At first glance, maternal deaths might appear to be a straightforward issue to quantify and measure. Deaths occurring at the time of birth in a hospital as a result of conditions clearly related to delivery, such as postpartum hemorrhage, are easy to classify as pregnancy-related. However, deaths on the day of birth constitute less than one in seven maternal deaths in the United States, with the majority happening in the year following delivery, and almost a third of postpartum deaths occurring outside a medical facility. Determining, for example, whether pregnancy contributed to an overdose death 8 months after a birth can be a very complicated task. This has left researchers around the world grappling with how to answer the question, “Would this person have died if they hadn’t been pregnant?”
How we document and measure maternal mortality has major implications in the United States, where maternal deaths are the highest among high-income countries and rates are two to three times higher for Black and Native American women. Efforts are being made to stem the crisis through improved protocols for dealing with medical emergencies at birth, expanded insurance coverage for doulas to advocate for pregnant people, and extended postpartum Medicaid coverage for pregnant women. However, these programs require accurate and comparable data to determine if, for example, the better performance of one state’s program over another is a sign of success or just different approaches to counting maternal deaths.
The Benefits and Pitfalls of the Pregnancy Checkbox
In the late 1990s, growing concern that countries were missing significant numbers of maternal deaths in early pregnancy or the months following birth led to an international effort to improve ascertainment. This included introducing the World Health Organization–recommended strategy of a checkbox on official death records to indicate if the deceased was pregnant or had been within 365 days of their death. U.S. states began phasing in the checkbox in the early 2000s, with all states using it by 2017. In a sense, the checkbox worked as hoped, with one Centers for Disease Control and Prevention (CDC) study finding that without the pregnancy checkbox, states would have missed approximately 50 percent of deaths that occurred during pregnancy. But improved detection came with greater potential for errors, including mistakenly checked boxes on the death certificates of women of reproductive age who weren’t pregnant or postpartum. Over 60,000 women ages 15 to 44 die annually in the U.S., with pregnancy-related causes accounting for about 1,200 of these deaths. Even a tiny “false positive” in classifying those 60,000 deaths (such as 0.5%, or 300 deaths) could increase the count of maternal deaths by 25 percent.
A national study from the National Center for Health Statistics found abundant evidence of false positives, especially when involving women over age 40. A 2024 study found that relying only on the pregnancy checkbox, without confirmation using the cause of death on the death certificate, leads to overcounts of as much as 37 percent. However, limiting maternal deaths to situations where pregnancy was clearly listed as a cause of death results in maternal mortality estimates far lower than the official rates from the National Vital Statistics System (NVSS). Studies in the U.S. and England found this approach underestimated the actual maternal mortality rate.
Are autopsies the answer? Not likely, since only 40 percent of maternal deaths in the U.S. involve an autopsy to accurately determine the cause of death. In the other 60 percent of cases, those reporting the cause of death, such as medical examiners and funeral directors, sometimes fail to note the decedent had been pregnant. Such underreporting is why the World Health Organization (WHO) and CDC recommended adding a pregnancy checkbox in the first place.
How to Improve Maternal Mortality Measurement in the U.S.
There’s no perfect tool to measure maternal mortality, so the best way forward may be to draw from the strengths of the three separate systems in use in the U.S. NVSS relies most heavily on the pregnancy checkbox, which is supplemented with cause-of-death information examined by trained medical classification experts (nosologists) who convert the codes on death certificates to causes consistent with WHO reporting guidelines. NVSS publishes the official rate for the U.S. in WHO cross-national comparisons and it is by far the timeliest of the three data systems, having already released 2023 results.
The CDC’s Pregnancy Mortality Surveillance System (PMSS) dates back to the late 1980s and solicits birth, NVSS death data, and other records such as obituaries from states. Medical epidemiologists review the greater information available to them and eliminate cases that could not be confirmed, resulting in significantly lower rates of maternal deaths than the official system — 24.3 per 100,000 births in 2021 compared to the NVSS rate of 32.9. Unlike the NVSS, the PMSS also reported no notable increases in the decade prior to the pandemic, but their reports have never garnered much media attention.
The third system, a network of state maternal mortality review committees (MMRCs), holds the most promise for accurately documenting maternal deaths and developing solutions, but they require sufficient time and resources to conduct. Almost every state now has a committee that typically includes physicians, midwives, public health researchers, and medical specialists in areas such as substance use disorder and cardiology. These groups review potential pregnancy-related deaths to determine their cause and relationship (if any) to pregnancy, and whether they were preventable. A 38-state MMRC report found that 84 percent were preventable. They also report much lower state rates than the NVSS (24.5 compared to 39.6 in Louisiana and 30.1 compared to 45.9 in Georgia). MMRCs also recommend statewide clinical and community-level changes to prevent future deaths.
It is important to note, however, that measurement problems don’t account for the U.S.’s poor performance internationally — even the more conservative PMSS and MMRC estimates show the U.S. significantly lagging other high-income countries on maternal deaths, and all U.S. measurement systems report wide racial and ethnic disparities.
Continued investment is critical to understand where we are on maternal health. The timeliness of the NVSS system gives us a valuable first look at maternal mortality in the U.S., while the PMSS, reporting one to two years later, provides a richer and more nuanced review of deaths out to a year postpartum. In creating solutions, states should look at the specific, localized recommendations made by their MMRCs. The goal is community-led initiatives to improve maternal health, and we can draw on all three systems to develop solutions that allow us to act quickly and effectively to prevent maternal deaths.
