NR 503 Week 4 Maternal Mortality Data Source Review Example

One Field on the U.S. Death Certificate: The Pregnancy Status Checkbox Behind National Maternal Mortality Figures, Who Ticks It, How NCHS Recodes It, and What the Suppression Rule Hides

Student Name

College of Nursing, Chamberlain University

NR 503: Population Health, Epidemiology and Statistical Principles

Instructor Name

Month Day, Year

One Field on the U.S. Death Certificate: The Pregnancy Status Checkbox Behind National Maternal Mortality Figures, Who Ticks It, How NCHS Recodes It, and What the Suppression Rule Hides

The Question and the Source

My question is whether maternal deaths in my state have risen since 2018 and whether any rise is concentrated in women over 35. The source for official U.S. maternal mortality figures is the National Vital Statistics System (NVSS), run by the National Center for Health Statistics (NCHS), which compiles every death certificate filed in the states. The latest national report gives a maternal mortality rate for 2023 of 18.6 deaths per 100,000 live births, with the rate for women aged 40 and older at 59.8 (Hoyert, 2025).

A review that stopped there would describe the agency and quote the number. This one leaves the front page for the technical documentation, because my question depends on a single field on the death certificate, and the documentation is where that field's history, its errors and the way NCHS handles it are written down.

The Documentation, Not the Front Page

The key document is a 2020 NCHS report explaining how maternal deaths are coded and why the method changed in 2018 (Hoyert & Miniño, 2020). The rate page says only that a maternal death is one occurring during pregnancy or in the 42 days after it, caused or made worse by the pregnancy or its care, excluding accidents and incidental causes. The coding report explains how a death actually gets into that count, and it shows that the answer depends heavily on one item on the certificate.

The report also explains a gap that would otherwise puzzle anyone looking at the trend: NCHS published no official national maternal mortality rate between 2007 and 2018. The reason is that the checkbox was adopted state by state over about 14 years, and combining states that had it with states that did not produced an apparent rise that mainly reflected the rollout rather than any real change in deaths.

One Field, Chosen by the Question

The field is the pregnancy status checkbox, added to the U.S. Standard Certificate of Death in its 2003 revision. For a female decedent, the certifier indicates whether she was not pregnant within the past year, pregnant at the time of death, not pregnant but pregnant within 42 days of death, not pregnant but pregnant 43 days to one year before death, or unknown. The checkbox was added because earlier studies showed that cause-of-death statements alone missed many deaths linked to pregnancy (Hoyert & Miniño, 2020).

My question turns on this field for two reasons. First, the checkbox accounts for a large share of the deaths now counted as maternal. Second, and more important for my interest in older women, the errors in the checkbox are not spread evenly by age.

Who Fills It, and When

The checkbox is part of the medical certification of death, completed by the certifier: the attending physician, or a medical examiner or coroner when the death is referred to them. It is filled in after the death, often by a clinician who may not have known the woman well, may be completing several certificates at once, and may have no record of a pregnancy that ended months earlier in another facility. The report notes that the cause of the errors is unclear, but it describes a pattern consistent with certifiers selecting the wrong box.

The evidence for error is striking. In 2013 data, 797 deaths of women aged 40 and older were marked on the checkbox as pregnant at death or within the past year, including 147 women over 85. An internal NCHS analysis linking hospital records to death certificates in selected hospitals suggested false-positive rates of 54% and 56% for the checkbox in 2014 and 2016, though those hospitals were not nationally representative. Research in four states found that only 19% of decedents aged 45 to 49 with a positive checkbox had any other evidence of pregnancy (Hoyert & Miniño, 2020).

Coding as a Second Author

The number that reaches the published rate is not simply what the certifier ticked. NCHS applies coding rules that decide how the checkbox is used, and those rules changed. Under the 2003 to 2017 method, any indication of pregnancy, including the checkbox alone, led to a maternal code for decedents aged 10 to 54. Under the 2018 method, the checkbox is applied only to decedents aged 10 to 44, and when the checkbox is the only indication of pregnancy on the certificate, the maternal code is assigned to the underlying cause alone (Hoyert & Miniño, 2020).

The coder is, in effect, a second author of the field. A 48-year-old woman whose certificate says only "pregnant within 42 days" on the checkbox, with no mention of pregnancy in the cause-of-death statement, was counted as a maternal death under the old method and is not counted under the new one. For my question about women over 35, this matters directly: a rise or fall in rates for women in their forties across the 2017 to 2018 boundary could reflect the coding change rather than any change in deaths. Comparisons have to use a single method across the whole period, and the report provides checkbox-free rates for some years for that purpose.

What Suppression Removes

The last feature of the source is what it chooses not to show. NCHS does not publish a mortality rate based on fewer than 20 deaths, replacing it with an asterisk, because such rates are statistically unreliable (Hoyert & Miniño, 2020). For national figures, this rarely matters. For my state, it matters a great deal. A state with a few dozen maternal deaths a year will have suppressed rates for most age groups and racial and ethnic groups in any single year, and the group I care most about, women aged 40 and older, is small.

Suppression protects against over-reading noisy numbers, but it also means that the source cannot answer my question for one year at a time. To get a stable rate for older women in my state, I would need to pool several years, and every year pooled must use the same coding method.

What the Source Can and Cannot Answer

NVSS can tell me the official maternal mortality rate for the nation and, pooled across years, for my state. It can tell me that the rate for women 40 and over is several times higher than for younger women. It cannot tell me, from the checkbox alone, how many of those older women's deaths were truly related to pregnancy, because the field's errors are concentrated in that age group, and it cannot give me reliable single-year rates for small groups.

For my question, the practical conclusion is to use NVSS rates only for 2018 onward, with the 2018 method throughout, to pool at least three years for any age-specific rate in my state, and to turn to my state's maternal mortality review committee, which examines each death individually, before concluding that pregnancy-related deaths among older women have truly risen.

References

Hoyert, D. L. (2025). Maternal mortality rates in the United States, 2023 (NCHS Health E-Stats). National Center for Health Statistics. https://doi.org/10.15620/cdc/174577

Hoyert, D. L., & Miniño, A. M. (2020). Maternal mortality in the United States: Changes in coding, publication, and data release, 2018. National Vital Statistics Reports, 69(2). National Center for Health Statistics.