{A} t a recent campaign stop, Vice President Kamala Harris, speaking of pro-life advocates, said that “while these extremists say they are motivated by the health and the well-being of the women and children of America, in reality, they ignore the crisis of maternal mortality. The top ten states in our country with the highest rates of maternal mortality all have abortion bans. The hypocrisy abounds” (emphasis added).
It’s an interesting claim for two reasons. First, it’s testable, and second, it appears to imply two different conclusions. One is that abortion bans perhaps cause maternal mortality, and the other that abortion bans combined with maternal mortality point to gross disregard for women’s health.
How do we count maternal mortality?To test the claim, we first have to understand maternal mortality. I have outlined elsewhere a very basic fact: The U.S. maternal-mortality measurement is unusual in global comparison and has changed considerably over the past few decades. Because the U.S. uses a “pregnancy checkbox” system, we capture an expansive range of pregnancy-related deaths, including those uncaused by pregnancy and those occurring subsequent to births but also to abortions and miscarriages.
Second, it’s important to grasp just how rare maternal mortality is. Even at the inflated rates the CDC reports, 32 maternal deaths per 100,000 births implies that many states would have single-digit numbers of maternal deaths annually, since the U.S. has only a few million births in total each year. This is important, because it means that, for many states, maternal mortality cannot be directly estimated, because for privacy reasons the CDC censors low-frequency deaths. Indeed, because of privacy rules, the CDC in its latest official estimate of state maternal mortality declines to provide an estimated rate for 14 states.
Right off the bat, then, testing Harris’s claim turns out to be a bit tricky. Fourteen states have no “official” maternal-mortality rate at all because their populations are too low to yield enough maternal deaths to overcome the CDC’s privacy thresholds, and so we don’t know what the ten states with the highest ranking are (since some of those small states might have quite high maternal-mortality rates even with low absolute numbers). And of course, your ranking of states will depend in the first place on how you want to measure maternal deaths.
To get at this question more precisely, I use two measures. First, I use the maximally expansive definition of pregnancy-related mortality: any death, of any underlying cause, which was reported to have any pregnancy-related factors associated with it, could be counted as maternal mortality. These can be thought of as “every death that happened within a year of a woman’s being pregnant.” We’ll call this “high count” maternal mortality. Second, I use the maximally restrictive definition of pregnancy mortality: only deaths that occurred within 42 days of pregnancy, had a pregnancy-related primary cause, and had no other reported causes of death. This is close to the approach that the CDC used before 2003. I call this “low count” maternal mortality.
Finally, to assess abortion bans, I use data from the pro-abortion-rights Guttmacher Institute. It isn’t exactly clear what Harris meant by an “abortion ban,” but I count together any of the states that Guttmacher identifies as “very” or “most” restrictive, and then any of the states Guttmacher identifies as “protective” or higher.
Where is maternal mortality highest?
To begin with, high-count maternal mortality is highest in the south as well as in a few western states. It peaks at 72 deaths per 100,000 live births in Alabama and Mississippi.
Of the ten states (or, more precisely, nine states and the District of Columbia: Mississippi, Alabama, the District of Columbia, Louisiana, Georgia, Arkansas, Montana, Tennessee, Wyoming, and North Carolina) with the highest “high count” maternal-mortality rates, only five are rated by Guttmacher as “most restrictive” (i.e. a ban, in Mississippi, Alabama, Louisiana, Arkansas, and Tennessee). Two more are “very restrictive” (Georgia and North Carolina), one is just “restrictive” (Wyoming), and two are “Protected” (D.C. and Montana).
So using the most expansive definition of pregnancy-related mortality, Harris’s claim is wrong.
But what if we use the more restrictive, low-count definition?
Low-count maternal mortality is, unsurprisingly, much lower. It also is less concentrated in the south: it is highest in D.C. and Alaska. The top ten states (D.C., Alaska, Tennessee, Mississippi, Oklahoma, Wyoming, South Carolina, Louisiana, Kentucky, Arkansas) are, again, not all abortion-ban states. Using the low-count approach, we find that seven states have the “most restrictive” rating, one (Wyoming) has the “restrictive” rating, and two (D.C. and Alaska) have the “protective” rating. Again, Harris’s claim fails.
What if we instead turn to the CDC’s official published data? Although 14 states don’t have reliable figures, we can use the CDC-flagged unreliable data (due to privacy constraints) to guess what their MMR rates might have been.
This measure again shows high concentration in the south, peaking at 44 maternal deaths per 100,000 live births in Arkansas. Of the ten states with the highest mortality, eight are “most restrictive,” according to Guttmacher, but two (Georgia and Arizona) are “very restrictive” — i.e., they do not have blanket bans on abortion. With eight out of ten states having abortion bans when using the CDC’s official data from 2018 through 2021, this approach comes closest to Harris’s meaning. But even here, it doesn’t seem to be true that the ten states with the worst maternal-mortality rates all have abortion bans: according to Guttmacher, two of them don’t.
Harris’s claim is, therefore, wrong, regardless of what measure is used.
Counting maternal deaths rightly
But it’s worth investigating a couple other related issues. Some states did much better on the “high count” or “low count” method. This is because states differ greatly in how expansively they code pregnancy-related death causes. The figure below shows what share of “high count” deaths are made up of “low count” deaths — i.e. what share of deaths “with pregnancy” are conclusively and decisively deaths “of pregnancy.”
Some states, including Alaska, Hawaii, Nevada, California, and Oklahoma, capture more than a third of their “with pregnancy” deaths as “of pregnancy” deaths. Other states, including South Dakota, Maine, and New Hampshire, capture single-digit percentage points. In general, states with abortion restrictions count about 20 percent of “with pregnancy” deaths as “of pregnancy,” while states with abortion protections count about 25 percent.
I note this because it tells us something important. Combined with the high variance in “high count” maternal mortality rates, the inescapable conclusion is that states including Alaska, Hawaii, Nevada, Oklahoma, and Harris’s home state of California are probably misclassifying a lot of “with pregnancy” deaths. That is, they fail to inquire enough about the pregnancy histories of recent deaths, thereby missing a large number of pregnancy-related deaths. This probably artificially suppresses their overall maternal-mortality statistics. Notably, California’s pregnancy checkbox remains to this day slightly out of sync with CDC guidelines for death certificates.
One way to determine which states might be misreporting maternal mortality rates is to start with one simple fact: Around the world, maternal mortality is closely correlated with mortality rates for reproductive-age women generally. If a state reports very low pregnancy-related mortality but high mortality for reproductive-age women generally, we may be able to assume that pregnancy-related mortality is underreported. On the other hand, if pregnancy-related mortality is very high and yet women in general have low mortality, a state may be overreporting pregnancy. (CDC follow-up studies have found that a nontrivial number of death certificates with pregnancies reported were for women who did not in fact have any pregnancy.)
A few states have very stark outliers. In particular, Maine, Nevada, Rhode Island, and California report high-count maternal-mortality rates 30 percent lower or more than what would be expected from their mortality rates for reproductive-age women. California, Rhode Island, and Nevada are also states with unusually low high-count rates compared with low-count rates, strongly suggesting that these three states may be underreporting pregnancy-associated mortality in comparison to the reporting norms of other states.
On the other hand, D.C. has a fairly massive excess of high-count maternal mortality, fully 95 percent higher than would be expected from its baseline mortality rates. Virginia, Georgia, Wyoming, and Alabama are also all 30 percent or more above expectations.
If instead we compare low-count mortality to general mortality rates, D.C. still shows up as having unusually high maternal mortality (159 percent above expectations), while Wyoming, Oklahoma, Tennessee, Hawaii, and Alaska are all 30 percent or more above expectations. That D.C. and Wyoming are so high for both high-count and low-count methods could suggest that they are just genuinely high maternal-mortality states, states where expectant mothers are less healthy, exposed to higher risks, or receive a worse quality of care.
On the other hand, the list of states with unusually low low-count mortality consists of Maine, South Dakota, New Hampshire, Vermont, and Delaware. In Maine, which showed up here and on the high-count list, expectant mothers may have unusually good health outcomes. But that California and Nevada did not show up on the list of low low-count states reinforces the notion that California and Nevada are probably undercounting pregnancy-related deaths.
This whole exercise may seem esoteric, but it has a point: Vice President Harris should perhaps encourage her home state of California to do pregnant women the basic dignity of correctly counting their tombstones before trying to (incorrectly!) dunk on states with higher maternal mortality. One neat way that California and some other big liberal states have managed to achieve low maternal-mortality rates is by simply not counting all pregnancy-related deaths.
What explains maternal-mortality rates?
Finally, it’s important to take a moment to assess why maternal-mortality rates vary. The implication carried in Harris’s words is that abortion restrictions may be harming women. The evidence for this, however, is pretty thin. We have maternal mortality data for 2023 and can compare maternal-mortality rates in states with abortion protections with the rates in states that imposed restrictions.
States that imposed abortion restrictions after Dobbs had a low-count rate of 9.6 deaths per 100,000 live births in 2021, compared with 7.7 in states that adopted abortion protections. In 2023, those numbers were 7.7 and 6.1. In other words, after the Dobbs decision, the maternal-mortality gap converged slightly, from 1.9 to 1.6. Far from killing moms, states that adopted abortion restrictions saw a bigger decline in maternal mortality compared with their maternal mortality in 2021 and especially in 2022.
And so abortion restrictions probably are not to blame for maternal mortality, and states that imposed restrictions may have seen a modest improvement in their mortality rates.
So what does cause maternal mortality? As already noted above, maternal mortality is closely correlated with general female mortality. The leading causes of death for reproductive-age women are overdoses, cancers, suicide, murders, diabetes, liver diseases, heart attacks, heart disease, mood and behavioral disorders, and accidents, including car accidents. The two greatest causes of death for women with any pregnancy-related contributing cause are pregnancy conditions, but the third-ranking cause is overdoses; the sixth, car accidents; the seventh, murder; and tenth, suicide. The general societal risk factors that threaten all women also threaten pregnant women. In states with more crime, obesity, and drug use, more people die. Among those who die, some are women; and among those women who die, some, in a particular tragedy, are expectant or recent mothers.
This principle, that maternal mortality primarily reflects background mortality risks for groups of women, not state-specific maternity-related factors, extends to the fraught issue of racial health disparities. Non-Hispanic white women average a low-count maternal mortality rate of about 6 or 7 deaths per 100,000 live births — non-Hispanic black women average about 20. The maps of race-specific maternal mortality rates make clear how universal this difference is:
While the gap is largest in two southern states (Arkansas and Mississippi), the racial gap in maternal-mortality rates is the same in New York and South Carolina, Louisiana and California, Massachusetts and Florida, Illinois and Georgia, Kentucky and Colorado. The scourge of excess prime-age mortality among black women has many causes, and it spills over into maternal mortality as well, but what can be said with some confidence is that state-specific social and cultural factors don’t seem to be the main cause.
Indeed, most state-level differences in maternal mortality rates turn out to be biases of composition. Some states have more women who, for various reasons, are in higher-mortality-risk populations. There is, tragically, not much evidence that any state has found a way to reliably, at the population level, push maternal mortality rates for higher-risk populations downward.
When Vice President Harris drew a facile linkage between state-specific policies and headline maternal-mortality rates, she not only did so incorrectly, but she obscured the far harder truth: Black women in the most liberal, progressive states die in pregnancy at the same excessive rate as in other states, and none of the policy measures yet adopted by states have managed to address this egregious inequity in health outcomes. High maternal mortality is not a product of one or two villainous policies that can be changed by a political savior but is rather the product of differences — far harder to solve and more deeply embedded — in health, economic opportunities, political structures, cultural norms, environments, and behaviors.
Politicians will use the lines they want to use. But for those who, unlike Harris, care to understand the tragic case of high maternal mortality in the United States, we would do well to focus on the massive health inequalities driving major differences in outcomes rather than render maternal mortality a second front in the conflict over abortion.