The failure of United States maternal mortality reporting and its impact on women’s lives

INTRODUCTION

Maternal mortality is a significant public health problem and an important indicator of the quality of health care both nationally and internationally. The death of a mother during pregnancy, childbirth, or postpartum is one of the greatest tragedies that can occur within a family, with wide-ranging consequences for the index child, other family members, and the larger society. Maternal mortality also represents the ultimate failure of perinatal medical care. In the United States, the National Vital Statistics System (NVSS) is the source of official maternal mortality statistics used for both subnational and international comparisons.¹ However, for the past decade, the data system has been unable to supply accurate estimates of United States maternal mortality as a result of problems in both reporting and coding of the data.^2–4 A brief summary of how the data got to this unfortunate state follows.

RECENT HISTORY OF VITAL STATISTICS MATERNAL MORTALITY DATA IN THE UNITED STATES

Studies based on data from the 1980s and 1990s identified significant underreporting of maternal deaths in the NVSS.^5,6. To improve the reporting of maternal deaths, a pregnancy question was added to the 2003 revision of the United States standard death certificate.¹ The question has checkboxes to ascertain whether female decedents were not pregnant within the past year, pregnant at the time of death, not pregnant but pregnant within 42 days of death, pregnant 43 days–1 year before death, or unknown whether pregnant within the past year.¹ The addition of the standard pregnancy question resulted in a substantial increase in reporting of maternal deaths.² However, delays in states’ adopting the standard pregnancy question, together with the use of nonstandard questions, created a situation where in any given data year, some states were using the United States standard question, others were using questions incompatible with the United States standard, and still others had no pregnancy question on their death certificates.² Because of the difficulties in disentangling these effects, the United States has not published an “official” maternal mortality rate since 2007 when half the states had the standard question and half did not.². By 2015, all states except Alabama, California, and West Virginia had adopted the standard pregnancy question, raising the possibility of again being able to publish close to-national estimates of United States maternal mortality. However, more recent studies identified significant overreporting of maternal deaths from states, using the pregnancy question. For example, a recent Centers for Disease Control and Prevention (CDC) report from maternal mortality review committees in four states found that 15% (97/650) of reported maternal deaths were not maternal deaths at all, since the women involved were confirmed to be not pregnant or postpartum within 1 year of death.⁷ The same study also found that the checkbox identified cases, particularly during pregnancy or late postpartum, that were identified only because of the checkbox, and with no other evidence that the case was a maternal death.⁷ Thus, the errors of overcounting were predominantly because of errors in the pregnancy checkbox.⁷ Obviously, it is insufficient to just add a question to a data collection form; training and quality control are essential to make sure that the information is reported accurately. Unfortunately, data that go unanalyzed and unpublished are likely to receive less scrutiny and quality control, and to deteriorate further in quality. For example, an accompanying paper in this issue of Birth found a near-doubling in the Texas maternal mortality rate from 2010 to 2012, which might have been investigated sooner and in greater detail if the Federal government had not discontinued publishing these data.⁴

The problems in reporting of pregnancy status are compounded by United States coding rules that code every death with the pregnancy or postpartum checkbox checked to maternal causes, regardless of what is written in the cause-of-death section.^8,9 The only exception is for external causes of injury (ie, accident, suicide, or homicide) which are coded to nonmaternal causes.^8,9 This coding scheme makes the checkbox information essentially the sole factor in deciding whether a death is maternal or nonmaternal. For example, right now, if “sunburn” is written as the cause of death, and if the pregnancy or postpartum checkbox is checked, United States coding rules code this as a maternal death.^8,9 This coding is clearly not in keeping with the spirit of the World Health Organization maternal mortality definition of maternal death as: The death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and the site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management, but not from accidental or incidental causes.¹⁰

These types of cases are often coded to nonspecific causes of death within the maternal mortality chapter of the International Classification of Diseases (ICD-10), which provide little or no information as to what the true cause of death is.^3,4 For example, in a paper published in this issue of the journal, we found that more than one-half of all reported maternal deaths in Texas in 2011–2015 were attributed to these nonspecific causes.⁴ Given documented inaccuracies in the reporting of the pregnancy checkbox data, this overreliance on the pregnancy checkbox leads to seriously flawed United States maternal mortality data. As a result of these data problems, there is currently no clear picture available of United States’ maternal mortality levels and trends. This lack of information has a clear influence on related data systems, and on the ability of the United States to prevent these tragic deaths.

IMPACT ON RELATED DATA SYSTEMS

Some clinicians comment that they are not overly concerned about the accuracy of vital statistics maternal mortality data because the United States has another supplemental data system that examines pregnancy-related deaths: the Pregnancy Mortality Surveillance System (PMSS). However, the NVSS is the main data source used to identify maternal deaths for inclusion in the PMSS, and the problems in the NVSS have created challenges for PMSS case ascertainment. For example, the PMSS has identified recent increases in pregnancy-related mortality in the United States;¹¹ however, since this system is based in large part on official vital statistics data (together with supplementary reports), these data could also have been influenced by the improved ascertainment or overascertainment of maternal deaths in the NVSS. The PMSS data are also confidential data held by the Federal government and are not available to researchers, thus limiting our ability to expand knowledge of maternal and pregnancy-related deaths in the United States. The PMSS is also not permitted to report rates at the state level, inhibiting intranational comparisons.

The United States is also working to expand its state and local maternal mortality review system, and most states now have maternal mortality review committees.¹² However, again, maternal deaths to be examined by these committees are largely identified from the NVSS; thus, the problems in the NVSS lead to considerable wasted time and effort as these volunteer-based committees examine deaths reported as maternal which are not maternal at all.¹³

IMPACT ON UNITED STATES AND INTERNATIONAL MATERNAL MORTALITY PREVENTION EFFORTS

The failure of the United States government to publish an official maternal mortality rate since 2007 has also led to a deficit of information both nationally and internationally, at a time when greater attention has been focused on maternal mortality than ever before.^14–17 United Nations’ Millennium Development Goal 5a was to: Reduce by three-quarters between 1990 and 2015, the maternal mortality ratio.¹⁸ This goal led to a huge international effort to reduce maternal mortality worldwide, with declines in 157 out of 183 countries.¹⁵. In fact, maternal mortality declined by 44% worldwide from 1990 to 2015 and by 48% for industrialized countries.¹⁹ In contrast, the best estimates of the United States maternal mortality during this time period show a 27% increase in the maternal mortality rate in 48 states and DC from 2000 to 2014.²

It is estimated that there are about 800 maternal deaths per year in the United States,2 and a recent study found that 59% of maternal deaths were preventable.⁷ The estimated maternal mortality rate of 23.8 for 48 states and DC in 20142 is several times higher than the 2014 maternal mortality rates in the United Kingdom (3.9), France (3.5), and Sweden (2.2).²⁰ Compared with the huge efforts in most other countries, there was no large, nationwide effort to reduce United States maternal mortality linked to the Millennium Development Goals during this period—in part because no one knew what the true levels or trends in United States maternal mortality were.

Accurate measurement of maternal mortality is an essential first step in any prevention program: to identify at-risk populations, to target prevention efforts to the most vulnerable groups, and to measure the progress of prevention programs. There were prevention efforts in California linked to the California Maternal Quality Care Collaborative during this period, with documented declines in maternal mortality compared with increases in other states.^2,21 These efforts have recently been expanded into the Alliance for Innovation in Maternal Health initiative, currently working in eight states to reduce maternal mortality.²² However, the magnitude of the lost opportunity of the United States fully participating in the Millennium Development Goal efforts from 2000 to 2015 cannot be overestimated.

RECOMMENDATIONS FOR IMPROVEMENT

Quality improvement efforts need to focus on improving the quality and validity of the pregnancy checkbox data. Periodic validation studies and the implementation of data quality checks at both the state and national levels are essential to improving reporting. A percentage of records, including 100% of records for women 40 years and older or coded to nonspecific causes should be routinely queried back to the certifier to confirm the fact of pregnancy.

We also need to develop and disseminate educational tools to inform people who complete death certificates as to both the importance of, and methods for, correct completion of the pregnancy checkbox and cause-of-death information for maternal deaths.

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 National Center for Health Statistics (the agency responsible for collecting and disseminating NVSS data) should undertake a systematic evaluation of current coding methods for maternal deaths, and develop scientifically defensible alternative methods, which are compatible with international standards.

Methods could also be developed to update NVSS information on maternal deaths with more detailed information from the PMSS and/or from state maternal mortality reviews. The CDC’s efforts to combine information from state maternal mortality reviews into a national system (MMRIA) also hold considerable promise.²³

THE PROBLEM OF POLITICAL WILL

The United States tracks many detailed measures of the health of its citizens, from infant mortality to cancer registries, to hospital performance statistics.²⁴ Compared with the detailed information available from other data systems, fixing the United States maternal mortality data is comparatively straightforward. In fact, Dr. William Callaghan, who leads the CDC’s maternal and infant health branch and is a prominent maternal mortality researcher, recently assessed the difficulty of collecting accurate maternal mortality data as about a 3 on a scale of 1 (easy) to 10 (hard).²⁵ The National Center for Health Statistics, the government agency responsible for the NVSS, has been aware of the problems in reporting and coding of maternal mortality data for some time. However, as of this writing (December 2017), no concrete action has been taken to address the data problems or to improve data quality.

Despite measurement issues, it is clear that the United States maternal mortality rate is considerably higher than in most industrialized countries, and that the majority of these deaths are preventable. Without accurate data to measure the magnitude of the problem and to identify at-risk populations, the efficacy of maternal mortality prevention efforts are severely compromised.

Simply put, if accurate maternal mortality data are not available, prevention efforts are scattered and unfocused … and more women die. This commentary is a plea for action to improve the measurement of maternal mortality in the United States, which is the essential first step in saving women’s lives.