If you’re an American woman, your chances of dying from pregnancy-related causes are higher than someone who lives in Sweden, Macedonia, Korea, Poland, or Libya. And a new study reveals how frustratingly little we know about why that’s the case.
Over the past few years, Texas has become the emblem of America’s maternal health crisis, after a 2016 study in Obstetrics & Gynecology reported that an astounding 148 women died from pregnancy-related causes in Texas, representing a 40% increase in maternal deaths in Texas between 2010 and 2012. Now, a new study reveals that the increase wasn’t real; it was the result of errors in the way the data was collected.
On its face, this should be excellent news, but it actually reveals something alarming about how little we know about maternal mortality in the United States in general.
Back in 2016, when the original Texas statistics were published, these numbers made global news. How could a jump that large happen in such a short timespan? The study authors urged caution, but nonetheless, the data seemed to fit within a larger trend: By all estimates, the United States is home to one of the highest maternal mortality rates — that is, the number of women who die in pregnancy or shortly after childbirth for every 100,000 births — in the developed world. Many connected the terrifying rise in Texas deaths to changes in the state’s funding: In 2011, in an effort to defund Planned Parenthood, legislators voted to cut the family planning budget by two-thirds, and followed that up in 2013 with an omnibus anti-abortion bill that was so onerous it closed half the state’s abortion clinics. (The Supreme Court stepped in to strike it down in 2016, but by then the damage was done.)
But now, the new study, also published in Obstetrics & Gynecology, shows that the actual number of women who died of pregnancy-related causes in Texas in 2012 was 56 — not 148. The cause for the discrepancy? Human error by doctors and medical examiners whose job it is to input data into the state’s electronic death registration system, researchers from the Texas Department of Health Services Centre found. Translation: We were completely wrong about the scope of a systemic problem in women’s healthcare because of accidental box-checking.
To be fair, it’s not hard for this kind of error to happen: The use of electronic death certificates in Texas increased from from 63% in 2010 to 91% in 2012, which suggests that people might have made more errors because they were getting used to a new way of doing things.
For about half of the deaths originally coded as pregnancy-related, the researchers could find no record that the women who died had been pregnant at all.
In an effort to determine the true cause of the rise in Texas, the researchers for the most recent study took the standard method a step further: They went back to the deaths counted in 2012 and tried to match them with other health records, as a way to verify that pregnancy or delivery was related.
In the end, the researchers could only confirm that 56 women had died during pregnancy or within 42 days of giving birth (the official definition of maternal death). For about half of the deaths originally coded as pregnancy-related, the researchers could find no record that the women who died had been pregnant at all. In the process of matching death records with birth and foetal death data, the researchers also found an additional nine deaths that had never been counted.
This makes the revised maternal mortality rate in Texas 14.6 maternal deaths per 100,000 live births, down from 38.4 deaths per 100,000 births. The good news, of course, is that the scope of the problem in Texas is probably much smaller than we thought. The bad news, the researchers point out, is that the data errors found in Texas point to a bigger data collection problem that’s likely nationwide. The original Texas numbers were calculated with the standard method (a.k.a. just looking at death certificates and counting how many deaths were linked to pregnancy), and this is the method used across the United States.
“Most times, when a maternal mortality rate is calculated, it relies on the codes on death certificates,” says Sonia Baeva, an epidemiologist for the state and the study’s lead author. “We know there are issues with maternal mortality data, nationally, and we encourage other states to take a closer look and use this enhanced method to get a clearer picture of maternal mortality in their area.”
The root of this data problem can actually be traced to 2003, when the federal government added a pregnancy or postpartum checkbox to the U.S. standard death certificate, reports The Washington Post. The idea was that this would improve maternal mortality data collection, but as Texas shows, the opposite can happen.
Like driver’s licenses for example, every state maintains control over their own death certificates, and so over the next 10 years, each state added their checkbox for maternal deaths in varying ways. “What we’re finding now is that it is often checked in error, just like any other checkbox on a big form,” Elliot Main, the medical director of the California Maternal Quality Care Collaborative, told The Washington Post. “Because pregnancy-related deaths are so uncommon, the frequency of the box being checked in error can significantly impact the maternal mortality rate reported.”
Meaning: A handful of accidentally checked boxes can completely throw off the scope of the problem on a state level. This creates confusion that affects women nation-wide: As extensive reporting from ProPublica and NPR revealed last fall, the federal government hasn’t even been able to release a national maternal mortality rate estimate since 2007, due to the differing reporting methods in each state.
What’s more, considering that the researchers in Texas also found deaths that had originally not been counted at all, it could be that in some places, we’re overestimating deaths while, in others, many are going uncounted. “The rates for other states are calculated using the standard method, so their rates could be inaccurate, too, and we don’t know by how much,” Baeva says. (On top of this, Baeva adds, it’s impossible to use the revised Texas rate to compare it to other states’ rates, since they were calculated using the standard method.)
The fact that we still figuring out the best way to measure the scope of the problem, and that the federal government has known about this confusion for a decade and has done nothing about it, suggests a galling cluelessness and lack of care for maternal health.
It remains true, even when you account for the data collection problems, that the overall trend in maternal mortality in the United States is an upward one, making the U.S. the lone outlier in comparison to developed countries like the United Kingdom, Canada, and France.
We also know that maternal mortality disproportionately affects Black mothers, who are between three and four times more likely to die from pregnancy-related causes than white women, even when you control for socio-economic factors like income and education levels, according to a report from the Center for Reproductive Rights and the Black Mamas Matter Alliance. Even in the revised Texas numbers this holds true: The maternal mortality rate for Black women was still double the overall rate, at 27.8 per 100,000 live births.
Despite what we know about these general trends, having a detailed and accurate accounting of the problem is crucial for the health of all people who plan to have kids. Maternal mortality is complex, and caused by many different factors, from access to care, racial discrimination, or underlying health issues (like high blood pressure) that can be exacerbated by pregnancy, just to name a few. In some places, the largest driver may be a lack of access to prenatal care, or it could be holes in doctors’ training that affect their abilities to identify women who are at risk. In others, it could be both these things, as well as a number of other factors.
The point is, until every state takes the steps to study what’s happening in their hospitals, and in their communities, we won’t be able to say for sure how to solve this.