example at Fig. 1.1 in a presentation by Ohsfeldt and Schneider.
 See Fig. 1.2 in the same presentation. For statistical wonks, the reason I say this line better fits the data is because the R2 for the standard model is .846-meaning that per capita GDP "explains" about 85 percent of the differences in per capita health spending across industrialized countries. The R2 for the improved model is .926.
 See Fig. 19.4b, which are based on the same OECD dataset used by McKinsey analysts, with the additional of official statistics (in countries where these are available) on sub-national areas.
 This is shown in Fig. 19.7a in my book, which is based on a study by Wharton School researchers Patricia Danzon and Michael Furukawa.
 This $1 billion figure reflects the extraordinary riskiness of pharmaceutical R&D. That is, it includes the cost of funding R&D for the literally thousands of "dry holes" that are explored for every product that turns out to be marketable. Moreover, even getting to market is no guarantee of profits. One careful analysis of the pharmaceutical industry showed that on average "only 3 out of every 10 of their products generate after-tax returns (measured in present value terms) in excess of average, after-tax R and D costs."
 Let me be clear that this ROI does not represent the return on investment from the standpoint of pharmaceutical companies. It is a "social" rate of return based on calculations by Columbia economist Frank Lichtenberg on how many lives have been saved by the myriad drugs produced by the massive investments in R&D made by such companies over decades. Each added year of life is monetized at $150,000, representing societal willingness-to-pay.
 See Fig. 4 in Kevin A. Hassett, "Price Controls and the Evolution of Pharmaceutical Markets," American Enterprise Institute, July 22, 2004. Statin use is three times as high among eligible patients in the U.S. (56%) as in Italy (17%) and more than twice as high compared to Germany (26%) and the UK (23%). If statin use were the same as in the U.S., this would avert over 11,000 preventable deaths over five years in Germany, another 12,000 in Italy, and 8,000 in the UK.