Accuracy of Maternal Mortality Rates and Updated NCHS Methodologies: Data Quality Matters

Posted on

By: Alison Dingwall, Principal Social Behavioral Scientist, MITRE; Rachel Mayer, Senior Health Systems Analyst, MITRE; Samuel Volkin, Healthcare Policy Analyst, MITRE

Alison Dingwall
Rachel Mayer
Samuel Volkin

Maternal mortality is a growing concern for women in the United States. The upward trend of pregnancy-related causes of death is alarming and requires attention to data surveillance and analysis. This commentary discusses the history of maternal mortality data quality issues and considers whether

the new methodology from the National Center for Health Statistics (NCHS) within the Centers for Disease Control and Prevention (CDC) sufficiently addresses these issues and what work still needs to be done.

Acknowledging flaws in how maternal mortality data is collected and analyzed, The NCHS’ National Vital Statistics System (NVSS) suspended its publication of maternal mortality data in 2007. Since then, substantial literature emerged detailing common errors in state-level maternal mortality data collection and reporting. Based on these findings, the NCHS released an updated methodology for coding maternal deaths and applied it to 2018 maternal mortality data, which was released in January 2020.

Errors in maternal mortality data begin with inaccurate information on death certificates. The cause-of-death section of the death certificate is used to assign International Classification of Disease (ICD) codes, which contain a list of codes specific to pregnancy. According to the former NCHS 2003–2017 method for coding maternal deaths, an ICD maternal code is assigned if there is indication of pregnancy on the death certificate through any one of the three occurrences:

  1. when causes specific to pregnancy, delivery, or the puerperium are reported in the cause-of-death section of the death certificate;
  2. when pregnancy is mentioned among the medical conditions reported on the death certificate;
  3. when there is a response in the checkbox item that indicates the woman had been pregnant at the time of death or in the preceding year.

Despite the NCHS’ need for accurate death certificate data, errors in cause of death certification occur in one out of every three death certificates. Of particular concern are the lack of and inconsistencies in formal death certificate training and in the reporting of cases of maternal mortality.

The NCHS’ New Approach

The NCHS 2018 method for coding maternal deaths introduced two changes to the 2003–2017 method. In response to the common occurrence of false-positive of maternal deaths amongst women ages 45 and older, the new NCHS method limits the use of the  pregnancy checkbox to women between the ages of 10–44.

The second NCHS change intended to capture additional details related to a pregnant woman’s cause-of-death.  In the 2003–2017 method, if a pregnancy checkbox was marked, all the medical conditions reported on the death certificate would automatically be assigned maternal ICD codes. This made it difficult for NCHS scientists to discern whether the cause of death written on the death certificate was related to maternity or not, which is important information to have in order to confirm whether the pregnancy checkbox was marked correctly. With this change, the 2018 method better reflects what was recorded on the death certificate and allows NCHS scientists to flag files more easily that may have been incorrectly labeled as maternal.

Looking Forward

The changes introduced by the NCHS revised 2018 method may result in an underreporting of maternal deaths in women ages 45 and older. Incremental changes made by NCHS are important and leave the maternal health community with many outstanding opportunities to improve pregnancy-related death statistics.

The onus of improving the quality of maternal mortality data does not rest on the shoulders of the NCHS alone. The NCHS is the federal entity responsible for national mortality data. However, since local jurisdictions including states, territories, and cities own the data and have purview over both collection and reporting of vital statistics, the federal government cannot unilaterally impose a solution. Instead, federal-, state-, territory-, and city-level stakeholders could collaborate to review and improve systems used by each jurisdiction.

Local jurisdictions need to implement more rigorous and standardized death certification training to ensure accurate and reliable documentation. There must also be alignment of data reporting to NCHS and existing state-level maternal mortality review committees that review each case of maternal death. An alignment would increase the quality of data that NCHS receives. With accurate data, maternal health professionals and policy makers can be certain that they are focusing resources where needed to reduce maternal mortality rates in the United States.

Reprinted with consent from the MITRE Corporation