Among the 55 World Bank-identified high-income countries for which the World Health Organization (WHO) also provides maternal mortality estimates, the WHO reported a median maternal mortality rate of 7 maternal deaths per 100,000 live births in 2019. Some countries, such as Poland, Norway, and Israel, report figures at or below 3 deaths per 100,000. Within those rankings, the WHO put the United States 45th, at almost 20 maternal deaths per 100,000 births. That’s a dismal showing for so wealthy a country—and one that’s produced an enormous amount of commentary blaming the high rate on everything from cesarean sections to aging mothers to racism.
Among the 55 World Bank-identified high-income countries for which the World Health Organization (WHO) also provides maternal mortality estimates, the WHO reported a median maternal mortality rate of 7 maternal deaths per 100,000 live births in 2019. Some countries, such as Poland, Norway, and Israel, report figures at or below 3 deaths per 100,000. Within those rankings, the WHO put the United States 45th, at almost 20 maternal deaths per 100,000 births. That’s a dismal showing for so wealthy a country—and one that’s produced an enormous amount of commentary blaming the high rate on everything from cesarean sections to aging mothers to racism.
However, these figures are completely wrong, and they have been known to be wrong for many years now. The U.S. National Center for Health Statistics, the branch of the Centers for Disease Control (CDC) charged with collating health and vital statistics, has published three separate reports elaborating in excruciating detail on one crucial fact about U.S. maternal mortality: It is measured in a vastly more expansive way than anywhere else in the world.
As a result, U.S. maternal mortality is overestimated by two to three times. Properly measured, the real U.S. maternal mortality rate in 2019 was 9.9 maternal deaths per 100,000 births, which would put it at 36th place—still not impressive by comparison, but somewhat better than Canada and a bit worse than Finland or the United Kingdom. And as we’ll see, even that rate may have more to do with U.S. mortality as a whole than with specific pregnancy-related issues.
Historically, the United States and most countries have tracked maternal mortality using data based on the cause of death listed on death certificates. When a person dies and the cause is assessed by an examiner of some kind, certain “maternal” causes, such as “eclampsia” or “obstetric trauma,” are commonly tracked. If a woman has died due to one of these pregnancy-related causes, she is listed as a maternal death. This process is fairly straightforward and has been widely adopted across many countries.
But in 2003, the U.S. CDC decided to launch an improved death certificate form. Among the various changes proposed was the addition of a checkbox, wherein whoever filled out the paperwork would identify if the deceased had been pregnant in the last 42 days or the last year. The reason for this checkbox was that the CDC believed (correctly, as it turns out) that in only measuring “maternal causes of death,” it might be underestimating the true health hazards of pregnancy. Pregnancy might alter the course of other diseases and conditions or interact with them in important ways.
The CDC anticipated that the checkbox would increase measured maternal deaths; it did not anticipate just how much it would increase them. As it happens, the CDC’s own reporting, which I have confirmed elsewhere, shows that the addition of the checkbox approximately doubled maternal mortality rates.
You might think a sudden doubling in maternal death rates would be obviously flagged as a data issue to correct, but this turns out not to be so. Because the United States has a federal system, individual states added the checkbox in different years. While individual state maternal deaths showed sharp level shifts, the national maternal death count drifted upward gradually as states added checkboxes to their death certificates: California in 2003, Florida in 2005, Texas in 2006, Ohio in 2007, Tennessee in 2012, etc.
In 2018, further modifications were made to the data-processing protocols used by the National Center for Health Statistics for pregnancy-related checkbox deaths, leading to more thorough inclusion of them. The result was a massive but gradual artificial inflation of maternal mortality.
This doesn’t mean that the American way of measuring death is wrong. It’s just quite different from the countries that it’s being compared to.
There are, of course, many ways to measure maternal mortality. Most death certificates list multiple contributing causes of death. Conventional maternal mortality data only includes deaths whose primary or underlying cause was some maternal cause of death; the CDC’s multiple-cause mortality files show that if every pregnancy-related death in the United States were included in the maternal mortality rate, 2019 would have had a rate of 48 maternal deaths per 100,000 births—a shockingly high number, higher than almost any high-income country. A country’s maternal mortality rate thus depends in large part on measurement choices.
But the U.S. case is particularly beguiling, since the United States now tracks all deaths of women who were pregnant, not only women who gave birth. Women who miscarried early or had abortions—whether officially reported or not—are also counted in the checkbox method. As a result, the United States may be the only country in the world where central vital records systems track all pregnancy-related mortality, not just maternal mortality.
This is an important point, because pregnancy-related mortality is a far more important concept for the public than maternal mortality. If miscarriage or abortion has health consequences, most people would consider that a part of the maternal experience, since many women have miscarriages or abortions on the way to having eventual live births.
Counting only those deaths subsequent to live births (or, in some countries, also officially reported abortions) as “maternal” mortality is a far cry from the public’s understanding of maternal mortality as a measure of the risk that women shoulder when they become pregnant. Because most high-income countries do not measure every deceased person’s pregnancy history, they implicitly exclude all of these potentially pregnancy-related deaths from their maternal mortality data.
Far from being “wrong,” then, the United States’ elevated maternal mortality rates are arguably more correct than putatively comparable estimates in Europe.
Most European maternal mortality estimates essentially capture just the first bar shown in the figure above. Official U.S. maternal mortality data since 2018 reflects the second column. But the checkbox system and one-year look-back make it possible to identify a vastly greater sweep of pregnancy-related deaths: The U.S. approach ultimately identifies nearly five times as many possibly pregnancy-related deaths as the usual benchmark approach.
Even the corrected U.S. maternal mortality rate, at 9.9 deaths per 100,000 births in 2019, can seem a bit on the high side compared to the United States’ relative income levels (Sweden, for example, has a rate under 5 deaths).
This is a bit misleading, however. The graph below shows WHO-estimated maternal mortality vs. U.N.-estimated mortality rates (from the latest U.N. World Population Prospects report) of all causes for women ages 15-50; the U.S. estimate from the WHO is highlighted in orange. Highlighted in red is my estimate for the corrected U.S. rate, as discussed above; all figures are from 2019.
The trend line shows that maternal death rates are extremely tightly correlated with overall female mortality. In countries where death rates are higher for reproductive-age women from any cause, maternal death is also higher.
American women die at high rates (37 deaths per in 2019) from many causes: suicide, homicide, drug abuse, and many natural causes as well; it is no surprise that in a generally more mortality-prone population than, say, Sweden (16 deaths per 100,000 women for general mortality, 5 deaths per 100,000 births for maternal mortality), maternal mortality is also higher.
What is striking, however, is the deviation from the trend. Both the orange (unadjusted maternal mortality) and red (adjusted to fix the checkbox issue) markers for the United States are below the trend line. U.S. maternal mortality is actually rather low compared to what we would expect based on the general prevailing mortality conditions for American women. It’s true that the country’s maternal mortality is on the high side for rich countries, but this is because its mortality in general is on the high side for rich countries. When it comes to taking care of pregnant women relative to the general health environment, the United States is actually doing very well. It’s farther below the trend line than most countries are.
Far from being a serious laggard in maternal health, U.S. maternal mortality rates appear pretty normal when measured in a truly comparable fashion, and the United States actually has much lower maternal mortality than might be expected from its overall mortality rates for women.
This means that international organizations such as the WHO are guilty of “garbage in, garbage out”—they have not adequately wrestled with the unique data-generating process at work due to the U.S. pregnancy checkbox, and as a result, they systematically overstate U.S. maternal mortality in comparison to other countries.
However, while U.S. maternal mortality is overstated in comparison to other countries, this is really a reflection of a serious deficiency in the vital records infrastructure in the rest of the world. Pregnancy-related mortality—which to any reasonable person should be classified as maternal mortality even though there may not be a live birth, since the experience of motherhood begins long before delivery—is by and large ignored in most countries.
Likewise, pregnancy-related mortality should not be limited to explicitly pregnancy-related causes of death: If pregnancy alters the course of some other disease or condition and a mother dies not “of” pregnancy but “with” pregnancy, then that is validly considered a form of maternal mortality!
Rather than panning the United States for its high measured rates, other countries should do the work of improving their statistical systems to cast as wide a net as the U.S. one, and as expansively measure the health and mortality risks that pregnant women bear.
login
who are we?
contact us
qosheapp
Toi Staff
Gideon Levy
Belen Fernandez
Mort Laitner
Rami G Khouri
Ali Fathollah-Nejad
Nikkei Editorial
visit website
