Is Overreporting of Maternal Mortality Key to High US Rate?  

Tara Haelle

June 02, 2017

As maternal mortality rates have dropped 44% globally over the past quarter century, researchers and public health officials have been struggling to understand why rates in the United States have actually worsened over that time.

In addition to speculation about the contribution of increasing obesity rates, chronic disease, and cesarean deliveries, a study published late last year in Obstetrics & Gynecology suggested that coding changes and improved surveillance accounted for a substantial part of the increase. Now, another study, published in the May issue of the same journal, suggests that both a change in coding and overreporting may be contributing to the apparent rise.

Whereas the new data partly concur with the previous study's findings regarding the effect of coding changes on reported rates, the new study goes a bit further and suggests some of those surveillance changes may have led to some inflation in maternal mortality statistics.

"Large increases in maternal mortality rates for older women and among nonspecific causes suggest possible data quality problems that may be worsening over time," concluded Marian MacDorman, PhD, a research professor at the University of Maryland in College Park, and colleagues.

Much of the discussion in both studies focuses on the introduction, starting in 2003, of five checkboxes on the US standard death certificate: "not pregnant within past year," "pregnant at time of death," "not pregnant but pregnant within 42 days of death," "not pregnant but pregnant 43 days to 1 year before death," and "unknown if pregnant within the past year." Only the boxes indicating pregnancy at the time of death or within the previous 42 days are counted in maternal mortality statistics, and a lot is riding on those boxes — even though individual states only adopted them slowly, if at all.

"Reporting has improved over time, primarily with adding this checkbox question," Dr MacDorman told Medscape Medical News. "The problem is that there's now substantial evidence of overreporting of maternal death" because the wrong boxes can be inadvertently checked, she says. If one of those two boxes is checked, the death will definitely be coded as a maternal death unless it was attributable to an accident, suicide, or homicide, she explains.

The World Health Organization defines maternal deaths as those resulting from pregnancy but not from "accidental or incidental causes." US coding procedures, however, generally neglect to exclude "incidental" causes. "If it's written as sunburn or brain tumor, it's still going to be coded as maternal death," if one of those two boxes is checked, she says. 

Like the authors of the previous paper, Dr MacDorman's team found that switching from the 9th to the 10th edition of the International Classification of Diseases (ICD) (the 10th added more maternal categories) also contributed to the increase in maternal mortality rates. However, unlike the previous authors, who concluded that ICD-10 coding more accurately captures actual maternal mortality rates, Dr MacDorman believes the evidence points instead to a misclassification of incidental causes of deaths as maternal deaths.

"We think a lot of the increase is reporting [health] problems that are getting worse over time," she told Medscape Medical News.

Slight Overreporting Could Tip the Scale

To better understand how the checkbox question and ICD-10 coding has influenced reporting, Dr MacDorman's team retrospectively analyzed the change in maternal mortality rates in 27 states and the District of Columbia during the 5-year period from 2008–2009 to 2013–2014. They used data from the National Center for Health Statistics and the Centers for Disease Control and Prevention WONDER database, which they stratified according to maternal age (5-year increments and before/after age 40 years), race and ethnicity, and detailed causes of death. Unlike the previous study, which covered a much longer period but had to account for incremental state adoption of the checkbox, Dr MacDorman's study included only states that had adopted the checkbox by 2008 and used consistent, comparable reporting procedures during that full 5-year timespan.

The 7,369,966 live births the researchers investigated made up 45% of all US births in 2008–2009 and 2013–2014 and included 1687 maternal deaths. Between the two time periods, the maternal death rate increased 23%, from 20.6 to 25.4 maternal deaths per 100,000 live births.

However, a careful parsing of the data revealed eyebrow-raising trends. In 2008–2009, the maternal mortality rate among women age 40 years and older — at 141.9 per 100,000 live births — was 10 times greater than the 14.1 rate seen among women 25 to 29 years old (the lowest-risk group). Just 5 years later, the maternal mortality rate for older women had nearly doubled, increasing 90% to 269.9 deaths per 100,000 live births. Yet, the rate for women aged 25 to 29 barely budged at 14.7 in 2013–2014, making the older women's rate 18 times that of the younger women. That difference alone appears entirely responsible for the overall increase in the maternal mortality rate over the 5-year period.

"Clearly the rates for women 40 and older are implausible," Dr MacDorman said. "We also found large increases in maternal mortality for nonspecific causes of death."

These are deaths lacking a particular maternal code in the ICD-10. They include "other specified pregnancy-related conditions" (O26.8), "other specified diseases and conditions complicating pregnancy, childbirth and the puerperium" (O99.8), and "obstetric death of unspecified cause" (O95). "It is more difficult from a clinical perspective to determine the specific cause of death for these cases," the authors write.

Maternal mortality rates for these combined causes of death climbed 47.9% from 2008–2009 to 2013–2014 and accounted for 83% of the overall increase in maternal deaths during that time. They made up half of all maternal deaths among women older than age 40 years in 2013–2014, which was 20 times greater than the deaths coded accordingly among women under 40. By contrast, specific causes of death among women over 40 were only 12 times greater than specific causes for women under 40.

Although the overall increase in maternal deaths among women under 40 during the 5-year span was not statistically significant, deaths coded in these three categories increased 26% during that time. Further, excluding these nonspecific deaths from the statistics resulted in no statistically significant increase in the maternal mortality rate at all during those 5 years. Moreover, almost two thirds of the overall increase in maternal deaths over those 5 years came from deaths coded as O26.8, while just over a third came from deaths coded as O99.8, the only specific subcategory that significantly increased.

To explore the extent to which overreporting might influence mortality rates in different age groups, Dr MacDorman's team conducted a sensitivity analysis to see the effect of 0.5%, 1%, and 1.5% false-positives (deaths coded as maternal when the woman was neither pregnant nor postpartum). Just 1% of overreporting maternal deaths among women older than age 40 more than tripled their mortality rate (232%), compared with just a 14% to 23% increase for women in their 20s and early 30s, who have fewer deaths but many more births. A 0.5% false-positive rate doubled maternal mortality rates for women over 40.

Dr MacDorman pointed out that a recent CDC report, which investigated maternal deaths in Colorado, Delaware, Georgia, and Ohio for 2- to 5-year periods, found that 14.9% of the 650 maternal deaths in their data set were for women confirmed not to be pregnant or postpartum at time of death.

"Given concerns about overreporting with the pregnancy checkbox, it is illogical to continue to use it as the sole means of identifying maternal deaths," the authors conclude. "Efforts to improve reporting for the pregnancy checkbox and to modify coding procedures to place less reliance on the checkbox are essential to improving vital statistics maternal mortality data."

They add that not all countries use the pregnancy checkbox on death certificates, which may mean significant underreporting in nonindustrialized countries with allegedly better maternal mortality rates than the United States.

"Other places don't have the tools we have," Nancy Chescheir, MD, editor in chief of Obstetrics & Gynecology, told Medscape Medical News, agreeing that deaths in a number of countries "are probably actually a lot worse than ours" regardless of what's reported.

Statistics Do Not Erase Concern

Regardless of the probable overreporting, there is still reason for concern, Dr MacDorman says.

"Our country is really this crazy outlier in that we haven't improved maternal mortality," she told Medscape Medical News, adding that the United States also has large race and ethnic disparities. In both 2008–2009 and 2013–2014, non-Hispanic black women had a maternal mortality rate almost triple that of non-Hispanic white women.

Compared with the United Kingdom, with its maternal mortality rate of 3.9, and France, with its rate of 3.5, the United States still has a rate of 17.8 when only women under 40 are included, or a rate of 15 for all ages when indirect causes are excluded. "Even if we say a lot of the increase is due to the reporting, it's certainly not getting better, and that's a really big problem," Dr MacDorman said.

Dr Chescheir questions the assertion that overreporting is skewing the data because there could also be underreporting. However, it is not possible to double-check all the women of reproductive age who died pregnant or postpartum but did not have the appropriate box checked. Yet she sees the results of Dr McDorman's paper as another important piece of the puzzle in deciphering what US maternal mortality rates really are and how to address them.

"I don't think either of these papers [the previous study and this one] tells the whole story," Dr Chescheir told Medscape Medical News. "The real crux of this to me is definitions and data sources. These two groups of researchers are trying to parse that out in different ways."

The issue as a whole, Dr Chescheir says, is among the most important facing obstetrician/gynecologists in the United States today, with the maternal mortality rate reminding physicians that women do die of pregnancy and those at elevated risk for death include women over 40, those with preexisting comorbidities, and non-Hispanic black women.

"When they're caring for those women, they need to be attuned to perhaps more nuanced presentations of illness and to get consultation as appropriate to take care of those patients when there are additional morbidities," Dr Chescheir said..

The enormous amount of scrutiny on maternal mortality is justified, she added, but we currently lack the tools necessary to fully assess the problem and understand how to prevent these deaths.

"It seems like a really easy thing to say, 'Did she die because of pregnancy or childbirth, yes or no?'" Dr Chescheir said. "But there are so many different pieces that go into this that make it really tough. I think we're going to continue to see very smart epidemiologists argue back and forth in the published literature about how we answer this question."

The study received partial funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The authors have disclosed no relevant financial relationships.

Obstet Gynecol. 2017;129:811-818. Abstract

For more news, join us on Facebook and Twitter


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.